{
  "id": "7768a9bb-116e-4c0a-a8a2-c20202d4ccf5",
  "revision": 0,
  "last_node_id": 96,
  "last_link_id": 240,
  "nodes": [
    {
      "id": 1,
      "type": "CLIPLoader",
      "pos": [
        1169.74560546875,
        974.6312255859375
      ],
      "size": [
        318.0999450683594,
        106
      ],
      "flags": {},
      "order": 0,
      "mode": 0,
      "inputs": [
        {
          "name": "clip_name",
          "type": "COMBO",
          "widget": {
            "name": "clip_name"
          }
        },
        {
          "name": "type",
          "type": "COMBO",
          "widget": {
            "name": "type"
          }
        },
        {
          "name": "device",
          "shape": 7,
          "type": "COMBO",
          "widget": {
            "name": "device"
          }
        }
      ],
      "outputs": [
        {
          "label": "CLIP",
          "name": "CLIP",
          "type": "CLIP",
          "links": [
            155
          ]
        }
      ],
      "properties": {
        "cnr_id": "comfy-core",
        "ver": "0.29.0",
        "Node name for S&R": "CLIPLoader",
        "models": [
          {
            "name": "qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors",
            "directory": "text_encoders",
            "url": "https://huggingface.co/Comfy-Org/MiniMax-H3/resolve/main/text_encoders/qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors"
          }
        ],
        "widget_ue_connectable": {}
      },
      "widgets_values": [
        "qwen3vl_32b_minimax_h3_int8_convrot.safetensors",
        "minimax",
        "default"
      ],
      "widgets_values_named": {
        "clip_name": "qwen3vl_32b_minimax_h3_int8_convrot.safetensors",
        "type": "minimax",
        "device": "default"
      }
    },
    {
      "id": 2,
      "type": "VAELoader",
      "pos": [
        1555.703857421875,
        1033.6461181640625
      ],
      "size": [
        284.4346923828125,
        58
      ],
      "flags": {},
      "order": 1,
      "mode": 0,
      "inputs": [
        {
          "name": "vae_name",
          "type": "COMBO",
          "widget": {
            "name": "vae_name"
          }
        }
      ],
      "outputs": [
        {
          "name": "VAE",
          "type": "VAE",
          "links": [
            154,
            157,
            180,
            181
          ]
        }
      ],
      "properties": {
        "cnr_id": "comfy-core",
        "ver": "0.29.0",
        "Node name for S&R": "VAELoader",
        "models": [
          {
            "directory_invalid": true,
            "name": "minimax_h3_audio_vae_fp32.safetensors",
            "directory": "vae",
            "url": "https://huggingface.co/Comfy-Org/MiniMax-H3/resolve/main/vae/minimax_h3_audio_vae_fp32.safetensors"
          }
        ],
        "widget_ue_connectable": {}
      },
      "widgets_values": [
        "minimax_h3_audio_vae_fp32.safetensors"
      ],
      "widgets_values_named": {
        "vae_name": "minimax_h3_audio_vae_fp32.safetensors"
      }
    },
    {
      "id": 6,
      "type": "UniBlockSwap",
      "pos": [
        1917.879638671875,
        566.4727783203125
      ],
      "size": [
        270,
        58
      ],
      "flags": {},
      "order": 20,
      "mode": 0,
      "inputs": [
        {
          "label": "model",
          "name": "model",
          "type": "MODEL",
          "link": 222
        },
        {
          "label": "num_blocks",
          "name": "num_blocks",
          "shape": 7,
          "type": "INT",
          "widget": {
            "name": "num_blocks"
          }
        }
      ],
      "outputs": [
        {
          "label": "model",
          "name": "model",
          "type": "MODEL",
          "links": [
            8
          ]
        }
      ],
      "properties": {
        "cnr_id": "uniblockswap",
        "ver": "a019e82c45185beb5f3a2273a6b47b9f88e001b9",
        "Node name for S&R": "UniBlockSwap",
        "widget_ue_connectable": {}
      },
      "widgets_values": [
        -1
      ],
      "widgets_values_named": {
        "num_blocks": -1
      }
