[ { "name": "infinitetalk_0001.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0001.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/10044526923_2.jpg", "driving_audio": "3_audio/infinitetalk_0001.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p293/p293_001.wav", "generated_video": "1_fake/infinitetalk_0001.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_009.mp4", "output_index": 9, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 1, "name_before": "10044526923_2__p293_001.mp4" }, { "name": "infinitetalk_0002.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0002.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/10082989283_0.jpg", "driving_audio": "3_audio/infinitetalk_0002.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p294/p294_322.wav", "generated_video": "1_fake/infinitetalk_0002.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_044.mp4", "output_index": 44, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 2, "name_before": "10082989283_0__p294_322.mp4" }, { "name": "infinitetalk_0003.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0003.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/10349803155_4.jpg", "driving_audio": "3_audio/infinitetalk_0003.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p300/p300_241.wav", "generated_video": "1_fake/infinitetalk_0003.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_006.mp4", "output_index": 6, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 3, "name_before": "10349803155_4__p300_241.mp4" }, { "name": "infinitetalk_0004.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0004.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/10677166173_0.jpg", "driving_audio": "3_audio/infinitetalk_0004.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p297/p297_336.wav", "generated_video": "1_fake/infinitetalk_0004.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_029.mp4", "output_index": 29, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 4, "name_before": "10677166173_0__p297_336.mp4" }, { "name": "infinitetalk_0005.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0005.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/11066178864_2.jpg", "driving_audio": "3_audio/infinitetalk_0005.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p271/p271_217.wav", "generated_video": "1_fake/infinitetalk_0005.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_042.mp4", "output_index": 42, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 5, "name_before": "11066178864_2__p271_217.mp4" }, { "name": "infinitetalk_0006.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0006.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/12479470275_0.jpg", "driving_audio": "3_audio/infinitetalk_0006.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p303/p303_191.wav", "generated_video": "1_fake/infinitetalk_0006.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_005.mp4", "output_index": 5, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 6, "name_before": "12479470275_0__p303_191.mp4" }, { "name": "infinitetalk_0007.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0007.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/13513147434_0.jpg", "driving_audio": "3_audio/infinitetalk_0007.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p361/p361_059.wav", "generated_video": "1_fake/infinitetalk_0007.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_023.mp4", "output_index": 23, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 7, "name_before": "13513147434_0__p361_059.mp4" }, { "name": "infinitetalk_0008.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0008.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/149196124_1.jpg", "driving_audio": "3_audio/infinitetalk_0008.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p318/p318_352.wav", "generated_video": "1_fake/infinitetalk_0008.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_007.mp4", "output_index": 7, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 8, "name_before": "149196124_1__p318_352.mp4" }, { "name": "infinitetalk_0009.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0009.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/2239641055_0.jpg", "driving_audio": "3_audio/infinitetalk_0009.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p312/p312_396.wav", "generated_video": "1_fake/infinitetalk_0009.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_035.mp4", "output_index": 35, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 9, "name_before": "2239641055_0__p312_396.mp4" }, { "name": "infinitetalk_0010.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0010.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/228827093_0.jpg", "driving_audio": "3_audio/infinitetalk_0010.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p283/p283_109.wav", "generated_video": "1_fake/infinitetalk_0010.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_010.mp4", "output_index": 10, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 10, "name_before": "228827093_0__p283_109.mp4" }, { "name": "infinitetalk_0011.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0011.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/2451041562_0.jpg", "driving_audio": "3_audio/infinitetalk_0011.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p335/p335_161.wav", "generated_video": "1_fake/infinitetalk_0011.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_045.mp4", "output_index": 45, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 11, "name_before": "2451041562_0__p335_161.mp4" }, { "name": "infinitetalk_0012.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0012.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/2478990176_0.jpg", "driving_audio": "3_audio/infinitetalk_0012.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p264/p264_252.wav", "generated_video": "1_fake/infinitetalk_0012.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_020.mp4", "output_index": 20, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 12, "name_before": "2478990176_0__p264_252.mp4" }, { "name": "infinitetalk_0013.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0013.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/2569042969_0.jpg", "driving_audio": "3_audio/infinitetalk_0013.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p333/p333_238.wav", "generated_video": "1_fake/infinitetalk_0013.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_018.mp4", "output_index": 18, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 13, "name_before": "2569042969_0__p333_238.mp4" }, { "name": "infinitetalk_0014.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0014.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/2766788906_0.jpg", "driving_audio": "3_audio/infinitetalk_0014.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p312/p312_006.wav", "generated_video": "1_fake/infinitetalk_0014.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_003.mp4", "output_index": 3, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 14, "name_before": "2766788906_0__p312_006.mp4" }, { "name": "infinitetalk_0015.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0015.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/2813746757_2.jpg", "driving_audio": "3_audio/infinitetalk_0015.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p234/p234_297.wav", "generated_video": "1_fake/infinitetalk_0015.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_008.mp4", "output_index": 8, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 15, "name_before": "2813746757_2__p234_297.mp4" }, { "name": "infinitetalk_0016.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0016.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/2989261763_4.jpg", "driving_audio": "3_audio/infinitetalk_0016.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p278/p278_240.wav", "generated_video": "1_fake/infinitetalk_0016.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_039.mp4", "output_index": 39, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 16, "name_before": "2989261763_4__p278_240.mp4" }, { "name": "infinitetalk_0017.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0017.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/3023646602_3.jpg", "driving_audio": "3_audio/infinitetalk_0017.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p308/p308_042.wav", "generated_video": "1_fake/infinitetalk_0017.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_001.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 17, "name_before": "3023646602_3__p308_042.mp4" }, { "name": "infinitetalk_0018.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0018.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/3056395949_1.jpg", "driving_audio": "3_audio/infinitetalk_0018.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p279/p279_309.wav", "generated_video": "1_fake/infinitetalk_0018.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_000.mp4", "output_index": 0, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 18, "name_before": "3056395949_1__p279_309.mp4" }, { "name": "infinitetalk_0019.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0019.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/3470970468_0.jpg", "driving_audio": "3_audio/infinitetalk_0019.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p360/p360_264.wav", "generated_video": "1_fake/infinitetalk_0019.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_011.mp4", "output_index": 11, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 19, "name_before": "3470970468_0__p360_264.mp4" }, { "name": "infinitetalk_0020.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0020.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/3792584456_0.jpg", "driving_audio": "3_audio/infinitetalk_0020.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p277/p277_084.wav", "generated_video": "1_fake/infinitetalk_0020.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_022.mp4", "output_index": 22, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 20, "name_before": "3792584456_0__p277_084.mp4" }, { "name": "infinitetalk_0021.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0021.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/384148537_2.jpg", "driving_audio": "3_audio/infinitetalk_0021.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p294/p294_099.wav", "generated_video": "1_fake/infinitetalk_0021.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_043.mp4", "output_index": 43, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 21, "name_before": "384148537_2__p294_099.mp4" }, { "name": "infinitetalk_0022.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0022.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/418495103_6.jpg", "driving_audio": "3_audio/infinitetalk_0022.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p276/p276_382.wav", "generated_video": "1_fake/infinitetalk_0022.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_015.mp4", "output_index": 15, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 22, "name_before": "418495103_6__p276_382.mp4" }, { "name": "infinitetalk_0023.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0023.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/4893363770_5.jpg", "driving_audio": "3_audio/infinitetalk_0023.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p341/p341_093.wav", "generated_video": "1_fake/infinitetalk_0023.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_030.mp4", "output_index": 30, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 23, "name_before": "4893363770_5__p341_093.mp4" }, { "name": "infinitetalk_0024.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0024.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/602283745_4.jpg", "driving_audio": "3_audio/infinitetalk_0024.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p228/p228_163.wav", "generated_video": "1_fake/infinitetalk_0024.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_038.mp4", "output_index": 38, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 24, "name_before": "602283745_4__p228_163.mp4" }, { "name": "infinitetalk_0025.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0025.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/6275888182_0.jpg", "driving_audio": "3_audio/infinitetalk_0025.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p345/p345_137.wav", "generated_video": "1_fake/infinitetalk_0025.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_040.mp4", "output_index": 40, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 25, "name_before": "6275888182_0__p345_137.mp4" }, { "name": "infinitetalk_0026.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0026.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/6946329065_1.jpg", "driving_audio": "3_audio/infinitetalk_0026.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p300/p300_288.wav", "generated_video": "1_fake/infinitetalk_0026.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_017.mp4", "output_index": 17, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 26, "name_before": "6946329065_1__p300_288.mp4" }, { "name": "infinitetalk_0027.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0027.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/7410812438_8.jpg", "driving_audio": "3_audio/infinitetalk_0027.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p258/p258_327.wav", "generated_video": "1_fake/infinitetalk_0027.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_012.mp4", "output_index": 12, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 27, "name_before": "7410812438_8__p258_327.mp4" }, { "name": "infinitetalk_0028.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0028.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/9079336453_4.jpg", "driving_audio": "3_audio/infinitetalk_0028.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p229/p229_098.wav", "generated_video": "1_fake/infinitetalk_0028.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_037.mp4", "output_index": 37, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 28, "name_before": "9079336453_4__p229_098.mp4" }, { "name": "infinitetalk_0029.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0029.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/9509653613_0.jpg", "driving_audio": "3_audio/infinitetalk_0029.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p316/p316_101.wav", "generated_video": "1_fake/infinitetalk_0029.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_024.mp4", "output_index": 24, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 29, "name_before": "9509653613_0__p316_101.mp4" }, { "name": "infinitetalk_0030.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0030.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/AFW_1204062305_1_1.jpg", "driving_audio": "3_audio/infinitetalk_0030.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p239/p239_358.wav", "generated_video": "1_fake/infinitetalk_0030.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_027.mp4", "output_index": 27, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 30, "name_before": "AFW_1204062305_1_1__p239_358.mp4" }, { "name": "infinitetalk_0031.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0031.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/AFW_213654866_1_0.jpg", "driving_audio": "3_audio/infinitetalk_0031.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p329/p329_192.wav", "generated_video": "1_fake/infinitetalk_0031.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_032.mp4", "output_index": 32, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 31, "name_before": "AFW_213654866_1_0__p329_192.mp4" }, { "name": "infinitetalk_0032.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0032.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/HELEN_1021890651_1_0.jpg", "driving_audio": "3_audio/infinitetalk_0032.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p334/p334_016.wav", "generated_video": "1_fake/infinitetalk_0032.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_046.mp4", "output_index": 46, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 32, "name_before": "HELEN_1021890651_1_0__p334_016.mp4" }, { "name": "infinitetalk_0033.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0033.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/HELEN_1270798682_2_1.jpg", "driving_audio": "3_audio/infinitetalk_0033.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p230/p230_045.wav", "generated_video": "1_fake/infinitetalk_0033.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_014.mp4", "output_index": 14, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 33, "name_before": "HELEN_1270798682_2_1__p230_045.mp4" }, { "name": "infinitetalk_0034.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0034.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/HELEN_128244378_1_1.jpg", "driving_audio": "3_audio/infinitetalk_0034.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p248/p248_076.wav", "generated_video": "1_fake/infinitetalk_0034.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_013.mp4", "output_index": 13, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 34, "name_before": "HELEN_128244378_1_1__p248_076.mp4" }, { "name": "infinitetalk_0035.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0035.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/HELEN_221238636_1_2.jpg", "driving_audio": "3_audio/infinitetalk_0035.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p264/p264_300.wav", "generated_video": "1_fake/infinitetalk_0035.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_034.mp4", "output_index": 34, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 35, "name_before": "HELEN_221238636_1_2__p264_300.mp4" }, { "name": "infinitetalk_0036.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0036.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/HELEN_2251139211_2_3.jpg", "driving_audio": "3_audio/infinitetalk_0036.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p247/p247_257.wav", "generated_video": "1_fake/infinitetalk_0036.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_016.mp4", "output_index": 16, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 36, "name_before": "HELEN_2251139211_2_3__p247_257.mp4" }, { "name": "infinitetalk_0037.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0037.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/HELEN_2255538912_1_1.jpg", "driving_audio": "3_audio/infinitetalk_0037.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p283/p283_107.wav", "generated_video": "1_fake/infinitetalk_0037.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_002.mp4", "output_index": 2, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 37, "name_before": "HELEN_2255538912_1_1__p283_107.mp4" }, { "name": "infinitetalk_0038.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0038.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/HELEN_2386704064_1_3.jpg", "driving_audio": "3_audio/infinitetalk_0038.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p266/p266_277.wav", "generated_video": "1_fake/infinitetalk_0038.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_048.mp4", "output_index": 48, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 38, "name_before": "HELEN_2386704064_1_3__p266_277.mp4" }, { "name": "infinitetalk_0039.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0039.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/HELEN_2402052185_1_7.jpg", "driving_audio": "3_audio/infinitetalk_0039.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p233/p233_261.wav", "generated_video": "1_fake/infinitetalk_0039.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_004.mp4", "output_index": 4, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 39, "name_before": "HELEN_2402052185_1_7__p233_261.mp4" }, { "name": "infinitetalk_0040.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0040.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/HELEN_2815405614_1_5.jpg", "driving_audio": "3_audio/infinitetalk_0040.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p262/p262_014.wav", "generated_video": "1_fake/infinitetalk_0040.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_031.mp4", "output_index": 31, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 40, "name_before": "HELEN_2815405614_1_5__p262_014.mp4" }, { "name": "infinitetalk_0041.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0041.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/HELEN_2877095014_1_1.jpg", "driving_audio": "3_audio/infinitetalk_0041.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p249/p249_231.wav", "generated_video": "1_fake/infinitetalk_0041.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_021.mp4", "output_index": 21, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 41, "name_before": "HELEN_2877095014_1_1__p249_231.mp4" }, { "name": "infinitetalk_0042.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0042.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/HELEN_3026829129_1_6.jpg", "driving_audio": "3_audio/infinitetalk_0042.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p278/p278_357.wav", "generated_video": "1_fake/infinitetalk_0042.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_036.mp4", "output_index": 36, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 42, "name_before": "HELEN_3026829129_1_6__p278_357.mp4" }, { "name": "infinitetalk_0043.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0043.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/HELEN_3036412907_2_0.jpg", "driving_audio": "3_audio/infinitetalk_0043.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p341/p341_075.wav", "generated_video": "1_fake/infinitetalk_0043.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_026.mp4", "output_index": 26, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 43, "name_before": "HELEN_3036412907_2_0__p341_075.mp4" }, { "name": "infinitetalk_0044.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0044.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/HELEN_3048153244_1_0.jpg", "driving_audio": "3_audio/infinitetalk_0044.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p283/p283_006.wav", "generated_video": "1_fake/infinitetalk_0044.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_047.mp4", "output_index": 47, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 44, "name_before": "HELEN_3048153244_1_0__p283_006.mp4" }, { "name": "infinitetalk_0045.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0045.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/LFPW_image_train_0004_0.jpg", "driving_audio": "3_audio/infinitetalk_0045.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p240/p240_196.wav", "generated_video": "1_fake/infinitetalk_0045.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_019.mp4", "output_index": 19, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 45, "name_before": "LFPW_image_train_0004_0__p240_196.mp4" }, { "name": "infinitetalk_0046.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0046.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/LFPW_image_train_0435_2.jpg", "driving_audio": "3_audio/infinitetalk_0046.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p269/p269_189.wav", "generated_video": "1_fake/infinitetalk_0046.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_041.mp4", "output_index": 41, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 46, "name_before": "LFPW_image_train_0435_2__p269_189.mp4" }, { "name": "infinitetalk_0047.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0047.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/LFPW_image_train_0567_3.jpg", "driving_audio": "3_audio/infinitetalk_0047.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p241/p241_233.wav", "generated_video": "1_fake/infinitetalk_0047.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_025.mp4", "output_index": 25, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 47, "name_before": "LFPW_image_train_0567_3__p241_233.mp4" }, { "name": "infinitetalk_0048.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0048.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/LFPW_image_train_0837_4.jpg", "driving_audio": "3_audio/infinitetalk_0048.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p261/p261_032.wav", "generated_video": "1_fake/infinitetalk_0048.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_028.mp4", "output_index": 28, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 48, "name_before": "LFPW_image_train_0837_4__p261_032.mp4" }, { "name": "infinitetalk_0049.mp4", "variant": "lapa_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0049.jpg", "identity_image_from": "/data/lzd/sty/LaPa_images/raw/LaPa/train/images/LFPW_image_train_0853_2.jpg", "driving_audio": "3_audio/infinitetalk_0049.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p279/p279_048.wav", "generated_video": "1_fake/infinitetalk_0049.mp4", "generated_from": "/archive/lzd/InfiniteTalk/kdtalker_batch_033.mp4", "output_index": 33, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch was generated with a short frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 49, "name_before": "LFPW_image_train_0853_2__p279_048.mp4" }, { "name": "infinitetalk_0050.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0050.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+-KXks9ifwwk+P0+C0+F98086-98431.jpg", "driving_audio": "3_audio/infinitetalk_0050.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p295/p295_013.wav", "generated_video": "1_fake/infinitetalk_0050.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+-KXks9ifwwk+P0+C0+F98086-98431__p295_013.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 50, "real_video": "0_real/infinitetalk_0050.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+-KXks9ifwwk+P0+C0+F98086-98431.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+-KXks9ifwwk+P0+C0+F98086-98431__p295_013.mp4" }, { "name": "infinitetalk_0051.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0051.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+-LU6SAtmBK0+P0+C0+F53495-53624.jpg", "driving_audio": "3_audio/infinitetalk_0051.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p340/p340_143.wav", "generated_video": "1_fake/infinitetalk_0051.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+-LU6SAtmBK0+P0+C0+F53495-53624__p340_143.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 51, "real_video": "0_real/infinitetalk_0051.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+-LU6SAtmBK0+P0+C0+F53495-53624.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+-LU6SAtmBK0+P0+C0+F53495-53624__p340_143.mp4" }, { "name": "infinitetalk_0052.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0052.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+-mGEays8BTI+P0+C0+F28182-28388.jpg", "driving_audio": "3_audio/infinitetalk_0052.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p279/p279_063.wav", "generated_video": "1_fake/infinitetalk_0052.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+-mGEays8BTI+P0+C0+F28182-28388__p279_063.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 52, "real_video": "0_real/infinitetalk_0052.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+-mGEays8BTI+P0+C0+F28182-28388.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+-mGEays8BTI+P0+C0+F28182-28388__p279_063.mp4" }, { "name": "infinitetalk_0053.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0053.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+0XHF4V0qWoY+P0+C0+F3147-3475.jpg", "driving_audio": "3_audio/infinitetalk_0053.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p310/p310_206.wav", "generated_video": "1_fake/infinitetalk_0053.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+0XHF4V0qWoY+P0+C0+F3147-3475__p310_206.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 53, "real_video": "0_real/infinitetalk_0053.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+0XHF4V0qWoY+P0+C0+F3147-3475.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+0XHF4V0qWoY+P0+C0+F3147-3475__p310_206.mp4" }, { "name": "infinitetalk_0054.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0054.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+1E4xk_DzU8c+P0+C0+F9901-10622.jpg", "driving_audio": "3_audio/infinitetalk_0054.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p228/p228_299.wav", "generated_video": "1_fake/infinitetalk_0054.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+1E4xk_DzU8c+P0+C0+F9901-10622__p228_299.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 54, "real_video": "0_real/infinitetalk_0054.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+1E4xk_DzU8c+P0+C0+F9901-10622.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+1E4xk_DzU8c+P0+C0+F9901-10622__p228_299.mp4" }, { "name": "infinitetalk_0055.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0055.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+1epA83fCeXc+P0+C0+F19350-19458.jpg", "driving_audio": "3_audio/infinitetalk_0055.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p261/p261_060.wav", "generated_video": "1_fake/infinitetalk_0055.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+1epA83fCeXc+P0+C0+F19350-19458__p261_060.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 55, "real_video": "0_real/infinitetalk_0055.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+1epA83fCeXc+P0+C0+F19350-19458.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+1epA83fCeXc+P0+C0+F19350-19458__p261_060.mp4" }, { "name": "infinitetalk_0056.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0056.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+2DfczSnGEKI+P0+C1+F3704-3825.jpg", "driving_audio": "3_audio/infinitetalk_0056.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p233/p233_289.wav", "generated_video": "1_fake/infinitetalk_0056.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+2DfczSnGEKI+P0+C1+F3704-3825__p233_289.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 56, "real_video": "0_real/infinitetalk_0056.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+2DfczSnGEKI+P0+C1+F3704-3825.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+2DfczSnGEKI+P0+C1+F3704-3825__p233_289.mp4" }, { "name": "infinitetalk_0057.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0057.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+3qypcEj611U+P2+C0+F15183-15472.jpg", "driving_audio": "3_audio/infinitetalk_0057.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p292/p292_274.wav", "generated_video": "1_fake/infinitetalk_0057.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+3qypcEj611U+P2+C0+F15183-15472__p292_274.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 57, "real_video": "0_real/infinitetalk_0057.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+3qypcEj611U+P2+C0+F15183-15472.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+3qypcEj611U+P2+C0+F15183-15472__p292_274.mp4" }, { "name": "infinitetalk_0058.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0058.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+4G6H7rHdlp0+P0+C0+F8585-8743.jpg", "driving_audio": "3_audio/infinitetalk_0058.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p292/p292_112.wav", "generated_video": "1_fake/infinitetalk_0058.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+4G6H7rHdlp0+P0+C0+F8585-8743__p292_112.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 58, "real_video": "0_real/infinitetalk_0058.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+4G6H7rHdlp0+P0+C0+F8585-8743.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+4G6H7rHdlp0+P0+C0+F8585-8743__p292_112.mp4" }, { "name": "infinitetalk_0059.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0059.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+4dBdfUudsIk+P0+C2+F26873-27093.jpg", "driving_audio": "3_audio/infinitetalk_0059.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p360/p360_249.wav", "generated_video": "1_fake/infinitetalk_0059.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+4dBdfUudsIk+P0+C2+F26873-27093__p360_249.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 59, "real_video": "0_real/infinitetalk_0059.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+4dBdfUudsIk+P0+C2+F26873-27093.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+4dBdfUudsIk+P0+C2+F26873-27093__p360_249.mp4" }, { "name": "infinitetalk_0060.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0060.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+6CxP-EV9gCU+P0+C2+F9475-9617.jpg", "driving_audio": "3_audio/infinitetalk_0060.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p283/p283_256.wav", "generated_video": "1_fake/infinitetalk_0060.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+6CxP-EV9gCU+P0+C2+F9475-9617__p283_256.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 60, "real_video": "0_real/infinitetalk_0060.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+6CxP-EV9gCU+P0+C2+F9475-9617.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+6CxP-EV9gCU+P0+C2+F9475-9617__p283_256.mp4" }, { "name": "infinitetalk_0061.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0061.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+6FUXuBunfQ0+P0+C2+F29837-30231.jpg", "driving_audio": "3_audio/infinitetalk_0061.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p239/p239_302.wav", "generated_video": "1_fake/infinitetalk_0061.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+6FUXuBunfQ0+P0+C2+F29837-30231__p239_302.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 61, "real_video": "0_real/infinitetalk_0061.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+6FUXuBunfQ0+P0+C2+F29837-30231.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+6FUXuBunfQ0+P0+C2+F29837-30231__p239_302.mp4" }, { "name": "infinitetalk_0062.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0062.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+6r6JroR8R0I+P0+C2+F5374-5500.jpg", "driving_audio": "3_audio/infinitetalk_0062.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p255/p255_055.wav", "generated_video": "1_fake/infinitetalk_0062.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+6r6JroR8R0I+P0+C2+F5374-5500__p255_055.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 62, "real_video": "0_real/infinitetalk_0062.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+6r6JroR8R0I+P0+C2+F5374-5500.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+6r6JroR8R0I+P0+C2+F5374-5500__p255_055.mp4" }, { "name": "infinitetalk_0063.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0063.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+80Vq0ZEd-o4+P0+C0+F2014-2920.jpg", "driving_audio": "3_audio/infinitetalk_0063.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p229/p229_362.wav", "generated_video": "1_fake/infinitetalk_0063.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+80Vq0ZEd-o4+P0+C0+F2014-2920__p229_362.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 63, "real_video": "0_real/infinitetalk_0063.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+80Vq0ZEd-o4+P0+C0+F2014-2920.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+80Vq0ZEd-o4+P0+C0+F2014-2920__p229_362.mp4" }, { "name": "infinitetalk_0064.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0064.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+8FoWhyh1-HU+P0+C1+F23141-23285.jpg", "driving_audio": "3_audio/infinitetalk_0064.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p228/p228_284.wav", "generated_video": "1_fake/infinitetalk_0064.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+8FoWhyh1-HU+P0+C1+F23141-23285__p228_284.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 64, "real_video": "0_real/infinitetalk_0064.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+8FoWhyh1-HU+P0+C1+F23141-23285.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+8FoWhyh1-HU+P0+C1+F23141-23285__p228_284.mp4" }, { "name": "infinitetalk_0065.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0065.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+8Z8txKVtBlk+P0+C2+F11908-12525.jpg", "driving_audio": "3_audio/infinitetalk_0065.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p234/p234_348.wav", "generated_video": "1_fake/infinitetalk_0065.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+8Z8txKVtBlk+P0+C2+F11908-12525__p234_348.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 65, "real_video": "0_real/infinitetalk_0065.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+8Z8txKVtBlk+P0+C2+F11908-12525.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+8Z8txKVtBlk+P0+C2+F11908-12525__p234_348.mp4" }, { "name": "infinitetalk_0066.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0066.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+91roKxXmTX0+P0+C0+F465-576.jpg", "driving_audio": "3_audio/infinitetalk_0066.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p287/p287_217.wav", "generated_video": "1_fake/infinitetalk_0066.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+91roKxXmTX0+P0+C0+F465-576__p287_217.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 66, "real_video": "0_real/infinitetalk_0066.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+91roKxXmTX0+P0+C0+F465-576.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+91roKxXmTX0+P0+C0+F465-576__p287_217.mp4" }, { "name": "infinitetalk_0067.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0067.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+9_K0u85y_Cs+P1+C0+F2100-2291.jpg", "driving_audio": "3_audio/infinitetalk_0067.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p285/p285_247.wav", "generated_video": "1_fake/infinitetalk_0067.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+9_K0u85y_Cs+P1+C0+F2100-2291__p285_247.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 67, "real_video": "0_real/infinitetalk_0067.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+9_K0u85y_Cs+P1+C0+F2100-2291.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+9_K0u85y_Cs+P1+C0+F2100-2291__p285_247.mp4" }, { "name": "infinitetalk_0068.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0068.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+AnPF6eVGgzU+P0+C0+F45495-45601.jpg", "driving_audio": "3_audio/infinitetalk_0068.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p253/p253_278.wav", "generated_video": "1_fake/infinitetalk_0068.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+AnPF6eVGgzU+P0+C0+F45495-45601__p253_278.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 68, "real_video": "0_real/infinitetalk_0068.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+AnPF6eVGgzU+P0+C0+F45495-45601.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+AnPF6eVGgzU+P0+C0+F45495-45601__p253_278.mp4" }, { "name": "infinitetalk_0069.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0069.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+BZUCDBO4GBU+P0+C1+F543-691.jpg", "driving_audio": "3_audio/infinitetalk_0069.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p334/p334_266.wav", "generated_video": "1_fake/infinitetalk_0069.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+BZUCDBO4GBU+P0+C1+F543-691__p334_266.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 69, "real_video": "0_real/infinitetalk_0069.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+BZUCDBO4GBU+P0+C1+F543-691.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+BZUCDBO4GBU+P0+C1+F543-691__p334_266.mp4" }, { "name": "infinitetalk_0070.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0070.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+BiuHKtBpDqw+P0+C2+F33039-33210.jpg", "driving_audio": "3_audio/infinitetalk_0070.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p283/p283_124.wav", "generated_video": "1_fake/infinitetalk_0070.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+BiuHKtBpDqw+P0+C2+F33039-33210__p283_124.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 70, "real_video": "0_real/infinitetalk_0070.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+BiuHKtBpDqw+P0+C2+F33039-33210.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+BiuHKtBpDqw+P0+C2+F33039-33210__p283_124.mp4" }, { "name": "infinitetalk_0071.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0071.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+EPy9_Q1eX1I+P1+C0+F69755-69985.jpg", "driving_audio": "3_audio/infinitetalk_0071.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p262/p262_358.wav", "generated_video": "1_fake/infinitetalk_0071.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+EPy9_Q1eX1I+P1+C0+F69755-69985__p262_358.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 71, "real_video": "0_real/infinitetalk_0071.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+EPy9_Q1eX1I+P1+C0+F69755-69985.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+EPy9_Q1eX1I+P1+C0+F69755-69985__p262_358.mp4" }, { "name": "infinitetalk_0072.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0072.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+EyVCNDPz-Ok+P0+C1+F137-302.jpg", "driving_audio": "3_audio/infinitetalk_0072.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p268/p268_104.wav", "generated_video": "1_fake/infinitetalk_0072.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+EyVCNDPz-Ok+P0+C1+F137-302__p268_104.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 72, "real_video": "0_real/infinitetalk_0072.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+EyVCNDPz-Ok+P0+C1+F137-302.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+EyVCNDPz-Ok+P0+C1+F137-302__p268_104.mp4" }, { "name": "infinitetalk_0073.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0073.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+FBgwEsM1yU8+P0+C2+F3164-3270.jpg", "driving_audio": "3_audio/infinitetalk_0073.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p228/p228_321.wav", "generated_video": "1_fake/infinitetalk_0073.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+FBgwEsM1yU8+P0+C2+F3164-3270__p228_321.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 73, "real_video": "0_real/infinitetalk_0073.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+FBgwEsM1yU8+P0+C2+F3164-3270.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+FBgwEsM1yU8+P0+C2+F3164-3270__p228_321.mp4" }, { "name": "infinitetalk_0074.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0074.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+Fli2X24CfSI+P0+C0+F5759-5955.jpg", "driving_audio": "3_audio/infinitetalk_0074.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p301/p301_275.wav", "generated_video": "1_fake/infinitetalk_0074.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+Fli2X24CfSI+P0+C0+F5759-5955__p301_275.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 74, "real_video": "0_real/infinitetalk_0074.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Fli2X24CfSI+P0+C0+F5759-5955.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+Fli2X24CfSI+P0+C0+F5759-5955__p301_275.mp4" }, { "name": "infinitetalk_0075.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0075.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+GcwLIj3p8Ak+P0+C0+F26512-26635.jpg", "driving_audio": "3_audio/infinitetalk_0075.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p363/p363_255.wav", "generated_video": "1_fake/infinitetalk_0075.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+GcwLIj3p8Ak+P0+C0+F26512-26635__p363_255.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 75, "real_video": "0_real/infinitetalk_0075.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+GcwLIj3p8Ak+P0+C0+F26512-26635.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+GcwLIj3p8Ak+P0+C0+F26512-26635__p363_255.mp4" }, { "name": "infinitetalk_0076.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0076.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+GlQPrN6bNCM+P0+C2+F12093-12213.jpg", "driving_audio": "3_audio/infinitetalk_0076.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p240/p240_185.wav", "generated_video": "1_fake/infinitetalk_0076.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+GlQPrN6bNCM+P0+C2+F12093-12213__p240_185.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 76, "real_video": "0_real/infinitetalk_0076.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+GlQPrN6bNCM+P0+C2+F12093-12213.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+GlQPrN6bNCM+P0+C2+F12093-12213__p240_185.mp4" }, { "name": "infinitetalk_0077.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0077.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+GlQPrN6bNCM+P1+C0+F23828-24031.jpg", "driving_audio": "3_audio/infinitetalk_0077.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p274/p274_066.wav", "generated_video": "1_fake/infinitetalk_0077.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+GlQPrN6bNCM+P1+C0+F23828-24031__p274_066.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 77, "real_video": "0_real/infinitetalk_0077.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+GlQPrN6bNCM+P1+C0+F23828-24031.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+GlQPrN6bNCM+P1+C0+F23828-24031__p274_066.mp4" }, { "name": "infinitetalk_0078.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0078.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+Hfd8oZtUomk+P0+C2+F28589-28830.jpg", "driving_audio": "3_audio/infinitetalk_0078.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p251/p251_111.wav", "generated_video": "1_fake/infinitetalk_0078.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+Hfd8oZtUomk+P0+C2+F28589-28830__p251_111.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 78, "real_video": "0_real/infinitetalk_0078.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Hfd8oZtUomk+P0+C2+F28589-28830.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+Hfd8oZtUomk+P0+C2+F28589-28830__p251_111.mp4" }, { "name": "infinitetalk_0079.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0079.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+I-O0ztG8Jcc+P0+C0+F13304-14035.jpg", "driving_audio": "3_audio/infinitetalk_0079.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p232/p232_073.wav", "generated_video": "1_fake/infinitetalk_0079.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+I-O0ztG8Jcc+P0+C0+F13304-14035__p232_073.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 79, "real_video": "0_real/infinitetalk_0079.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+I-O0ztG8Jcc+P0+C0+F13304-14035.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+I-O0ztG8Jcc+P0+C0+F13304-14035__p232_073.mp4" }, { "name": "infinitetalk_0080.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0080.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+IZ_T-mFATxU+P1+C1+F51580-51964.jpg", "driving_audio": "3_audio/infinitetalk_0080.