    },
    {
      "id": 35,
      "type": "easy mergeVideosFromPaths",
      "pos": [
        2703.69677734375,
        176.0471649169922
      ],
      "size": [
        270,
        102
      ],
      "flags": {},
      "order": 49,
      "mode": 0,
      "inputs": [
        {
          "label": "paths",
          "name": "paths",
          "type": "STRING",
          "link": 60
        },
        {
          "label": "audio",
          "name": "audio",
          "shape": 7,
          "type": "AUDIO"
        },
        {
          "label": "frame_count",
          "name": "frame_count",
          "type": "INT",
          "widget": {
            "name": "frame_count"
          }
        },
        {
          "label": "audio_mode",
          "name": "audio_mode",
          "type": "COMBO",
          "widget": {
            "name": "audio_mode"
          }
        }
      ],
      "outputs": [
        {
          "label": "VIDEO",
          "name": "VIDEO",
          "type": "VIDEO",
          "links": [
            226
          ]
        }
      ],
      "properties": {
        "cnr_id": "comfyui-easy-media",
        "ver": "1254e0e0afa9afee268027b9540fdc91c888ebc5",
        "Node name for S&R": "easy mergeVideosFromPaths",
        "widget_ue_connectable": {}
      },
      "widgets_values": [
        -1,
        "merge"
      ],
      "widgets_values_named": {
        "frame_count": -1,
        "audio_mode": "merge"
      }
    },
    {
      "id": 41,
      "type": "ComfyMathExpression",
      "pos": [
        2321.67822265625,
        137.3386688232422
      ],
      "size": [
        276.9344787597656,
        128.14254760742188
      ],
      "flags": {
        "collapsed": true
      },
      "order": 24,
      "mode": 0,
      "inputs": [
        {
          "label": "values.a",
          "name": "values.a",
          "type": "FLOAT,INT,BOOLEAN",
          "link": 74
        },
        {
          "label": "b",
          "name": "values.b",
          "shape": 7,
          "type": "FLOAT,INT,BOOLEAN",
          "link": 75
        },
        {
          "label": "c",
          "name": "values.c",
          "shape": 7,
          "type": "FLOAT,INT,BOOLEAN"
        },
        {
          "label": "expression",
          "name": "expression",
          "type": "STRING",
          "widget": {
            "name": "expression"
          }
        }
      ],
      "outputs": [
        {
          "label": "FLOAT",
          "name": "FLOAT",
          "type": "FLOAT"
        },
        {
          "label": "INT",
          "name": "INT",
          "type": "INT",
          "links": [
            76
          ]
        },
        {
          "label": "BOOL",
          "name": "BOOL",
          "type": "BOOLEAN"
        }
      ],
      "properties": {
        "cnr_id": "comfy-core",
        "ver": "0.30.0",
        "Node name for S&R": "ComfyMathExpression",
        "widget_ue_connectable": {}
      },
      "widgets_values": [
        "(a<=-1)*b+(a>-1)*a"
      ],
      "widgets_values_named": {
        "expression": "(a<=-1)*b+(a>-1)*a"
      }
    },
    {
      "id": 31,
      "type": "easy forLoopEnd",
      "pos": [
        2504.874755859375,
        151.6279296875
      ],
      "size": [
        163.08984375,
        166
      ],
      "flags": {},
      "order": 48,
      "mode": 0,
      "inputs": [
        {
          "label": "流",
          "name": "flow",
          "shape": 5,
          "type": "FLOW_CONTROL",
          "link": 53
        },
        {
          "label": "初始值1",
          "name": "initial_value1",
          "shape": 7,
          "type": "*",
          "link": 59
        },
        {
          "label": "初始值2",
          "name": "initial_value2",
          "type": "*"
        },
        {
          "label": "初始值3",
          "name": "initial_value3",
          "type": "*"
        },
        {
          "label": "初始值4",
          "name": "initial_value4",
          "type": "*"
        },
        {
          "label": "初始值5",
          "name": "initial_value5",
          "type": "*"
        },
        {
          "name": "initial_value6",
          "type": "*"