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p362/p362_081.wav", "generated_video": "1_fake/infinitetalk_0080.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+IZ_T-mFATxU+P1+C1+F51580-51964__p362_081.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 80, "real_video": "0_real/infinitetalk_0080.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+IZ_T-mFATxU+P1+C1+F51580-51964.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+IZ_T-mFATxU+P1+C1+F51580-51964__p362_081.mp4" }, { "name": "infinitetalk_0081.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0081.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+IbAuiVLrA-s+P0+C0+F5671-5795.jpg", "driving_audio": "3_audio/infinitetalk_0081.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p283/p283_337.wav", "generated_video": "1_fake/infinitetalk_0081.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+IbAuiVLrA-s+P0+C0+F5671-5795__p283_337.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 81, "real_video": "0_real/infinitetalk_0081.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+IbAuiVLrA-s+P0+C0+F5671-5795.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+IbAuiVLrA-s+P0+C0+F5671-5795__p283_337.mp4" }, { "name": "infinitetalk_0082.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0082.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+KCr8s57hdzY+P0+C0+F18777-18902.jpg", "driving_audio": "3_audio/infinitetalk_0082.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p306/p306_223.wav", "generated_video": "1_fake/infinitetalk_0082.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+KCr8s57hdzY+P0+C0+F18777-18902__p306_223.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 82, "real_video": "0_real/infinitetalk_0082.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+KCr8s57hdzY+P0+C0+F18777-18902.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+KCr8s57hdzY+P0+C0+F18777-18902__p306_223.mp4" }, { "name": "infinitetalk_0083.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0083.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+KZpHl8LnmvA+P0+C0+F2648-3040.jpg", "driving_audio": "3_audio/infinitetalk_0083.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p294/p294_295.wav", "generated_video": "1_fake/infinitetalk_0083.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+KZpHl8LnmvA+P0+C0+F2648-3040__p294_295.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 83, "real_video": "0_real/infinitetalk_0083.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+KZpHl8LnmvA+P0+C0+F2648-3040.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+KZpHl8LnmvA+P0+C0+F2648-3040__p294_295.mp4" }, { "name": "infinitetalk_0084.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0084.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+KzQTrL_F30o+P3+C2+F6542-6677.jpg", "driving_audio": "3_audio/infinitetalk_0084.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p232/p232_332.wav", "generated_video": "1_fake/infinitetalk_0084.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+KzQTrL_F30o+P3+C2+F6542-6677__p232_332.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 84, "real_video": "0_real/infinitetalk_0084.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+KzQTrL_F30o+P3+C2+F6542-6677.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+KzQTrL_F30o+P3+C2+F6542-6677__p232_332.mp4" }, { "name": "infinitetalk_0085.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0085.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+L0BWuwT-4hY+P0+C1+F4158-4532.jpg", "driving_audio": "3_audio/infinitetalk_0085.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p255/p255_273.wav", "generated_video": "1_fake/infinitetalk_0085.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+L0BWuwT-4hY+P0+C1+F4158-4532__p255_273.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 85, "real_video": "0_real/infinitetalk_0085.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+L0BWuwT-4hY+P0+C1+F4158-4532.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+L0BWuwT-4hY+P0+C1+F4158-4532__p255_273.mp4" }, { "name": "infinitetalk_0086.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0086.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+LQZawXaSeNw+P0+C2+F86259-86380.jpg", "driving_audio": "3_audio/infinitetalk_0086.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p363/p363_154.wav", "generated_video": "1_fake/infinitetalk_0086.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+LQZawXaSeNw+P0+C2+F86259-86380__p363_154.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 86, "real_video": "0_real/infinitetalk_0086.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+LQZawXaSeNw+P0+C2+F86259-86380.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+LQZawXaSeNw+P0+C2+F86259-86380__p363_154.mp4" }, { "name": "infinitetalk_0087.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0087.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+Lg2XLkA41BA+P0+C2+F6670-6831.jpg", "driving_audio": "3_audio/infinitetalk_0087.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p266/p266_100.wav", "generated_video": "1_fake/infinitetalk_0087.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+Lg2XLkA41BA+P0+C2+F6670-6831__p266_100.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 87, "real_video": "0_real/infinitetalk_0087.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Lg2XLkA41BA+P0+C2+F6670-6831.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+Lg2XLkA41BA+P0+C2+F6670-6831__p266_100.mp4" }, { "name": "infinitetalk_0088.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0088.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+M5YKRv8jYoA+P0+C2+F18950-19055.jpg", "driving_audio": "3_audio/infinitetalk_0088.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p254/p254_156.wav", "generated_video": "1_fake/infinitetalk_0088.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+M5YKRv8jYoA+P0+C2+F18950-19055__p254_156.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 88, "real_video": "0_real/infinitetalk_0088.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+M5YKRv8jYoA+P0+C2+F18950-19055.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+M5YKRv8jYoA+P0+C2+F18950-19055__p254_156.mp4" }, { "name": "infinitetalk_0089.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0089.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+M6_v8DgBliY+P0+C1+F412-547.jpg", "driving_audio": "3_audio/infinitetalk_0089.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p297/p297_137.wav", "generated_video": "1_fake/infinitetalk_0089.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+M6_v8DgBliY+P0+C1+F412-547__p297_137.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 89, "real_video": "0_real/infinitetalk_0089.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+M6_v8DgBliY+P0+C1+F412-547.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+M6_v8DgBliY+P0+C1+F412-547__p297_137.mp4" }, { "name": "infinitetalk_0090.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0090.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+MLRFAICNkJM+P0+C2+F10159-10810.jpg", "driving_audio": "3_audio/infinitetalk_0090.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p341/p341_339.wav", "generated_video": "1_fake/infinitetalk_0090.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+MLRFAICNkJM+P0+C2+F10159-10810__p341_339.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 90, "real_video": "0_real/infinitetalk_0090.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+MLRFAICNkJM+P0+C2+F10159-10810.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+MLRFAICNkJM+P0+C2+F10159-10810__p341_339.mp4" }, { "name": "infinitetalk_0091.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0091.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+MTLgEwJx2gk+P0+C2+F7850-8030.jpg", "driving_audio": "3_audio/infinitetalk_0091.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p361/p361_112.wav", "generated_video": "1_fake/infinitetalk_0091.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+MTLgEwJx2gk+P0+C2+F7850-8030__p361_112.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 91, "real_video": "0_real/infinitetalk_0091.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+MTLgEwJx2gk+P0+C2+F7850-8030.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+MTLgEwJx2gk+P0+C2+F7850-8030__p361_112.mp4" }, { "name": "infinitetalk_0092.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0092.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+MksPBLEcGZg+P0+C1+F45324-45463.jpg", "driving_audio": "3_audio/infinitetalk_0092.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p237/p237_005.wav", "generated_video": "1_fake/infinitetalk_0092.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+MksPBLEcGZg+P0+C1+F45324-45463__p237_005.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 92, "real_video": "0_real/infinitetalk_0092.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+MksPBLEcGZg+P0+C1+F45324-45463.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+MksPBLEcGZg+P0+C1+F45324-45463__p237_005.mp4" }, { "name": "infinitetalk_0093.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0093.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+N2O9lveUsyg+P0+C0+F1041-1217.jpg", "driving_audio": "3_audio/infinitetalk_0093.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p231/p231_466.wav", "generated_video": "1_fake/infinitetalk_0093.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+N2O9lveUsyg+P0+C0+F1041-1217__p231_466.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 93, "real_video": "0_real/infinitetalk_0093.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+N2O9lveUsyg+P0+C0+F1041-1217.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+N2O9lveUsyg+P0+C0+F1041-1217__p231_466.mp4" }, { "name": "infinitetalk_0094.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0094.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+N_GkQ3D5wUc+P1+C2+F41914-42040.jpg", "driving_audio": "3_audio/infinitetalk_0094.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p280/p280_254.wav", "generated_video": "1_fake/infinitetalk_0094.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+N_GkQ3D5wUc+P1+C2+F41914-42040__p280_254.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 94, "real_video": "0_real/infinitetalk_0094.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+N_GkQ3D5wUc+P1+C2+F41914-42040.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+N_GkQ3D5wUc+P1+C2+F41914-42040__p280_254.mp4" }, { "name": "infinitetalk_0095.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0095.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+OGHWMS8xnIU+P0+C1+F27805-27918.jpg", "driving_audio": "3_audio/infinitetalk_0095.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p264/p264_460.wav", "generated_video": "1_fake/infinitetalk_0095.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+OGHWMS8xnIU+P0+C1+F27805-27918__p264_460.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 95, "real_video": "0_real/infinitetalk_0095.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+OGHWMS8xnIU+P0+C1+F27805-27918.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+OGHWMS8xnIU+P0+C1+F27805-27918__p264_460.mp4" }, { "name": "infinitetalk_0096.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0096.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+OSI5DLrFxw4+P1+C0+F18082-18216.jpg", "driving_audio": "3_audio/infinitetalk_0096.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p294/p294_365.wav", "generated_video": "1_fake/infinitetalk_0096.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+OSI5DLrFxw4+P1+C0+F18082-18216__p294_365.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 96, "real_video": "0_real/infinitetalk_0096.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+OSI5DLrFxw4+P1+C0+F18082-18216.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+OSI5DLrFxw4+P1+C0+F18082-18216__p294_365.mp4" }, { "name": "infinitetalk_0097.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0097.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+OqymHbKFJ-c+P0+C1+F27049-27320.jpg", "driving_audio": "3_audio/infinitetalk_0097.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p335/p335_067.wav", "generated_video": "1_fake/infinitetalk_0097.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+OqymHbKFJ-c+P0+C1+F27049-27320__p335_067.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 97, "real_video": "0_real/infinitetalk_0097.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+OqymHbKFJ-c+P0+C1+F27049-27320.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+OqymHbKFJ-c+P0+C1+F27049-27320__p335_067.mp4" }, { "name": "infinitetalk_0098.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0098.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+PG5qCrNS6ec+P1+C2+F29469-29660.jpg", "driving_audio": "3_audio/infinitetalk_0098.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p274/p274_435.wav", "generated_video": "1_fake/infinitetalk_0098.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+PG5qCrNS6ec+P1+C2+F29469-29660__p274_435.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 98, "real_video": "0_real/infinitetalk_0098.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+PG5qCrNS6ec+P1+C2+F29469-29660.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+PG5qCrNS6ec+P1+C2+F29469-29660__p274_435.mp4" }, { "name": "infinitetalk_0099.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0099.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+Pm4LrNwr6jU+P0+C1+F2121-2230.jpg", "driving_audio": "3_audio/infinitetalk_0099.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p314/p314_264.wav", "generated_video": "1_fake/infinitetalk_0099.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+Pm4LrNwr6jU+P0+C1+F2121-2230__p314_264.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 99, "real_video": "0_real/infinitetalk_0099.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Pm4LrNwr6jU+P0+C1+F2121-2230.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+Pm4LrNwr6jU+P0+C1+F2121-2230__p314_264.mp4" }, { "name": "infinitetalk_0100.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0100.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+QUcWwJgnEFk+P0+C0+F2724-2850.jpg", "driving_audio": "3_audio/infinitetalk_0100.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p269/p269_403.wav", "generated_video": "1_fake/infinitetalk_0100.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+QUcWwJgnEFk+P0+C0+F2724-2850__p269_403.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 100, "real_video": "0_real/infinitetalk_0100.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+QUcWwJgnEFk+P0+C0+F2724-2850.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+QUcWwJgnEFk+P0+C0+F2724-2850__p269_403.mp4" }, { "name": "infinitetalk_0101.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0101.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+RcPyjffurHw+P0+C1+F43817-44463.jpg", "driving_audio": "3_audio/infinitetalk_0101.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p329/p329_330.wav", "generated_video": "1_fake/infinitetalk_0101.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+RcPyjffurHw+P0+C1+F43817-44463__p329_330.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 101, "real_video": "0_real/infinitetalk_0101.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+RcPyjffurHw+P0+C1+F43817-44463.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+RcPyjffurHw+P0+C1+F43817-44463__p329_330.mp4" }, { "name": "infinitetalk_0102.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0102.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+SAeDER3hAA8+P0+C2+F4445-4684.jpg", "driving_audio": "3_audio/infinitetalk_0102.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p276/p276_275.wav", "generated_video": "1_fake/infinitetalk_0102.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+SAeDER3hAA8+P0+C2+F4445-4684__p276_275.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 102, "real_video": "0_real/infinitetalk_0102.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+SAeDER3hAA8+P0+C2+F4445-4684.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+SAeDER3hAA8+P0+C2+F4445-4684__p276_275.mp4" }, { "name": "infinitetalk_0103.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0103.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+SBNuQNqgJnw+P0+C2+F27319-27423.jpg", "driving_audio": "3_audio/infinitetalk_0103.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p275/p275_276.wav", "generated_video": "1_fake/infinitetalk_0103.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+SBNuQNqgJnw+P0+C2+F27319-27423__p275_276.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 103, "real_video": "0_real/infinitetalk_0103.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+SBNuQNqgJnw+P0+C2+F27319-27423.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+SBNuQNqgJnw+P0+C2+F27319-27423__p275_276.mp4" }, { "name": "infinitetalk_0104.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0104.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+Sz5wAJ5fBEA+P0+C1+F2665-2790.jpg", "driving_audio": "3_audio/infinitetalk_0104.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p265/p265_299.wav", "generated_video": "1_fake/infinitetalk_0104.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+Sz5wAJ5fBEA+P0+C1+F2665-2790__p265_299.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 104, "real_video": "0_real/infinitetalk_0104.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Sz5wAJ5fBEA+P0+C1+F2665-2790.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+Sz5wAJ5fBEA+P0+C1+F2665-2790__p265_299.mp4" }, { "name": "infinitetalk_0105.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0105.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+TcDFFj9JlNI+P1+C1+F8252-8387.jpg", "driving_audio": "3_audio/infinitetalk_0105.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p336/p336_074.wav", "generated_video": "1_fake/infinitetalk_0105.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+TcDFFj9JlNI+P1+C1+F8252-8387__p336_074.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 105, "real_video": "0_real/infinitetalk_0105.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+TcDFFj9JlNI+P1+C1+F8252-8387.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+TcDFFj9JlNI+P1+C1+F8252-8387__p336_074.mp4" }, { "name": "infinitetalk_0106.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0106.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+U-fZfB05AN0+P1+C1+F3625-3799.jpg", "driving_audio": "3_audio/infinitetalk_0106.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p240/p240_249.wav", "generated_video": "1_fake/infinitetalk_0106.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+U-fZfB05AN0+P1+C1+F3625-3799__p240_249.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 106, "real_video": "0_real/infinitetalk_0106.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+U-fZfB05AN0+P1+C1+F3625-3799.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+U-fZfB05AN0+P1+C1+F3625-3799__p240_249.mp4" }, { "name": "infinitetalk_0107.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0107.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+UKCb960G_jo+P0+C2+F11657-11779.jpg", "driving_audio": "3_audio/infinitetalk_0107.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p362/p362_277.wav", "generated_video": "1_fake/infinitetalk_0107.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+UKCb960G_jo+P0+C2+F11657-11779__p362_277.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 107, "real_video": "0_real/infinitetalk_0107.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+UKCb960G_jo+P0+C2+F11657-11779.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+UKCb960G_jo+P0+C2+F11657-11779__p362_277.mp4" }, { "name": "infinitetalk_0108.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0108.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+URZnSGJt6Cs+P0+C1+F7099-7365.jpg", "driving_audio": "3_audio/infinitetalk_0108.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p257/p257_369.wav", "generated_video": "1_fake/infinitetalk_0108.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+URZnSGJt6Cs+P0+C1+F7099-7365__p257_369.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 108, "real_video": "0_real/infinitetalk_0108.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+URZnSGJt6Cs+P0+C1+F7099-7365.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+URZnSGJt6Cs+P0+C1+F7099-7365__p257_369.mp4" }, { "name": "infinitetalk_0109.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0109.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+Upqf0RSbr-4+P0+C0+F850-1089.jpg", "driving_audio": "3_audio/infinitetalk_0109.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p326/p326_025.wav", "generated_video": "1_fake/infinitetalk_0109.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+Upqf0RSbr-4+P0+C0+F850-1089__p326_025.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 109, "real_video": "0_real/infinitetalk_0109.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Upqf0RSbr-4+P0+C0+F850-1089.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+Upqf0RSbr-4+P0+C0+F850-1089__p326_025.mp4" }, { "name": "infinitetalk_0110.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0110.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+VCeyqlbe6aI+P0+C0+F12828-13182.jpg", "driving_audio": "3_audio/infinitetalk_0110.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p269/p269_304.wav", "generated_video": "1_fake/infinitetalk_0110.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+VCeyqlbe6aI+P0+C0+F12828-13182__p269_304.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 110, "real_video": "0_real/infinitetalk_0110.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+VCeyqlbe6aI+P0+C0+F12828-13182.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+VCeyqlbe6aI+P0+C0+F12828-13182__p269_304.mp4" }, { "name": "infinitetalk_0111.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0111.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+VlWnh5HD5dU+P0+C1+F4950-5209.jpg", "driving_audio": "3_audio/infinitetalk_0111.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p257/p257_130.wav", "generated_video": "1_fake/infinitetalk_0111.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+VlWnh5HD5dU+P0+C1+F4950-5209__p257_130.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 111, "real_video": "0_real/infinitetalk_0111.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+VlWnh5HD5dU+P0+C1+F4950-5209.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+VlWnh5HD5dU+P0+C1+F4950-5209__p257_130.mp4" }, { "name": "infinitetalk_0112.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0112.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+Vzt2p4XkKco+P0+C2+F2148-2345.jpg", "driving_audio": "3_audio/infinitetalk_0112.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p232/p232_220.wav", "generated_video": "1_fake/infinitetalk_0112.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+Vzt2p4XkKco+P0+C2+F2148-2345__p232_220.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 112, "real_video": "0_real/infinitetalk_0112.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Vzt2p4XkKco+P0+C2+F2148-2345.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+Vzt2p4XkKco+P0+C2+F2148-2345__p232_220.mp4" }, { "name": "infinitetalk_0113.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0113.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+WbGymZprEMs+P0+C0+F32705-32946.jpg", "driving_audio": "3_audio/infinitetalk_0113.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p333/p333_134.wav", "generated_video": "1_fake/infinitetalk_0113.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+WbGymZprEMs+P0+C0+F32705-32946__p333_134.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 113, "real_video": "0_real/infinitetalk_0113.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+WbGymZprEMs+P0+C0+F32705-32946.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+WbGymZprEMs+P0+C0+F32705-32946__p333_134.mp4" }, { "name": "infinitetalk_0114.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0114.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+WdxyeU2gzLQ+P0+C0+F5759-5897.jpg", "driving_audio": "3_audio/infinitetalk_0114.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p262/p262_384.wav", "generated_video": "1_fake/infinitetalk_0114.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+WdxyeU2gzLQ+P0+C0+F5759-5897__p262_384.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 114, "real_video": "0_real/infinitetalk_0114.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+WdxyeU2gzLQ+P0+C0+F5759-5897.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+WdxyeU2gzLQ+P0+C0+F5759-5897__p262_384.mp4" }, { "name": "infinitetalk_0115.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0115.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+WzHMRG7jhpQ+P0+C0+F2100-2239.jpg", "driving_audio": "3_audio/infinitetalk_0115.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p268/p268_242.wav", "generated_video": "1_fake/infinitetalk_0115.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+WzHMRG7jhpQ+P0+C0+F2100-2239__p268_242.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 115, "real_video": "0_real/infinitetalk_0115.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+WzHMRG7jhpQ+P0+C0+F2100-2239.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+WzHMRG7jhpQ+P0+C0+F2100-2239__p268_242.mp4" }, { "name": "infinitetalk_0116.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0116.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+X8CZwq5EjtY+P0+C2+F6284-6745.jpg", "driving_audio": "3_audio/infinitetalk_0116.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p339/p339_159.wav", "generated_video": "1_fake/infinitetalk_0116.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+X8CZwq5EjtY+P0+C2+F6284-6745__p339_159.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 116, "real_video": "0_real/infinitetalk_0116.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+X8CZwq5EjtY+P0+C2+F6284-6745.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+X8CZwq5EjtY+P0+C2+F6284-6745__p339_159.mp4" }, { "name": "infinitetalk_0117.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0117.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+Xd5ySYjaom8+P0+C2+F8140-8435.jpg", "driving_audio": "3_audio/infinitetalk_0117.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p295/p295_173.wav", "generated_video": "1_fake/infinitetalk_0117.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+Xd5ySYjaom8+P0+C2+F8140-8435__p295_173.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 117, "real_video": "0_real/infinitetalk_0117.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Xd5ySYjaom8+P0+C2+F8140-8435.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+Xd5ySYjaom8+P0+C2+F8140-8435__p295_173.mp4" }, { "name": "infinitetalk_0118.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0118.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+XuYruSTavgY+P0+C1+F7749-7905.jpg", "driving_audio": "3_audio/infinitetalk_0118.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p258/p258_327.wav", "generated_video": "1_fake/infinitetalk_0118.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+XuYruSTavgY+P0+C1+F7749-7905__p258_327.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 118, "real_video": "0_real/infinitetalk_0118.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+XuYruSTavgY+P0+C1+F7749-7905.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+XuYruSTavgY+P0+C1+F7749-7905__p258_327.mp4" }, { "name": "infinitetalk_0119.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0119.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+Y58xQAgZCIk+P0+C1+F5243-5476.jpg", "driving_audio": "3_audio/infinitetalk_0119.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p361/p361_089.wav", "generated_video": "1_fake/infinitetalk_0119.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+Y58xQAgZCIk+P0+C1+F5243-5476__p361_089.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 119, "real_video": "0_real/infinitetalk_0119.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Y58xQAgZCIk+P0+C1+F5243-5476.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+Y58xQAgZCIk+P0+C1+F5243-5476__p361_089.mp4" }, { "name": "infinitetalk_0120.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0120.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+YCWCzB_PNq8+P1+C1+F5788-6020.jpg", "driving_audio": "3_audio/infinitetalk_0120.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p301/p301_266.wav", "generated_video": "1_fake/infinitetalk_0120.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+YCWCzB_PNq8+P1+C1+F5788-6020__p301_266.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 120, "real_video": "0_real/infinitetalk_0120.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+YCWCzB_PNq8+P1+C1+F5788-6020.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+YCWCzB_PNq8+P1+C1+F5788-6020__p301_266.mp4" }, { "name": "infinitetalk_0121.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0121.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+YD3VmX0chU0+P0+C1+F46558-46682.jpg", "driving_audio": "3_audio/infinitetalk_0121.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p232/p232_224.wav", "generated_video": "1_fake/infinitetalk_0121.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+YD3VmX0chU0+P0+C1+F46558-46682__p232_224.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 121, "real_video": "0_real/infinitetalk_0121.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+YD3VmX0chU0+P0+C1+F46558-46682.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+YD3VmX0chU0+P0+C1+F46558-46682__p232_224.mp4" }, { "name": "infinitetalk_0122.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0122.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+YGU2N3Cgwqw+P0+C1+F48261-48676.jpg", "driving_audio": "3_audio/infinitetalk_0122.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p297/p297_336.wav", "generated_video": "1_fake/infinitetalk_0122.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+YGU2N3Cgwqw+P0+C1+F48261-48676__p297_336.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 122, "real_video": "0_real/infinitetalk_0122.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+YGU2N3Cgwqw+P0+C1+F48261-48676.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+YGU2N3Cgwqw+P0+C1+F48261-48676__p297_336.mp4" }, { "name": "infinitetalk_0123.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0123.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+_gXFzH7VFXo+P0+C1+F11940-12048.jpg", "driving_audio": "3_audio/infinitetalk_0123.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p261/p261_129.wav", "generated_video": "1_fake/infinitetalk_0123.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+_gXFzH7VFXo+P0+C1+F11940-12048__p261_129.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 123, "real_video": "0_real/infinitetalk_0123.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+_gXFzH7VFXo+P0+C1+F11940-12048.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+_gXFzH7VFXo+P0+C1+F11940-12048__p261_129.mp4" }, { "name": "infinitetalk_0124.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0124.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+a7Ams-DoJMI+P0+C1+F4152-4413.jpg", "driving_audio": "3_audio/infinitetalk_0124.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p303/p303_289.wav", "generated_video": "1_fake/infinitetalk_0124.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+a7Ams-DoJMI+P0+C1+F4152-4413__p303_289.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 124, "real_video": "0_real/infinitetalk_0124.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+a7Ams-DoJMI+P0+C1+F4152-4413.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+a7Ams-DoJMI+P0+C1+F4152-4413__p303_289.mp4" }, { "name": "infinitetalk_0125.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0125.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+bJnSZ8kxOis+P0+C1+F39919-40046.jpg", "driving_audio": "3_audio/infinitetalk_0125.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p323/p323_096.wav", "generated_video": "1_fake/infinitetalk_0125.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+bJnSZ8kxOis+P0+C1+F39919-40046__p323_096.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 125, "real_video": "0_real/infinitetalk_0125.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+bJnSZ8kxOis+P0+C1+F39919-40046.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+bJnSZ8kxOis+P0+C1+F39919-40046__p323_096.mp4" }, { "name": "infinitetalk_0126.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0126.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+bJnSZ8kxOis+P0+C2+F51270-51439.jpg", "driving_audio": "3_audio/infinitetalk_0126.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p248/p248_224.wav", "generated_video": "1_fake/infinitetalk_0126.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+bJnSZ8kxOis+P0+C2+F51270-51439__p248_224.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 126, "real_video": "0_real/infinitetalk_0126.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+bJnSZ8kxOis+P0+C2+F51270-51439.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+bJnSZ8kxOis+P0+C2+F51270-51439__p248_224.mp4" }, { "name": "infinitetalk_0127.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0127.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+bqT-NE98gSI+P0+C2+F10191-10383.jpg", "driving_audio": "3_audio/infinitetalk_0127.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p258/p258_063.wav", "generated_video": "1_fake/infinitetalk_0127.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+bqT-NE98gSI+P0+C2+F10191-10383__p258_063.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 127, "real_video": "0_real/infinitetalk_0127.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+bqT-NE98gSI+P0+C2+F10191-10383.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+bqT-NE98gSI+P0+C2+F10191-10383__p258_063.mp4" }, { "name": "infinitetalk_0128.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0128.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+buRmXeNJsmw+P1+C0+F4377-4484.jpg", "driving_audio": "3_audio/infinitetalk_0128.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p233/p233_290.wav", "generated_video": "1_fake/infinitetalk_0128.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+buRmXeNJsmw+P1+C0+F4377-4484__p233_290.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 128, "real_video": "0_real/infinitetalk_0128.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+buRmXeNJsmw+P1+C0+F4377-4484.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+buRmXeNJsmw+P1+C0+F4377-4484__p233_290.mp4" }, { "name": "infinitetalk_0129.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0129.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+c1fYLfQjUDo+P0+C1+F5821-6199.jpg", "driving_audio": "3_audio/infinitetalk_0129.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p277/p277_132.wav", "generated_video": "1_fake/infinitetalk_0129.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+c1fYLfQjUDo+P0+C1+F5821-6199__p277_132.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 129, "real_video": "0_real/infinitetalk_0129.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+c1fYLfQjUDo+P0+C1+F5821-6199.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+c1fYLfQjUDo+P0+C1+F5821-6199__p277_132.mp4" }, { "name": "infinitetalk_0130.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0130.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+cXbzljh83l8+P0+C2+F5557-5759.jpg", "driving_audio": "3_audio/infinitetalk_0130.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p250/p250_196.wav", "generated_video": "1_fake/infinitetalk_0130.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+cXbzljh83l8+P0+C2+F5557-5759__p250_196.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 130, "real_video": "0_real/infinitetalk_0130.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+cXbzljh83l8+P0+C2+F5557-5759.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+cXbzljh83l8+P0+C2+F5557-5759__p250_196.mp4" }, { "name": "infinitetalk_0131.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0131.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+d5nr2VnTIAA+P0+C0+F2330-2476.jpg", "driving_audio": "3_audio/infinitetalk_0131.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p340/p340_360.wav", "generated_video": "1_fake/infinitetalk_0131.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+d5nr2VnTIAA+P0+C0+F2330-2476__p340_360.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 131, "real_video": "0_real/infinitetalk_0131.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+d5nr2VnTIAA+P0+C0+F2330-2476.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+d5nr2VnTIAA+P0+C0+F2330-2476__p340_360.mp4" }, { "name": "infinitetalk_0132.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0132.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+dDbJEVR6pSU+P0+C0+F3141-3298.jpg", "driving_audio": "3_audio/infinitetalk_0132.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p239/p239_192.wav", "generated_video": "1_fake/infinitetalk_0132.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+dDbJEVR6pSU+P0+C0+F3141-3298__p239_192.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 132, "real_video": "0_real/infinitetalk_0132.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+dDbJEVR6pSU+P0+C0+F3141-3298.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+dDbJEVR6pSU+P0+C0+F3141-3298__p239_192.mp4" }, { "name": "infinitetalk_0133.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0133.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+dK90OM26uW8+P0+C0+F4090-4196.jpg", "driving_audio": "3_audio/infinitetalk_0133.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p278/p278_319.wav", "generated_video": "1_fake/infinitetalk_0133.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+dK90OM26uW8+P0+C0+F4090-4196__p278_319.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 133, "real_video": "0_real/infinitetalk_0133.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+dK90OM26uW8+P0+C0+F4090-4196.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+dK90OM26uW8+P0+C0+F4090-4196__p278_319.mp4" }, { "name": "infinitetalk_0134.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0134.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+dOv3XKBu1Dw+P1+C2+F43406-43537.jpg", "driving_audio": "3_audio/infinitetalk_0134.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p295/p295_157.wav", "generated_video": "1_fake/infinitetalk_0134.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+dOv3XKBu1Dw+P1+C2+F43406-43537__p295_157.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 134, "real_video": "0_real/infinitetalk_0134.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+dOv3XKBu1Dw+P1+C2+F43406-43537.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+dOv3XKBu1Dw+P1+C2+F43406-43537__p295_157.mp4" }, { "name": "infinitetalk_0135.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0135.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+dY2ujYs6pVM+P0+C1+F5064-5209.jpg", "driving_audio": "3_audio/infinitetalk_0135.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p336/p336_307.wav", "generated_video": "1_fake/infinitetalk_0135.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+dY2ujYs6pVM+P0+C1+F5064-5209__p336_307.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 135, "real_video": "0_real/infinitetalk_0135.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+dY2ujYs6pVM+P0+C1+F5064-5209.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+dY2ujYs6pVM+P0+C1+F5064-5209__p336_307.mp4" }, { "name": "infinitetalk_0136.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0136.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+eco-Lvt0Pj0+P1+C0+F3634-3735.jpg", "driving_audio": "3_audio/infinitetalk_0136.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p306/p306_078.wav", "generated_video": "1_fake/infinitetalk_0136.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+eco-Lvt0Pj0+P1+C0+F3634-3735__p306_078.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 136, "real_video": "0_real/infinitetalk_0136.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+eco-Lvt0Pj0+P1+C0+F3634-3735.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+eco-Lvt0Pj0+P1+C0+F3634-3735__p306_078.mp4" }, { "name": "infinitetalk_0137.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0137.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+ej0v1iwHwpI+P0+C1+F3150-3343.jpg", "driving_audio": "3_audio/infinitetalk_0137.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p270/p270_391.wav", "generated_video": "1_fake/infinitetalk_0137.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+ej0v1iwHwpI+P0+C1+F3150-3343__p270_391.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 137, "real_video": "0_real/infinitetalk_0137.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+ej0v1iwHwpI+P0+C1+F3150-3343.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+ej0v1iwHwpI+P0+C1+F3150-3343__p270_391.mp4" }, { "name": "infinitetalk_0138.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0138.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+gTksiG6TptA+P1+C2+F62839-62950.jpg", "driving_audio": "3_audio/infinitetalk_0138.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p282/p282_150.wav", "generated_video": "1_fake/infinitetalk_0138.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+gTksiG6TptA+P1+C2+F62839-62950__p282_150.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 138, "real_video": "0_real/infinitetalk_0138.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+gTksiG6TptA+P1+C2+F62839-62950.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+gTksiG6TptA+P1+C2+F62839-62950__p282_150.mp4" }, { "name": "infinitetalk_0139.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0139.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+glNJHVJgI2Y+P9+C2+F28522-28671.jpg", "driving_audio": "3_audio/infinitetalk_0139.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p230/p230_397.wav", "generated_video": "1_fake/infinitetalk_0139.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+glNJHVJgI2Y+P9+C2+F28522-28671__p230_397.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 139, "real_video": "0_real/infinitetalk_0139.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+glNJHVJgI2Y+P9+C2+F28522-28671.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+glNJHVJgI2Y+P9+C2+F28522-28671__p230_397.mp4" }, { "name": "infinitetalk_0140.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0140.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+i1QZDA1BFUo+P1+C2+F35785-35925.jpg", "driving_audio": "3_audio/infinitetalk_0140.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p236/p236_201.wav", "generated_video": "1_fake/infinitetalk_0140.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+i1QZDA1BFUo+P1+C2+F35785-35925__p236_201.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 140, "real_video": "0_real/infinitetalk_0140.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+i1QZDA1BFUo+P1+C2+F35785-35925.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+i1QZDA1BFUo+P1+C2+F35785-35925__p236_201.mp4" }, { "name": "infinitetalk_0141.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0141.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+jhpPb003PP4+P0+C0+F2349-2455.jpg", "driving_audio": "3_audio/infinitetalk_0141.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p295/p295_123.wav", "generated_video": "1_fake/infinitetalk_0141.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+jhpPb003PP4+P0+C0+F2349-2455__p295_123.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 141, "real_video": "0_real/infinitetalk_0141.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+jhpPb003PP4+P0+C0+F2349-2455.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+jhpPb003PP4+P0+C0+F2349-2455__p295_123.mp4" }, { "name": "infinitetalk_0142.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0142.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+ke6WQxVTtM0+P2+C0+F13585-13801.jpg", "driving_audio": "3_audio/infinitetalk_0142.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p266/p266_018.wav", "generated_video": "1_fake/infinitetalk_0142.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+ke6WQxVTtM0+P2+C0+F13585-13801__p266_018.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 142, "real_video": "0_real/infinitetalk_0142.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+ke6WQxVTtM0+P2+C0+F13585-13801.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+ke6WQxVTtM0+P2+C0+F13585-13801__p266_018.mp4" }, { "name": "infinitetalk_0143.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0143.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+kjrtv5o1aDs+P0+C0+F3130-3252.jpg", "driving_audio": "3_audio/infinitetalk_0143.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p249/p249_324.wav", "generated_video": "1_fake/infinitetalk_0143.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+kjrtv5o1aDs+P0+C0+F3130-3252__p249_324.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 143, "real_video": "0_real/infinitetalk_0143.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+kjrtv5o1aDs+P0+C0+F3130-3252.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+kjrtv5o1aDs+P0+C0+F3130-3252__p249_324.mp4" }, { "name": "infinitetalk_0144.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0144.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+nA1PvPfT4uQ+P0+C1+F4816-4959.jpg", "driving_audio": "3_audio/infinitetalk_0144.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p245/p245_060.wav", "generated_video": "1_fake/infinitetalk_0144.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+nA1PvPfT4uQ+P0+C1+F4816-4959__p245_060.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 144, "real_video": "0_real/infinitetalk_0144.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+nA1PvPfT4uQ+P0+C1+F4816-4959.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+nA1PvPfT4uQ+P0+C1+F4816-4959__p245_060.mp4" }, { "name": "infinitetalk_0145.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0145.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+nmoxSjkyn8M+P1+C1+F18133-18410.jpg", "driving_audio": "3_audio/infinitetalk_0145.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p238/p238_299.wav", "generated_video": "1_fake/infinitetalk_0145.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+nmoxSjkyn8M+P1+C1+F18133-18410__p238_299.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 145, "real_video": "0_real/infinitetalk_0145.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+nmoxSjkyn8M+P1+C1+F18133-18410.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+nmoxSjkyn8M+P1+C1+F18133-18410__p238_299.mp4" }, { "name": "infinitetalk_0146.