        },
        {
          "name": "initial_value7",
          "type": "*",
          "link": null
        }
      ],
      "outputs": [
        {
          "label": "值1",
          "name": "value1",
          "type": "*",
          "links": [
            60
          ]
        },
        {
          "label": "值2",
          "name": "value2",
          "type": "*"
        },
        {
          "label": "值3",
          "name": "value3",
          "type": "*"
        },
        {
          "label": "值4",
          "name": "value4",
          "type": "*"
        },
        {
          "label": "值5",
          "name": "value5",
          "type": "*"
        },
        {
          "name": "value6",
          "type": "*"
        },
        {
          "name": "value7",
          "type": "*",
          "links": null
        }
      ],
      "properties": {
        "cnr_id": "comfyui-easy-use",
        "ver": "595e0738a9e3f8d0d9c4d875461b2d2c9e7559c7",
        "Node name for S&R": "easy forLoopEnd",
        "widget_ue_connectable": {}
      },
      "color": "#223",
      "bgcolor": "#335"
    },
    {
      "id": 38,
      "type": "ComfyMathExpression",
      "pos": [
        2043.7784423828125,
        98.46134948730469
      ],
      "size": [
        276.9344787597656,
        128
      ],
      "flags": {
        "collapsed": true
      },
      "order": 19,
      "mode": 0,
      "inputs": [
        {
          "label": "values.a",
          "name": "values.a",
          "type": "FLOAT,INT,BOOLEAN",
          "link": 71
        },
        {
          "label": "b",
          "name": "values.b",
          "shape": 7,
          "type": "FLOAT,INT,BOOLEAN",
          "link": 72
        },
        {
          "label": "c",
          "name": "values.c",
          "shape": 7,
          "type": "FLOAT,INT,BOOLEAN"
        },
        {
          "label": "expression",
          "name": "expression",
          "type": "STRING",
          "widget": {
            "name": "expression"
          }
        }
      ],
      "outputs": [
        {
          "label": "FLOAT",
          "name": "FLOAT",
          "type": "FLOAT"
        },
        {
          "label": "INT",
          "name": "INT",
          "type": "INT",
          "links": [
            73
          ]
        },
        {
          "label": "BOOL",
          "name": "BOOL",
          "type": "BOOLEAN"
        }
      ],
      "properties": {
        "cnr_id": "comfy-core",
        "ver": "0.30.0",
        "Node name for S&R": "ComfyMathExpression",
        "widget_ue_connectable": {}
      },
      "widgets_values": [
        "(a<=-1)*b+(a>-1)*1"
      ],
      "widgets_values_named": {
        "expression": "(a<=-1)*b+(a>-1)*1"
      }
    },
    {
      "id": 32,
      "type": "easy forLoopStart",
      "pos": [
        2021.247802734375,
        142.70419311523438
      ],
      "size": [
        270,
        218
      ],
      "flags": {},
      "order": 21,
      "mode": 0,
      "inputs": [
        {
          "label": "初始值1",
          "name": "initial_value1",
          "shape": 7,
          "type": "*"
        },
        {
          "label": "总量",
          "name": "total",
          "type": "INT",
          "widget": {
            "name": "total"
          },
          "link": 73
        },
        {
          "label": "初始值2",
          "name": "initial_value2",
          "type": "*"
        },
        {
          "label": "初始值3",
          "name": "initial_value3",
          "type": "*"
        },
        {
          "label": "初始值4",
          "name": "initial_value4",
          "type": "*"
        },
        {
          "label": "初始值5",
          "name": "initial_value5",
          "type": "*"
        },
        {
          "name": "initial_value6",
          "type": "*"
        },
        {
          "name": "initial_value7",
          "type": "*",
          "link": null
        }
      ],
      "outputs": [
        {
          "label": "流",
          "name": "flow",
          "shape": 5,
          "type": "FLOW_CONTROL",
          "links": [
            53
          ]
        },
        {
          "label": "索引",
          "name": "index",
          "type": "INT",
          "links": [
            66,
            75
          ]
        },
        {
          "label": "值1",
          "name": "value1",
          "type": "*",
          "links": [
            58
          ]
        },
        {
          "label": "值2",
          "name": "value2",
          "type": "*"
        },
        {
          "label": "值3",
          "name": "value3",
          "type": "*"
        },
        {
          "label": "值4",
          "name": "value4",
          "type": "*"
        },
        {
          "label": "值5",
          "name": "value5",
          "type": "*"
        },
        {
          "name": "value6",
          "type": "*"
        },
        {
          "name": "value7",
          "type": "*",
          "links": null
        }
      ],
      "properties": {
        "cnr_id": "comfyui-easy-use",
        "ver": "595e0738a9e3f8d0d9c4d875461b2d2c9e7559c7",
        "Node name for S&R": "easy forLoopStart",
        "widget_ue_connectable": {}
      },
      "widgets_values": [
        1
      ],
      "widgets_values_named": {
        "total": 1
      },
      "color": "#223",
      "bgcolor": "#335"
    },
    {
      "id": 68,
      "type": "KSamplerSelect",
      "pos": [
        3303.470703125,
        518.082763671875
      ],
      "size": [
        210,
        58
      ],
      "flags": {},
      "order": 2,
      "mode": 0,
      "inputs": [
        {
          "name": "sampler_name",
          "type": "COMBO",
          "widget": {
            "name": "sampler_name"
          }
        }
      ],
      "outputs": [
        {
          "name": "SAMPLER",
          "type": "SAMPLER",
          "links": [
            125
          ]
        }
      ],
      "properties": {
        "cnr_id": "comfy-core",
        "ver": "0.31.0",
        "Node name for S&R": "KSamplerSelect",
        "widget_ue_connectable": {
          "sampler_name": true
        }
      },
      "widgets_values": [
        "euler"
      ],
      "widgets_values_named": {
        "sampler_name": "euler"
      }
    },
    {
      "id": 65,
      "type": "BasicGuider",
      "pos": [
        3343.397216796875,
        648.375732421875
      ],
      "size": [
        140,
        46
      ],
      "flags": {},
      "order": 34,
      "mode": 0,
      "inputs": [
        {
          "label": "模型",
          "name": "model",
          "type": "MODEL",
          "link": 179
        },
        {
          "label": "条件",
          "name": "conditioning",
          "type": "CONDITIONING",
          "link": 164
        }
      ],
      "outputs": [
        {
          "label": "引导",
          "name": "GUIDER",
          "type": "GUIDER",
          "links": [
            124
          ]
        }
      ],
      "properties": {
        "cnr_id": "comfy-core",
        "ver": "0.30.0",
        "Node name for S&R": "BasicGuider",
        "widget_ue_connectable": {}
      }
    },
    {
      "id": 7,
      "type": "easy ifElse",
      "pos": [
        1919.099365234375,
        684.06982421875
      ],
      "size": [
        270,
        78
      ],
      "flags": {},
      "order": 22,
      "mode": 0,
      "inputs": [
        {
          "label": "真时",
          "name": "on_true",
          "type": "*",
          "link": 8
        },
        {
          "label": "假时",
          "name": "on_false",
          "type": "*",
          "link": 223
        },
        {
          "label": "布尔",
          "name": "boolean",
          "type": "BOOLEAN",
          "widget": {
            "name": "boolean"
          },
          "link": 10
        }
      ],
      "outputs": [
        {
          "label": "*",
          "name": "*",
          "type": "*",
          "links": [
            46,
            176,
            179
          ]
        }
      ],
      "properties": {
        "cnr_id": "comfyui-easy-use",
        "ver": "595e0738a9e3f8d0d9c4d875461b2d2c9e7559c7",
        "Node name for S&R": "easy ifElse",
        "widget_ue_connectable": {}
      },
      "widgets_values": [
        false
      ],
      "widgets_values_named": {
        "boolean": false