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0146.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+nqGJkOyVCTk+P0+C0+F2085-2366.jpg", "driving_audio": "3_audio/infinitetalk_0146.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p341/p341_113.wav", "generated_video": "1_fake/infinitetalk_0146.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+nqGJkOyVCTk+P0+C0+F2085-2366__p341_113.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 146, "real_video": "0_real/infinitetalk_0146.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+nqGJkOyVCTk+P0+C0+F2085-2366.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+nqGJkOyVCTk+P0+C0+F2085-2366__p341_113.mp4" }, { "name": "infinitetalk_0147.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0147.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+oegwLrZdk8w+P0+C0+F11505-11621.jpg", "driving_audio": "3_audio/infinitetalk_0147.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p376/p376_101.wav", "generated_video": "1_fake/infinitetalk_0147.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+oegwLrZdk8w+P0+C0+F11505-11621__p376_101.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 147, "real_video": "0_real/infinitetalk_0147.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+oegwLrZdk8w+P0+C0+F11505-11621.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+oegwLrZdk8w+P0+C0+F11505-11621__p376_101.mp4" }, { "name": "infinitetalk_0148.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0148.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+of-_HgJGJo8+P0+C1+F70733-71202.jpg", "driving_audio": "3_audio/infinitetalk_0148.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p230/p230_078.wav", "generated_video": "1_fake/infinitetalk_0148.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+of-_HgJGJo8+P0+C1+F70733-71202__p230_078.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 148, "real_video": "0_real/infinitetalk_0148.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+of-_HgJGJo8+P0+C1+F70733-71202.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+of-_HgJGJo8+P0+C1+F70733-71202__p230_078.mp4" }, { "name": "infinitetalk_0149.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0149.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+pnkKGbDArcA+P0+C0+F6011-6132.jpg", "driving_audio": "3_audio/infinitetalk_0149.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p341/p341_307.wav", "generated_video": "1_fake/infinitetalk_0149.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+pnkKGbDArcA+P0+C0+F6011-6132__p341_307.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 149, "real_video": "0_real/infinitetalk_0149.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+pnkKGbDArcA+P0+C0+F6011-6132.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+pnkKGbDArcA+P0+C0+F6011-6132__p341_307.mp4" }, { "name": "infinitetalk_0150.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0150.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+pt6LLNHnIfg+P0+C1+F30890-31048.jpg", "driving_audio": "3_audio/infinitetalk_0150.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p282/p282_281.wav", "generated_video": "1_fake/infinitetalk_0150.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+pt6LLNHnIfg+P0+C1+F30890-31048__p282_281.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 150, "real_video": "0_real/infinitetalk_0150.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+pt6LLNHnIfg+P0+C1+F30890-31048.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+pt6LLNHnIfg+P0+C1+F30890-31048__p282_281.mp4" }, { "name": "infinitetalk_0151.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0151.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+qAkGk3au-8A+P0+C1+F14504-14900.jpg", "driving_audio": "3_audio/infinitetalk_0151.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p233/p233_080.wav", "generated_video": "1_fake/infinitetalk_0151.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+qAkGk3au-8A+P0+C1+F14504-14900__p233_080.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 151, "real_video": "0_real/infinitetalk_0151.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+qAkGk3au-8A+P0+C1+F14504-14900.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+qAkGk3au-8A+P0+C1+F14504-14900__p233_080.mp4" }, { "name": "infinitetalk_0152.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0152.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+qi7APGihOVA+P0+C2+F4829-5210.jpg", "driving_audio": "3_audio/infinitetalk_0152.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p243/p243_164.wav", "generated_video": "1_fake/infinitetalk_0152.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+qi7APGihOVA+P0+C2+F4829-5210__p243_164.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 152, "real_video": "0_real/infinitetalk_0152.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+qi7APGihOVA+P0+C2+F4829-5210.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+qi7APGihOVA+P0+C2+F4829-5210__p243_164.mp4" }, { "name": "infinitetalk_0153.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0153.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+rBy18DRzBu4+P0+C1+F2278-2460.jpg", "driving_audio": "3_audio/infinitetalk_0153.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p339/p339_240.wav", "generated_video": "1_fake/infinitetalk_0153.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+rBy18DRzBu4+P0+C1+F2278-2460__p339_240.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 153, "real_video": "0_real/infinitetalk_0153.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+rBy18DRzBu4+P0+C1+F2278-2460.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+rBy18DRzBu4+P0+C1+F2278-2460__p339_240.mp4" }, { "name": "infinitetalk_0154.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0154.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+rJR12OgN5dc+P0+C1+F2055-2455.jpg", "driving_audio": "3_audio/infinitetalk_0154.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p292/p292_235.wav", "generated_video": "1_fake/infinitetalk_0154.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+rJR12OgN5dc+P0+C1+F2055-2455__p292_235.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 154, "real_video": "0_real/infinitetalk_0154.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+rJR12OgN5dc+P0+C1+F2055-2455.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+rJR12OgN5dc+P0+C1+F2055-2455__p292_235.mp4" }, { "name": "infinitetalk_0155.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0155.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+szowfCkYQ8g+P1+C0+F8979-9080.jpg", "driving_audio": "3_audio/infinitetalk_0155.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p308/p308_100.wav", "generated_video": "1_fake/infinitetalk_0155.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+szowfCkYQ8g+P1+C0+F8979-9080__p308_100.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 155, "real_video": "0_real/infinitetalk_0155.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+szowfCkYQ8g+P1+C0+F8979-9080.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+szowfCkYQ8g+P1+C0+F8979-9080__p308_100.mp4" }, { "name": "infinitetalk_0156.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0156.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+tCzQERoQ21E+P0+C2+F2033-2157.jpg", "driving_audio": "3_audio/infinitetalk_0156.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p330/p330_408.wav", "generated_video": "1_fake/infinitetalk_0156.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+tCzQERoQ21E+P0+C2+F2033-2157__p330_408.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 156, "real_video": "0_real/infinitetalk_0156.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+tCzQERoQ21E+P0+C2+F2033-2157.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+tCzQERoQ21E+P0+C2+F2033-2157__p330_408.mp4" }, { "name": "infinitetalk_0157.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0157.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+tNdOw-yNIUM+P0+C0+F28139-28458.jpg", "driving_audio": "3_audio/infinitetalk_0157.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p255/p255_081.wav", "generated_video": "1_fake/infinitetalk_0157.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+tNdOw-yNIUM+P0+C0+F28139-28458__p255_081.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 157, "real_video": "0_real/infinitetalk_0157.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+tNdOw-yNIUM+P0+C0+F28139-28458.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+tNdOw-yNIUM+P0+C0+F28139-28458__p255_081.mp4" }, { "name": "infinitetalk_0158.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0158.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+uVmOfr4720E+P0+C0+F74099-74230.jpg", "driving_audio": "3_audio/infinitetalk_0158.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p345/p345_117.wav", "generated_video": "1_fake/infinitetalk_0158.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+uVmOfr4720E+P0+C0+F74099-74230__p345_117.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 158, "real_video": "0_real/infinitetalk_0158.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+uVmOfr4720E+P0+C0+F74099-74230.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+uVmOfr4720E+P0+C0+F74099-74230__p345_117.mp4" }, { "name": "infinitetalk_0159.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0159.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+vqMwq8RMxeM+P1+C0+F33116-33224.jpg", "driving_audio": "3_audio/infinitetalk_0159.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p310/p310_372.wav", "generated_video": "1_fake/infinitetalk_0159.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+vqMwq8RMxeM+P1+C0+F33116-33224__p310_372.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 159, "real_video": "0_real/infinitetalk_0159.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+vqMwq8RMxeM+P1+C0+F33116-33224.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+vqMwq8RMxeM+P1+C0+F33116-33224__p310_372.mp4" }, { "name": "infinitetalk_0160.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0160.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+vuWPoGC6b1I+P0+C0+F5537-5647.jpg", "driving_audio": "3_audio/infinitetalk_0160.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p323/p323_164.wav", "generated_video": "1_fake/infinitetalk_0160.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+vuWPoGC6b1I+P0+C0+F5537-5647__p323_164.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 160, "real_video": "0_real/infinitetalk_0160.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+vuWPoGC6b1I+P0+C0+F5537-5647.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+vuWPoGC6b1I+P0+C0+F5537-5647__p323_164.mp4" }, { "name": "infinitetalk_0161.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0161.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+wOFnqM3fXhg+P0+C1+F39546-39689.jpg", "driving_audio": "3_audio/infinitetalk_0161.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p351/p351_389.wav", "generated_video": "1_fake/infinitetalk_0161.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+wOFnqM3fXhg+P0+C1+F39546-39689__p351_389.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 161, "real_video": "0_real/infinitetalk_0161.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+wOFnqM3fXhg+P0+C1+F39546-39689.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+wOFnqM3fXhg+P0+C1+F39546-39689__p351_389.mp4" }, { "name": "infinitetalk_0162.mp4", "variant": "vfhq_vctk", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0162.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+xHb3_PCNsR0+P0+C1+F3246-3625.jpg", "driving_audio": "3_audio/infinitetalk_0162.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p276/p276_176.wav", "generated_video": "1_fake/infinitetalk_0162.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch2/Clip+xHb3_PCNsR0+P0+C1+F3246-3625__p276_176.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk animates a still image driven by audio. This batch uses a VFHQ frame as identity and a VCTK wav as audio, generated with a 49-frame cap (~2s), shorter than the full driving wav; 3_audio is the muxed segment so video and audio align. Original full driving wav recorded in driving_audio_full_from. Identity is an image; no 0_real.", "index": 162, "real_video": "0_real/infinitetalk_0162.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+xHb3_PCNsR0+P0+C1+F3246-3625.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+xHb3_PCNsR0+P0+C1+F3246-3625__p276_176.mp4" }, { "name": "infinitetalk_0163.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0163.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+Qgq0NqN6a9A+P0+C2+F2375-2545.jpg", "driving_audio": "3_audio/infinitetalk_0163.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p276/p276_396.wav", "generated_video": "1_fake/infinitetalk_0163.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+Qgq0NqN6a9A+P0+C2+F2375-2545__p276_396.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 163, "real_video": "0_real/infinitetalk_0163.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Qgq0NqN6a9A+P0+C2+F2375-2545.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+Qgq0NqN6a9A+P0+C2+F2375-2545__p276_396.mp4" }, { "name": "infinitetalk_0164.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0164.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+3ZZkROFfgBY+P0+C0+F3760-3888.jpg", "driving_audio": "3_audio/infinitetalk_0164.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p297/p297_236.wav", "generated_video": "1_fake/infinitetalk_0164.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+3ZZkROFfgBY+P0+C0+F3760-3888__p297_236.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 164, "real_video": "0_real/infinitetalk_0164.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+3ZZkROFfgBY+P0+C0+F3760-3888.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+3ZZkROFfgBY+P0+C0+F3760-3888__p297_236.mp4" }, { "name": "infinitetalk_0165.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0165.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+0QEsnOorioQ+P0+C0+F1663-1784.jpg", "driving_audio": "3_audio/infinitetalk_0165.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p304/p304_022.wav", "generated_video": "1_fake/infinitetalk_0165.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+0QEsnOorioQ+P0+C0+F1663-1784__p304_022.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 165, "real_video": "0_real/infinitetalk_0165.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+0QEsnOorioQ+P0+C0+F1663-1784.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+0QEsnOorioQ+P0+C0+F1663-1784__p304_022.mp4" }, { "name": "infinitetalk_0166.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0166.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+-vEY2DV36Os+P0+C1+F5778-5895.jpg", "driving_audio": "3_audio/infinitetalk_0166.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p329/p329_231.wav", "generated_video": "1_fake/infinitetalk_0166.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+-vEY2DV36Os+P0+C1+F5778-5895__p329_231.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 166, "real_video": "0_real/infinitetalk_0166.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+-vEY2DV36Os+P0+C1+F5778-5895.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+-vEY2DV36Os+P0+C1+F5778-5895__p329_231.mp4" }, { "name": "infinitetalk_0167.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0167.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+WwN0Vgo4SkI+P0+C2+F957-1178.jpg", "driving_audio": "3_audio/infinitetalk_0167.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p300/p300_197.wav", "generated_video": "1_fake/infinitetalk_0167.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+WwN0Vgo4SkI+P0+C2+F957-1178__p300_197.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 167, "real_video": "0_real/infinitetalk_0167.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+WwN0Vgo4SkI+P0+C2+F957-1178.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+WwN0Vgo4SkI+P0+C2+F957-1178__p300_197.mp4" }, { "name": "infinitetalk_0168.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0168.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+QMDGYXd_VzM+P0+C0+F14-570.jpg", "driving_audio": "3_audio/infinitetalk_0168.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p286/p286_275.wav", "generated_video": "1_fake/infinitetalk_0168.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+QMDGYXd_VzM+P0+C0+F14-570__p286_275.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 168, "real_video": "0_real/infinitetalk_0168.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+QMDGYXd_VzM+P0+C0+F14-570.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+QMDGYXd_VzM+P0+C0+F14-570__p286_275.mp4" }, { "name": "infinitetalk_0169.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0169.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+hot5ixctq70+P0+C1+F5266-5406.jpg", "driving_audio": "3_audio/infinitetalk_0169.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p275/p275_171.wav", "generated_video": "1_fake/infinitetalk_0169.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+hot5ixctq70+P0+C1+F5266-5406__p275_171.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 169, "real_video": "0_real/infinitetalk_0169.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+hot5ixctq70+P0+C1+F5266-5406.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+hot5ixctq70+P0+C1+F5266-5406__p275_171.mp4" }, { "name": "infinitetalk_0170.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0170.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+ycPOl8ifMgU+P0+C0+F1234-1368.jpg", "driving_audio": "3_audio/infinitetalk_0170.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p278/p278_036.wav", "generated_video": "1_fake/infinitetalk_0170.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+ycPOl8ifMgU+P0+C0+F1234-1368__p278_036.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 170, "real_video": "0_real/infinitetalk_0170.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+ycPOl8ifMgU+P0+C0+F1234-1368.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+ycPOl8ifMgU+P0+C0+F1234-1368__p278_036.mp4" }, { "name": "infinitetalk_0171.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0171.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+Ch9y34ah7UY+P0+C0+F1624-1985.jpg", "driving_audio": "3_audio/infinitetalk_0171.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p362/p362_296.wav", "generated_video": "1_fake/infinitetalk_0171.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+Ch9y34ah7UY+P0+C0+F1624-1985__p362_296.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 171, "real_video": "0_real/infinitetalk_0171.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Ch9y34ah7UY+P0+C0+F1624-1985.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+Ch9y34ah7UY+P0+C0+F1624-1985__p362_296.mp4" }, { "name": "infinitetalk_0172.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0172.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+JvXO7Aj9ZEs+P0+C2+F7000-7349.jpg", "driving_audio": "3_audio/infinitetalk_0172.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p340/p340_046.wav", "generated_video": "1_fake/infinitetalk_0172.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+JvXO7Aj9ZEs+P0+C2+F7000-7349__p340_046.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 172, "real_video": "0_real/infinitetalk_0172.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+JvXO7Aj9ZEs+P0+C2+F7000-7349.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+JvXO7Aj9ZEs+P0+C2+F7000-7349__p340_046.mp4" }, { "name": "infinitetalk_0173.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0173.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+XNZvmoVaVgU+P0+C0+F12220-12344.jpg", "driving_audio": "3_audio/infinitetalk_0173.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p318/p318_100.wav", "generated_video": "1_fake/infinitetalk_0173.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+XNZvmoVaVgU+P0+C0+F12220-12344__p318_100.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 173, "real_video": "0_real/infinitetalk_0173.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+XNZvmoVaVgU+P0+C0+F12220-12344.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+XNZvmoVaVgU+P0+C0+F12220-12344__p318_100.mp4" }, { "name": "infinitetalk_0174.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0174.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+ilk2AWaCQjE+P0+C1+F2443-2595.jpg", "driving_audio": "3_audio/infinitetalk_0174.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p304/p304_307.wav", "generated_video": "1_fake/infinitetalk_0174.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+ilk2AWaCQjE+P0+C1+F2443-2595__p304_307.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 174, "real_video": "0_real/infinitetalk_0174.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+ilk2AWaCQjE+P0+C1+F2443-2595.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+ilk2AWaCQjE+P0+C1+F2443-2595__p304_307.mp4" }, { "name": "infinitetalk_0175.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0175.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+aaDW3bZCvvA+P0+C0+F2930-3040.jpg", "driving_audio": "3_audio/infinitetalk_0175.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p281/p281_038.wav", "generated_video": "1_fake/infinitetalk_0175.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+aaDW3bZCvvA+P0+C0+F2930-3040__p281_038.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 175, "real_video": "0_real/infinitetalk_0175.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+aaDW3bZCvvA+P0+C0+F2930-3040.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+aaDW3bZCvvA+P0+C0+F2930-3040__p281_038.mp4" }, { "name": "infinitetalk_0176.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0176.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+j_itcamB8N8+P2+C0+F2269-2615.jpg", "driving_audio": "3_audio/infinitetalk_0176.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p351/p351_357.wav", "generated_video": "1_fake/infinitetalk_0176.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+j_itcamB8N8+P2+C0+F2269-2615__p351_357.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 176, "real_video": "0_real/infinitetalk_0176.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+j_itcamB8N8+P2+C0+F2269-2615.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+j_itcamB8N8+P2+C0+F2269-2615__p351_357.mp4" }, { "name": "infinitetalk_0177.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0177.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+buaqlAjjfn0+P0+C0+F1080-1692.jpg", "driving_audio": "3_audio/infinitetalk_0177.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p334/p334_019.wav", "generated_video": "1_fake/infinitetalk_0177.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+buaqlAjjfn0+P0+C0+F1080-1692__p334_019.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 177, "real_video": "0_real/infinitetalk_0177.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+buaqlAjjfn0+P0+C0+F1080-1692.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+buaqlAjjfn0+P0+C0+F1080-1692__p334_019.mp4" }, { "name": "infinitetalk_0178.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0178.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+4vfqZDelVEs+P1+C1+F2770-2910.jpg", "driving_audio": "3_audio/infinitetalk_0178.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p230/p230_084.wav", "generated_video": "1_fake/infinitetalk_0178.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+4vfqZDelVEs+P1+C1+F2770-2910__p230_084.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 178, "real_video": "0_real/infinitetalk_0178.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+4vfqZDelVEs+P1+C1+F2770-2910.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+4vfqZDelVEs+P1+C1+F2770-2910__p230_084.mp4" }, { "name": "infinitetalk_0179.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0179.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+DaaM6TKW5k4+P0+C0+F3449-3655.jpg", "driving_audio": "3_audio/infinitetalk_0179.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p266/p266_184.wav", "generated_video": "1_fake/infinitetalk_0179.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+DaaM6TKW5k4+P0+C0+F3449-3655__p266_184.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 179, "real_video": "0_real/infinitetalk_0179.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+DaaM6TKW5k4+P0+C0+F3449-3655.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+DaaM6TKW5k4+P0+C0+F3449-3655__p266_184.mp4" }, { "name": "infinitetalk_0180.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0180.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+OgrmBdlIvGs+P1+C0+F2586-3235.jpg", "driving_audio": "3_audio/infinitetalk_0180.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p241/p241_057.wav", "generated_video": "1_fake/infinitetalk_0180.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+OgrmBdlIvGs+P1+C0+F2586-3235__p241_057.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 180, "real_video": "0_real/infinitetalk_0180.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+OgrmBdlIvGs+P1+C0+F2586-3235.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+OgrmBdlIvGs+P1+C0+F2586-3235__p241_057.mp4" }, { "name": "infinitetalk_0181.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0181.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+65qIgbRmzqE+P1+C0+F7650-7913.jpg", "driving_audio": "3_audio/infinitetalk_0181.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p376/p376_294.wav", "generated_video": "1_fake/infinitetalk_0181.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+65qIgbRmzqE+P1+C0+F7650-7913__p376_294.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 181, "real_video": "0_real/infinitetalk_0181.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+65qIgbRmzqE+P1+C0+F7650-7913.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+65qIgbRmzqE+P1+C0+F7650-7913__p376_294.mp4" }, { "name": "infinitetalk_0182.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0182.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+U5M8p_LDU0s+P0+C1+F44504-44720.jpg", "driving_audio": "3_audio/infinitetalk_0182.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p317/p317_299.wav", "generated_video": "1_fake/infinitetalk_0182.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+U5M8p_LDU0s+P0+C1+F44504-44720__p317_299.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 182, "real_video": "0_real/infinitetalk_0182.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+U5M8p_LDU0s+P0+C1+F44504-44720.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+U5M8p_LDU0s+P0+C1+F44504-44720__p317_299.mp4" }, { "name": "infinitetalk_0183.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0183.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+Xqq5VUe1Ic8+P0+C1+F15479-15950.jpg", "driving_audio": "3_audio/infinitetalk_0183.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p280/p280_251.wav", "generated_video": "1_fake/infinitetalk_0183.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+Xqq5VUe1Ic8+P0+C1+F15479-15950__p280_251.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 183, "real_video": "0_real/infinitetalk_0183.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Xqq5VUe1Ic8+P0+C1+F15479-15950.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+Xqq5VUe1Ic8+P0+C1+F15479-15950__p280_251.mp4" }, { "name": "infinitetalk_0184.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0184.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+KrSUiuQovdw+P0+C2+F4524-4645.jpg", "driving_audio": "3_audio/infinitetalk_0184.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p347/p347_349.wav", "generated_video": "1_fake/infinitetalk_0184.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+KrSUiuQovdw+P0+C2+F4524-4645__p347_349.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 184, "real_video": "0_real/infinitetalk_0184.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+KrSUiuQovdw+P0+C2+F4524-4645.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+KrSUiuQovdw+P0+C2+F4524-4645__p347_349.mp4" }, { "name": "infinitetalk_0185.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0185.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+hCukZ2M_VvU+P0+C0+F14820-14928.jpg", "driving_audio": "3_audio/infinitetalk_0185.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p347/p347_229.wav", "generated_video": "1_fake/infinitetalk_0185.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+hCukZ2M_VvU+P0+C0+F14820-14928__p347_229.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 185, "real_video": "0_real/infinitetalk_0185.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+hCukZ2M_VvU+P0+C0+F14820-14928.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+hCukZ2M_VvU+P0+C0+F14820-14928__p347_229.mp4" }, { "name": "infinitetalk_0186.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0186.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+gN_WuwHSoXM+P0+C1+F16611-16718.jpg", "driving_audio": "3_audio/infinitetalk_0186.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p275/p275_220.wav", "generated_video": "1_fake/infinitetalk_0186.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+gN_WuwHSoXM+P0+C1+F16611-16718__p275_220.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 186, "real_video": "0_real/infinitetalk_0186.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+gN_WuwHSoXM+P0+C1+F16611-16718.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+gN_WuwHSoXM+P0+C1+F16611-16718__p275_220.mp4" }, { "name": "infinitetalk_0187.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0187.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+VXl2yN_BiXs+P1+C1+F21883-22051.jpg", "driving_audio": "3_audio/infinitetalk_0187.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p312/p312_418.wav", "generated_video": "1_fake/infinitetalk_0187.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+VXl2yN_BiXs+P1+C1+F21883-22051__p312_418.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 187, "real_video": "0_real/infinitetalk_0187.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+VXl2yN_BiXs+P1+C1+F21883-22051.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+VXl2yN_BiXs+P1+C1+F21883-22051__p312_418.mp4" }, { "name": "infinitetalk_0188.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0188.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+piTpiBgtaGA+P0+C0+F75859-76049.jpg", "driving_audio": "3_audio/infinitetalk_0188.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p264/p264_249.wav", "generated_video": "1_fake/infinitetalk_0188.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+piTpiBgtaGA+P0+C0+F75859-76049__p264_249.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 188, "real_video": "0_real/infinitetalk_0188.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+piTpiBgtaGA+P0+C0+F75859-76049.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+piTpiBgtaGA+P0+C0+F75859-76049__p264_249.mp4" }, { "name": "infinitetalk_0189.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0189.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+Cz6ZhQ3NpOY+P0+C1+F34749-34922.jpg", "driving_audio": "3_audio/infinitetalk_0189.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p269/p269_161.wav", "generated_video": "1_fake/infinitetalk_0189.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+Cz6ZhQ3NpOY+P0+C1+F34749-34922__p269_161.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 189, "real_video": "0_real/infinitetalk_0189.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Cz6ZhQ3NpOY+P0+C1+F34749-34922.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+Cz6ZhQ3NpOY+P0+C1+F34749-34922__p269_161.mp4" }, { "name": "infinitetalk_0190.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0190.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+tRUQRW16p90+P0+C0+F71391-71841.jpg", "driving_audio": "3_audio/infinitetalk_0190.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p278/p278_012.wav", "generated_video": "1_fake/infinitetalk_0190.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+tRUQRW16p90+P0+C0+F71391-71841__p278_012.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 190, "real_video": "0_real/infinitetalk_0190.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+tRUQRW16p90+P0+C0+F71391-71841.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+tRUQRW16p90+P0+C0+F71391-71841__p278_012.mp4" }, { "name": "infinitetalk_0191.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0191.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+wtALXV7bLX0+P1+C0+F2984-3115.jpg", "driving_audio": "3_audio/infinitetalk_0191.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p227/p227_187.wav", "generated_video": "1_fake/infinitetalk_0191.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+wtALXV7bLX0+P1+C0+F2984-3115__p227_187.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 191, "real_video": "0_real/infinitetalk_0191.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+wtALXV7bLX0+P1+C0+F2984-3115.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+wtALXV7bLX0+P1+C0+F2984-3115__p227_187.mp4" }, { "name": "infinitetalk_0192.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0192.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+XXZvxnbrl6c+P0+C0+F36268-36391.jpg", "driving_audio": "3_audio/infinitetalk_0192.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p256/p256_172.wav", "generated_video": "1_fake/infinitetalk_0192.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+XXZvxnbrl6c+P0+C0+F36268-36391__p256_172.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 192, "real_video": "0_real/infinitetalk_0192.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+XXZvxnbrl6c+P0+C0+F36268-36391.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+XXZvxnbrl6c+P0+C0+F36268-36391__p256_172.mp4" }, { "name": "infinitetalk_0193.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0193.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+idS6e9dck7A+P0+C0+F3154-3270.jpg", "driving_audio": "3_audio/infinitetalk_0193.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p243/p243_394.wav", "generated_video": "1_fake/infinitetalk_0193.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+idS6e9dck7A+P0+C0+F3154-3270__p243_394.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 193, "real_video": "0_real/infinitetalk_0193.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+idS6e9dck7A+P0+C0+F3154-3270.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+idS6e9dck7A+P0+C0+F3154-3270__p243_394.mp4" }, { "name": "infinitetalk_0194.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0194.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+VOxdReoQjvg+P2+C0+F33806-33993.jpg", "driving_audio": "3_audio/infinitetalk_0194.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p275/p275_121.wav", "generated_video": "1_fake/infinitetalk_0194.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+VOxdReoQjvg+P2+C0+F33806-33993__p275_121.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 194, "real_video": "0_real/infinitetalk_0194.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+VOxdReoQjvg+P2+C0+F33806-33993.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+VOxdReoQjvg+P2+C0+F33806-33993__p275_121.mp4" }, { "name": "infinitetalk_0195.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0195.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+K_qgp9uiLmY+P0+C1+F32984-33813.jpg", "driving_audio": "3_audio/infinitetalk_0195.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p306/p306_064.wav", "generated_video": "1_fake/infinitetalk_0195.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+K_qgp9uiLmY+P0+C1+F32984-33813__p306_064.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 195, "real_video": "0_real/infinitetalk_0195.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+K_qgp9uiLmY+P0+C1+F32984-33813.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+K_qgp9uiLmY+P0+C1+F32984-33813__p306_064.mp4" }, { "name": "infinitetalk_0196.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0196.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+M9SAzpwNuxY+P0+C2+F8709-8836.jpg", "driving_audio": "3_audio/infinitetalk_0196.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p266/p266_238.wav", "generated_video": "1_fake/infinitetalk_0196.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+M9SAzpwNuxY+P0+C2+F8709-8836__p266_238.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 196, "real_video": "0_real/infinitetalk_0196.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+M9SAzpwNuxY+P0+C2+F8709-8836.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+M9SAzpwNuxY+P0+C2+F8709-8836__p266_238.mp4" }, { "name": "infinitetalk_0197.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0197.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+LgQTkCO5MD8+P0+C1+F3061-3286.jpg", "driving_audio": "3_audio/infinitetalk_0197.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p311/p311_281.wav", "generated_video": "1_fake/infinitetalk_0197.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+LgQTkCO5MD8+P0+C1+F3061-3286__p311_281.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 197, "real_video": "0_real/infinitetalk_0197.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+LgQTkCO5MD8+P0+C1+F3061-3286.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+LgQTkCO5MD8+P0+C1+F3061-3286__p311_281.mp4" }, { "name": "infinitetalk_0198.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0198.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+jAFV36xrpng+P0+C1+F6917-7235.jpg", "driving_audio": "3_audio/infinitetalk_0198.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p241/p241_205.wav", "generated_video": "1_fake/infinitetalk_0198.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+jAFV36xrpng+P0+C1+F6917-7235__p241_205.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 198, "real_video": "0_real/infinitetalk_0198.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+jAFV36xrpng+P0+C1+F6917-7235.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+jAFV36xrpng+P0+C1+F6917-7235__p241_205.mp4" }, { "name": "infinitetalk_0199.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0199.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+B6ur6AX9JgI+P0+C1+F6381-6709.jpg", "driving_audio": "3_audio/infinitetalk_0199.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p269/p269_007.wav", "generated_video": "1_fake/infinitetalk_0199.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+B6ur6AX9JgI+P0+C1+F6381-6709__p269_007.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 199, "real_video": "0_real/infinitetalk_0199.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+B6ur6AX9JgI+P0+C1+F6381-6709.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+B6ur6AX9JgI+P0+C1+F6381-6709__p269_007.mp4" }, { "name": "infinitetalk_0200.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0200.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+xje1oM17Vk8+P1+C2+F24615-24805.jpg", "driving_audio": "3_audio/infinitetalk_0200.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p284/p284_009.wav", "generated_video": "1_fake/infinitetalk_0200.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+xje1oM17Vk8+P1+C2+F24615-24805__p284_009.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 200, "real_video": "0_real/infinitetalk_0200.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+xje1oM17Vk8+P1+C2+F24615-24805.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+xje1oM17Vk8+P1+C2+F24615-24805__p284_009.mp4" }, { "name": "infinitetalk_0201.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0201.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+bLSOn0W1Qck+P0+C0+F10019-10400.jpg", "driving_audio": "3_audio/infinitetalk_0201.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p268/p268_288.wav", "generated_video": "1_fake/infinitetalk_0201.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+bLSOn0W1Qck+P0+C0+F10019-10400__p268_288.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 201, "real_video": "0_real/infinitetalk_0201.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+bLSOn0W1Qck+P0+C0+F10019-10400.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+bLSOn0W1Qck+P0+C0+F10019-10400__p268_288.mp4" }, { "name": "infinitetalk_0202.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0202.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+-IW8wEAPglw+P0+C2+F16520-16757.jpg", "driving_audio": "3_audio/infinitetalk_0202.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p305/p305_338.wav", "generated_video": "1_fake/infinitetalk_0202.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+-IW8wEAPglw+P0+C2+F16520-16757__p305_338.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 202, "real_video": "0_real/infinitetalk_0202.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+-IW8wEAPglw+P0+C2+F16520-16757.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+-IW8wEAPglw+P0+C2+F16520-16757__p305_338.mp4" }, { "name": "infinitetalk_0203.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0203.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+gTksiG6TptA+P1+C1+F58909-59090.jpg", "driving_audio": "3_audio/infinitetalk_0203.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p279/p279_169.wav", "generated_video": "1_fake/infinitetalk_0203.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+gTksiG6TptA+P1+C1+F58909-59090__p279_169.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 203, "real_video": "0_real/infinitetalk_0203.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+gTksiG6TptA+P1+C1+F58909-59090.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+gTksiG6TptA+P1+C1+F58909-59090__p279_169.mp4" }, { "name": "infinitetalk_0204.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0204.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+Oju0Ur-SOOQ+P0+C0+F4254-4725.jpg", "driving_audio": "3_audio/infinitetalk_0204.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p247/p247_436.wav", "generated_video": "1_fake/infinitetalk_0204.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+Oju0Ur-SOOQ+P0+C0+F4254-4725__p247_436.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 204, "real_video": "0_real/infinitetalk_0204.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Oju0Ur-SOOQ+P0+C0+F4254-4725.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+Oju0Ur-SOOQ+P0+C0+F4254-4725__p247_436.mp4" }, { "name": "infinitetalk_0205.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0205.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+lgBgFhFnI4M+P0+C1+F22934-23172.jpg", "driving_audio": "3_audio/infinitetalk_0205.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p262/p262_327.wav", "generated_video": "1_fake/infinitetalk_0205.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+lgBgFhFnI4M+P0+C1+F22934-23172__p262_327.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 205, "real_video": "0_real/infinitetalk_0205.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+lgBgFhFnI4M+P0+C1+F22934-23172.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+lgBgFhFnI4M+P0+C1+F22934-23172__p262_327.mp4" }, { "name": "infinitetalk_0206.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0206.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+4--y5nQdVE0+P1+C0+F30817-30991.jpg", "driving_audio": "3_audio/infinitetalk_0206.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p256/p256_146.wav", "generated_video": "1_fake/infinitetalk_0206.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+4--y5nQdVE0+P1+C0+F30817-30991__p256_146.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 206, "real_video": "0_real/infinitetalk_0206.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+4--y5nQdVE0+P1+C0+F30817-30991.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+4--y5nQdVE0+P1+C0+F30817-30991__p256_146.mp4" }, { "name": "infinitetalk_0207.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0207.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+X8a_kw0qls4+P1+C2+F67830-67988.jpg", "driving_audio": "3_audio/infinitetalk_0207.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p231/p231_219.wav", "generated_video": "1_fake/infinitetalk_0207.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+X8a_kw0qls4+P1+C2+F67830-67988__p231_219.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 207, "real_video": "0_real/infinitetalk_0207.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+X8a_kw0qls4+P1+C2+F67830-67988.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+X8a_kw0qls4+P1+C2+F67830-67988__p231_219.mp4" }, { "name": "infinitetalk_0208.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0208.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+Q6VpqqXn4iE+P0+C1+F9000-9110.jpg", "driving_audio": "3_audio/infinitetalk_0208.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p261/p261_005.wav", "generated_video": "1_fake/infinitetalk_0208.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+Q6VpqqXn4iE+P0+C1+F9000-9110__p261_005.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 208, "real_video": "0_real/infinitetalk_0208.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Q6VpqqXn4iE+P0+C1+F9000-9110.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+Q6VpqqXn4iE+P0+C1+F9000-9110__p261_005.mp4" }, { "name": "infinitetalk_0209.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0209.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+2_tBBEA3cxA+P0+C2+F5769-5895.jpg", "driving_audio": "3_audio/infinitetalk_0209.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p240/p240_027.wav", "generated_video": "1_fake/infinitetalk_0209.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+2_tBBEA3cxA+P0+C2+F5769-5895__p240_027.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 209, "real_video": "0_real/infinitetalk_0209.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+2_tBBEA3cxA+P0+C2+F5769-5895.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+2_tBBEA3cxA+P0+C2+F5769-5895__p240_027.mp4" }, { "name": "infinitetalk_0210.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0210.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+rtl4dj9MTh8+P1+C0+F4489-4620.jpg", "driving_audio": "3_audio/infinitetalk_0210.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p376/p376_254.wav", "generated_video": "1_fake/infinitetalk_0210.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+rtl4dj9MTh8+P1+C0+F4489-4620__p376_254.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 210, "real_video": "0_real/infinitetalk_0210.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+rtl4dj9MTh8+P1+C0+F4489-4620.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+rtl4dj9MTh8+P1+C0+F4489-4620__p376_254.mp4" }, { "name": "infinitetalk_0211.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0211.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+k0-VL7EdI4o+P1+C2+F29363-29497.jpg", "driving_audio": "3_audio/infinitetalk_0211.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p239/p239_015.wav", "generated_video": "1_fake/infinitetalk_0211.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+k0-VL7EdI4o+P1+C2+F29363-29497__p239_015.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 211, "real_video": "0_real/infinitetalk_0211.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+k0-VL7EdI4o+P1+C2+F29363-29497.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+k0-VL7EdI4o+P1+C2+F29363-29497__p239_015.mp4" }, { "name": "infinitetalk_0212.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0212.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+Sxgyljamz9U+P0+C1+F37579-37700.jpg", "driving_audio": "3_audio/infinitetalk_0212.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p230/p230_063.wav", "generated_video": "1_fake/infinitetalk_0212.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+Sxgyljamz9U+P0+C1+F37579-37700__p230_063.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 212, "real_video": "0_real/infinitetalk_0212.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Sxgyljamz9U+P0+C1+F37579-37700.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+Sxgyljamz9U+P0+C1+F37579-37700__p230_063.mp4" }, { "name": "infinitetalk_0213.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0213.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+QVY19B-p4iA+P1+C0+F7269-7370.jpg", "driving_audio": "3_audio/infinitetalk_0213.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p252/p252_299.wav", "generated_video": "1_fake/infinitetalk_0213.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+QVY19B-p4iA+P1+C0+F7269-7370__p252_299.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 213, "real_video": "0_real/infinitetalk_0213.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+QVY19B-p4iA+P1+C0+F7269-7370.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+QVY19B-p4iA+P1+C0+F7269-7370__p252_299.mp4" }, { "name": "infinitetalk_0214.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0214.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+gkeU7JLIY5g+P0+C1+F5020-5265.jpg", "driving_audio": "3_audio/infinitetalk_0214.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p256/p256_084.wav", "generated_video": "1_fake/infinitetalk_0214.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+gkeU7JLIY5g+P0+C1+F5020-5265__p256_084.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 214, "real_video": "0_real/infinitetalk_0214.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+gkeU7JLIY5g+P0+C1+F5020-5265.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+gkeU7JLIY5g+P0+C1+F5020-5265__p256_084.mp4" }, { "name": "infinitetalk_0215.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0215.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+gpWPtu10e5E+P0+C2+F10609-10740.jpg", "driving_audio": "3_audio/infinitetalk_0215.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p256/p256_233.wav", "generated_video": "1_fake/infinitetalk_0215.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+gpWPtu10e5E+P0+C2+F10609-10740__p256_233.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 215, "real_video": "0_real/infinitetalk_0215.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+gpWPtu10e5E+P0+C2+F10609-10740.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+gpWPtu10e5E+P0+C2+F10609-10740__p256_233.mp4" }, { "name": "infinitetalk_0216.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0216.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+KDmVoKJXEKE+P5+C1+F83144-83665.jpg", "driving_audio": "3_audio/infinitetalk_0216.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p272/p272_158.wav", "generated_video": "1_fake/infinitetalk_0216.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+KDmVoKJXEKE+P5+C1+F83144-83665__p272_158.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 216, "real_video": "0_real/infinitetalk_0216.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+KDmVoKJXEKE+P5+C1+F83144-83665.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+KDmVoKJXEKE+P5+C1+F83144-83665__p272_158.mp4" }, { "name": "infinitetalk_0217.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0217.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+OSO0bzns5dQ+P1+C1+F25665-25860.jpg", "driving_audio": "3_audio/infinitetalk_0217.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p277/p277_426.wav", "generated_video": "1_fake/infinitetalk_0217.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+OSO0bzns5dQ+P1+C1+F25665-25860__p277_426.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 217, "real_video": "0_real/infinitetalk_0217.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+OSO0bzns5dQ+P1+C1+F25665-25860.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+OSO0bzns5dQ+P1+C1+F25665-25860__p277_426.mp4" }, { "name": "infinitetalk_0218.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0218.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+rsCw8tKhGe4+P0+C1+F30994-31345.jpg", "driving_audio": "3_audio/infinitetalk_0218.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p285/p285_217.wav", "generated_video": "1_fake/infinitetalk_0218.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+rsCw8tKhGe4+P0+C1+F30994-31345__p285_217.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 218, "real_video": "0_real/infinitetalk_0218.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+rsCw8tKhGe4+P0+C1+F30994-31345.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+rsCw8tKhGe4+P0+C1+F30994-31345__p285_217.mp4" }, { "name": "infinitetalk_0219.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0219.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+n_kK7GJPG7g+P0+C2+F5194-5425.jpg", "driving_audio": "3_audio/infinitetalk_0219.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p258/p258_181.wav", "generated_video": "1_fake/infinitetalk_0219.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+n_kK7GJPG7g+P0+C2+F5194-5425__p258_181.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 219, "real_video": "0_real/infinitetalk_0219.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+n_kK7GJPG7g+P0+C2+F5194-5425.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+n_kK7GJPG7g+P0+C2+F5194-5425__p258_181.mp4" }, { "name": "infinitetalk_0220.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0220.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+GvWfIudFqFY+P0+C0+F17390-17565.jpg", "driving_audio": "3_audio/infinitetalk_0220.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p329/p329_016.wav", "generated_video": "1_fake/infinitetalk_0220.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+GvWfIudFqFY+P0+C0+F17390-17565__p329_016.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 220, "real_video": "0_real/infinitetalk_0220.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+GvWfIudFqFY+P0+C0+F17390-17565.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+GvWfIudFqFY+P0+C0+F17390-17565__p329_016.mp4" }, { "name": "infinitetalk_0221.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0221.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+IYgpXMbdOco+P0+C0+F9844-10385.jpg", "driving_audio": "3_audio/infinitetalk_0221.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p243/p243_004.wav", "generated_video": "1_fake/infinitetalk_0221.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+IYgpXMbdOco+P0+C0+F9844-10385__p243_004.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 221, "real_video": "0_real/infinitetalk_0221.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+IYgpXMbdOco+P0+C0+F9844-10385.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+IYgpXMbdOco+P0+C0+F9844-10385__p243_004.mp4" }, { "name": "infinitetalk_0222.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0222.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+e8d5_syRoIo+P0+C2+F2785-2953.jpg", "driving_audio": "3_audio/infinitetalk_0222.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p314/p314_243.wav", "generated_video": "1_fake/infinitetalk_0222.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+e8d5_syRoIo+P0+C2+F2785-2953__p314_243.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 222, "real_video": "0_real/infinitetalk_0222.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+e8d5_syRoIo+P0+C2+F2785-2953.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+e8d5_syRoIo+P0+C2+F2785-2953__p314_243.mp4" }, { "name": "infinitetalk_0223.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0223.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+-K1N7OXS14U+P0+C1+F4559-4825.jpg", "driving_audio": "3_audio/infinitetalk_0223.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p330/p330_301.wav", "generated_video": "1_fake/infinitetalk_0223.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+-K1N7OXS14U+P0+C1+F4559-4825__p330_301.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 223, "real_video": "0_real/infinitetalk_0223.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+-K1N7OXS14U+P0+C1+F4559-4825.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+-K1N7OXS14U+P0+C1+F4559-4825__p330_301.mp4" }, { "name": "infinitetalk_0224.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0224.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+A2-SXpIKsZY+P0+C0+F4809-4977.jpg", "driving_audio": "3_audio/infinitetalk_0224.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p347/p347_283.wav", "generated_video": "1_fake/infinitetalk_0224.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+A2-SXpIKsZY+P0+C0+F4809-4977__p347_283.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 224, "real_video": "0_real/infinitetalk_0224.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+A2-SXpIKsZY+P0+C0+F4809-4977.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+A2-SXpIKsZY+P0+C0+F4809-4977__p347_283.mp4" }, { "name": "infinitetalk_0225.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0225.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+EVMJcgEc9nQ+P0+C0+F2493-2632.jpg", "driving_audio": "3_audio/infinitetalk_0225.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p247/p247_115.wav", "generated_video": "1_fake/infinitetalk_0225.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+EVMJcgEc9nQ+P0+C0+F2493-2632__p247_115.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 225, "real_video": "0_real/infinitetalk_0225.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+EVMJcgEc9nQ+P0+C0+F2493-2632.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+EVMJcgEc9nQ+P0+C0+F2493-2632__p247_115.mp4" }, { "name": "infinitetalk_0226.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0226.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+hVbN81bhusc+P0+C2+F41457-41563.jpg", "driving_audio": "3_audio/infinitetalk_0226.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p286/p286_013.wav", "generated_video": "1_fake/infinitetalk_0226.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+hVbN81bhusc+P0+C2+F41457-41563__p286_013.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 226, "real_video": "0_real/infinitetalk_0226.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+hVbN81bhusc+P0+C2+F41457-41563.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+hVbN81bhusc+P0+C2+F41457-41563__p286_013.mp4" }, { "name": "infinitetalk_0227.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0227.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+L6iZXqt4e7Y+P0+C0+F6885-7430.jpg", "driving_audio": "3_audio/infinitetalk_0227.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p274/p274_392.wav", "generated_video": "1_fake/infinitetalk_0227.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+L6iZXqt4e7Y+P0+C0+F6885-7430__p274_392.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 227, "real_video": "0_real/infinitetalk_0227.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+L6iZXqt4e7Y+P0+C0+F6885-7430.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+L6iZXqt4e7Y+P0+C0+F6885-7430__p274_392.mp4" }, { "name": "infinitetalk_0228.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0228.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+VB-duzW00Ow+P0+C0+F52776-53101.jpg", "driving_audio": "3_audio/infinitetalk_0228.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p299/p299_296.wav", "generated_video": "1_fake/infinitetalk_0228.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+VB-duzW00Ow+P0+C0+F52776-53101__p299_296.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 228, "real_video": "0_real/infinitetalk_0228.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+VB-duzW00Ow+P0+C0+F52776-53101.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+VB-duzW00Ow+P0+C0+F52776-53101__p299_296.mp4" }, { "name": "infinitetalk_0229.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0229.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+wOizifrfpXM+P0+C1+F7404-7540.jpg", "driving_audio": "3_audio/infinitetalk_0229.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p297/p297_221.wav", "generated_video": "1_fake/infinitetalk_0229.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+wOizifrfpXM+P0+C1+F7404-7540__p297_221.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 229, "real_video": "0_real/infinitetalk_0229.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+wOizifrfpXM+P0+C1+F7404-7540.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+wOizifrfpXM+P0+C1+F7404-7540__p297_221.mp4" }, { "name": "infinitetalk_0230.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0230.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+bGZ-fjfldKE+P0+C2+F25744-25850.jpg", "driving_audio": "3_audio/infinitetalk_0230.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p238/p238_144.wav", "generated_video": "1_fake/infinitetalk_0230.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+bGZ-fjfldKE+P0+C2+F25744-25850__p238_144.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 230, "real_video": "0_real/infinitetalk_0230.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+bGZ-fjfldKE+P0+C2+F25744-25850.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+bGZ-fjfldKE+P0+C2+F25744-25850__p238_144.mp4" }, { "name": "infinitetalk_0231.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0231.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+0Ji2p_XBNRk+P0+C2+F23595-23756.jpg", "driving_audio": "3_audio/infinitetalk_0231.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p267/p267_385.wav", "generated_video": "1_fake/infinitetalk_0231.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+0Ji2p_XBNRk+P0+C2+F23595-23756__p267_385.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 231, "real_video": "0_real/infinitetalk_0231.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+0Ji2p_XBNRk+P0+C2+F23595-23756.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+0Ji2p_XBNRk+P0+C2+F23595-23756__p267_385.mp4" }, { "name": "infinitetalk_0232.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0232.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+EwKxc9E7MsY+P0+C1+F5198-5876.jpg", "driving_audio": "3_audio/infinitetalk_0232.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p295/p295_284.wav", "generated_video": "1_fake/infinitetalk_0232.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+EwKxc9E7MsY+P0+C1+F5198-5876__p295_284.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 232, "real_video": "0_real/infinitetalk_0232.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+EwKxc9E7MsY+P0+C1+F5198-5876.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+EwKxc9E7MsY+P0+C1+F5198-5876__p295_284.mp4" }, { "name": "infinitetalk_0233.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0233.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+git2XTfLREk+P1+C0+F928-1160.jpg", "driving_audio": "3_audio/infinitetalk_0233.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p314/p314_194.wav", "generated_video": "1_fake/infinitetalk_0233.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+git2XTfLREk+P1+C0+F928-1160__p314_194.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 233, "real_video": "0_real/infinitetalk_0233.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+git2XTfLREk+P1+C0+F928-1160.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+git2XTfLREk+P1+C0+F928-1160__p314_194.mp4" }, { "name": "infinitetalk_0234.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0234.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+rKvfTRGVue4+P0+C1+F10114-10290.jpg", "driving_audio": "3_audio/infinitetalk_0234.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p255/p255_099.wav", "generated_video": "1_fake/infinitetalk_0234.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+rKvfTRGVue4+P0+C1+F10114-10290__p255_099.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 234, "real_video": "0_real/infinitetalk_0234.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+rKvfTRGVue4+P0+C1+F10114-10290.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+rKvfTRGVue4+P0+C1+F10114-10290__p255_099.mp4" }, { "name": "infinitetalk_0235.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0235.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+5YYsRIJ1Yh8+P1+C0+F10154-10970.jpg", "driving_audio": "3_audio/infinitetalk_0235.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p257/p257_229.wav", "generated_video": "1_fake/infinitetalk_0235.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+5YYsRIJ1Yh8+P1+C0+F10154-10970__p257_229.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 235, "real_video": "0_real/infinitetalk_0235.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+5YYsRIJ1Yh8+P1+C0+F10154-10970.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+5YYsRIJ1Yh8+P1+C0+F10154-10970__p257_229.mp4" }, { "name": "infinitetalk_0236.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0236.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+15JLvW2URIk+P0+C2+F9224-9333.jpg", "driving_audio": "3_audio/infinitetalk_0236.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p270/p270_032.wav", "generated_video": "1_fake/infinitetalk_0236.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+15JLvW2URIk+P0+C2+F9224-9333__p270_032.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 236, "real_video": "0_real/infinitetalk_0236.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+15JLvW2URIk+P0+C2+F9224-9333.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+15JLvW2URIk+P0+C2+F9224-9333__p270_032.mp4" }, { "name": "infinitetalk_0237.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0237.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+Iekxojt3vyw+P0+C2+F43486-43760.jpg", "driving_audio": "3_audio/infinitetalk_0237.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p267/p267_317.wav", "generated_video": "1_fake/infinitetalk_0237.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+Iekxojt3vyw+P0+C2+F43486-43760__p267_317.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 237, "real_video": "0_real/infinitetalk_0237.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Iekxojt3vyw+P0+C2+F43486-43760.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+Iekxojt3vyw+P0+C2+F43486-43760__p267_317.mp4" }, { "name": "infinitetalk_0238.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0238.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+Gqe4qjrjgUw+P0+C2+F8930-9202.jpg", "driving_audio": "3_audio/infinitetalk_0238.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p293/p293_002.wav", "generated_video": "1_fake/infinitetalk_0238.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+Gqe4qjrjgUw+P0+C2+F8930-9202__p293_002.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 238, "real_video": "0_real/infinitetalk_0238.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Gqe4qjrjgUw+P0+C2+F8930-9202.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+Gqe4qjrjgUw+P0+C2+F8930-9202__p293_002.mp4" }, { "name": "infinitetalk_0239.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0239.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+mTvdmosy3q0+P0+C0+F12870-13056.jpg", "driving_audio": "3_audio/infinitetalk_0239.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p243/p243_073.wav", "generated_video": "1_fake/infinitetalk_0239.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+mTvdmosy3q0+P0+C0+F12870-13056__p243_073.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 239, "real_video": "0_real/infinitetalk_0239.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+mTvdmosy3q0+P0+C0+F12870-13056.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+mTvdmosy3q0+P0+C0+F12870-13056__p243_073.mp4" }, { "name": "infinitetalk_0240.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0240.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+4pG1f3-n7pA+P0+C2+F4435-4932.jpg", "driving_audio": "3_audio/infinitetalk_0240.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p256/p256_094.wav", "generated_video": "1_fake/infinitetalk_0240.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+4pG1f3-n7pA+P0+C2+F4435-4932__p256_094.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 240, "real_video": "0_real/infinitetalk_0240.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+4pG1f3-n7pA+P0+C2+F4435-4932.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+4pG1f3-n7pA+P0+C2+F4435-4932__p256_094.mp4" }, { "name": "infinitetalk_0241.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0241.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+l8GamLk0HT0+P1+C1+F6316-6642.jpg", "driving_audio": "3_audio/infinitetalk_0241.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p241/p241_141.wav", "generated_video": "1_fake/infinitetalk_0241.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+l8GamLk0HT0+P1+C1+F6316-6642__p241_141.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 241, "real_video": "0_real/infinitetalk_0241.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+l8GamLk0HT0+P1+C1+F6316-6642.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+l8GamLk0HT0+P1+C1+F6316-6642__p241_141.mp4" }, { "name": "infinitetalk_0242.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0242.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+nNGf0_c3U_s+P1+C2+F10574-10764.jpg", "driving_audio": "3_audio/infinitetalk_0242.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p225/p225_044.wav", "generated_video": "1_fake/infinitetalk_0242.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+nNGf0_c3U_s+P1+C2+F10574-10764__p225_044.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 242, "real_video": "0_real/infinitetalk_0242.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+nNGf0_c3U_s+P1+C2+F10574-10764.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+nNGf0_c3U_s+P1+C2+F10574-10764__p225_044.mp4" }, { "name": "infinitetalk_0243.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0243.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+sxFRbikGN-M+P0+C0+F13193-13370.jpg", "driving_audio": "3_audio/infinitetalk_0243.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p281/p281_294.wav", "generated_video": "1_fake/infinitetalk_0243.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+sxFRbikGN-M+P0+C0+F13193-13370__p281_294.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 243, "real_video": "0_real/infinitetalk_0243.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+sxFRbikGN-M+P0+C0+F13193-13370.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+sxFRbikGN-M+P0+C0+F13193-13370__p281_294.mp4" }, { "name": "infinitetalk_0244.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0244.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+VxxxXykcNds+P1+C0+F13930-14107.jpg", "driving_audio": "3_audio/infinitetalk_0244.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p313/p313_070.wav", "generated_video": "1_fake/infinitetalk_0244.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+VxxxXykcNds+P1+C0+F13930-14107__p313_070.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 244, "real_video": "0_real/infinitetalk_0244.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+VxxxXykcNds+P1+C0+F13930-14107.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+VxxxXykcNds+P1+C0+F13930-14107__p313_070.mp4" }, { "name": "infinitetalk_0245.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0245.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+8JbMZhEjEnA+P0+C0+F10808-10982.jpg", "driving_audio": "3_audio/infinitetalk_0245.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p253/p253_379.wav", "generated_video": "1_fake/infinitetalk_0245.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+8JbMZhEjEnA+P0+C0+F10808-10982__p253_379.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 245, "real_video": "0_real/infinitetalk_0245.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+8JbMZhEjEnA+P0+C0+F10808-10982.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+8JbMZhEjEnA+P0+C0+F10808-10982__p253_379.mp4" }, { "name": "infinitetalk_0246.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0246.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+OEmkyPFllEw+P0+C0+F9325-9537.jpg", "driving_audio": "3_audio/infinitetalk_0246.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p247/p247_397.wav", "generated_video": "1_fake/infinitetalk_0246.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+OEmkyPFllEw+P0+C0+F9325-9537__p247_397.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 246, "real_video": "0_real/infinitetalk_0246.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+OEmkyPFllEw+P0+C0+F9325-9537.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+OEmkyPFllEw+P0+C0+F9325-9537__p247_397.mp4" }, { "name": "infinitetalk_0247.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0247.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+3OsQhtLHuZA+P0+C1+F59390-59503.jpg", "driving_audio": "3_audio/infinitetalk_0247.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p298/p298_177.wav", "generated_video": "1_fake/infinitetalk_0247.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+3OsQhtLHuZA+P0+C1+F59390-59503__p298_177.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 247, "real_video": "0_real/infinitetalk_0247.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+3OsQhtLHuZA+P0+C1+F59390-59503.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+3OsQhtLHuZA+P0+C1+F59390-59503__p298_177.mp4" }, { "name": "infinitetalk_0248.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0248.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+GhgMpvBxOoo+P0+C0+F3954-4935.jpg", "driving_audio": "3_audio/infinitetalk_0248.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p316/p316_178.wav", "generated_video": "1_fake/infinitetalk_0248.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+GhgMpvBxOoo+P0+C0+F3954-4935__p316_178.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 248, "real_video": "0_real/infinitetalk_0248.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+GhgMpvBxOoo+P0+C0+F3954-4935.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+GhgMpvBxOoo+P0+C0+F3954-4935__p316_178.mp4" }, { "name": "infinitetalk_0249.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0249.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+q2d6PlHg_Is+P0+C0+F12654-12920.jpg", "driving_audio": "3_audio/infinitetalk_0249.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p283/p283_022.wav", "generated_video": "1_fake/infinitetalk_0249.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+q2d6PlHg_Is+P0+C0+F12654-12920__p283_022.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 249, "real_video": "0_real/infinitetalk_0249.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+q2d6PlHg_Is+P0+C0+F12654-12920.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+q2d6PlHg_Is+P0+C0+F12654-12920__p283_022.mp4" }, { "name": "infinitetalk_0250.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0250.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+UfSn5Yri410+P0+C0+F4864-5100.jpg", "driving_audio": "3_audio/infinitetalk_0250.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p361/p361_397.wav", "generated_video": "1_fake/infinitetalk_0250.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+UfSn5Yri410+P0+C0+F4864-5100__p361_397.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 250, "real_video": "0_real/infinitetalk_0250.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+UfSn5Yri410+P0+C0+F4864-5100.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+UfSn5Yri410+P0+C0+F4864-5100__p361_397.mp4" }, { "name": "infinitetalk_0251.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0251.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+tB-75j9Denw+P0+C0+F13895-14035.jpg", "driving_audio": "3_audio/infinitetalk_0251.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p256/p256_194.wav", "generated_video": "1_fake/infinitetalk_0251.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+tB-75j9Denw+P0+C0+F13895-14035__p256_194.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 251, "real_video": "0_real/infinitetalk_0251.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+tB-75j9Denw+P0+C0+F13895-14035.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+tB-75j9Denw+P0+C0+F13895-14035__p256_194.mp4" }, { "name": "infinitetalk_0252.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0252.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+rinoibFVoXg+P0+C1+F2145-3023.jpg", "driving_audio": "3_audio/infinitetalk_0252.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p336/p336_003.wav", "generated_video": "1_fake/infinitetalk_0252.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+rinoibFVoXg+P0+C1+F2145-3023__p336_003.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 252, "real_video": "0_real/infinitetalk_0252.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+rinoibFVoXg+P0+C1+F2145-3023.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+rinoibFVoXg+P0+C1+F2145-3023__p336_003.mp4" }, { "name": "infinitetalk_0253.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0253.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+xRkgNkWxvoY+P1+C2+F7484-7745.jpg", "driving_audio": "3_audio/infinitetalk_0253.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p283/p283_426.wav", "generated_video": "1_fake/infinitetalk_0253.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+xRkgNkWxvoY+P1+C2+F7484-7745__p283_426.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 253, "real_video": "0_real/infinitetalk_0253.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+xRkgNkWxvoY+P1+C2+F7484-7745.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+xRkgNkWxvoY+P1+C2+F7484-7745__p283_426.mp4" }, { "name": "infinitetalk_0254.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0254.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+_B9r3YjLaqU+P0+C1+F26209-26350.jpg", "driving_audio": "3_audio/infinitetalk_0254.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p329/p329_022.wav", "generated_video": "1_fake/infinitetalk_0254.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+_B9r3YjLaqU+P0+C1+F26209-26350__p329_022.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 254, "real_video": "0_real/infinitetalk_0254.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+_B9r3YjLaqU+P0+C1+F26209-26350.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+_B9r3YjLaqU+P0+C1+F26209-26350__p329_022.mp4" }, { "name": "infinitetalk_0255.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0255.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+rDKivNmPcLg+P0+C1+F3971-4085.jpg", "driving_audio": "3_audio/infinitetalk_0255.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p323/p323_296.wav", "generated_video": "1_fake/infinitetalk_0255.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+rDKivNmPcLg+P0+C1+F3971-4085__p323_296.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 255, "real_video": "0_real/infinitetalk_0255.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+rDKivNmPcLg+P0+C1+F3971-4085.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+rDKivNmPcLg+P0+C1+F3971-4085__p323_296.mp4" }, { "name": "infinitetalk_0256.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0256.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+6MYdH91DZgg+P0+C2+F19059-19258.jpg", "driving_audio": "3_audio/infinitetalk_0256.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p271/p271_196.wav", "generated_video": "1_fake/infinitetalk_0256.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+6MYdH91DZgg+P0+C2+F19059-19258__p271_196.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 256, "real_video": "0_real/infinitetalk_0256.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+6MYdH91DZgg+P0+C2+F19059-19258.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+6MYdH91DZgg+P0+C2+F19059-19258__p271_196.mp4" }, { "name": "infinitetalk_0257.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0257.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+KO5Zh11A3vI+P0+C1+F31609-31857.jpg", "driving_audio": "3_audio/infinitetalk_0257.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p303/p303_295.wav", "generated_video": "1_fake/infinitetalk_0257.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+KO5Zh11A3vI+P0+C1+F31609-31857__p303_295.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 257, "real_video": "0_real/infinitetalk_0257.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+KO5Zh11A3vI+P0+C1+F31609-31857.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+KO5Zh11A3vI+P0+C1+F31609-31857__p303_295.mp4" }, { "name": "infinitetalk_0258.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0258.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+E5mTDw2NpSE+P2+C0+F42384-42895.jpg", "driving_audio": "3_audio/infinitetalk_0258.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p340/p340_149.wav", "generated_video": "1_fake/infinitetalk_0258.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+E5mTDw2NpSE+P2+C0+F42384-42895__p340_149.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 258, "real_video": "0_real/infinitetalk_0258.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+E5mTDw2NpSE+P2+C0+F42384-42895.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+E5mTDw2NpSE+P2+C0+F42384-42895__p340_149.mp4" }, { "name": "infinitetalk_0259.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0259.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+BBq4Dfrwn6s+P0+C0+F6234-6353.jpg", "driving_audio": "3_audio/infinitetalk_0259.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p276/p276_114.wav", "generated_video": "1_fake/infinitetalk_0259.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+BBq4Dfrwn6s+P0+C0+F6234-6353__p276_114.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 259, "real_video": "0_real/infinitetalk_0259.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+BBq4Dfrwn6s+P0+C0+F6234-6353.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+BBq4Dfrwn6s+P0+C0+F6234-6353__p276_114.mp4" }, { "name": "infinitetalk_0260.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0260.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+s6LjK4GFdJ8+P1+C1+F40321-40530.jpg", "driving_audio": "3_audio/infinitetalk_0260.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p304/p304_381.wav", "generated_video": "1_fake/infinitetalk_0260.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+s6LjK4GFdJ8+P1+C1+F40321-40530__p304_381.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 260, "real_video": "0_real/infinitetalk_0260.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+s6LjK4GFdJ8+P1+C1+F40321-40530.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+s6LjK4GFdJ8+P1+C1+F40321-40530__p304_381.mp4" }, { "name": "infinitetalk_0261.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0261.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+kGg_6BGIHuM+P0+C0+F29242-29375.jpg", "driving_audio": "3_audio/infinitetalk_0261.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p294/p294_297.wav", "generated_video": "1_fake/infinitetalk_0261.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+kGg_6BGIHuM+P0+C0+F29242-29375__p294_297.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 261, "real_video": "0_real/infinitetalk_0261.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+kGg_6BGIHuM+P0+C0+F29242-29375.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+kGg_6BGIHuM+P0+C0+F29242-29375__p294_297.mp4" }, { "name": "infinitetalk_0262.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0262.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+WsPpuj9TCQ0+P0+C1+F37000-37407.jpg", "driving_audio": "3_audio/infinitetalk_0262.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p264/p264_491.wav", "generated_video": "1_fake/infinitetalk_0262.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+WsPpuj9TCQ0+P0+C1+F37000-37407__p264_491.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 262, "real_video": "0_real/infinitetalk_0262.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+WsPpuj9TCQ0+P0+C1+F37000-37407.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+WsPpuj9TCQ0+P0+C1+F37000-37407__p264_491.mp4" }, { "name": "infinitetalk_0263.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0263.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+puGB4PQlZ5U+P0+C0+F5664-5895.jpg", "driving_audio": "3_audio/infinitetalk_0263.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p274/p274_418.wav", "generated_video": "1_fake/infinitetalk_0263.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+puGB4PQlZ5U+P0+C0+F5664-5895__p274_418.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 263, "real_video": "0_real/infinitetalk_0263.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+puGB4PQlZ5U+P0+C0+F5664-5895.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+puGB4PQlZ5U+P0+C0+F5664-5895__p274_418.mp4" }, { "name": "infinitetalk_0264.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0264.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+3D5FOKEiY30+P0+C2+F2941-3060.jpg", "driving_audio": "3_audio/infinitetalk_0264.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p314/p314_083.wav", "generated_video": "1_fake/infinitetalk_0264.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+3D5FOKEiY30+P0+C2+F2941-3060__p314_083.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 264, "real_video": "0_real/infinitetalk_0264.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+3D5FOKEiY30+P0+C2+F2941-3060.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+3D5FOKEiY30+P0+C2+F2941-3060__p314_083.mp4" }, { "name": "infinitetalk_0265.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0265.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+O6VcW3aWwKc+P1+C1+F3599-4080.jpg", "driving_audio": "3_audio/infinitetalk_0265.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p345/p345_081.wav", "generated_video": "1_fake/infinitetalk_0265.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+O6VcW3aWwKc+P1+C1+F3599-4080__p345_081.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 265, "real_video": "0_real/infinitetalk_0265.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+O6VcW3aWwKc+P1+C1+F3599-4080.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+O6VcW3aWwKc+P1+C1+F3599-4080__p345_081.mp4" }, { "name": "infinitetalk_0266.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0266.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+yGN97WrbbvQ+P0+C2+F1812-2085.jpg", "driving_audio": "3_audio/infinitetalk_0266.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p260/p260_234.wav", "generated_video": "1_fake/infinitetalk_0266.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+yGN97WrbbvQ+P0+C2+F1812-2085__p260_234.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 266, "real_video": "0_real/infinitetalk_0266.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+yGN97WrbbvQ+P0+C2+F1812-2085.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+yGN97WrbbvQ+P0+C2+F1812-2085__p260_234.mp4" }, { "name": "infinitetalk_0267.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0267.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+ISoh38vR6_Y+P0+C0+F1419-1578.jpg", "driving_audio": "3_audio/infinitetalk_0267.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p237/p237_026.wav", "generated_video": "1_fake/infinitetalk_0267.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+ISoh38vR6_Y+P0+C0+F1419-1578__p237_026.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 267, "real_video": "0_real/infinitetalk_0267.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+ISoh38vR6_Y+P0+C0+F1419-1578.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+ISoh38vR6_Y+P0+C0+F1419-1578__p237_026.mp4" }, { "name": "infinitetalk_0268.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0268.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+BVn-602OWDI+P0+C2+F37780-38755.jpg", "driving_audio": "3_audio/infinitetalk_0268.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p254/p254_231.wav", "generated_video": "1_fake/infinitetalk_0268.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+BVn-602OWDI+P0+C2+F37780-38755__p254_231.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 268, "real_video": "0_real/infinitetalk_0268.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+BVn-602OWDI+P0+C2+F37780-38755.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+BVn-602OWDI+P0+C2+F37780-38755__p254_231.mp4" }, { "name": "infinitetalk_0269.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0269.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+8IyJEFbBhXo+P1+C0+F48337-48465.jpg", "driving_audio": "3_audio/infinitetalk_0269.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p240/p240_168.wav", "generated_video": "1_fake/infinitetalk_0269.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+8IyJEFbBhXo+P1+C0+F48337-48465__p240_168.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 269, "real_video": "0_real/infinitetalk_0269.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+8IyJEFbBhXo+P1+C0+F48337-48465.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+8IyJEFbBhXo+P1+C0+F48337-48465__p240_168.mp4" }, { "name": "infinitetalk_0270.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0270.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+bB2hSyyUlKQ+P1+C1+F68156-68284.jpg", "driving_audio": "3_audio/infinitetalk_0270.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p273/p273_188.wav", "generated_video": "1_fake/infinitetalk_0270.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+bB2hSyyUlKQ+P1+C1+F68156-68284__p273_188.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 270, "real_video": "0_real/infinitetalk_0270.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+bB2hSyyUlKQ+P1+C1+F68156-68284.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+bB2hSyyUlKQ+P1+C1+F68156-68284__p273_188.mp4" }, { "name": "infinitetalk_0271.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0271.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+AIwBtKGnBqk+P1+C0+F5803-6022.jpg", "driving_audio": "3_audio/infinitetalk_0271.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p275/p275_033.wav", "generated_video": "1_fake/infinitetalk_0271.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+AIwBtKGnBqk+P1+C0+F5803-6022__p275_033.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 271, "real_video": "0_real/infinitetalk_0271.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+AIwBtKGnBqk+P1+C0+F5803-6022.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+AIwBtKGnBqk+P1+C0+F5803-6022__p275_033.mp4" }, { "name": "infinitetalk_0272.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0272.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+Uu_NQ5GBVHk+P0+C2+F5259-5380.jpg", "driving_audio": "3_audio/infinitetalk_0272.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p323/p323_169.wav", "generated_video": "1_fake/infinitetalk_0272.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+Uu_NQ5GBVHk+P0+C2+F5259-5380__p323_169.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 272, "real_video": "0_real/infinitetalk_0272.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Uu_NQ5GBVHk+P0+C2+F5259-5380.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+Uu_NQ5GBVHk+P0+C2+F5259-5380__p323_169.mp4" }, { "name": "infinitetalk_0273.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0273.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+0-ewW6GskhY+P0+C0+F4999-5165.jpg", "driving_audio": "3_audio/infinitetalk_0273.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p225/p225_279.wav", "generated_video": "1_fake/infinitetalk_0273.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+0-ewW6GskhY+P0+C0+F4999-5165__p225_279.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 273, "real_video": "0_real/infinitetalk_0273.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+0-ewW6GskhY+P0+C0+F4999-5165.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+0-ewW6GskhY+P0+C0+F4999-5165__p225_279.mp4" }, { "name": "infinitetalk_0274.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0274.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+1pEpzvyZ_Y4+P0+C0+F6805-6925.jpg", "driving_audio": "3_audio/infinitetalk_0274.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p238/p238_031.wav", "generated_video": "1_fake/infinitetalk_0274.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+1pEpzvyZ_Y4+P0+C0+F6805-6925__p238_031.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 274, "real_video": "0_real/infinitetalk_0274.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+1pEpzvyZ_Y4+P0+C0+F6805-6925.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+1pEpzvyZ_Y4+P0+C0+F6805-6925__p238_031.mp4" }, { "name": "infinitetalk_0275.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0275.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+nZf4GWVNTrU+P0+C1+F4187-4353.jpg", "driving_audio": "3_audio/infinitetalk_0275.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p286/p286_117.wav", "generated_video": "1_fake/infinitetalk_0275.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+nZf4GWVNTrU+P0+C1+F4187-4353__p286_117.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 275, "real_video": "0_real/infinitetalk_0275.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+nZf4GWVNTrU+P0+C1+F4187-4353.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+nZf4GWVNTrU+P0+C1+F4187-4353__p286_117.mp4" }, { "name": "infinitetalk_0276.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0276.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+oB2r1gqqAvA+P0+C0+F74239-74435.jpg", "driving_audio": "3_audio/infinitetalk_0276.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p261/p261_286.wav", "generated_video": "1_fake/infinitetalk_0276.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+oB2r1gqqAvA+P0+C0+F74239-74435__p261_286.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 276, "real_video": "0_real/infinitetalk_0276.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+oB2r1gqqAvA+P0+C0+F74239-74435.