      }
    },
    {
      "id": 66,
      "type": "BasicGuider",
      "pos": [
        2822.558837890625,
        489.5196533203125
      ],
      "size": [
        140,
        46
      ],
      "flags": {},
      "order": 35,
      "mode": 0,
      "inputs": [
        {
          "label": "模型",
          "name": "model",
          "type": "MODEL",
          "link": 177
        },
        {
          "label": "条件",
          "name": "conditioning",
          "type": "CONDITIONING",
          "link": 165
        }
      ],
      "outputs": [
        {
          "label": "引导",
          "name": "GUIDER",
          "type": "GUIDER",
          "links": [
            145
          ]
        }
      ],
      "properties": {
        "cnr_id": "comfy-core",
        "ver": "0.30.0",
        "Node name for S&R": "BasicGuider",
        "widget_ue_connectable": {}
      }
    },
    {
      "id": 34,
      "type": "easy batchAnything",
      "pos": [
        2323.621826171875,
        250.47927856445312
      ],
      "size": [
        140,
        46
      ],
      "flags": {},
      "order": 47,
      "mode": 0,
      "inputs": [
        {
          "label": "任意_1",
          "name": "any_1",
          "type": "*",
          "link": 58
        },
        {
          "label": "任意_2",
          "name": "any_2",
          "type": "*",
          "link": 57
        }
      ],
      "outputs": [
        {
          "label": "batch",
          "name": "batch",
          "type": "*",
          "links": [
            59
          ]
        }
      ],
      "properties": {
        "cnr_id": "comfyui-easy-use",
        "ver": "595e0738a9e3f8d0d9c4d875461b2d2c9e7559c7",
        "Node name for S&R": "easy batchAnything",
        "widget_ue_connectable": {}
      }
    },
    {
      "id": 8,
      "type": "VAELoader",
      "pos": [
        1551.593505859375,
        915.9638671875
      ],
      "size": [
        284.6875,
        58
      ],
      "flags": {},
      "order": 3,
      "mode": 0,
      "inputs": [
        {
          "name": "vae_name",
          "type": "COMBO",
          "widget": {
            "name": "vae_name"
          }
        }
      ],
      "outputs": [
        {
          "name": "VAE",
          "type": "VAE",
          "links": [
            153,
            156,
            182,
            183
          ]
        }
      ],
      "properties": {
        "cnr_id": "comfy-core",
        "ver": "0.29.0",
        "Node name for S&R": "VAELoader",
        "models": [
          {
            "directory_invalid": true,
            "name": "minimax_h3_video_vae_fp16.safetensors",
            "directory": "vae",
            "url": "https://huggingface.co/Comfy-Org/MiniMax-H3/resolve/main/vae/minimax_h3_video_vae_fp16.safetensors"
          }
        ],
        "widget_ue_connectable": {}
      },
      "widgets_values": [
        "minimax_h3_video_vae_fp16.safetensors"
      ],
      "widgets_values_named": {
        "vae_name": "minimax_h3_video_vae_fp16.safetensors"
      }
    },
    {
      "id": 67,
      "type": "BasicScheduler",
      "pos": [
        3564.44775390625,
        668.78857421875
      ],
      "size": [
        210,
        106
      ],
      "flags": {},
      "order": 26,
      "mode": 0,
      "inputs": [
        {
          "name": "model",
          "type": "MODEL",
          "link": 176
        },
        {
          "name": "scheduler",
          "type": "COMBO",
          "widget": {
            "name": "scheduler"
          }
        },
        {
          "name": "steps",
          "type": "INT",
          "widget": {
            "name": "steps"
          }
        },
        {
          "name": "denoise",
          "type": "FLOAT",
          "widget": {
            "name": "denoise"
          }
        }
      ],
      "outputs": [
        {
          "name": "SIGMAS",
          "type": "SIGMAS",
          "links": [
            126
          ]
        }
      ],
      "properties": {
        "cnr_id": "comfy-core",
        "ver": "0.31.0",
        "Node name for S&R": "BasicScheduler",
        "widget_ue_connectable": {
          "scheduler": true,
          "denoise": true,
          "steps": true
        }
      },
      "widgets_values": [
        "beta",
        4,
        0.2
      ],
      "widgets_values_named": {
        "scheduler": "beta",
        "steps": 4,
        "denoise": 0.2
      }
    },
    {
      "id": 58,
      "type": "VAEDecodeAudio",
      "pos": [
        3590.927734375,
        1020.4344482421875
      ],
      "size": [
        158.71542358398438,
        46
      ],
      "flags": {
        "collapsed": false
      },
      "order": 44,
      "mode": 0,
      "inputs": [
        {
          "label": "Latent",
          "name": "samples",
          "type": "LATENT",
          "link": 131
        },
        {
          "label": "VAE",
          "name": "vae",
          "type": "VAE",
          "link": 180
        }
      ],
      "outputs": [
        {
          "label": "音频",
          "name": "AUDIO",
          "type": "AUDIO",
          "links": [
            175,
            204,
            239
          ]
        }
      ],
      "properties": {
        "cnr_id": "comfy-core",
        "ver": "0.29.0",
        "Node name for S&R": "VAEDecodeAudio",
        "widget_ue_connectable": {}
      }
    },
    {
      "id": 72,
      "type": "KSamplerSelect",
      "pos": [
        2296.680419921875,
        453.9066162109375
      ],
      "size": [
        234.31468200683594,
        58
      ],
      "flags": {},
      "order": 4,
      "mode": 0,
      "inputs": [
        {
          "name": "sampler_name",
          "type": "COMBO",
          "widget": {
            "name": "sampler_name"
          }
        }
      ],
      "outputs": [
        {
          "name": "SAMPLER",
          "type": "SAMPLER",
          "links": [
            146
          ]
        }
      ],
      "properties": {
        "cnr_id": "comfy-core",
        "ver": "0.31.0",
        "Node name for S&R": "KSamplerSelect",
        "widget_ue_connectable": {
          "sampler_name": true
        }
      },
      "widgets_values": [
        "er_sde"
      ],
      "widgets_values_named": {
        "sampler_name": "er_sde"
      }
    },
    {
      "id": 37,
      "type": "ComfyMathExpression",
      "pos": [
        2729.76953125,
        439.9652099609375
      ],
      "size": [
        210,
        128
      ],
      "flags": {
        "collapsed": true
      },
      "order": 23,
      "mode": 0,
      "inputs": [
        {
          "label": "values.a",
          "name": "values.a",
          "type": "FLOAT,INT,BOOLEAN",
          "link": 65
        },
        {
          "label": "b",
          "name": "values.b",
          "shape": 7,
          "type": "FLOAT,INT,BOOLEAN",
          "link": 66
        },
        {
          "label": "c",
          "name": "values.c",
          "shape": 7,
          "type": "FLOAT,INT,BOOLEAN"
        },
        {
          "label": "expression",
          "name": "expression",
          "type": "STRING",
          "widget": {
            "name": "expression"
          }
        }
      ],
      "outputs": [
        {
          "label": "FLOAT",
          "name": "FLOAT",
          "type": "FLOAT"
        },
        {
          "label": "INT",
          "name": "INT",
          "type": "INT",
          "links": [
            171
          ]
        },
        {
          "label": "BOOL",
          "name": "BOOL",
          "type": "BOOLEAN"
        }
      ],
      "properties": {
        "cnr_id": "comfy-core",
        "ver": "0.30.0",
        "Node name for S&R": "ComfyMathExpression",
        "widget_ue_connectable": {}
      },
      "widgets_values": [
        "a+b"
      ],
      "widgets_values_named": {
        "expression": "a+b"
      }
    },
    {
      "id": 83,
      "type": "MiniMaxH3HybridLoader",
      "pos": [
        1209.3497314453125,
        521.16552734375
      ],
      "size": [
        303.171875,
        250
      ],
      "flags": {},
      "order": 5,
      "mode": 0,
      "inputs": [
        {
          "name": "base_model",
          "type": "COMBO",
          "widget": {
            "name": "base_model"
          }
        },
        {