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+oB2r1gqqAvA+P0+C0+F74239-74435__p261_286.mp4" }, { "name": "infinitetalk_0277.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0277.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+Tv0lsmVwNs8+P0+C0+F9629-9785.jpg", "driving_audio": "3_audio/infinitetalk_0277.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p228/p228_219.wav", "generated_video": "1_fake/infinitetalk_0277.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+Tv0lsmVwNs8+P0+C0+F9629-9785__p228_219.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 277, "real_video": "0_real/infinitetalk_0277.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Tv0lsmVwNs8+P0+C0+F9629-9785.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+Tv0lsmVwNs8+P0+C0+F9629-9785__p228_219.mp4" }, { "name": "infinitetalk_0278.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0278.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+5q3OgZlSab0+P0+C0+F5835-5968.jpg", "driving_audio": "3_audio/infinitetalk_0278.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p249/p249_063.wav", "generated_video": "1_fake/infinitetalk_0278.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+5q3OgZlSab0+P0+C0+F5835-5968__p249_063.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 278, "real_video": "0_real/infinitetalk_0278.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+5q3OgZlSab0+P0+C0+F5835-5968.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+5q3OgZlSab0+P0+C0+F5835-5968__p249_063.mp4" }, { "name": "infinitetalk_0279.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0279.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+Uu_NQ5GBVHk+P0+C1+F13959-14135.jpg", "driving_audio": "3_audio/infinitetalk_0279.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p330/p330_336.wav", "generated_video": "1_fake/infinitetalk_0279.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+Uu_NQ5GBVHk+P0+C1+F13959-14135__p330_336.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 279, "real_video": "0_real/infinitetalk_0279.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Uu_NQ5GBVHk+P0+C1+F13959-14135.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+Uu_NQ5GBVHk+P0+C1+F13959-14135__p330_336.mp4" }, { "name": "infinitetalk_0280.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0280.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+U_D7yLiBezM+P1+C1+F12327-12925.jpg", "driving_audio": "3_audio/infinitetalk_0280.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p307/p307_047.wav", "generated_video": "1_fake/infinitetalk_0280.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+U_D7yLiBezM+P1+C1+F12327-12925__p307_047.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 280, "real_video": "0_real/infinitetalk_0280.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+U_D7yLiBezM+P1+C1+F12327-12925.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+U_D7yLiBezM+P1+C1+F12327-12925__p307_047.mp4" }, { "name": "infinitetalk_0281.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0281.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+BQIuTs2QZUU+P0+C0+F5004-5443.jpg", "driving_audio": "3_audio/infinitetalk_0281.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p336/p336_340.wav", "generated_video": "1_fake/infinitetalk_0281.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+BQIuTs2QZUU+P0+C0+F5004-5443__p336_340.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 281, "real_video": "0_real/infinitetalk_0281.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+BQIuTs2QZUU+P0+C0+F5004-5443.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+BQIuTs2QZUU+P0+C0+F5004-5443__p336_340.mp4" }, { "name": "infinitetalk_0282.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0282.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+usBo8-jWj5M+P0+C0+F3101-3257.jpg", "driving_audio": "3_audio/infinitetalk_0282.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p259/p259_296.wav", "generated_video": "1_fake/infinitetalk_0282.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+usBo8-jWj5M+P0+C0+F3101-3257__p259_296.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 282, "real_video": "0_real/infinitetalk_0282.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+usBo8-jWj5M+P0+C0+F3101-3257.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+usBo8-jWj5M+P0+C0+F3101-3257__p259_296.mp4" }, { "name": "infinitetalk_0283.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0283.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+6CriJKsGfwQ+P0+C0+F6099-6240.jpg", "driving_audio": "3_audio/infinitetalk_0283.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p269/p269_309.wav", "generated_video": "1_fake/infinitetalk_0283.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+6CriJKsGfwQ+P0+C0+F6099-6240__p269_309.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 283, "real_video": "0_real/infinitetalk_0283.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+6CriJKsGfwQ+P0+C0+F6099-6240.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+6CriJKsGfwQ+P0+C0+F6099-6240__p269_309.mp4" }, { "name": "infinitetalk_0284.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0284.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+4vJfXNkRtGc+P0+C2+F11018-11155.jpg", "driving_audio": "3_audio/infinitetalk_0284.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p361/p361_313.wav", "generated_video": "1_fake/infinitetalk_0284.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+4vJfXNkRtGc+P0+C2+F11018-11155__p361_313.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 284, "real_video": "0_real/infinitetalk_0284.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+4vJfXNkRtGc+P0+C2+F11018-11155.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+4vJfXNkRtGc+P0+C2+F11018-11155__p361_313.mp4" }, { "name": "infinitetalk_0285.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0285.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+yaj21Am7F_U+P0+C0+F8530-8960.jpg", "driving_audio": "3_audio/infinitetalk_0285.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p334/p334_219.wav", "generated_video": "1_fake/infinitetalk_0285.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+yaj21Am7F_U+P0+C0+F8530-8960__p334_219.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 285, "real_video": "0_real/infinitetalk_0285.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+yaj21Am7F_U+P0+C0+F8530-8960.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+yaj21Am7F_U+P0+C0+F8530-8960__p334_219.mp4" }, { "name": "infinitetalk_0286.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0286.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+zRwRmv6EkdY+P0+C0+F22658-22804.jpg", "driving_audio": "3_audio/infinitetalk_0286.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p270/p270_256.wav", "generated_video": "1_fake/infinitetalk_0286.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+zRwRmv6EkdY+P0+C0+F22658-22804__p270_256.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 286, "real_video": "0_real/infinitetalk_0286.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+zRwRmv6EkdY+P0+C0+F22658-22804.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+zRwRmv6EkdY+P0+C0+F22658-22804__p270_256.mp4" }, { "name": "infinitetalk_0287.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0287.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+kmBfcqamNUI+P0+C0+F23684-23855.jpg", "driving_audio": "3_audio/infinitetalk_0287.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p286/p286_025.wav", "generated_video": "1_fake/infinitetalk_0287.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+kmBfcqamNUI+P0+C0+F23684-23855__p286_025.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 287, "real_video": "0_real/infinitetalk_0287.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+kmBfcqamNUI+P0+C0+F23684-23855.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+kmBfcqamNUI+P0+C0+F23684-23855__p286_025.mp4" }, { "name": "infinitetalk_0288.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0288.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+Ig_dsNlZrog+P0+C0+F11409-11525.jpg", "driving_audio": "3_audio/infinitetalk_0288.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p298/p298_394.wav", "generated_video": "1_fake/infinitetalk_0288.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+Ig_dsNlZrog+P0+C0+F11409-11525__p298_394.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 288, "real_video": "0_real/infinitetalk_0288.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Ig_dsNlZrog+P0+C0+F11409-11525.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+Ig_dsNlZrog+P0+C0+F11409-11525__p298_394.mp4" }, { "name": "infinitetalk_0289.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0289.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+lWOtffgxdKc+P0+C0+F13479-14080.jpg", "driving_audio": "3_audio/infinitetalk_0289.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p311/p311_050.wav", "generated_video": "1_fake/infinitetalk_0289.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+lWOtffgxdKc+P0+C0+F13479-14080__p311_050.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 289, "real_video": "0_real/infinitetalk_0289.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+lWOtffgxdKc+P0+C0+F13479-14080.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+lWOtffgxdKc+P0+C0+F13479-14080__p311_050.mp4" }, { "name": "infinitetalk_0290.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0290.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+zWREWbDXNW4+P0+C2+F915-1052.jpg", "driving_audio": "3_audio/infinitetalk_0290.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p280/p280_098.wav", "generated_video": "1_fake/infinitetalk_0290.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+zWREWbDXNW4+P0+C2+F915-1052__p280_098.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 290, "real_video": "0_real/infinitetalk_0290.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+zWREWbDXNW4+P0+C2+F915-1052.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+zWREWbDXNW4+P0+C2+F915-1052__p280_098.mp4" }, { "name": "infinitetalk_0291.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0291.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+4WXe3I7j03Q+P0+C1+F19139-19255.jpg", "driving_audio": "3_audio/infinitetalk_0291.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p233/p233_073.wav", "generated_video": "1_fake/infinitetalk_0291.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+4WXe3I7j03Q+P0+C1+F19139-19255__p233_073.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 291, "real_video": "0_real/infinitetalk_0291.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+4WXe3I7j03Q+P0+C1+F19139-19255.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+4WXe3I7j03Q+P0+C1+F19139-19255__p233_073.mp4" }, { "name": "infinitetalk_0292.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0292.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+tRUQRW16p90+P0+C1+F71392-71552.jpg", "driving_audio": "3_audio/infinitetalk_0292.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p362/p362_087.wav", "generated_video": "1_fake/infinitetalk_0292.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+tRUQRW16p90+P0+C1+F71392-71552__p362_087.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 292, "real_video": "0_real/infinitetalk_0292.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+tRUQRW16p90+P0+C1+F71392-71552.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+tRUQRW16p90+P0+C1+F71392-71552__p362_087.mp4" }, { "name": "infinitetalk_0293.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0293.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+_hioGUzeFbM+P1+C1+F1544-1715.jpg", "driving_audio": "3_audio/infinitetalk_0293.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p376/p376_089.wav", "generated_video": "1_fake/infinitetalk_0293.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+_hioGUzeFbM+P1+C1+F1544-1715__p376_089.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 293, "real_video": "0_real/infinitetalk_0293.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+_hioGUzeFbM+P1+C1+F1544-1715.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+_hioGUzeFbM+P1+C1+F1544-1715__p376_089.mp4" }, { "name": "infinitetalk_0294.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0294.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+NDVM4JGSSD0+P4+C1+F5707-5993.jpg", "driving_audio": "3_audio/infinitetalk_0294.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p269/p269_337.wav", "generated_video": "1_fake/infinitetalk_0294.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+NDVM4JGSSD0+P4+C1+F5707-5993__p269_337.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 294, "real_video": "0_real/infinitetalk_0294.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+NDVM4JGSSD0+P4+C1+F5707-5993.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+NDVM4JGSSD0+P4+C1+F5707-5993__p269_337.mp4" }, { "name": "infinitetalk_0295.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0295.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+yVmVOdc9glA+P2+C2+F56876-56997.jpg", "driving_audio": "3_audio/infinitetalk_0295.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p269/p269_111.wav", "generated_video": "1_fake/infinitetalk_0295.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+yVmVOdc9glA+P2+C2+F56876-56997__p269_111.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 295, "real_video": "0_real/infinitetalk_0295.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+yVmVOdc9glA+P2+C2+F56876-56997.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+yVmVOdc9glA+P2+C2+F56876-56997__p269_111.mp4" }, { "name": "infinitetalk_0296.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0296.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+9TWuTKJNZIk+P0+C2+F5866-5994.jpg", "driving_audio": "3_audio/infinitetalk_0296.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p253/p253_319.wav", "generated_video": "1_fake/infinitetalk_0296.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+9TWuTKJNZIk+P0+C2+F5866-5994__p253_319.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 296, "real_video": "0_real/infinitetalk_0296.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+9TWuTKJNZIk+P0+C2+F5866-5994.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+9TWuTKJNZIk+P0+C2+F5866-5994__p253_319.mp4" }, { "name": "infinitetalk_0297.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0297.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+DYxhlanOfl4+P0+C2+F3191-3298.jpg", "driving_audio": "3_audio/infinitetalk_0297.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p259/p259_050.wav", "generated_video": "1_fake/infinitetalk_0297.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+DYxhlanOfl4+P0+C2+F3191-3298__p259_050.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 297, "real_video": "0_real/infinitetalk_0297.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+DYxhlanOfl4+P0+C2+F3191-3298.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+DYxhlanOfl4+P0+C2+F3191-3298__p259_050.mp4" }, { "name": "infinitetalk_0298.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0298.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+gyRkCUgtJRk+P0+C1+F6534-6640.jpg", "driving_audio": "3_audio/infinitetalk_0298.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p297/p297_304.wav", "generated_video": "1_fake/infinitetalk_0298.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+gyRkCUgtJRk+P0+C1+F6534-6640__p297_304.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 298, "real_video": "0_real/infinitetalk_0298.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+gyRkCUgtJRk+P0+C1+F6534-6640.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+gyRkCUgtJRk+P0+C1+F6534-6640__p297_304.mp4" }, { "name": "infinitetalk_0299.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0299.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+ho1-fsVil9Q+P0+C0+F2511-2645.jpg", "driving_audio": "3_audio/infinitetalk_0299.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p264/p264_285.wav", "generated_video": "1_fake/infinitetalk_0299.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+ho1-fsVil9Q+P0+C0+F2511-2645__p264_285.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 299, "real_video": "0_real/infinitetalk_0299.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+ho1-fsVil9Q+P0+C0+F2511-2645.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+ho1-fsVil9Q+P0+C0+F2511-2645__p264_285.mp4" }, { "name": "infinitetalk_0300.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0300.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+I-qZshJW8r4+P0+C0+F74501-74906.jpg", "driving_audio": "3_audio/infinitetalk_0300.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p364/p364_085.wav", "generated_video": "1_fake/infinitetalk_0300.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+I-qZshJW8r4+P0+C0+F74501-74906__p364_085.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 300, "real_video": "0_real/infinitetalk_0300.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+I-qZshJW8r4+P0+C0+F74501-74906.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+I-qZshJW8r4+P0+C0+F74501-74906__p364_085.mp4" }, { "name": "infinitetalk_0301.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0301.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+Luha_Hc0m7Q+P0+C2+F10830-11168.jpg", "driving_audio": "3_audio/infinitetalk_0301.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p280/p280_079.wav", "generated_video": "1_fake/infinitetalk_0301.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+Luha_Hc0m7Q+P0+C2+F10830-11168__p280_079.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 301, "real_video": "0_real/infinitetalk_0301.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Luha_Hc0m7Q+P0+C2+F10830-11168.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+Luha_Hc0m7Q+P0+C2+F10830-11168__p280_079.mp4" }, { "name": "infinitetalk_0302.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0302.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+LJuNIYJqt1U+P0+C0+F3601-4182.jpg", "driving_audio": "3_audio/infinitetalk_0302.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p228/p228_251.wav", "generated_video": "1_fake/infinitetalk_0302.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+LJuNIYJqt1U+P0+C0+F3601-4182__p228_251.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 302, "real_video": "0_real/infinitetalk_0302.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+LJuNIYJqt1U+P0+C0+F3601-4182.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+LJuNIYJqt1U+P0+C0+F3601-4182__p228_251.mp4" }, { "name": "infinitetalk_0303.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0303.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+kGg_6BGIHuM+P0+C1+F16923-17035.jpg", "driving_audio": "3_audio/infinitetalk_0303.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p227/p227_192.wav", "generated_video": "1_fake/infinitetalk_0303.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+kGg_6BGIHuM+P0+C1+F16923-17035__p227_192.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 303, "real_video": "0_real/infinitetalk_0303.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+kGg_6BGIHuM+P0+C1+F16923-17035.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+kGg_6BGIHuM+P0+C1+F16923-17035__p227_192.mp4" }, { "name": "infinitetalk_0304.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0304.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+dg0LzcKiONg+P0+C2+F5282-5491.jpg", "driving_audio": "3_audio/infinitetalk_0304.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p273/p273_240.wav", "generated_video": "1_fake/infinitetalk_0304.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+dg0LzcKiONg+P0+C2+F5282-5491__p273_240.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 304, "real_video": "0_real/infinitetalk_0304.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+dg0LzcKiONg+P0+C2+F5282-5491.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+dg0LzcKiONg+P0+C2+F5282-5491__p273_240.mp4" }, { "name": "infinitetalk_0305.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0305.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+tzb4XSmtI1s+P0+C1+F5895-6022.jpg", "driving_audio": "3_audio/infinitetalk_0305.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p286/p286_383.wav", "generated_video": "1_fake/infinitetalk_0305.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+tzb4XSmtI1s+P0+C1+F5895-6022__p286_383.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 305, "real_video": "0_real/infinitetalk_0305.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+tzb4XSmtI1s+P0+C1+F5895-6022.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+tzb4XSmtI1s+P0+C1+F5895-6022__p286_383.mp4" }, { "name": "infinitetalk_0306.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0306.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+Ftqq_gLj4f4+P0+C0+F4504-4650.jpg", "driving_audio": "3_audio/infinitetalk_0306.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p256/p256_168.wav", "generated_video": "1_fake/infinitetalk_0306.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+Ftqq_gLj4f4+P0+C0+F4504-4650__p256_168.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 306, "real_video": "0_real/infinitetalk_0306.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Ftqq_gLj4f4+P0+C0+F4504-4650.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+Ftqq_gLj4f4+P0+C0+F4504-4650__p256_168.mp4" }, { "name": "infinitetalk_0307.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0307.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+7jgQyLpsAhc+P0+C2+F2343-2635.jpg", "driving_audio": "3_audio/infinitetalk_0307.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p264/p264_326.wav", "generated_video": "1_fake/infinitetalk_0307.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+7jgQyLpsAhc+P0+C2+F2343-2635__p264_326.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 307, "real_video": "0_real/infinitetalk_0307.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+7jgQyLpsAhc+P0+C2+F2343-2635.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+7jgQyLpsAhc+P0+C2+F2343-2635__p264_326.mp4" }, { "name": "infinitetalk_0308.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0308.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+Eo43VxFTUU8+P0+C0+F7973-8339.jpg", "driving_audio": "3_audio/infinitetalk_0308.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p298/p298_403.wav", "generated_video": "1_fake/infinitetalk_0308.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+Eo43VxFTUU8+P0+C0+F7973-8339__p298_403.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 308, "real_video": "0_real/infinitetalk_0308.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Eo43VxFTUU8+P0+C0+F7973-8339.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+Eo43VxFTUU8+P0+C0+F7973-8339__p298_403.mp4" }, { "name": "infinitetalk_0309.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0309.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+sf8bUrzYmGA+P0+C2+F1040-1289.jpg", "driving_audio": "3_audio/infinitetalk_0309.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p323/p323_112.wav", "generated_video": "1_fake/infinitetalk_0309.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+sf8bUrzYmGA+P0+C2+F1040-1289__p323_112.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 309, "real_video": "0_real/infinitetalk_0309.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+sf8bUrzYmGA+P0+C2+F1040-1289.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+sf8bUrzYmGA+P0+C2+F1040-1289__p323_112.mp4" }, { "name": "infinitetalk_0310.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0310.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+HelswnwVZ-o+P0+C1+F15424-15580.jpg", "driving_audio": "3_audio/infinitetalk_0310.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p339/p339_143.wav", "generated_video": "1_fake/infinitetalk_0310.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+HelswnwVZ-o+P0+C1+F15424-15580__p339_143.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 310, "real_video": "0_real/infinitetalk_0310.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+HelswnwVZ-o+P0+C1+F15424-15580.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+HelswnwVZ-o+P0+C1+F15424-15580__p339_143.mp4" }, { "name": "infinitetalk_0311.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0311.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+cliMs-2H8eE+P0+C1+F8382-8535.jpg", "driving_audio": "3_audio/infinitetalk_0311.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p305/p305_418.wav", "generated_video": "1_fake/infinitetalk_0311.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+cliMs-2H8eE+P0+C1+F8382-8535__p305_418.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 311, "real_video": "0_real/infinitetalk_0311.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+cliMs-2H8eE+P0+C1+F8382-8535.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+cliMs-2H8eE+P0+C1+F8382-8535__p305_418.mp4" }, { "name": "infinitetalk_0312.mp4", "variant": "vfhq_vctk_batch3", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0312.jpg", "identity_image_from": "/archive/lzd/_gen_inputs/identity_frames/Clip+Ixrocx7v4u4+P1+C1+F5675-5850.jpg", "driving_audio": "3_audio/infinitetalk_0312.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p245/p245_267.wav", "generated_video": "1_fake/infinitetalk_0312.mp4", "generated_from": "/archive/lzd/InfiniteTalk/gen_batch3/Clip+Ixrocx7v4u4+P1+C1+F5675-5850__p245_267.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk batch3. VFHQ frame identity and VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 312, "real_video": "0_real/infinitetalk_0312.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Ixrocx7v4u4+P1+C1+F5675-5850.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "reconstructed_vfhq", "name_before": "Clip+Ixrocx7v4u4+P1+C1+F5675-5850__p245_267.mp4" }, { "name": "infinitetalk_0313.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0313.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+hdI3mhblTgQ+P2+C0+F61554-61677.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+hdI3mhblTgQ+P2+C0+F61554-61677.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0313.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p300/p300_124.wav", "generated_video": "1_fake/infinitetalk_0313.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+hdI3mhblTgQ+P2+C0+F61554-61677__p300_124.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 313, "real_video": "0_real/infinitetalk_0313.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+hdI3mhblTgQ+P2+C0+F61554-61677.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+hdI3mhblTgQ+P2+C0+F61554-61677__p300_124.mp4" }, { "name": "infinitetalk_0314.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0314.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+32abllGf77g+P0+C2+F5399-5737.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+32abllGf77g+P0+C2+F5399-5737.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0314.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p295/p295_394.wav", "generated_video": "1_fake/infinitetalk_0314.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+32abllGf77g+P0+C2+F5399-5737__p295_394.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 314, "real_video": "0_real/infinitetalk_0314.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+32abllGf77g+P0+C2+F5399-5737.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+32abllGf77g+P0+C2+F5399-5737__p295_394.mp4" }, { "name": "infinitetalk_0315.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0315.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+ZM1G7uCHdxE+P0+C2+F7956-8100.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+ZM1G7uCHdxE+P0+C2+F7956-8100.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0315.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p303/p303_134.wav", "generated_video": "1_fake/infinitetalk_0315.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+ZM1G7uCHdxE+P0+C2+F7956-8100__p303_134.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 315, "real_video": "0_real/infinitetalk_0315.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+ZM1G7uCHdxE+P0+C2+F7956-8100.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+ZM1G7uCHdxE+P0+C2+F7956-8100__p303_134.mp4" }, { "name": "infinitetalk_0316.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0316.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+sosfsCDOg9Q+P0+C1+F6944-7070.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+sosfsCDOg9Q+P0+C1+F6944-7070.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0316.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p280/p280_130.wav", "generated_video": "1_fake/infinitetalk_0316.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+sosfsCDOg9Q+P0+C1+F6944-7070__p280_130.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 316, "real_video": "0_real/infinitetalk_0316.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+sosfsCDOg9Q+P0+C1+F6944-7070.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+sosfsCDOg9Q+P0+C1+F6944-7070__p280_130.mp4" }, { "name": "infinitetalk_0317.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0317.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+KJcG2RZ8YIE+P0+C0+F401-531.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+KJcG2RZ8YIE+P0+C0+F401-531.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0317.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p274/p274_220.wav", "generated_video": "1_fake/infinitetalk_0317.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+KJcG2RZ8YIE+P0+C0+F401-531__p274_220.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 317, "real_video": "0_real/infinitetalk_0317.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+KJcG2RZ8YIE+P0+C0+F401-531.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+KJcG2RZ8YIE+P0+C0+F401-531__p274_220.mp4" }, { "name": "infinitetalk_0318.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0318.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+bqT-NE98gSI+P0+C1+F10710-11028.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+bqT-NE98gSI+P0+C1+F10710-11028.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0318.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p279/p279_175.wav", "generated_video": "1_fake/infinitetalk_0318.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+bqT-NE98gSI+P0+C1+F10710-11028__p279_175.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 318, "real_video": "0_real/infinitetalk_0318.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+bqT-NE98gSI+P0+C1+F10710-11028.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+bqT-NE98gSI+P0+C1+F10710-11028__p279_175.mp4" }, { "name": "infinitetalk_0319.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0319.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+GUKVh9OpxJc+P0+C2+F3167-3428.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+GUKVh9OpxJc+P0+C2+F3167-3428.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0319.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p228/p228_282.wav", "generated_video": "1_fake/infinitetalk_0319.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+GUKVh9OpxJc+P0+C2+F3167-3428__p228_282.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 319, "real_video": "0_real/infinitetalk_0319.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+GUKVh9OpxJc+P0+C2+F3167-3428.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+GUKVh9OpxJc+P0+C2+F3167-3428__p228_282.mp4" }, { "name": "infinitetalk_0320.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0320.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+DB8ROEyQal0+P0+C0+F3831-3945.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+DB8ROEyQal0+P0+C0+F3831-3945.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0320.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p313/p313_215.wav", "generated_video": "1_fake/infinitetalk_0320.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+DB8ROEyQal0+P0+C0+F3831-3945__p313_215.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 320, "real_video": "0_real/infinitetalk_0320.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+DB8ROEyQal0+P0+C0+F3831-3945.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+DB8ROEyQal0+P0+C0+F3831-3945__p313_215.mp4" }, { "name": "infinitetalk_0321.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0321.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+5exxOVSOBps+P0+C1+F79431-79719.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+5exxOVSOBps+P0+C1+F79431-79719.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0321.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p251/p251_076.wav", "generated_video": "1_fake/infinitetalk_0321.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+5exxOVSOBps+P0+C1+F79431-79719__p251_076.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 321, "real_video": "0_real/infinitetalk_0321.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+5exxOVSOBps+P0+C1+F79431-79719.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+5exxOVSOBps+P0+C1+F79431-79719__p251_076.mp4" }, { "name": "infinitetalk_0322.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0322.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+CBdWZC-XXR0+P0+C0+F2298-2536.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+CBdWZC-XXR0+P0+C0+F2298-2536.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0322.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p282/p282_086.wav", "generated_video": "1_fake/infinitetalk_0322.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+CBdWZC-XXR0+P0+C0+F2298-2536__p282_086.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 322, "real_video": "0_real/infinitetalk_0322.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+CBdWZC-XXR0+P0+C0+F2298-2536.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+CBdWZC-XXR0+P0+C0+F2298-2536__p282_086.mp4" }, { "name": "infinitetalk_0323.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0323.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+Yg0JvTWnnNE+P0+C0+F13944-14140.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Yg0JvTWnnNE+P0+C0+F13944-14140.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0323.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p260/p260_252.wav", "generated_video": "1_fake/infinitetalk_0323.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+Yg0JvTWnnNE+P0+C0+F13944-14140__p260_252.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 323, "real_video": "0_real/infinitetalk_0323.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Yg0JvTWnnNE+P0+C0+F13944-14140.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+Yg0JvTWnnNE+P0+C0+F13944-14140__p260_252.mp4" }, { "name": "infinitetalk_0324.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0324.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+-IW8wEAPglw+P0+C1+F18284-18465.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+-IW8wEAPglw+P0+C1+F18284-18465.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0324.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p362/p362_160.wav", "generated_video": "1_fake/infinitetalk_0324.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+-IW8wEAPglw+P0+C1+F18284-18465__p362_160.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 324, "real_video": "0_real/infinitetalk_0324.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+-IW8wEAPglw+P0+C1+F18284-18465.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+-IW8wEAPglw+P0+C1+F18284-18465__p362_160.mp4" }, { "name": "infinitetalk_0325.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0325.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+quytFBVK_h8+P0+C0+F3467-3645.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+quytFBVK_h8+P0+C0+F3467-3645.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0325.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p276/p276_092.wav", "generated_video": "1_fake/infinitetalk_0325.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+quytFBVK_h8+P0+C0+F3467-3645__p276_092.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 325, "real_video": "0_real/infinitetalk_0325.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+quytFBVK_h8+P0+C0+F3467-3645.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+quytFBVK_h8+P0+C0+F3467-3645__p276_092.mp4" }, { "name": "infinitetalk_0326.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0326.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+ySQpXmEpP7s+P0+C1+F20070-20209.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+ySQpXmEpP7s+P0+C1+F20070-20209.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0326.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p288/p288_285.wav", "generated_video": "1_fake/infinitetalk_0326.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+ySQpXmEpP7s+P0+C1+F20070-20209__p288_285.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 326, "real_video": "0_real/infinitetalk_0326.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+ySQpXmEpP7s+P0+C1+F20070-20209.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+ySQpXmEpP7s+P0+C1+F20070-20209__p288_285.mp4" }, { "name": "infinitetalk_0327.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0327.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+HFBr-nBVx6g+P0+C2+F2573-2689.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+HFBr-nBVx6g+P0+C2+F2573-2689.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0327.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p272/p272_009.wav", "generated_video": "1_fake/infinitetalk_0327.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+HFBr-nBVx6g+P0+C2+F2573-2689__p272_009.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 327, "real_video": "0_real/infinitetalk_0327.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+HFBr-nBVx6g+P0+C2+F2573-2689.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+HFBr-nBVx6g+P0+C2+F2573-2689__p272_009.mp4" }, { "name": "infinitetalk_0328.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0328.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+PgVTCFMF-8g+P0+C0+F4729-5710.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+PgVTCFMF-8g+P0+C0+F4729-5710.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0328.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p312/p312_080.wav", "generated_video": "1_fake/infinitetalk_0328.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+PgVTCFMF-8g+P0+C0+F4729-5710__p312_080.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 328, "real_video": "0_real/infinitetalk_0328.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+PgVTCFMF-8g+P0+C0+F4729-5710.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+PgVTCFMF-8g+P0+C0+F4729-5710__p312_080.mp4" }, { "name": "infinitetalk_0329.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0329.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+FIIQG_w19Tg+P0+C2+F7431-8009.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+FIIQG_w19Tg+P0+C2+F7431-8009.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0329.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p295/p295_195.wav", "generated_video": "1_fake/infinitetalk_0329.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+FIIQG_w19Tg+P0+C2+F7431-8009__p295_195.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 329, "real_video": "0_real/infinitetalk_0329.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+FIIQG_w19Tg+P0+C2+F7431-8009.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+FIIQG_w19Tg+P0+C2+F7431-8009__p295_195.mp4" }, { "name": "infinitetalk_0330.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0330.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+qzQyhu2RI_g+P0+C2+F14974-15090.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+qzQyhu2RI_g+P0+C2+F14974-15090.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0330.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p343/p343_167.wav", "generated_video": "1_fake/infinitetalk_0330.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+qzQyhu2RI_g+P0+C2+F14974-15090__p343_167.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 330, "real_video": "0_real/infinitetalk_0330.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+qzQyhu2RI_g+P0+C2+F14974-15090.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+qzQyhu2RI_g+P0+C2+F14974-15090__p343_167.mp4" }, { "name": "infinitetalk_0331.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0331.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+CsSXUiAh1Bs+P0+C2+F2074-3060.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+CsSXUiAh1Bs+P0+C2+F2074-3060.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0331.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p307/p307_332.wav", "generated_video": "1_fake/infinitetalk_0331.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+CsSXUiAh1Bs+P0+C2+F2074-3060__p307_332.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 331, "real_video": "0_real/infinitetalk_0331.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+CsSXUiAh1Bs+P0+C2+F2074-3060.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+CsSXUiAh1Bs+P0+C2+F2074-3060__p307_332.mp4" }, { "name": "infinitetalk_0332.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0332.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+4BBJA8OB3RY+P0+C2+F10856-11045.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+4BBJA8OB3RY+P0+C2+F10856-11045.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0332.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p280/p280_172.wav", "generated_video": "1_fake/infinitetalk_0332.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+4BBJA8OB3RY+P0+C2+F10856-11045__p280_172.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 332, "real_video": "0_real/infinitetalk_0332.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+4BBJA8OB3RY+P0+C2+F10856-11045.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+4BBJA8OB3RY+P0+C2+F10856-11045__p280_172.mp4" }, { "name": "infinitetalk_0333.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0333.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+Ojv-wQX85do+P0+C0+F55760-55938.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Ojv-wQX85do+P0+C0+F55760-55938.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0333.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p302/p302_066.wav", "generated_video": "1_fake/infinitetalk_0333.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+Ojv-wQX85do+P0+C0+F55760-55938__p302_066.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 333, "real_video": "0_real/infinitetalk_0333.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Ojv-wQX85do+P0+C0+F55760-55938.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+Ojv-wQX85do+P0+C0+F55760-55938__p302_066.mp4" }, { "name": "infinitetalk_0334.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0334.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+tg8yUmmfRrU+P0+C0+F18145-18355.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+tg8yUmmfRrU+P0+C0+F18145-18355.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0334.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p260/p260_035.wav", "generated_video": "1_fake/infinitetalk_0334.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+tg8yUmmfRrU+P0+C0+F18145-18355__p260_035.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 334, "real_video": "0_real/infinitetalk_0334.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+tg8yUmmfRrU+P0+C0+F18145-18355.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+tg8yUmmfRrU+P0+C0+F18145-18355__p260_035.mp4" }, { "name": "infinitetalk_0335.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0335.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+IvAtrT9XKqY+P0+C2+F2237-2405.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+IvAtrT9XKqY+P0+C2+F2237-2405.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0335.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p361/p361_265.wav", "generated_video": "1_fake/infinitetalk_0335.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+IvAtrT9XKqY+P0+C2+F2237-2405__p361_265.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 335, "real_video": "0_real/infinitetalk_0335.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+IvAtrT9XKqY+P0+C2+F2237-2405.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+IvAtrT9XKqY+P0+C2+F2237-2405__p361_265.mp4" }, { "name": "infinitetalk_0336.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0336.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+VG9dERgCLJU+P0+C2+F3686-3915.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+VG9dERgCLJU+P0+C2+F3686-3915.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0336.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p301/p301_268.wav", "generated_video": "1_fake/infinitetalk_0336.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+VG9dERgCLJU+P0+C2+F3686-3915__p301_268.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 336, "real_video": "0_real/infinitetalk_0336.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+VG9dERgCLJU+P0+C2+F3686-3915.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+VG9dERgCLJU+P0+C2+F3686-3915__p301_268.mp4" }, { "name": "infinitetalk_0337.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0337.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+GUKVh9OpxJc+P0+C0+F3460-3590.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+GUKVh9OpxJc+P0+C0+F3460-3590.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0337.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p252/p252_172.wav", "generated_video": "1_fake/infinitetalk_0337.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+GUKVh9OpxJc+P0+C0+F3460-3590__p252_172.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 337, "real_video": "0_real/infinitetalk_0337.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+GUKVh9OpxJc+P0+C0+F3460-3590.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+GUKVh9OpxJc+P0+C0+F3460-3590__p252_172.mp4" }, { "name": "infinitetalk_0338.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0338.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+PTb_KQCDVN4+P0+C1+F2289-2775.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+PTb_KQCDVN4+P0+C1+F2289-2775.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0338.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p279/p279_223.wav", "generated_video": "1_fake/infinitetalk_0338.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+PTb_KQCDVN4+P0+C1+F2289-2775__p279_223.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 338, "real_video": "0_real/infinitetalk_0338.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+PTb_KQCDVN4+P0+C1+F2289-2775.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+PTb_KQCDVN4+P0+C1+F2289-2775__p279_223.mp4" }, { "name": "infinitetalk_0339.