          "name": "overlay_model",
          "type": "COMBO",
          "widget": {
            "name": "overlay_model"
          }
        },
        {
          "name": "overlay_preset",
          "type": "COMBO",
          "widget": {
            "name": "overlay_preset"
          }
        },
        {
          "name": "block_range_start",
          "shape": 7,
          "type": "INT",
          "widget": {
            "name": "block_range_start"
          }
        },
        {
          "name": "block_range_end",
          "shape": 7,
          "type": "INT",
          "widget": {
            "name": "block_range_end"
          }
        },
        {
          "name": "final_adaln_from_overlay",
          "shape": 7,
          "type": "BOOLEAN",
          "widget": {
            "name": "final_adaln_from_overlay"
          }
        },
        {
          "name": "custom_overlays",
          "shape": 7,
          "type": "STRING",
          "widget": {
            "name": "custom_overlays"
          }
        },
        {
          "name": "custom_base",
          "shape": 7,
          "type": "STRING",
          "widget": {
            "name": "custom_base"
          }
        },
        {
          "name": "weight_dtype",
          "shape": 7,
          "type": "COMBO",
          "widget": {
            "name": "weight_dtype"
          }
        }
      ],
      "outputs": [
        {
          "name": "model",
          "type": "MODEL",
          "links": [
            200
          ]
        }
      ],
      "properties": {
        "aux_id": "scottmudge/ComfyUI_MinimaxH3HybridLoader",
        "ver": "a44c69b02242e41fbd01e22abe2a492adc853038",
        "Node name for S&R": "MiniMaxH3HybridLoader",
        "widget_ue_connectable": {
          "weight_dtype": true,
          "base_model": true,
          "overlay_preset": true,
          "block_range_end": true,
          "final_adaln_from_overlay": true,
          "custom_overlays": true,
          "custom_base": true,
          "overlay_model": true,
          "block_range_start": true
        }
      },
      "widgets_values": [
        "minimax_h3_fl2va_pruned_int8_convrot.safetensors",
        "minimax_h3_ref2va_pruned_int8_convrot.safetensors",
        "block_range_adaln",
        25,
        49,
        false,
        "",
        "",
        "default"
      ],
      "widgets_values_named": {
        "base_model": "minimax_h3_fl2va_pruned_int8_convrot.safetensors",
        "overlay_model": "minimax_h3_ref2va_pruned_int8_convrot.safetensors",
        "overlay_preset": "block_range_adaln",
        "block_range_start": 25,
        "block_range_end": 49,
        "final_adaln_from_overlay": false,
        "custom_overlays": "",
        "custom_base": "",
        "weight_dtype": "default"
      }
    },
    {
      "id": 21,
      "type": "easy seed",
      "pos": [
        1990.9669189453125,
        1831.77294921875
      ],
      "size": [
        210,
        106
      ],
      "flags": {},
      "order": 6,
      "mode": 0,
      "inputs": [
        {
          "label": "随机种子",
          "name": "seed",
          "type": "INT",
          "widget": {
            "name": "seed"
          }
        }
      ],
      "outputs": [
        {
          "label": "随机种子",
          "name": "seed",
          "type": "INT",
          "links": [
            65
          ]
        }
      ],
      "properties": {
        "cnr_id": "comfyui-easy-use",
        "ver": "595e0738a9e3f8d0d9c4d875461b2d2c9e7559c7",
        "Node name for S&R": "easy seed",
        "widget_ue_connectable": {}
      },
      "widgets_values": [
        675084531497780,
        "fixed",
        null
      ],
      "widgets_values_named": {
        "seed": 675084531497780,
        "control_after_generate": "fixed",
        "🎲 Manual Random Seed": null
      }
    },
    {
      "id": 39,
      "type": "PrimitiveInt",
      "pos": [
        1983.380126953125,
        1669.67626953125
      ],
      "size": [
        224.23788452148438,
        82.27592468261719
      ],
      "flags": {},
      "order": 7,
      "mode": 0,
      "inputs": [
        {
          "label": "value",