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0339.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+BGKZXSPpseU+P0+C1+F5486-5873.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+BGKZXSPpseU+P0+C1+F5486-5873.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0339.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p248/p248_237.wav", "generated_video": "1_fake/infinitetalk_0339.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+BGKZXSPpseU+P0+C1+F5486-5873__p248_237.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 339, "real_video": "0_real/infinitetalk_0339.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+BGKZXSPpseU+P0+C1+F5486-5873.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+BGKZXSPpseU+P0+C1+F5486-5873__p248_237.mp4" }, { "name": "infinitetalk_0340.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0340.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+79vu9mgSorY+P0+C2+F2919-3040.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+79vu9mgSorY+P0+C2+F2919-3040.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0340.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p247/p247_444.wav", "generated_video": "1_fake/infinitetalk_0340.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+79vu9mgSorY+P0+C2+F2919-3040__p247_444.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 340, "real_video": "0_real/infinitetalk_0340.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+79vu9mgSorY+P0+C2+F2919-3040.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+79vu9mgSorY+P0+C2+F2919-3040__p247_444.mp4" }, { "name": "infinitetalk_0341.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0341.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+il-hwWZhUUU+P1+C1+F16856-17064.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+il-hwWZhUUU+P1+C1+F16856-17064.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0341.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p276/p276_052.wav", "generated_video": "1_fake/infinitetalk_0341.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+il-hwWZhUUU+P1+C1+F16856-17064__p276_052.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 341, "real_video": "0_real/infinitetalk_0341.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+il-hwWZhUUU+P1+C1+F16856-17064.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+il-hwWZhUUU+P1+C1+F16856-17064__p276_052.mp4" }, { "name": "infinitetalk_0342.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0342.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+C97PWbX-nDc+P0+C0+F5845-5984.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+C97PWbX-nDc+P0+C0+F5845-5984.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0342.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p299/p299_169.wav", "generated_video": "1_fake/infinitetalk_0342.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+C97PWbX-nDc+P0+C0+F5845-5984__p299_169.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 342, "real_video": "0_real/infinitetalk_0342.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+C97PWbX-nDc+P0+C0+F5845-5984.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+C97PWbX-nDc+P0+C0+F5845-5984__p299_169.mp4" }, { "name": "infinitetalk_0343.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0343.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+zqSJy6IyFXk+P0+C1+F4924-5039.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+zqSJy6IyFXk+P0+C1+F4924-5039.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0343.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p269/p269_077.wav", "generated_video": "1_fake/infinitetalk_0343.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+zqSJy6IyFXk+P0+C1+F4924-5039__p269_077.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 343, "real_video": "0_real/infinitetalk_0343.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+zqSJy6IyFXk+P0+C1+F4924-5039.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+zqSJy6IyFXk+P0+C1+F4924-5039__p269_077.mp4" }, { "name": "infinitetalk_0344.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0344.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+4mtwbI9xyL8+P0+C1+F3301-3459.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+4mtwbI9xyL8+P0+C1+F3301-3459.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0344.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p257/p257_118.wav", "generated_video": "1_fake/infinitetalk_0344.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+4mtwbI9xyL8+P0+C1+F3301-3459__p257_118.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 344, "real_video": "0_real/infinitetalk_0344.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+4mtwbI9xyL8+P0+C1+F3301-3459.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+4mtwbI9xyL8+P0+C1+F3301-3459__p257_118.mp4" }, { "name": "infinitetalk_0345.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0345.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+Nyp7c5UCpNI+P0+C0+F29072-29299.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Nyp7c5UCpNI+P0+C0+F29072-29299.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0345.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p245/p245_124.wav", "generated_video": "1_fake/infinitetalk_0345.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+Nyp7c5UCpNI+P0+C0+F29072-29299__p245_124.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 345, "real_video": "0_real/infinitetalk_0345.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Nyp7c5UCpNI+P0+C0+F29072-29299.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+Nyp7c5UCpNI+P0+C0+F29072-29299__p245_124.mp4" }, { "name": "infinitetalk_0346.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0346.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+EKDAV5Y5km0+P0+C0+F3729-4245.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+EKDAV5Y5km0+P0+C0+F3729-4245.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0346.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p313/p313_113.wav", "generated_video": "1_fake/infinitetalk_0346.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+EKDAV5Y5km0+P0+C0+F3729-4245__p313_113.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 346, "real_video": "0_real/infinitetalk_0346.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+EKDAV5Y5km0+P0+C0+F3729-4245.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+EKDAV5Y5km0+P0+C0+F3729-4245__p313_113.mp4" }, { "name": "infinitetalk_0347.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0347.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+DT5khKKq5pk+P1+C2+F710-953.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+DT5khKKq5pk+P1+C2+F710-953.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0347.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p314/p314_228.wav", "generated_video": "1_fake/infinitetalk_0347.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+DT5khKKq5pk+P1+C2+F710-953__p314_228.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 347, "real_video": "0_real/infinitetalk_0347.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+DT5khKKq5pk+P1+C2+F710-953.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+DT5khKKq5pk+P1+C2+F710-953__p314_228.mp4" }, { "name": "infinitetalk_0348.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0348.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+IX-KTrKnYZo+P0+C1+F8539-8740.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+IX-KTrKnYZo+P0+C1+F8539-8740.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0348.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p295/p295_235.wav", "generated_video": "1_fake/infinitetalk_0348.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+IX-KTrKnYZo+P0+C1+F8539-8740__p295_235.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 348, "real_video": "0_real/infinitetalk_0348.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+IX-KTrKnYZo+P0+C1+F8539-8740.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+IX-KTrKnYZo+P0+C1+F8539-8740__p295_235.mp4" }, { "name": "infinitetalk_0349.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0349.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+6GuVGVsTXo8+P0+C0+F2642-2757.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+6GuVGVsTXo8+P0+C0+F2642-2757.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0349.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p272/p272_054.wav", "generated_video": "1_fake/infinitetalk_0349.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+6GuVGVsTXo8+P0+C0+F2642-2757__p272_054.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 349, "real_video": "0_real/infinitetalk_0349.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+6GuVGVsTXo8+P0+C0+F2642-2757.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+6GuVGVsTXo8+P0+C0+F2642-2757__p272_054.mp4" }, { "name": "infinitetalk_0350.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0350.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+Y8ARilIKJ_k+P0+C2+F1350-1459.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Y8ARilIKJ_k+P0+C2+F1350-1459.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0350.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p301/p301_215.wav", "generated_video": "1_fake/infinitetalk_0350.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+Y8ARilIKJ_k+P0+C2+F1350-1459__p301_215.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 350, "real_video": "0_real/infinitetalk_0350.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Y8ARilIKJ_k+P0+C2+F1350-1459.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+Y8ARilIKJ_k+P0+C2+F1350-1459__p301_215.mp4" }, { "name": "infinitetalk_0351.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0351.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+DVn-VkVYuQw+P0+C2+F925-1568.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+DVn-VkVYuQw+P0+C2+F925-1568.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0351.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p288/p288_025.wav", "generated_video": "1_fake/infinitetalk_0351.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+DVn-VkVYuQw+P0+C2+F925-1568__p288_025.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 351, "real_video": "0_real/infinitetalk_0351.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+DVn-VkVYuQw+P0+C2+F925-1568.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+DVn-VkVYuQw+P0+C2+F925-1568__p288_025.mp4" }, { "name": "infinitetalk_0352.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0352.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+5nqqv1I4nog+P0+C1+F1664-2015.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+5nqqv1I4nog+P0+C1+F1664-2015.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0352.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p281/p281_400.wav", "generated_video": "1_fake/infinitetalk_0352.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+5nqqv1I4nog+P0+C1+F1664-2015__p281_400.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 352, "real_video": "0_real/infinitetalk_0352.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+5nqqv1I4nog+P0+C1+F1664-2015.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+5nqqv1I4nog+P0+C1+F1664-2015__p281_400.mp4" }, { "name": "infinitetalk_0353.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0353.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+7IAUCD4P0OM+P0+C0+F1851-2044.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+7IAUCD4P0OM+P0+C0+F1851-2044.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0353.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p251/p251_226.wav", "generated_video": "1_fake/infinitetalk_0353.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+7IAUCD4P0OM+P0+C0+F1851-2044__p251_226.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 353, "real_video": "0_real/infinitetalk_0353.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+7IAUCD4P0OM+P0+C0+F1851-2044.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+7IAUCD4P0OM+P0+C0+F1851-2044__p251_226.mp4" }, { "name": "infinitetalk_0354.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0354.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+8JbMZhEjEnA+P0+C1+F7739-7955.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+8JbMZhEjEnA+P0+C1+F7739-7955.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0354.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p283/p283_385.wav", "generated_video": "1_fake/infinitetalk_0354.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+8JbMZhEjEnA+P0+C1+F7739-7955__p283_385.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 354, "real_video": "0_real/infinitetalk_0354.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+8JbMZhEjEnA+P0+C1+F7739-7955.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+8JbMZhEjEnA+P0+C1+F7739-7955__p283_385.mp4" }, { "name": "infinitetalk_0355.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0355.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+mQl3XRDYuaU+P0+C0+F11209-11505.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+mQl3XRDYuaU+P0+C0+F11209-11505.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0355.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p277/p277_361.wav", "generated_video": "1_fake/infinitetalk_0355.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+mQl3XRDYuaU+P0+C0+F11209-11505__p277_361.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 355, "real_video": "0_real/infinitetalk_0355.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+mQl3XRDYuaU+P0+C0+F11209-11505.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+mQl3XRDYuaU+P0+C0+F11209-11505__p277_361.mp4" }, { "name": "infinitetalk_0356.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0356.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+_mO5W7VHxlM+P0+C2+F15694-15802.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+_mO5W7VHxlM+P0+C2+F15694-15802.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0356.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p277/p277_077.wav", "generated_video": "1_fake/infinitetalk_0356.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+_mO5W7VHxlM+P0+C2+F15694-15802__p277_077.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 356, "real_video": "0_real/infinitetalk_0356.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+_mO5W7VHxlM+P0+C2+F15694-15802.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+_mO5W7VHxlM+P0+C2+F15694-15802__p277_077.mp4" }, { "name": "infinitetalk_0357.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0357.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+INiib5Qs1mo+P0+C2+F4281-4592.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+INiib5Qs1mo+P0+C2+F4281-4592.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0357.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p261/p261_046.wav", "generated_video": "1_fake/infinitetalk_0357.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+INiib5Qs1mo+P0+C2+F4281-4592__p261_046.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 357, "real_video": "0_real/infinitetalk_0357.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+INiib5Qs1mo+P0+C2+F4281-4592.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+INiib5Qs1mo+P0+C2+F4281-4592__p261_046.mp4" }, { "name": "infinitetalk_0358.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0358.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+jGWSFajpjWs+P2+C0+F55234-55360.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+jGWSFajpjWs+P2+C0+F55234-55360.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0358.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p260/p260_248.wav", "generated_video": "1_fake/infinitetalk_0358.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+jGWSFajpjWs+P2+C0+F55234-55360__p260_248.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 358, "real_video": "0_real/infinitetalk_0358.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+jGWSFajpjWs+P2+C0+F55234-55360.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+jGWSFajpjWs+P2+C0+F55234-55360__p260_248.mp4" }, { "name": "infinitetalk_0359.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0359.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+hot5ixctq70+P0+C2+F5750-6053.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+hot5ixctq70+P0+C2+F5750-6053.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0359.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p276/p276_272.wav", "generated_video": "1_fake/infinitetalk_0359.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+hot5ixctq70+P0+C2+F5750-6053__p276_272.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 359, "real_video": "0_real/infinitetalk_0359.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+hot5ixctq70+P0+C2+F5750-6053.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+hot5ixctq70+P0+C2+F5750-6053__p276_272.mp4" }, { "name": "infinitetalk_0360.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0360.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+S_qjwyhBo7c+P0+C2+F2299-2705.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+S_qjwyhBo7c+P0+C2+F2299-2705.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0360.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p262/p262_258.wav", "generated_video": "1_fake/infinitetalk_0360.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+S_qjwyhBo7c+P0+C2+F2299-2705__p262_258.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 360, "real_video": "0_real/infinitetalk_0360.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+S_qjwyhBo7c+P0+C2+F2299-2705.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+S_qjwyhBo7c+P0+C2+F2299-2705__p262_258.mp4" }, { "name": "infinitetalk_0361.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0361.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+S6ouZz2pqT4+P0+C0+F13614-13770.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+S6ouZz2pqT4+P0+C0+F13614-13770.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0361.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p293/p293_246.wav", "generated_video": "1_fake/infinitetalk_0361.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+S6ouZz2pqT4+P0+C0+F13614-13770__p293_246.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 361, "real_video": "0_real/infinitetalk_0361.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+S6ouZz2pqT4+P0+C0+F13614-13770.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+S6ouZz2pqT4+P0+C0+F13614-13770__p293_246.mp4" }, { "name": "infinitetalk_0362.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0362.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+mmsasEoupO4+P0+C0+F170746-170867.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+mmsasEoupO4+P0+C0+F170746-170867.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0362.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p307/p307_424.wav", "generated_video": "1_fake/infinitetalk_0362.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+mmsasEoupO4+P0+C0+F170746-170867__p307_424.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 362, "real_video": "0_real/infinitetalk_0362.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+mmsasEoupO4+P0+C0+F170746-170867.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+mmsasEoupO4+P0+C0+F170746-170867__p307_424.mp4" }, { "name": "infinitetalk_0363.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0363.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+4ZygAG6lBkw+P0+C2+F26138-26699.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+4ZygAG6lBkw+P0+C2+F26138-26699.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0363.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p300/p300_316.wav", "generated_video": "1_fake/infinitetalk_0363.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+4ZygAG6lBkw+P0+C2+F26138-26699__p300_316.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 363, "real_video": "0_real/infinitetalk_0363.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+4ZygAG6lBkw+P0+C2+F26138-26699.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+4ZygAG6lBkw+P0+C2+F26138-26699__p300_316.mp4" }, { "name": "infinitetalk_0364.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0364.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+DVn-VkVYuQw+P0+C0+F2584-2798.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+DVn-VkVYuQw+P0+C0+F2584-2798.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0364.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p275/p275_014.wav", "generated_video": "1_fake/infinitetalk_0364.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+DVn-VkVYuQw+P0+C0+F2584-2798__p275_014.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 364, "real_video": "0_real/infinitetalk_0364.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+DVn-VkVYuQw+P0+C0+F2584-2798.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+DVn-VkVYuQw+P0+C0+F2584-2798__p275_014.mp4" }, { "name": "infinitetalk_0365.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0365.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+-QJw7ZjxAbw+P0+C0+F5142-5276.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+-QJw7ZjxAbw+P0+C0+F5142-5276.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0365.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p278/p278_195.wav", "generated_video": "1_fake/infinitetalk_0365.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+-QJw7ZjxAbw+P0+C0+F5142-5276__p278_195.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 365, "real_video": "0_real/infinitetalk_0365.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+-QJw7ZjxAbw+P0+C0+F5142-5276.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+-QJw7ZjxAbw+P0+C0+F5142-5276__p278_195.mp4" }, { "name": "infinitetalk_0366.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0366.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+ODeh8GR5Bb0+P0+C0+F5671-5929.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+ODeh8GR5Bb0+P0+C0+F5671-5929.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0366.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p260/p260_297.wav", "generated_video": "1_fake/infinitetalk_0366.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+ODeh8GR5Bb0+P0+C0+F5671-5929__p260_297.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 366, "real_video": "0_real/infinitetalk_0366.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+ODeh8GR5Bb0+P0+C0+F5671-5929.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+ODeh8GR5Bb0+P0+C0+F5671-5929__p260_297.mp4" }, { "name": "infinitetalk_0367.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0367.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+2vmjNDcz6mA+P1+C0+F11769-11896.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+2vmjNDcz6mA+P1+C0+F11769-11896.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0367.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p347/p347_103.wav", "generated_video": "1_fake/infinitetalk_0367.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+2vmjNDcz6mA+P1+C0+F11769-11896__p347_103.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 367, "real_video": "0_real/infinitetalk_0367.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+2vmjNDcz6mA+P1+C0+F11769-11896.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+2vmjNDcz6mA+P1+C0+F11769-11896__p347_103.mp4" }, { "name": "infinitetalk_0368.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0368.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+LnhYWO-hWhU+P0+C2+F10908-11060.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+LnhYWO-hWhU+P0+C2+F10908-11060.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0368.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p274/p274_426.wav", "generated_video": "1_fake/infinitetalk_0368.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+LnhYWO-hWhU+P0+C2+F10908-11060__p274_426.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 368, "real_video": "0_real/infinitetalk_0368.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+LnhYWO-hWhU+P0+C2+F10908-11060.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+LnhYWO-hWhU+P0+C2+F10908-11060__p274_426.mp4" }, { "name": "infinitetalk_0369.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0369.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+Gy-LkE72Wzw+P1+C0+F42343-42455.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Gy-LkE72Wzw+P1+C0+F42343-42455.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0369.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p285/p285_354.wav", "generated_video": "1_fake/infinitetalk_0369.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+Gy-LkE72Wzw+P1+C0+F42343-42455__p285_354.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 369, "real_video": "0_real/infinitetalk_0369.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Gy-LkE72Wzw+P1+C0+F42343-42455.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+Gy-LkE72Wzw+P1+C0+F42343-42455__p285_354.mp4" }, { "name": "infinitetalk_0370.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0370.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+g1EcQaWtP0E+P0+C2+F30929-31075.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+g1EcQaWtP0E+P0+C2+F30929-31075.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0370.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p254/p254_209.wav", "generated_video": "1_fake/infinitetalk_0370.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+g1EcQaWtP0E+P0+C2+F30929-31075__p254_209.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 370, "real_video": "0_real/infinitetalk_0370.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+g1EcQaWtP0E+P0+C2+F30929-31075.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+g1EcQaWtP0E+P0+C2+F30929-31075__p254_209.mp4" }, { "name": "infinitetalk_0371.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0371.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+4Wv_EEku11s+P0+C1+F3872-4011.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+4Wv_EEku11s+P0+C1+F3872-4011.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0371.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p230/p230_097.wav", "generated_video": "1_fake/infinitetalk_0371.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+4Wv_EEku11s+P0+C1+F3872-4011__p230_097.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 371, "real_video": "0_real/infinitetalk_0371.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+4Wv_EEku11s+P0+C1+F3872-4011.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+4Wv_EEku11s+P0+C1+F3872-4011__p230_097.mp4" }, { "name": "infinitetalk_0372.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0372.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+cef3h7wiIYM+P0+C0+F6674-6930.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+cef3h7wiIYM+P0+C0+F6674-6930.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0372.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p297/p297_134.wav", "generated_video": "1_fake/infinitetalk_0372.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+cef3h7wiIYM+P0+C0+F6674-6930__p297_134.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 372, "real_video": "0_real/infinitetalk_0372.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+cef3h7wiIYM+P0+C0+F6674-6930.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+cef3h7wiIYM+P0+C0+F6674-6930__p297_134.mp4" }, { "name": "infinitetalk_0373.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0373.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+93lbE8oM2r8+P0+C0+F24095-24479.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+93lbE8oM2r8+P0+C0+F24095-24479.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0373.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p300/p300_220.wav", "generated_video": "1_fake/infinitetalk_0373.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+93lbE8oM2r8+P0+C0+F24095-24479__p300_220.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 373, "real_video": "0_real/infinitetalk_0373.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+93lbE8oM2r8+P0+C0+F24095-24479.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+93lbE8oM2r8+P0+C0+F24095-24479__p300_220.mp4" }, { "name": "infinitetalk_0374.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0374.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+3LPklp0rf4M+P3+C0+F10779-10943.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+3LPklp0rf4M+P3+C0+F10779-10943.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0374.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p233/p233_171.wav", "generated_video": "1_fake/infinitetalk_0374.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+3LPklp0rf4M+P3+C0+F10779-10943__p233_171.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 374, "real_video": "0_real/infinitetalk_0374.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+3LPklp0rf4M+P3+C0+F10779-10943.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+3LPklp0rf4M+P3+C0+F10779-10943__p233_171.mp4" }, { "name": "infinitetalk_0375.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0375.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+Ng0mAMJKaaY+P0+C2+F985-1263.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Ng0mAMJKaaY+P0+C2+F985-1263.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0375.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p336/p336_404.wav", "generated_video": "1_fake/infinitetalk_0375.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+Ng0mAMJKaaY+P0+C2+F985-1263__p336_404.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 375, "real_video": "0_real/infinitetalk_0375.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Ng0mAMJKaaY+P0+C2+F985-1263.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+Ng0mAMJKaaY+P0+C2+F985-1263__p336_404.mp4" }, { "name": "infinitetalk_0376.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0376.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+_tAVz5mqACY+P0+C2+F40033-40224.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+_tAVz5mqACY+P0+C2+F40033-40224.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0376.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p301/p301_340.wav", "generated_video": "1_fake/infinitetalk_0376.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+_tAVz5mqACY+P0+C2+F40033-40224__p301_340.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 376, "real_video": "0_real/infinitetalk_0376.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+_tAVz5mqACY+P0+C2+F40033-40224.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+_tAVz5mqACY+P0+C2+F40033-40224__p301_340.mp4" }, { "name": "infinitetalk_0377.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0377.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+sFzCikcEjZY+P5+C2+F55113-55233.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+sFzCikcEjZY+P5+C2+F55113-55233.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0377.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p231/p231_226.wav", "generated_video": "1_fake/infinitetalk_0377.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+sFzCikcEjZY+P5+C2+F55113-55233__p231_226.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 377, "real_video": "0_real/infinitetalk_0377.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+sFzCikcEjZY+P5+C2+F55113-55233.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+sFzCikcEjZY+P5+C2+F55113-55233__p231_226.mp4" }, { "name": "infinitetalk_0378.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0378.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+xHyVum7bzXo+P1+C0+F7361-7471.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+xHyVum7bzXo+P1+C0+F7361-7471.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0378.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p255/p255_363.wav", "generated_video": "1_fake/infinitetalk_0378.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+xHyVum7bzXo+P1+C0+F7361-7471__p255_363.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 378, "real_video": "0_real/infinitetalk_0378.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+xHyVum7bzXo+P1+C0+F7361-7471.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+xHyVum7bzXo+P1+C0+F7361-7471__p255_363.mp4" }, { "name": "infinitetalk_0379.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0379.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+TFU1BrFuyEk+P0+C1+F7081-7275.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+TFU1BrFuyEk+P0+C1+F7081-7275.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0379.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p228/p228_028.wav", "generated_video": "1_fake/infinitetalk_0379.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+TFU1BrFuyEk+P0+C1+F7081-7275__p228_028.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 379, "real_video": "0_real/infinitetalk_0379.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+TFU1BrFuyEk+P0+C1+F7081-7275.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+TFU1BrFuyEk+P0+C1+F7081-7275__p228_028.mp4" }, { "name": "infinitetalk_0380.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0380.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+hDnzpiF32RI+P0+C0+F1520-1660.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+hDnzpiF32RI+P0+C0+F1520-1660.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0380.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p318/p318_118.wav", "generated_video": "1_fake/infinitetalk_0380.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+hDnzpiF32RI+P0+C0+F1520-1660__p318_118.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 380, "real_video": "0_real/infinitetalk_0380.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+hDnzpiF32RI+P0+C0+F1520-1660.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+hDnzpiF32RI+P0+C0+F1520-1660__p318_118.mp4" }, { "name": "infinitetalk_0381.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0381.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+QM-G15znRNA+P0+C0+F3002-3148.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+QM-G15znRNA+P0+C0+F3002-3148.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0381.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p243/p243_077.wav", "generated_video": "1_fake/infinitetalk_0381.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+QM-G15znRNA+P0+C0+F3002-3148__p243_077.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 381, "real_video": "0_real/infinitetalk_0381.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+QM-G15znRNA+P0+C0+F3002-3148.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+QM-G15znRNA+P0+C0+F3002-3148__p243_077.mp4" }, { "name": "infinitetalk_0382.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0382.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+GcwLIj3p8Ak+P0+C1+F84324-84508.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+GcwLIj3p8Ak+P0+C1+F84324-84508.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0382.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p376/p376_180.wav", "generated_video": "1_fake/infinitetalk_0382.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+GcwLIj3p8Ak+P0+C1+F84324-84508__p376_180.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 382, "real_video": "0_real/infinitetalk_0382.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+GcwLIj3p8Ak+P0+C1+F84324-84508.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+GcwLIj3p8Ak+P0+C1+F84324-84508__p376_180.mp4" }, { "name": "infinitetalk_0383.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0383.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+JvXO7Aj9ZEs+P0+C1+F2012-2992.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+JvXO7Aj9ZEs+P0+C1+F2012-2992.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0383.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p326/p326_302.wav", "generated_video": "1_fake/infinitetalk_0383.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+JvXO7Aj9ZEs+P0+C1+F2012-2992__p326_302.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 383, "real_video": "0_real/infinitetalk_0383.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+JvXO7Aj9ZEs+P0+C1+F2012-2992.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+JvXO7Aj9ZEs+P0+C1+F2012-2992__p326_302.mp4" }, { "name": "infinitetalk_0384.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0384.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+3VSAMrNdIqg+P0+C1+F20760-20880.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+3VSAMrNdIqg+P0+C1+F20760-20880.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0384.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p263/p263_377.wav", "generated_video": "1_fake/infinitetalk_0384.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+3VSAMrNdIqg+P0+C1+F20760-20880__p263_377.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 384, "real_video": "0_real/infinitetalk_0384.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+3VSAMrNdIqg+P0+C1+F20760-20880.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+3VSAMrNdIqg+P0+C1+F20760-20880__p263_377.mp4" }, { "name": "infinitetalk_0385.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0385.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+W0l4KqEVg1Y+P0+C1+F5096-5300.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+W0l4KqEVg1Y+P0+C1+F5096-5300.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0385.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p302/p302_156.wav", "generated_video": "1_fake/infinitetalk_0385.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+W0l4KqEVg1Y+P0+C1+F5096-5300__p302_156.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 385, "real_video": "0_real/infinitetalk_0385.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+W0l4KqEVg1Y+P0+C1+F5096-5300.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+W0l4KqEVg1Y+P0+C1+F5096-5300__p302_156.mp4" }, { "name": "infinitetalk_0386.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0386.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+bchyWTgiiq8+P0+C1+F23619-23760.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+bchyWTgiiq8+P0+C1+F23619-23760.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0386.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p262/p262_167.wav", "generated_video": "1_fake/infinitetalk_0386.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+bchyWTgiiq8+P0+C1+F23619-23760__p262_167.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 386, "real_video": "0_real/infinitetalk_0386.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+bchyWTgiiq8+P0+C1+F23619-23760.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+bchyWTgiiq8+P0+C1+F23619-23760__p262_167.mp4" }, { "name": "infinitetalk_0387.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0387.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+ryAXMAfdOTE+P0+C0+F6694-6885.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+ryAXMAfdOTE+P0+C0+F6694-6885.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0387.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p262/p262_253.wav", "generated_video": "1_fake/infinitetalk_0387.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+ryAXMAfdOTE+P0+C0+F6694-6885__p262_253.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 387, "real_video": "0_real/infinitetalk_0387.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+ryAXMAfdOTE+P0+C0+F6694-6885.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+ryAXMAfdOTE+P0+C0+F6694-6885__p262_253.mp4" }, { "name": "infinitetalk_0388.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0388.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+hot5ixctq70+P0+C0+F8311-8473.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+hot5ixctq70+P0+C0+F8311-8473.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0388.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p305/p305_323.wav", "generated_video": "1_fake/infinitetalk_0388.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+hot5ixctq70+P0+C0+F8311-8473__p305_323.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 388, "real_video": "0_real/infinitetalk_0388.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+hot5ixctq70+P0+C0+F8311-8473.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+hot5ixctq70+P0+C0+F8311-8473__p305_323.mp4" }, { "name": "infinitetalk_0389.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0389.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+Qyj5IpY9rUE+P0+C0+F47704-47860.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Qyj5IpY9rUE+P0+C0+F47704-47860.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0389.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p295/p295_197.wav", "generated_video": "1_fake/infinitetalk_0389.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+Qyj5IpY9rUE+P0+C0+F47704-47860__p295_197.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 389, "real_video": "0_real/infinitetalk_0389.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Qyj5IpY9rUE+P0+C0+F47704-47860.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+Qyj5IpY9rUE+P0+C0+F47704-47860__p295_197.mp4" }, { "name": "infinitetalk_0390.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0390.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+ta25I2CbDVw+P0+C2+F3173-3360.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+ta25I2CbDVw+P0+C2+F3173-3360.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0390.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p345/p345_400.wav", "generated_video": "1_fake/infinitetalk_0390.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+ta25I2CbDVw+P0+C2+F3173-3360__p345_400.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 390, "real_video": "0_real/infinitetalk_0390.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+ta25I2CbDVw+P0+C2+F3173-3360.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+ta25I2CbDVw+P0+C2+F3173-3360__p345_400.mp4" }, { "name": "infinitetalk_0391.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0391.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+0olJlK-mlg8+P0+C1+F5561-6247.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+0olJlK-mlg8+P0+C1+F5561-6247.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0391.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p233/p233_246.wav", "generated_video": "1_fake/infinitetalk_0391.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+0olJlK-mlg8+P0+C1+F5561-6247__p233_246.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 391, "real_video": "0_real/infinitetalk_0391.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+0olJlK-mlg8+P0+C1+F5561-6247.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+0olJlK-mlg8+P0+C1+F5561-6247__p233_246.mp4" }, { "name": "infinitetalk_0392.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0392.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+y98EvmMIK4A+P0+C0+F5367-5530.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+y98EvmMIK4A+P0+C0+F5367-5530.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0392.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p341/p341_185.wav", "generated_video": "1_fake/infinitetalk_0392.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+y98EvmMIK4A+P0+C0+F5367-5530__p341_185.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 392, "real_video": "0_real/infinitetalk_0392.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+y98EvmMIK4A+P0+C0+F5367-5530.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+y98EvmMIK4A+P0+C0+F5367-5530__p341_185.mp4" }, { "name": "infinitetalk_0393.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0393.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+Mp7qW0tWaIs+P0+C1+F7960-8127.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Mp7qW0tWaIs+P0+C1+F7960-8127.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0393.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p285/p285_080.wav", "generated_video": "1_fake/infinitetalk_0393.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+Mp7qW0tWaIs+P0+C1+F7960-8127__p285_080.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 393, "real_video": "0_real/infinitetalk_0393.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Mp7qW0tWaIs+P0+C1+F7960-8127.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+Mp7qW0tWaIs+P0+C1+F7960-8127__p285_080.mp4" }, { "name": "infinitetalk_0394.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0394.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+TuznyUDYIZk+P2+C1+F4122-4292.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+TuznyUDYIZk+P2+C1+F4122-4292.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0394.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p247/p247_276.wav", "generated_video": "1_fake/infinitetalk_0394.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+TuznyUDYIZk+P2+C1+F4122-4292__p247_276.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 394, "real_video": "0_real/infinitetalk_0394.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+TuznyUDYIZk+P2+C1+F4122-4292.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+TuznyUDYIZk+P2+C1+F4122-4292__p247_276.mp4" }, { "name": "infinitetalk_0395.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0395.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+83wzksqKvMg+P0+C1+F34300-34472.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+83wzksqKvMg+P0+C1+F34300-34472.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0395.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p300/p300_137.wav", "generated_video": "1_fake/infinitetalk_0395.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+83wzksqKvMg+P0+C1+F34300-34472__p300_137.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 395, "real_video": "0_real/infinitetalk_0395.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+83wzksqKvMg+P0+C1+F34300-34472.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+83wzksqKvMg+P0+C1+F34300-34472__p300_137.mp4" }, { "name": "infinitetalk_0396.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0396.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+aKVt6DI-Tv8+P1+C0+F2304-2535.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+aKVt6DI-Tv8+P1+C0+F2304-2535.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0396.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p276/p276_159.wav", "generated_video": "1_fake/infinitetalk_0396.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+aKVt6DI-Tv8+P1+C0+F2304-2535__p276_159.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 396, "real_video": "0_real/infinitetalk_0396.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+aKVt6DI-Tv8+P1+C0+F2304-2535.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+aKVt6DI-Tv8+P1+C0+F2304-2535__p276_159.mp4" }, { "name": "infinitetalk_0397.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0397.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+AUKdoTHJ0mQ+P0+C1+F13876-14007.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+AUKdoTHJ0mQ+P0+C1+F13876-14007.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0397.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p261/p261_428.wav", "generated_video": "1_fake/infinitetalk_0397.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+AUKdoTHJ0mQ+P0+C1+F13876-14007__p261_428.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 397, "real_video": "0_real/infinitetalk_0397.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+AUKdoTHJ0mQ+P0+C1+F13876-14007.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+AUKdoTHJ0mQ+P0+C1+F13876-14007__p261_428.mp4" }, { "name": "infinitetalk_0398.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0398.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+j4JTcMkyF54+P1+C0+F11444-12145.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+j4JTcMkyF54+P1+C0+F11444-12145.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0398.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p268/p268_387.wav", "generated_video": "1_fake/infinitetalk_0398.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+j4JTcMkyF54+P1+C0+F11444-12145__p268_387.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 398, "real_video": "0_real/infinitetalk_0398.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+j4JTcMkyF54+P1+C0+F11444-12145.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+j4JTcMkyF54+P1+C0+F11444-12145__p268_387.mp4" }, { "name": "infinitetalk_0399.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0399.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+sGUTxWYoUUo+P4+C0+F32913-33027.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+sGUTxWYoUUo+P4+C0+F32913-33027.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0399.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p323/p323_087.wav", "generated_video": "1_fake/infinitetalk_0399.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+sGUTxWYoUUo+P4+C0+F32913-33027__p323_087.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 399, "real_video": "0_real/infinitetalk_0399.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+sGUTxWYoUUo+P4+C0+F32913-33027.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+sGUTxWYoUUo+P4+C0+F32913-33027__p323_087.mp4" }, { "name": "infinitetalk_0400.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0400.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+blbFHqMnu6U+P0+C0+F3721-4085.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+blbFHqMnu6U+P0+C0+F3721-4085.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0400.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p297/p297_367.wav", "generated_video": "1_fake/infinitetalk_0400.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+blbFHqMnu6U+P0+C0+F3721-4085__p297_367.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 400, "real_video": "0_real/infinitetalk_0400.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+blbFHqMnu6U+P0+C0+F3721-4085.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+blbFHqMnu6U+P0+C0+F3721-4085__p297_367.mp4" }, { "name": "infinitetalk_0401.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0401.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+odt-a6EMOjU+P0+C0+F43468-43598.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+odt-a6EMOjU+P0+C0+F43468-43598.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0401.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p303/p303_305.wav", "generated_video": "1_fake/infinitetalk_0401.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+odt-a6EMOjU+P0+C0+F43468-43598__p303_305.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 401, "real_video": "0_real/infinitetalk_0401.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+odt-a6EMOjU+P0+C0+F43468-43598.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+odt-a6EMOjU+P0+C0+F43468-43598__p303_305.mp4" }, { "name": "infinitetalk_0402.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0402.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+xe7LkQv5Ev0+P0+C0+F20439-20567.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+xe7LkQv5Ev0+P0+C0+F20439-20567.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0402.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p246/p246_062.wav", "generated_video": "1_fake/infinitetalk_0402.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+xe7LkQv5Ev0+P0+C0+F20439-20567__p246_062.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 402, "real_video": "0_real/infinitetalk_0402.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+xe7LkQv5Ev0+P0+C0+F20439-20567.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+xe7LkQv5Ev0+P0+C0+F20439-20567__p246_062.mp4" }, { "name": "infinitetalk_0403.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0403.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+X5IUgpYr3qc+P0+C2+F53651-54600.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+X5IUgpYr3qc+P0+C2+F53651-54600.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0403.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p308/p308_310.wav", "generated_video": "1_fake/infinitetalk_0403.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+X5IUgpYr3qc+P0+C2+F53651-54600__p308_310.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 403, "real_video": "0_real/infinitetalk_0403.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+X5IUgpYr3qc+P0+C2+F53651-54600.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+X5IUgpYr3qc+P0+C2+F53651-54600__p308_310.mp4" }, { "name": "infinitetalk_0404.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0404.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+jvFUoeXJRM0+P0+C1+F5444-5565.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+jvFUoeXJRM0+P0+C1+F5444-5565.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0404.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p240/p240_084.wav", "generated_video": "1_fake/infinitetalk_0404.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+jvFUoeXJRM0+P0+C1+F5444-5565__p240_084.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 404, "real_video": "0_real/infinitetalk_0404.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+jvFUoeXJRM0+P0+C1+F5444-5565.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+jvFUoeXJRM0+P0+C1+F5444-5565__p240_084.mp4" }, { "name": "infinitetalk_0405.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0405.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+7_fPy-yBzfk+P2+C0+F35813-36007.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+7_fPy-yBzfk+P2+C0+F35813-36007.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0405.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p240/p240_175.wav", "generated_video": "1_fake/infinitetalk_0405.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+7_fPy-yBzfk+P2+C0+F35813-36007__p240_175.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 405, "real_video": "0_real/infinitetalk_0405.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+7_fPy-yBzfk+P2+C0+F35813-36007.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+7_fPy-yBzfk+P2+C0+F35813-36007__p240_175.mp4" }, { "name": "infinitetalk_0406.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0406.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+wvkKTbQWiM8+P0+C1+F84004-84195.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+wvkKTbQWiM8+P0+C1+F84004-84195.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0406.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p268/p268_372.wav", "generated_video": "1_fake/infinitetalk_0406.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+wvkKTbQWiM8+P0+C1+F84004-84195__p268_372.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 406, "real_video": "0_real/infinitetalk_0406.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+wvkKTbQWiM8+P0+C1+F84004-84195.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+wvkKTbQWiM8+P0+C1+F84004-84195__p268_372.mp4" }, { "name": "infinitetalk_0407.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0407.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+z3jjUE6zWEk+P0+C1+F44148-44501.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+z3jjUE6zWEk+P0+C1+F44148-44501.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0407.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p318/p318_289.wav", "generated_video": "1_fake/infinitetalk_0407.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+z3jjUE6zWEk+P0+C1+F44148-44501__p318_289.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 407, "real_video": "0_real/infinitetalk_0407.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+z3jjUE6zWEk+P0+C1+F44148-44501.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+z3jjUE6zWEk+P0+C1+F44148-44501__p318_289.mp4" }, { "name": "infinitetalk_0408.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0408.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+e8dWioL5_oQ+P8+C1+F52219-52575.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+e8dWioL5_oQ+P8+C1+F52219-52575.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0408.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p234/p234_331.wav", "generated_video": "1_fake/infinitetalk_0408.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+e8dWioL5_oQ+P8+C1+F52219-52575__p234_331.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 408, "real_video": "0_real/infinitetalk_0408.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+e8dWioL5_oQ+P8+C1+F52219-52575.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+e8dWioL5_oQ+P8+C1+F52219-52575__p234_331.mp4" }, { "name": "infinitetalk_0409.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0409.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+0GQaq8aywy0+P0+C0+F4881-4991.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+0GQaq8aywy0+P0+C0+F4881-4991.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0409.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p241/p241_222.wav", "generated_video": "1_fake/infinitetalk_0409.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+0GQaq8aywy0+P0+C0+F4881-4991__p241_222.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 409, "real_video": "0_real/infinitetalk_0409.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+0GQaq8aywy0+P0+C0+F4881-4991.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+0GQaq8aywy0+P0+C0+F4881-4991__p241_222.mp4" }, { "name": "infinitetalk_0410.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0410.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+PvDS1-BEjRQ+P0+C1+F7258-7677.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+PvDS1-BEjRQ+P0+C1+F7258-7677.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0410.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p360/p360_393.wav", "generated_video": "1_fake/infinitetalk_0410.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+PvDS1-BEjRQ+P0+C1+F7258-7677__p360_393.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 410, "real_video": "0_real/infinitetalk_0410.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+PvDS1-BEjRQ+P0+C1+F7258-7677.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+PvDS1-BEjRQ+P0+C1+F7258-7677__p360_393.mp4" }, { "name": "infinitetalk_0411.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0411.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+roM_F5RqTSI+P0+C0+F1060-1979.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+roM_F5RqTSI+P0+C0+F1060-1979.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0411.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p274/p274_387.wav", "generated_video": "1_fake/infinitetalk_0411.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+roM_F5RqTSI+P0+C0+F1060-1979__p274_387.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 411, "real_video": "0_real/infinitetalk_0411.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+roM_F5RqTSI+P0+C0+F1060-1979.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+roM_F5RqTSI+P0+C0+F1060-1979__p274_387.mp4" }, { "name": "infinitetalk_0412.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0412.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+ojTV0lbV3kY+P0+C0+F11801-11906.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+ojTV0lbV3kY+P0+C0+F11801-11906.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0412.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p314/p314_067.wav", "generated_video": "1_fake/infinitetalk_0412.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+ojTV0lbV3kY+P0+C0+F11801-11906__p314_067.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 412, "real_video": "0_real/infinitetalk_0412.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+ojTV0lbV3kY+P0+C0+F11801-11906.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+ojTV0lbV3kY+P0+C0+F11801-11906__p314_067.mp4" }, { "name": "infinitetalk_0413.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0413.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+FDPMwEXexUc+P1+C0+F75124-75235.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+FDPMwEXexUc+P1+C0+F75124-75235.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0413.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p266/p266_132.wav", "generated_video": "1_fake/infinitetalk_0413.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+FDPMwEXexUc+P1+C0+F75124-75235__p266_132.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 413, "real_video": "0_real/infinitetalk_0413.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+FDPMwEXexUc+P1+C0+F75124-75235.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+FDPMwEXexUc+P1+C0+F75124-75235__p266_132.mp4" }, { "name": "infinitetalk_0414.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0414.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+w_dSsBuufHk+P1+C2+F14834-15000.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+w_dSsBuufHk+P1+C2+F14834-15000.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0414.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p314/p314_095.wav", "generated_video": "1_fake/infinitetalk_0414.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+w_dSsBuufHk+P1+C2+F14834-15000__p314_095.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 414, "real_video": "0_real/infinitetalk_0414.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+w_dSsBuufHk+P1+C2+F14834-15000.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+w_dSsBuufHk+P1+C2+F14834-15000__p314_095.mp4" }, { "name": "infinitetalk_0415.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0415.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+OpekCi6BGX8+P0+C2+F4704-4925.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+OpekCi6BGX8+P0+C2+F4704-4925.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0415.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p305/p305_069.wav", "generated_video": "1_fake/infinitetalk_0415.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+OpekCi6BGX8+P0+C2+F4704-4925__p305_069.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 415, "real_video": "0_real/infinitetalk_0415.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+OpekCi6BGX8+P0+C2+F4704-4925.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+OpekCi6BGX8+P0+C2+F4704-4925__p305_069.mp4" }, { "name": "infinitetalk_0416.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0416.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+VHWsaXNUll4+P4+C0+F24520-24751.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+VHWsaXNUll4+P4+C0+F24520-24751.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0416.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p246/p246_135.wav", "generated_video": "1_fake/infinitetalk_0416.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+VHWsaXNUll4+P4+C0+F24520-24751__p246_135.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 416, "real_video": "0_real/infinitetalk_0416.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+VHWsaXNUll4+P4+C0+F24520-24751.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+VHWsaXNUll4+P4+C0+F24520-24751__p246_135.mp4" }, { "name": "infinitetalk_0417.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0417.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+3S7MXGCd0tE+P0+C0+F4304-4440.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+3S7MXGCd0tE+P0+C0+F4304-4440.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0417.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p249/p249_258.wav", "generated_video": "1_fake/infinitetalk_0417.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+3S7MXGCd0tE+P0+C0+F4304-4440__p249_258.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 417, "real_video": "0_real/infinitetalk_0417.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+3S7MXGCd0tE+P0+C0+F4304-4440.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+3S7MXGCd0tE+P0+C0+F4304-4440__p249_258.mp4" }, { "name": "infinitetalk_0418.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0418.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+e4zqWh94VQE+P0+C0+F9773-9895.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+e4zqWh94VQE+P0+C0+F9773-9895.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0418.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p233/p233_212.wav", "generated_video": "1_fake/infinitetalk_0418.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+e4zqWh94VQE+P0+C0+F9773-9895__p233_212.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 418, "real_video": "0_real/infinitetalk_0418.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+e4zqWh94VQE+P0+C0+F9773-9895.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+e4zqWh94VQE+P0+C0+F9773-9895__p233_212.mp4" }, { "name": "infinitetalk_0419.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0419.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+itmFZ91PZ_I+P0+C2+F10170-10289.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+itmFZ91PZ_I+P0+C2+F10170-10289.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0419.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p273/p273_154.wav", "generated_video": "1_fake/infinitetalk_0419.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+itmFZ91PZ_I+P0+C2+F10170-10289__p273_154.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 419, "real_video": "0_real/infinitetalk_0419.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+itmFZ91PZ_I+P0+C2+F10170-10289.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+itmFZ91PZ_I+P0+C2+F10170-10289__p273_154.mp4" }, { "name": "infinitetalk_0420.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0420.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+tslCj-aDDlI+P1+C2+F20900-21170.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+tslCj-aDDlI+P1+C2+F20900-21170.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0420.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p300/p300_377.wav", "generated_video": "1_fake/infinitetalk_0420.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+tslCj-aDDlI+P1+C2+F20900-21170__p300_377.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 420, "real_video": "0_real/infinitetalk_0420.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+tslCj-aDDlI+P1+C2+F20900-21170.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+tslCj-aDDlI+P1+C2+F20900-21170__p300_377.mp4" }, { "name": "infinitetalk_0421.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0421.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+yOpi_okNVeE+P0+C1+F53184-53505.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+yOpi_okNVeE+P0+C1+F53184-53505.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0421.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p270/p270_358.wav", "generated_video": "1_fake/infinitetalk_0421.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+yOpi_okNVeE+P0+C1+F53184-53505__p270_358.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 421, "real_video": "0_real/infinitetalk_0421.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+yOpi_okNVeE+P0+C1+F53184-53505.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+yOpi_okNVeE+P0+C1+F53184-53505__p270_358.mp4" }, { "name": "infinitetalk_0422.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0422.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+RGzTaTMRRPw+P0+C2+F16278-16520.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+RGzTaTMRRPw+P0+C2+F16278-16520.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0422.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p232/p232_029.wav", "generated_video": "1_fake/infinitetalk_0422.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+RGzTaTMRRPw+P0+C2+F16278-16520__p232_029.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 422, "real_video": "0_real/infinitetalk_0422.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+RGzTaTMRRPw+P0+C2+F16278-16520.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+RGzTaTMRRPw+P0+C2+F16278-16520__p232_029.mp4" }, { "name": "infinitetalk_0423.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0423.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+tRUQRW16p90+P0+C2+F71556-71841.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+tRUQRW16p90+P0+C2+F71556-71841.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0423.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p284/p284_351.wav", "generated_video": "1_fake/infinitetalk_0423.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+tRUQRW16p90+P0+C2+F71556-71841__p284_351.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 423, "real_video": "0_real/infinitetalk_0423.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+tRUQRW16p90+P0+C2+F71556-71841.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+tRUQRW16p90+P0+C2+F71556-71841__p284_351.mp4" }, { "name": "infinitetalk_0424.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0424.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+XZTInSL9FfQ+P0+C0+F5473-5623.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+XZTInSL9FfQ+P0+C0+F5473-5623.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0424.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p339/p339_424.wav", "generated_video": "1_fake/infinitetalk_0424.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+XZTInSL9FfQ+P0+C0+F5473-5623__p339_424.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 424, "real_video": "0_real/infinitetalk_0424.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+XZTInSL9FfQ+P0+C0+F5473-5623.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+XZTInSL9FfQ+P0+C0+F5473-5623__p339_424.mp4" }, { "name": "infinitetalk_0425.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0425.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+NEWWPJQL3ys+P0+C1+F28459-28685.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+NEWWPJQL3ys+P0+C1+F28459-28685.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0425.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p252/p252_217.wav", "generated_video": "1_fake/infinitetalk_0425.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+NEWWPJQL3ys+P0+C1+F28459-28685__p252_217.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 425, "real_video": "0_real/infinitetalk_0425.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+NEWWPJQL3ys+P0+C1+F28459-28685.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+NEWWPJQL3ys+P0+C1+F28459-28685__p252_217.mp4" }, { "name": "infinitetalk_0426.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0426.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+yI4IwGD52bc+P2+C1+F9532-9696.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+yI4IwGD52bc+P2+C1+F9532-9696.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0426.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p335/p335_220.wav", "generated_video": "1_fake/infinitetalk_0426.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+yI4IwGD52bc+P2+C1+F9532-9696__p335_220.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 426, "real_video": "0_real/infinitetalk_0426.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+yI4IwGD52bc+P2+C1+F9532-9696.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+yI4IwGD52bc+P2+C1+F9532-9696__p335_220.mp4" }, { "name": "infinitetalk_0427.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0427.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+4U6gwo0HE28+P0+C0+F44-188.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+4U6gwo0HE28+P0+C0+F44-188.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0427.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p272/p272_341.wav", "generated_video": "1_fake/infinitetalk_0427.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+4U6gwo0HE28+P0+C0+F44-188__p272_341.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 427, "real_video": "0_real/infinitetalk_0427.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+4U6gwo0HE28+P0+C0+F44-188.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+4U6gwo0HE28+P0+C0+F44-188__p272_341.mp4" }, { "name": "infinitetalk_0428.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0428.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+PdVwbZPthhE+P0+C2+F23375-23486.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+PdVwbZPthhE+P0+C2+F23375-23486.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0428.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p240/p240_007.wav", "generated_video": "1_fake/infinitetalk_0428.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+PdVwbZPthhE+P0+C2+F23375-23486__p240_007.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 428, "real_video": "0_real/infinitetalk_0428.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+PdVwbZPthhE+P0+C2+F23375-23486.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+PdVwbZPthhE+P0+C2+F23375-23486__p240_007.mp4" }, { "name": "infinitetalk_0429.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0429.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+qx6KUhgF_6g+P0+C1+F6184-7170.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+qx6KUhgF_6g+P0+C1+F6184-7170.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0429.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p294/p294_257.wav", "generated_video": "1_fake/infinitetalk_0429.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+qx6KUhgF_6g+P0+C1+F6184-7170__p294_257.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 429, "real_video": "0_real/infinitetalk_0429.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+qx6KUhgF_6g+P0+C1+F6184-7170.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+qx6KUhgF_6g+P0+C1+F6184-7170__p294_257.mp4" }, { "name": "infinitetalk_0430.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0430.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+dirExLhh9mg+P0+C0+F5304-5550.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+dirExLhh9mg+P0+C0+F5304-5550.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0430.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p232/p232_032.wav", "generated_video": "1_fake/infinitetalk_0430.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+dirExLhh9mg+P0+C0+F5304-5550__p232_032.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 430, "real_video": "0_real/infinitetalk_0430.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+dirExLhh9mg+P0+C0+F5304-5550.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+dirExLhh9mg+P0+C0+F5304-5550__p232_032.mp4" }, { "name": "infinitetalk_0431.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0431.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+fBamjRRShRM+P0+C1+F3025-3237.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+fBamjRRShRM+P0+C1+F3025-3237.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0431.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p262/p262_200.wav", "generated_video": "1_fake/infinitetalk_0431.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+fBamjRRShRM+P0+C1+F3025-3237__p262_200.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 431, "real_video": "0_real/infinitetalk_0431.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+fBamjRRShRM+P0+C1+F3025-3237.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+fBamjRRShRM+P0+C1+F3025-3237__p262_200.mp4" }, { "name": "infinitetalk_0432.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0432.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+Y6p3sMerhzs+P0+C1+F54140-54269.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Y6p3sMerhzs+P0+C1+F54140-54269.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0432.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p248/p248_083.wav", "generated_video": "1_fake/infinitetalk_0432.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+Y6p3sMerhzs+P0+C1+F54140-54269__p248_083.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 432, "real_video": "0_real/infinitetalk_0432.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Y6p3sMerhzs+P0+C1+F54140-54269.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+Y6p3sMerhzs+P0+C1+F54140-54269__p248_083.mp4" }, { "name": "infinitetalk_0433.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0433.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+6kMD9eZpduI+P0+C1+F10470-10611.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+6kMD9eZpduI+P0+C1+F10470-10611.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0433.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p262/p262_330.wav", "generated_video": "1_fake/infinitetalk_0433.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+6kMD9eZpduI+P0+C1+F10470-10611__p262_330.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 433, "real_video": "0_real/infinitetalk_0433.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+6kMD9eZpduI+P0+C1+F10470-10611.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+6kMD9eZpduI+P0+C1+F10470-10611__p262_330.mp4" }, { "name": "infinitetalk_0434.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0434.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+rgFmpm8gNt4+P0+C0+F3356-3569.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+rgFmpm8gNt4+P0+C0+F3356-3569.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0434.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p281/p281_433.wav", "generated_video": "1_fake/infinitetalk_0434.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+rgFmpm8gNt4+P0+C0+F3356-3569__p281_433.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 434, "real_video": "0_real/infinitetalk_0434.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+rgFmpm8gNt4+P0+C0+F3356-3569.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+rgFmpm8gNt4+P0+C0+F3356-3569__p281_433.mp4" }, { "name": "infinitetalk_0435.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0435.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+Hc6z8dwqJ0Q+P1+C0+F17656-17938.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Hc6z8dwqJ0Q+P1+C0+F17656-17938.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0435.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p247/p247_227.wav", "generated_video": "1_fake/infinitetalk_0435.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+Hc6z8dwqJ0Q+P1+C0+F17656-17938__p247_227.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 435, "real_video": "0_real/infinitetalk_0435.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Hc6z8dwqJ0Q+P1+C0+F17656-17938.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+Hc6z8dwqJ0Q+P1+C0+F17656-17938__p247_227.mp4" }, { "name": "infinitetalk_0436.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0436.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+tYZmdppbRFg+P1+C2+F101322-101483.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+tYZmdppbRFg+P1+C2+F101322-101483.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0436.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p287/p287_339.wav", "generated_video": "1_fake/infinitetalk_0436.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+tYZmdppbRFg+P1+C2+F101322-101483__p287_339.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 436, "real_video": "0_real/infinitetalk_0436.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+tYZmdppbRFg+P1+C2+F101322-101483.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+tYZmdppbRFg+P1+C2+F101322-101483__p287_339.mp4" }, { "name": "infinitetalk_0437.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0437.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+yevApvkdTq8+P3+C1+F3186-3326.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+yevApvkdTq8+P3+C1+F3186-3326.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0437.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p280/p280_206.wav", "generated_video": "1_fake/infinitetalk_0437.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+yevApvkdTq8+P3+C1+F3186-3326__p280_206.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 437, "real_video": "0_real/infinitetalk_0437.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+yevApvkdTq8+P3+C1+F3186-3326.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+yevApvkdTq8+P3+C1+F3186-3326__p280_206.mp4" }, { "name": "infinitetalk_0438.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0438.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+GBA9zOlDxx4+P0+C1+F9817-9976.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+GBA9zOlDxx4+P0+C1+F9817-9976.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0438.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p329/p329_292.wav", "generated_video": "1_fake/infinitetalk_0438.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+GBA9zOlDxx4+P0+C1+F9817-9976__p329_292.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 438, "real_video": "0_real/infinitetalk_0438.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+GBA9zOlDxx4+P0+C1+F9817-9976.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+GBA9zOlDxx4+P0+C1+F9817-9976__p329_292.mp4" }, { "name": "infinitetalk_0439.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0439.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+63JED-Kw1I0+P0+C0+F4379-4540.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+63JED-Kw1I0+P0+C0+F4379-4540.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0439.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p245/p245_017.wav", "generated_video": "1_fake/infinitetalk_0439.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+63JED-Kw1I0+P0+C0+F4379-4540__p245_017.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 439, "real_video": "0_real/infinitetalk_0439.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+63JED-Kw1I0+P0+C0+F4379-4540.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+63JED-Kw1I0+P0+C0+F4379-4540__p245_017.mp4" }, { "name": "infinitetalk_0440.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0440.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+jI37pgdjBAw+P0+C0+F12060-12181.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+jI37pgdjBAw+P0+C0+F12060-12181.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0440.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p326/p326_296.wav", "generated_video": "1_fake/infinitetalk_0440.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+jI37pgdjBAw+P0+C0+F12060-12181__p326_296.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 440, "real_video": "0_real/infinitetalk_0440.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+jI37pgdjBAw+P0+C0+F12060-12181.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+jI37pgdjBAw+P0+C0+F12060-12181__p326_296.mp4" }, { "name": "infinitetalk_0441.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0441.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+C6EOBCzb3Dk+P0+C2+F4684-4840.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+C6EOBCzb3Dk+P0+C2+F4684-4840.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0441.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p334/p334_015.wav", "generated_video": "1_fake/infinitetalk_0441.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+C6EOBCzb3Dk+P0+C2+F4684-4840__p334_015.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 441, "real_video": "0_real/infinitetalk_0441.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+C6EOBCzb3Dk+P0+C2+F4684-4840.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+C6EOBCzb3Dk+P0+C2+F4684-4840__p334_015.mp4" }, { "name": "infinitetalk_0442.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0442.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+BUK4vwqMPso+P0+C1+F2864-3385.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+BUK4vwqMPso+P0+C1+F2864-3385.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0442.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p241/p241_323.wav", "generated_video": "1_fake/infinitetalk_0442.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+BUK4vwqMPso+P0+C1+F2864-3385__p241_323.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 442, "real_video": "0_real/infinitetalk_0442.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+BUK4vwqMPso+P0+C1+F2864-3385.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+BUK4vwqMPso+P0+C1+F2864-3385__p241_323.mp4" }, { "name": "infinitetalk_0443.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0443.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+dSXRpVQAP4o+P1+C0+F7696-8012.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+dSXRpVQAP4o+P1+C0+F7696-8012.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0443.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p317/p317_174.wav", "generated_video": "1_fake/infinitetalk_0443.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+dSXRpVQAP4o+P1+C0+F7696-8012__p317_174.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 443, "real_video": "0_real/infinitetalk_0443.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+dSXRpVQAP4o+P1+C0+F7696-8012.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+dSXRpVQAP4o+P1+C0+F7696-8012__p317_174.mp4" }, { "name": "infinitetalk_0444.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0444.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+mhk0Ui6dCag+P0+C0+F16989-17102.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+mhk0Ui6dCag+P0+C0+F16989-17102.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0444.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p297/p297_355.wav", "generated_video": "1_fake/infinitetalk_0444.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+mhk0Ui6dCag+P0+C0+F16989-17102__p297_355.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 444, "real_video": "0_real/infinitetalk_0444.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+mhk0Ui6dCag+P0+C0+F16989-17102.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+mhk0Ui6dCag+P0+C0+F16989-17102__p297_355.mp4" }, { "name": "infinitetalk_0445.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0445.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+8fYTgdPprGQ+P0+C2+F25755-25926.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+8fYTgdPprGQ+P0+C2+F25755-25926.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0445.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p300/p300_076.wav", "generated_video": "1_fake/infinitetalk_0445.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+8fYTgdPprGQ+P0+C2+F25755-25926__p300_076.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 445, "real_video": "0_real/infinitetalk_0445.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+8fYTgdPprGQ+P0+C2+F25755-25926.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+8fYTgdPprGQ+P0+C2+F25755-25926__p300_076.mp4" }, { "name": "infinitetalk_0446.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0446.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+ql8LBb9iiYI+P0+C0+F4642-4795.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+ql8LBb9iiYI+P0+C0+F4642-4795.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0446.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p303/p303_095.wav", "generated_video": "1_fake/infinitetalk_0446.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+ql8LBb9iiYI+P0+C0+F4642-4795__p303_095.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 446, "real_video": "0_real/infinitetalk_0446.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+ql8LBb9iiYI+P0+C0+F4642-4795.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+ql8LBb9iiYI+P0+C0+F4642-4795__p303_095.mp4" }, { "name": "infinitetalk_0447.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0447.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+mh8EhstLyt0+P0+C0+F4884-5041.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+mh8EhstLyt0+P0+C0+F4884-5041.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0447.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p228/p228_034.wav", "generated_video": "1_fake/infinitetalk_0447.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+mh8EhstLyt0+P0+C0+F4884-5041__p228_034.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 447, "real_video": "0_real/infinitetalk_0447.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+mh8EhstLyt0+P0+C0+F4884-5041.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+mh8EhstLyt0+P0+C0+F4884-5041__p228_034.mp4" }, { "name": "infinitetalk_0448.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0448.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+rULn21v1a74+P0+C1+F35790-35946.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+rULn21v1a74+P0+C1+F35790-35946.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0448.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p288/p288_017.wav", "generated_video": "1_fake/infinitetalk_0448.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+rULn21v1a74+P0+C1+F35790-35946__p288_017.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 448, "real_video": "0_real/infinitetalk_0448.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+rULn21v1a74+P0+C1+F35790-35946.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+rULn21v1a74+P0+C1+F35790-35946__p288_017.mp4" }, { "name": "infinitetalk_0449.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0449.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+haeb50MFuso+P0+C1+F6397-6578.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+haeb50MFuso+P0+C1+F6397-6578.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0449.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p303/p303_129.wav", "generated_video": "1_fake/infinitetalk_0449.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+haeb50MFuso+P0+C1+F6397-6578__p303_129.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 449, "real_video": "0_real/infinitetalk_0449.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+haeb50MFuso+P0+C1+F6397-6578.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+haeb50MFuso+P0+C1+F6397-6578__p303_129.mp4" }, { "name": "infinitetalk_0450.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0450.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+j37p0e6daF0+P0+C0+F10254-10448.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+j37p0e6daF0+P0+C0+F10254-10448.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0450.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p279/p279_229.wav", "generated_video": "1_fake/infinitetalk_0450.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+j37p0e6daF0+P0+C0+F10254-10448__p279_229.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 450, "real_video": "0_real/infinitetalk_0450.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+j37p0e6daF0+P0+C0+F10254-10448.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+j37p0e6daF0+P0+C0+F10254-10448__p279_229.mp4" }, { "name": "infinitetalk_0451.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0451.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+bKQqr6UNVyw+P0+C1+F7639-7810.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+bKQqr6UNVyw+P0+C1+F7639-7810.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0451.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p228/p228_352.wav", "generated_video": "1_fake/infinitetalk_0451.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+bKQqr6UNVyw+P0+C1+F7639-7810__p228_352.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 451, "real_video": "0_real/infinitetalk_0451.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+bKQqr6UNVyw+P0+C1+F7639-7810.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+bKQqr6UNVyw+P0+C1+F7639-7810__p228_352.mp4" }, { "name": "infinitetalk_0452.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0452.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+ngmDlcgeeU4+P0+C1+F1706-1921.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+ngmDlcgeeU4+P0+C1+F1706-1921.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0452.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p248/p248_208.wav", "generated_video": "1_fake/infinitetalk_0452.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+ngmDlcgeeU4+P0+C1+F1706-1921__p248_208.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 452, "real_video": "0_real/infinitetalk_0452.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+ngmDlcgeeU4+P0+C1+F1706-1921.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+ngmDlcgeeU4+P0+C1+F1706-1921__p248_208.mp4" }, { "name": "infinitetalk_0453.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0453.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+27We1asHYfc+P0+C1+F24005-24161.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+27We1asHYfc+P0+C1+F24005-24161.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0453.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p316/p316_389.wav", "generated_video": "1_fake/infinitetalk_0453.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+27We1asHYfc+P0+C1+F24005-24161__p316_389.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 453, "real_video": "0_real/infinitetalk_0453.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+27We1asHYfc+P0+C1+F24005-24161.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+27We1asHYfc+P0+C1+F24005-24161__p316_389.mp4" }, { "name": "infinitetalk_0454.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0454.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+Zay8fFwgqYo+P2+C0+F5280-5770.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Zay8fFwgqYo+P2+C0+F5280-5770.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0454.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p255/p255_133.wav", "generated_video": "1_fake/infinitetalk_0454.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+Zay8fFwgqYo+P2+C0+F5280-5770__p255_133.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 454, "real_video": "0_real/infinitetalk_0454.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Zay8fFwgqYo+P2+C0+F5280-5770.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+Zay8fFwgqYo+P2+C0+F5280-5770__p255_133.mp4" }, { "name": "infinitetalk_0455.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0455.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+e4mfKNly8qI+P0+C1+F44226-44354.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+e4mfKNly8qI+P0+C1+F44226-44354.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0455.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p313/p313_352.wav", "generated_video": "1_fake/infinitetalk_0455.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+e4mfKNly8qI+P0+C1+F44226-44354__p313_352.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 455, "real_video": "0_real/infinitetalk_0455.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+e4mfKNly8qI+P0+C1+F44226-44354.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+e4mfKNly8qI+P0+C1+F44226-44354__p313_352.mp4" }, { "name": "infinitetalk_0456.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0456.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+xXvOqCuMrYo+P0+C1+F4290-4397.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+xXvOqCuMrYo+P0+C1+F4290-4397.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0456.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p262/p262_073.wav", "generated_video": "1_fake/infinitetalk_0456.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+xXvOqCuMrYo+P0+C1+F4290-4397__p262_073.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 456, "real_video": "0_real/infinitetalk_0456.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+xXvOqCuMrYo+P0+C1+F4290-4397.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+xXvOqCuMrYo+P0+C1+F4290-4397__p262_073.mp4" }, { "name": "infinitetalk_0457.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0457.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+A9nUBMk0mjg+P0+C1+F88584-88691.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+A9nUBMk0mjg+P0+C1+F88584-88691.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0457.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p270/p270_073.wav", "generated_video": "1_fake/infinitetalk_0457.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+A9nUBMk0mjg+P0+C1+F88584-88691__p270_073.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 457, "real_video": "0_real/infinitetalk_0457.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+A9nUBMk0mjg+P0+C1+F88584-88691.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+A9nUBMk0mjg+P0+C1+F88584-88691__p270_073.mp4" }, { "name": "infinitetalk_0458.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0458.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+ozOIt7S83g4+P1+C1+F32424-32586.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+ozOIt7S83g4+P1+C1+F32424-32586.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0458.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p255/p255_317.wav", "generated_video": "1_fake/infinitetalk_0458.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+ozOIt7S83g4+P1+C1+F32424-32586__p255_317.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 458, "real_video": "0_real/infinitetalk_0458.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+ozOIt7S83g4+P1+C1+F32424-32586.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+ozOIt7S83g4+P1+C1+F32424-32586__p255_317.mp4" }, { "name": "infinitetalk_0459.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0459.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+7FS8R10RfYk+P0+C0+F4435-4929.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+7FS8R10RfYk+P0+C0+F4435-4929.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0459.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p312/p312_086.wav", "generated_video": "1_fake/infinitetalk_0459.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+7FS8R10RfYk+P0+C0+F4435-4929__p312_086.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 459, "real_video": "0_real/infinitetalk_0459.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+7FS8R10RfYk+P0+C0+F4435-4929.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+7FS8R10RfYk+P0+C0+F4435-4929__p312_086.mp4" }, { "name": "infinitetalk_0460.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0460.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+08a0356cztg+P1+C1+F6558-6671.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+08a0356cztg+P1+C1+F6558-6671.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0460.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p307/p307_346.wav", "generated_video": "1_fake/infinitetalk_0460.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+08a0356cztg+P1+C1+F6558-6671__p307_346.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 460, "real_video": "0_real/infinitetalk_0460.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+08a0356cztg+P1+C1+F6558-6671.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+08a0356cztg+P1+C1+F6558-6671__p307_346.mp4" }, { "name": "infinitetalk_0461.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0461.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+bU5SLb4mbJY+P7+C0+F30708-30926.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+bU5SLb4mbJY+P7+C0+F30708-30926.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0461.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p264/p264_104.wav", "generated_video": "1_fake/infinitetalk_0461.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+bU5SLb4mbJY+P7+C0+F30708-30926__p264_104.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 461, "real_video": "0_real/infinitetalk_0461.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+bU5SLb4mbJY+P7+C0+F30708-30926.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+bU5SLb4mbJY+P7+C0+F30708-30926__p264_104.mp4" }, { "name": "infinitetalk_0462.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0462.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+OuzvnKr2JR4+P0+C1+F8026-8144.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+OuzvnKr2JR4+P0+C1+F8026-8144.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0462.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p280/p280_007.wav", "generated_video": "1_fake/infinitetalk_0462.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+OuzvnKr2JR4+P0+C1+F8026-8144__p280_007.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 462, "real_video": "0_real/infinitetalk_0462.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+OuzvnKr2JR4+P0+C1+F8026-8144.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+OuzvnKr2JR4+P0+C1+F8026-8144__p280_007.mp4" }, { "name": "infinitetalk_0463.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0463.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+m-gsT3YgTjg+P0+C0+F120-269.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+m-gsT3YgTjg+P0+C0+F120-269.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0463.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p270/p270_393.wav", "generated_video": "1_fake/infinitetalk_0463.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+m-gsT3YgTjg+P0+C0+F120-269__p270_393.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 463, "real_video": "0_real/infinitetalk_0463.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+m-gsT3YgTjg+P0+C0+F120-269.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+m-gsT3YgTjg+P0+C0+F120-269__p270_393.mp4" }, { "name": "infinitetalk_0464.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0464.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+dJduj33CsHk+P0+C0+F1545-1683.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+dJduj33CsHk+P0+C0+F1545-1683.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0464.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p248/p248_201.wav", "generated_video": "1_fake/infinitetalk_0464.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+dJduj33CsHk+P0+C0+F1545-1683__p248_201.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 464, "real_video": "0_real/infinitetalk_0464.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+dJduj33CsHk+P0+C0+F1545-1683.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+dJduj33CsHk+P0+C0+F1545-1683__p248_201.mp4" }, { "name": "infinitetalk_0465.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0465.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+ej0v1iwHwpI+P0+C0+F2370-2927.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+ej0v1iwHwpI+P0+C0+F2370-2927.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0465.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p252/p252_151.wav", "generated_video": "1_fake/infinitetalk_0465.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+ej0v1iwHwpI+P0+C0+F2370-2927__p252_151.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 465, "real_video": "0_real/infinitetalk_0465.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+ej0v1iwHwpI+P0+C0+F2370-2927.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+ej0v1iwHwpI+P0+C0+F2370-2927__p252_151.mp4" }, { "name": "infinitetalk_0466.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0466.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+PDrPDTcVKoI+P0+C0+F17004-17110.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+PDrPDTcVKoI+P0+C0+F17004-17110.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0466.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p293/p293_312.wav", "generated_video": "1_fake/infinitetalk_0466.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+PDrPDTcVKoI+P0+C0+F17004-17110__p293_312.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 466, "real_video": "0_real/infinitetalk_0466.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+PDrPDTcVKoI+P0+C0+F17004-17110.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+PDrPDTcVKoI+P0+C0+F17004-17110__p293_312.mp4" }, { "name": "infinitetalk_0467.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0467.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+JifNt_WDyvY+P0+C1+F11599-11855.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+JifNt_WDyvY+P0+C1+F11599-11855.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0467.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p279/p279_282.wav", "generated_video": "1_fake/infinitetalk_0467.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+JifNt_WDyvY+P0+C1+F11599-11855__p279_282.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 467, "real_video": "0_real/infinitetalk_0467.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+JifNt_WDyvY+P0+C1+F11599-11855.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+JifNt_WDyvY+P0+C1+F11599-11855__p279_282.mp4" }, { "name": "infinitetalk_0468.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0468.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+93Bqgkmoua0+P0+C0+F11001-11111.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+93Bqgkmoua0+P0+C0+F11001-11111.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0468.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p317/p317_066.wav", "generated_video": "1_fake/infinitetalk_0468.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+93Bqgkmoua0+P0+C0+F11001-11111__p317_066.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 468, "real_video": "0_real/infinitetalk_0468.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+93Bqgkmoua0+P0+C0+F11001-11111.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+93Bqgkmoua0+P0+C0+F11001-11111__p317_066.mp4" }, { "name": "infinitetalk_0469.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0469.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+my5BoIp4zBc+P0+C1+F2755-3044.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+my5BoIp4zBc+P0+C1+F2755-3044.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0469.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p278/p278_068.wav", "generated_video": "1_fake/infinitetalk_0469.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+my5BoIp4zBc+P0+C1+F2755-3044__p278_068.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 469, "real_video": "0_real/infinitetalk_0469.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+my5BoIp4zBc+P0+C1+F2755-3044.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+my5BoIp4zBc+P0+C1+F2755-3044__p278_068.mp4" }, { "name": "infinitetalk_0470.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0470.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+wE5MH4Lhznc+P0+C2+F11544-11720.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+wE5MH4Lhznc+P0+C2+F11544-11720.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0470.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p341/p341_143.wav", "generated_video": "1_fake/infinitetalk_0470.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+wE5MH4Lhznc+P0+C2+F11544-11720__p341_143.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 470, "real_video": "0_real/infinitetalk_0470.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+wE5MH4Lhznc+P0+C2+F11544-11720.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+wE5MH4Lhznc+P0+C2+F11544-11720__p341_143.mp4" }, { "name": "infinitetalk_0471.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0471.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+q2x6m2Lctk4+P0+C2+F4205-4405.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+q2x6m2Lctk4+P0+C2+F4205-4405.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0471.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p252/p252_208.wav", "generated_video": "1_fake/infinitetalk_0471.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+q2x6m2Lctk4+P0+C2+F4205-4405__p252_208.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 471, "real_video": "0_real/infinitetalk_0471.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+q2x6m2Lctk4+P0+C2+F4205-4405.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+q2x6m2Lctk4+P0+C2+F4205-4405__p252_208.mp4" }, { "name": "infinitetalk_0472.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0472.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+wcvNTAdUnCU+P0+C2+F5744-5891.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+wcvNTAdUnCU+P0+C2+F5744-5891.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0472.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p330/p330_096.wav", "generated_video": "1_fake/infinitetalk_0472.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+wcvNTAdUnCU+P0+C2+F5744-5891__p330_096.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 472, "real_video": "0_real/infinitetalk_0472.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+wcvNTAdUnCU+P0+C2+F5744-5891.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+wcvNTAdUnCU+P0+C2+F5744-5891__p330_096.mp4" }, { "name": "infinitetalk_0473.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0473.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+OgcJK0QP1Do+P0+C1+F5026-5279.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+OgcJK0QP1Do+P0+C1+F5026-5279.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0473.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p313/p313_067.wav", "generated_video": "1_fake/infinitetalk_0473.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+OgcJK0QP1Do+P0+C1+F5026-5279__p313_067.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 473, "real_video": "0_real/infinitetalk_0473.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+OgcJK0QP1Do+P0+C1+F5026-5279.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+OgcJK0QP1Do+P0+C1+F5026-5279__p313_067.mp4" }, { "name": "infinitetalk_0474.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0474.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+7Jsf8h_0I1M+P0+C0+F6049-6345.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+7Jsf8h_0I1M+P0+C0+F6049-6345.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0474.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p244/p244_148.wav", "generated_video": "1_fake/infinitetalk_0474.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+7Jsf8h_0I1M+P0+C0+F6049-6345__p244_148.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 474, "real_video": "0_real/infinitetalk_0474.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+7Jsf8h_0I1M+P0+C0+F6049-6345.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+7Jsf8h_0I1M+P0+C0+F6049-6345__p244_148.mp4" }, { "name": "infinitetalk_0475.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0475.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+pgerTXtAeF4+P0+C0+F5669-5945.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+pgerTXtAeF4+P0+C0+F5669-5945.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0475.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p268/p268_384.wav", "generated_video": "1_fake/infinitetalk_0475.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+pgerTXtAeF4+P0+C0+F5669-5945__p268_384.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 475, "real_video": "0_real/infinitetalk_0475.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+pgerTXtAeF4+P0+C0+F5669-5945.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+pgerTXtAeF4+P0+C0+F5669-5945__p268_384.mp4" }, { "name": "infinitetalk_0476.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0476.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+ThpcJDToBow+P0+C0+F22525-22633.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+ThpcJDToBow+P0+C0+F22525-22633.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0476.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p267/p267_267.wav", "generated_video": "1_fake/infinitetalk_0476.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+ThpcJDToBow+P0+C0+F22525-22633__p267_267.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 476, "real_video": "0_real/infinitetalk_0476.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+ThpcJDToBow+P0+C0+F22525-22633.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+ThpcJDToBow+P0+C0+F22525-22633__p267_267.mp4" }, { "name": "infinitetalk_0477.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0477.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+SHgAQn9rHeg+P0+C0+F15537-15655.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+SHgAQn9rHeg+P0+C0+F15537-15655.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0477.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p316/p316_055.wav", "generated_video": "1_fake/infinitetalk_0477.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+SHgAQn9rHeg+P0+C0+F15537-15655__p316_055.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 477, "real_video": "0_real/infinitetalk_0477.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+SHgAQn9rHeg+P0+C0+F15537-15655.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+SHgAQn9rHeg+P0+C0+F15537-15655__p316_055.mp4" }, { "name": "infinitetalk_0478.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0478.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+gTksiG6TptA+P1+C0+F32234-32365.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+gTksiG6TptA+P1+C0+F32234-32365.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0478.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p232/p232_347.wav", "generated_video": "1_fake/infinitetalk_0478.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+gTksiG6TptA+P1+C0+F32234-32365__p232_347.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 478, "real_video": "0_real/infinitetalk_0478.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+gTksiG6TptA+P1+C0+F32234-32365.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+gTksiG6TptA+P1+C0+F32234-32365__p232_347.mp4" }, { "name": "infinitetalk_0479.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0479.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+1xZFWoQPqbY+P1+C1+F6549-6680.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+1xZFWoQPqbY+P1+C1+F6549-6680.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0479.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p301/p301_090.wav", "generated_video": "1_fake/infinitetalk_0479.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+1xZFWoQPqbY+P1+C1+F6549-6680__p301_090.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 479, "real_video": "0_real/infinitetalk_0479.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+1xZFWoQPqbY+P1+C1+F6549-6680.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+1xZFWoQPqbY+P1+C1+F6549-6680__p301_090.mp4" }, { "name": "infinitetalk_0480.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0480.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+RCbzeTr4iYk+P0+C0+F25428-25535.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+RCbzeTr4iYk+P0+C0+F25428-25535.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0480.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p297/p297_414.wav", "generated_video": "1_fake/infinitetalk_0480.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+RCbzeTr4iYk+P0+C0+F25428-25535__p297_414.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 480, "real_video": "0_real/infinitetalk_0480.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+RCbzeTr4iYk+P0+C0+F25428-25535.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+RCbzeTr4iYk+P0+C0+F25428-25535__p297_414.mp4" }, { "name": "infinitetalk_0481.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0481.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+vHMMzERZ7DY+P0+C1+F3654-3885.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+vHMMzERZ7DY+P0+C1+F3654-3885.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0481.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p341/p341_387.wav", "generated_video": "1_fake/infinitetalk_0481.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+vHMMzERZ7DY+P0+C1+F3654-3885__p341_387.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 481, "real_video": "0_real/infinitetalk_0481.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+vHMMzERZ7DY+P0+C1+F3654-3885.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+vHMMzERZ7DY+P0+C1+F3654-3885__p341_387.mp4" }, { "name": "infinitetalk_0482.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0482.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+5FnC_cfb05Y+P0+C2+F20034-20170.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+5FnC_cfb05Y+P0+C2+F20034-20170.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0482.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p297/p297_115.wav", "generated_video": "1_fake/infinitetalk_0482.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+5FnC_cfb05Y+P0+C2+F20034-20170__p297_115.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 482, "real_video": "0_real/infinitetalk_0482.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+5FnC_cfb05Y+P0+C2+F20034-20170.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+5FnC_cfb05Y+P0+C2+F20034-20170__p297_115.mp4" }, { "name": "infinitetalk_0483.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0483.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+KZpHl8LnmvA+P1+C2+F4519-4706.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+KZpHl8LnmvA+P1+C2+F4519-4706.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0483.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p299/p299_387.wav", "generated_video": "1_fake/infinitetalk_0483.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+KZpHl8LnmvA+P1+C2+F4519-4706__p299_387.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 483, "real_video": "0_real/infinitetalk_0483.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+KZpHl8LnmvA+P1+C2+F4519-4706.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+KZpHl8LnmvA+P1+C2+F4519-4706__p299_387.mp4" }, { "name": "infinitetalk_0484.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0484.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+18hAMXc8gXA+P0+C0+F18598-18719.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+18hAMXc8gXA+P0+C0+F18598-18719.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0484.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p311/p311_309.wav", "generated_video": "1_fake/infinitetalk_0484.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+18hAMXc8gXA+P0+C0+F18598-18719__p311_309.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 484, "real_video": "0_real/infinitetalk_0484.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+18hAMXc8gXA+P0+C0+F18598-18719.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+18hAMXc8gXA+P0+C0+F18598-18719__p311_309.mp4" }, { "name": "infinitetalk_0485.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0485.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+IGYdGYJCE1E+P0+C0+F2047-2325.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+IGYdGYJCE1E+P0+C0+F2047-2325.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0485.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p335/p335_093.wav", "generated_video": "1_fake/infinitetalk_0485.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+IGYdGYJCE1E+P0+C0+F2047-2325__p335_093.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 485, "real_video": "0_real/infinitetalk_0485.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+IGYdGYJCE1E+P0+C0+F2047-2325.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+IGYdGYJCE1E+P0+C0+F2047-2325__p335_093.mp4" }, { "name": "infinitetalk_0486.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0486.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+iAch5Z9EeGM+P0+C2+F1152-1275.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+iAch5Z9EeGM+P0+C2+F1152-1275.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0486.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p239/p239_479.wav", "generated_video": "1_fake/infinitetalk_0486.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+iAch5Z9EeGM+P0+C2+F1152-1275__p239_479.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 486, "real_video": "0_real/infinitetalk_0486.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+iAch5Z9EeGM+P0+C2+F1152-1275.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+iAch5Z9EeGM+P0+C2+F1152-1275__p239_479.mp4" }, { "name": "infinitetalk_0487.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0487.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+GcJXwYx435A+P0+C2+F6524-6640.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+GcJXwYx435A+P0+C2+F6524-6640.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0487.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p307/p307_049.wav", "generated_video": "1_fake/infinitetalk_0487.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+GcJXwYx435A+P0+C2+F6524-6640__p307_049.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 487, "real_video": "0_real/infinitetalk_0487.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+GcJXwYx435A+P0+C2+F6524-6640.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+GcJXwYx435A+P0+C2+F6524-6640__p307_049.mp4" }, { "name": "infinitetalk_0488.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0488.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+jhpPb003PP4+P0+C1+F5466-5707.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+jhpPb003PP4+P0+C1+F5466-5707.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0488.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p244/p244_224.wav", "generated_video": "1_fake/infinitetalk_0488.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+jhpPb003PP4+P0+C1+F5466-5707__p244_224.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 488, "real_video": "0_real/infinitetalk_0488.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+jhpPb003PP4+P0+C1+F5466-5707.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+jhpPb003PP4+P0+C1+F5466-5707__p244_224.mp4" }, { "name": "infinitetalk_0489.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0489.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+TUS_dUNA25c+P0+C0+F14050-14234.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+TUS_dUNA25c+P0+C0+F14050-14234.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0489.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p230/p230_381.wav", "generated_video": "1_fake/infinitetalk_0489.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+TUS_dUNA25c+P0+C0+F14050-14234__p230_381.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 489, "real_video": "0_real/infinitetalk_0489.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+TUS_dUNA25c+P0+C0+F14050-14234.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+TUS_dUNA25c+P0+C0+F14050-14234__p230_381.mp4" }, { "name": "infinitetalk_0490.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0490.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+WsPpuj9TCQ0+P0+C2+F24188-24421.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+WsPpuj9TCQ0+P0+C2+F24188-24421.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0490.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p254/p254_174.wav", "generated_video": "1_fake/infinitetalk_0490.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+WsPpuj9TCQ0+P0+C2+F24188-24421__p254_174.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 490, "real_video": "0_real/infinitetalk_0490.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+WsPpuj9TCQ0+P0+C2+F24188-24421.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+WsPpuj9TCQ0+P0+C2+F24188-24421__p254_174.mp4" }, { "name": "infinitetalk_0491.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0491.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+ze1TZ1BJeCE+P0+C1+F19807-20208.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+ze1TZ1BJeCE+P0+C1+F19807-20208.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0491.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p284/p284_187.wav", "generated_video": "1_fake/infinitetalk_0491.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+ze1TZ1BJeCE+P0+C1+F19807-20208__p284_187.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 491, "real_video": "0_real/infinitetalk_0491.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+ze1TZ1BJeCE+P0+C1+F19807-20208.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+ze1TZ1BJeCE+P0+C1+F19807-20208__p284_187.mp4" }, { "name": "infinitetalk_0492.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0492.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+a5bnL0_nL_c+P3+C2+F47501-47683.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+a5bnL0_nL_c+P3+C2+F47501-47683.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0492.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p264/p264_171.wav", "generated_video": "1_fake/infinitetalk_0492.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+a5bnL0_nL_c+P3+C2+F47501-47683__p264_171.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 492, "real_video": "0_real/infinitetalk_0492.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+a5bnL0_nL_c+P3+C2+F47501-47683.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+a5bnL0_nL_c+P3+C2+F47501-47683__p264_171.mp4" }, { "name": "infinitetalk_0493.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0493.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+mP54BiCanLA+P0+C2+F24586-25433.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+mP54BiCanLA+P0+C2+F24586-25433.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0493.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p228/p228_166.wav", "generated_video": "1_fake/infinitetalk_0493.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+mP54BiCanLA+P0+C2+F24586-25433__p228_166.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 493, "real_video": "0_real/infinitetalk_0493.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+mP54BiCanLA+P0+C2+F24586-25433.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+mP54BiCanLA+P0+C2+F24586-25433__p228_166.mp4" }, { "name": "infinitetalk_0494.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0494.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+rhueZafQMLk+P0+C1+F1015-1351.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+rhueZafQMLk+P0+C1+F1015-1351.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0494.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p229/p229_273.wav", "generated_video": "1_fake/infinitetalk_0494.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+rhueZafQMLk+P0+C1+F1015-1351__p229_273.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 494, "real_video": "0_real/infinitetalk_0494.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+rhueZafQMLk+P0+C1+F1015-1351.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+rhueZafQMLk+P0+C1+F1015-1351__p229_273.mp4" }, { "name": "infinitetalk_0495.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0495.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+1-sqWG-gZhw+P0+C1+F4724-4845.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+1-sqWG-gZhw+P0+C1+F4724-4845.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0495.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p278/p278_236.wav", "generated_video": "1_fake/infinitetalk_0495.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+1-sqWG-gZhw+P0+C1+F4724-4845__p278_236.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 495, "real_video": "0_real/infinitetalk_0495.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+1-sqWG-gZhw+P0+C1+F4724-4845.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+1-sqWG-gZhw+P0+C1+F4724-4845__p278_236.mp4" }, { "name": "infinitetalk_0496.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0496.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+C0iYswxC-fs+P0+C0+F16264-16380.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+C0iYswxC-fs+P0+C0+F16264-16380.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0496.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p302/p302_199.wav", "generated_video": "1_fake/infinitetalk_0496.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+C0iYswxC-fs+P0+C0+F16264-16380__p302_199.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 496, "real_video": "0_real/infinitetalk_0496.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+C0iYswxC-fs+P0+C0+F16264-16380.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+C0iYswxC-fs+P0+C0+F16264-16380__p302_199.mp4" }, { "name": "infinitetalk_0497.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0497.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+niaszUkI4j8+P0+C2+F7183-7377.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+niaszUkI4j8+P0+C2+F7183-7377.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0497.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p278/p278_260.wav", "generated_video": "1_fake/infinitetalk_0497.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+niaszUkI4j8+P0+C2+F7183-7377__p278_260.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 497, "real_video": "0_real/infinitetalk_0497.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+niaszUkI4j8+P0+C2+F7183-7377.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+niaszUkI4j8+P0+C2+F7183-7377__p278_260.mp4" }, { "name": "infinitetalk_0498.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0498.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+BW2rPGtQCNo+P0+C1+F1356-1716.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+BW2rPGtQCNo+P0+C1+F1356-1716.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0498.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p345/p345_385.wav", "generated_video": "1_fake/infinitetalk_0498.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+BW2rPGtQCNo+P0+C1+F1356-1716__p345_385.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 498, "real_video": "0_real/infinitetalk_0498.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+BW2rPGtQCNo+P0+C1+F1356-1716.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+BW2rPGtQCNo+P0+C1+F1356-1716__p345_385.mp4" }, { "name": "infinitetalk_0499.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0499.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+qfBgRsro258+P0+C1+F3619-3748.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+qfBgRsro258+P0+C1+F3619-3748.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0499.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p318/p318_349.wav", "generated_video": "1_fake/infinitetalk_0499.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+qfBgRsro258+P0+C1+F3619-3748__p318_349.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 499, "real_video": "0_real/infinitetalk_0499.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+qfBgRsro258+P0+C1+F3619-3748.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+qfBgRsro258+P0+C1+F3619-3748__p318_349.mp4" }, { "name": "infinitetalk_0500.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0500.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+GUVxzJ1qiw0+P0+C0+F6054-6222.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+GUVxzJ1qiw0+P0+C0+F6054-6222.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0500.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p295/p295_117.wav", "generated_video": "1_fake/infinitetalk_0500.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+GUVxzJ1qiw0+P0+C0+F6054-6222__p295_117.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 500, "real_video": "0_real/infinitetalk_0500.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+GUVxzJ1qiw0+P0+C0+F6054-6222.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+GUVxzJ1qiw0+P0+C0+F6054-6222__p295_117.mp4" }, { "name": "infinitetalk_0501.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0501.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+hws7qG0QiLI+P0+C0+F2493-2624.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+hws7qG0QiLI+P0+C0+F2493-2624.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0501.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p374/p374_210.wav", "generated_video": "1_fake/infinitetalk_0501.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+hws7qG0QiLI+P0+C0+F2493-2624__p374_210.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 501, "real_video": "0_real/infinitetalk_0501.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+hws7qG0QiLI+P0+C0+F2493-2624.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+hws7qG0QiLI+P0+C0+F2493-2624__p374_210.mp4" }, { "name": "infinitetalk_0502.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0502.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+mhk0Ui6dCag+P0+C1+F16792-16983.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+mhk0Ui6dCag+P0+C1+F16792-16983.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0502.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p334/p334_313.wav", "generated_video": "1_fake/infinitetalk_0502.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+mhk0Ui6dCag+P0+C1+F16792-16983__p334_313.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 502, "real_video": "0_real/infinitetalk_0502.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+mhk0Ui6dCag+P0+C1+F16792-16983.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+mhk0Ui6dCag+P0+C1+F16792-16983__p334_313.mp4" }, { "name": "infinitetalk_0503.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0503.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+JiaVTeqLqJc+P0+C1+F20937-21290.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+JiaVTeqLqJc+P0+C1+F20937-21290.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0503.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p341/p341_222.wav", "generated_video": "1_fake/infinitetalk_0503.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+JiaVTeqLqJc+P0+C1+F20937-21290__p341_222.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 503, "real_video": "0_real/infinitetalk_0503.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+JiaVTeqLqJc+P0+C1+F20937-21290.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+JiaVTeqLqJc+P0+C1+F20937-21290__p341_222.mp4" }, { "name": "infinitetalk_0504.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0504.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+8-jvw2INT7E+P0+C1+F1264-2190.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+8-jvw2INT7E+P0+C1+F1264-2190.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0504.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p273/p273_350.wav", "generated_video": "1_fake/infinitetalk_0504.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+8-jvw2INT7E+P0+C1+F1264-2190__p273_350.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 504, "real_video": "0_real/infinitetalk_0504.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+8-jvw2INT7E+P0+C1+F1264-2190.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+8-jvw2INT7E+P0+C1+F1264-2190__p273_350.mp4" }, { "name": "infinitetalk_0505.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0505.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+Gso56m1VfcI+P3+C0+F3914-4035.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Gso56m1VfcI+P3+C0+F3914-4035.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0505.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p310/p310_300.wav", "generated_video": "1_fake/infinitetalk_0505.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+Gso56m1VfcI+P3+C0+F3914-4035__p310_300.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 505, "real_video": "0_real/infinitetalk_0505.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Gso56m1VfcI+P3+C0+F3914-4035.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+Gso56m1VfcI+P3+C0+F3914-4035__p310_300.mp4" }, { "name": "infinitetalk_0506.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0506.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+Xq_tov1qRZE+P0+C0+F1354-1638.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Xq_tov1qRZE+P0+C0+F1354-1638.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0506.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p302/p302_316.wav", "generated_video": "1_fake/infinitetalk_0506.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+Xq_tov1qRZE+P0+C0+F1354-1638__p302_316.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 506, "real_video": "0_real/infinitetalk_0506.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+Xq_tov1qRZE+P0+C0+F1354-1638.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+Xq_tov1qRZE+P0+C0+F1354-1638__p302_316.mp4" }, { "name": "infinitetalk_0507.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0507.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+bWkevSQn7Xg+P0+C1+F33960-34084.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+bWkevSQn7Xg+P0+C1+F33960-34084.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0507.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p276/p276_402.wav", "generated_video": "1_fake/infinitetalk_0507.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+bWkevSQn7Xg+P0+C1+F33960-34084__p276_402.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 507, "real_video": "0_real/infinitetalk_0507.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+bWkevSQn7Xg+P0+C1+F33960-34084.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+bWkevSQn7Xg+P0+C1+F33960-34084__p276_402.mp4" }, { "name": "infinitetalk_0508.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0508.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+wXR7rwscBdg+P0+C2+F17045-17235.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+wXR7rwscBdg+P0+C2+F17045-17235.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0508.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p298/p298_340.wav", "generated_video": "1_fake/infinitetalk_0508.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+wXR7rwscBdg+P0+C2+F17045-17235__p298_340.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 508, "real_video": "0_real/infinitetalk_0508.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+wXR7rwscBdg+P0+C2+F17045-17235.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+wXR7rwscBdg+P0+C2+F17045-17235__p298_340.mp4" }, { "name": "infinitetalk_0509.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0509.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+1djvVAo_Ef8+P1+C2+F7887-8002.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+1djvVAo_Ef8+P1+C2+F7887-8002.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0509.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p232/p232_328.wav", "generated_video": "1_fake/infinitetalk_0509.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+1djvVAo_Ef8+P1+C2+F7887-8002__p232_328.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 509, "real_video": "0_real/infinitetalk_0509.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+1djvVAo_Ef8+P1+C2+F7887-8002.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+1djvVAo_Ef8+P1+C2+F7887-8002__p232_328.mp4" }, { "name": "infinitetalk_0510.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0510.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+aHaf_vOwtQY+P0+C0+F7195-7808.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+aHaf_vOwtQY+P0+C0+F7195-7808.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0510.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p255/p255_368.wav", "generated_video": "1_fake/infinitetalk_0510.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+aHaf_vOwtQY+P0+C0+F7195-7808__p255_368.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 510, "real_video": "0_real/infinitetalk_0510.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+aHaf_vOwtQY+P0+C0+F7195-7808.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+aHaf_vOwtQY+P0+C0+F7195-7808__p255_368.mp4" }, { "name": "infinitetalk_0511.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0511.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+WaWyaJZ-Akk+P0+C1+F35039-35237.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+WaWyaJZ-Akk+P0+C1+F35039-35237.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0511.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p264/p264_133.wav", "generated_video": "1_fake/infinitetalk_0511.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+WaWyaJZ-Akk+P0+C1+F35039-35237__p264_133.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 511, "real_video": "0_real/infinitetalk_0511.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+WaWyaJZ-Akk+P0+C1+F35039-35237.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+WaWyaJZ-Akk+P0+C1+F35039-35237__p264_133.mp4" }, { "name": "infinitetalk_0512.mp4", "variant": "vfhq_vctk_expand200", "strategy": "identity image + driving audio -> generated video", "identity_image": "2_identity/infinitetalk_0512.png", "identity_image_from": "/archive/lzd/_gen_inputs/expand200_identity_frames/Clip+wejBoDRQukA+P0+C1+F1474-1615.png", "identity_video_from": "/data/lzd/sty/VFHQ_videos/Clip+wejBoDRQukA+P0+C1+F1474-1615.mp4", "source_frame_index": 0, "driving_audio": "3_audio/infinitetalk_0512.wav", "driving_audio_from": "audio muxed in the generated video (the segment actually used)", "driving_audio_full_from": "/data/lzd/sty/VCTK_audio/VCTK-Corpus/wav48/p226/p226_186.wav", "generated_video": "1_fake/infinitetalk_0512.mp4", "generated_from": "/archive/lzd/best30/InfiniteTalk/1_fake/Clip+wejBoDRQukA+P0+C1+F1474-1615__p226_186.mp4", "output_index": 1, "prompt": "A person is talking", "generated_codec": "audio muxed", "note": "InfiniteTalk expand200. VFHQ frame identity and unused VCTK audio, 49-frame cap (~2s). 3_audio is the muxed segment. No 0_real.", "index": 512, "real_video": "0_real/infinitetalk_0512.mp4", "real_video_from": "/data/lzd/sty/VFHQ_videos/Clip+wejBoDRQukA+P0+C1+F1474-1615.mp4", "real_video_role": "identity source video; frame 0 is 2_identity. Not a driving video (this method is audio-driven). Same source clip is hardlinked across methods when identity overlaps.", "real_source_field": "identity_video_from", "name_before": "Clip+wejBoDRQukA+P0+C1+F1474-1615__p226_186.mp4" } ]