          "name": "value",
          "type": "INT",
          "widget": {
            "name": "value"
          }
        }
      ],
      "outputs": [
        {
          "label": "INT",
          "name": "INT",
          "type": "INT",
          "links": [
            71,
            74
          ]
        }
      ],
      "title": "指定跑某索引片段（-1:全部）",
      "properties": {
        "cnr_id": "comfy-core",
        "ver": "0.30.0",
        "Node name for S&R": "PrimitiveInt",
        "widget_ue_connectable": {}
      },
      "widgets_values": [
        -1,
        "fixed"
      ],
      "widgets_values_named": {
        "value": -1,
        "fixed": "fixed"
      }
    },
    {
      "id": 85,
      "type": "PreviewAny",
      "pos": [
        2848.1171875,
        2172.07080078125
      ],
      "size": [
        280.9091491699219,
        527.2858276367188
      ],
      "flags": {},
      "order": 31,
      "mode": 0,
      "inputs": [
        {
          "name": "source",
          "type": "*",
          "link": 209
        }
      ],
      "outputs": [
        {
          "name": "STRING",
          "type": "STRING",
          "links": [
            211
          ]
        }
      ],
      "properties": {
        "cnr_id": "comfy-core",
        "ver": "0.32.0",
        "Node name for S&R": "PreviewAny",
        "widget_ue_connectable": {}
      },
      "widgets_values": [],
      "widgets_values_named": {}
    },
    {
      "id": 88,
      "type": "llama_cpp_model_loader",
      "pos": [
        2265.22900390625,
        2101.006591796875
      ],
      "size": [
        257.33758544921875,
        226
      ],
      "flags": {},
      "order": 8,
      "mode": 0,
      "inputs": [
        {
          "name": "model",
          "type": "COMBO",
          "widget": {
            "name": "model"
          }
        },
        {
          "name": "mmproj",
          "type": "COMBO",
          "widget": {
            "name": "mmproj"
          }
        },
        {
          "name": "chat_handler",
          "type": "COMBO",
          "widget": {
            "name": "chat_handler"
          }
        },
        {
          "name": "n_ctx",
          "type": "INT",
          "widget": {
            "name": "n_ctx"
          }
        },
        {
          "name": "vram_limit",
          "type": "INT",
          "widget": {
            "name": "vram_limit"
          }
        },
        {
          "name": "image_min_tokens",
          "type": "INT",
          "widget": {
            "name": "image_min_tokens"
          }
        },
        {
          "name": "image_max_tokens",
          "type": "INT",
          "widget": {
            "name": "image_max_tokens"
          }
        },
        {
          "name": "load_mtp",
          "type": "BOOLEAN",
          "widget": {
            "name": "load_mtp"
          }
        }
      ],
      "outputs": [
        {
          "name": "llama_model",
          "type": "LLAMACPPMODEL",
          "links": [
            210
          ]
        }
      ],
      "properties": {
        "cnr_id": "llama-cpp_vllm",
        "ver": "8b7c18bc437ae414e73d31723629b01f66a4ade4",
        "Node name for S&R": "llama_cpp_model_loader",
        "widget_ue_connectable": {
          "image_min_tokens": true,
          "load_mtp": true,
          "vram_limit": true,
          "mmproj": true,
          "model": true,
          "n_ctx": true,
          "chat_handler": true,
          "image_max_tokens": true
        },
        "aux_id": "yolain/ComfyUI-llama-cpp_vlm"
      },
      "widgets_values": [
        "Qwen3.8-27B-IQ4_XS.gguf",
        "Qwen3.8-27B-mmproj-BF16.gguf",
        "None",
        16384,
        -1,
        0,
        0,
        false
      ],
      "widgets_values_named": {
        "model": "Qwen3.8-27B-IQ4_XS.gguf",
        "mmproj": "Qwen3.8-27B-mmproj-BF16.gguf",
        "chat_handler": "None",
        "n_ctx": 16384,
        "vram_limit": -1,
        "image_min_tokens": 0,
        "image_max_tokens": 0,
        "load_mtp": false
      }
    },
    {
      "id": 86,
      "type": "llama_cpp_instruct_adv",
      "pos": [
        2545.2802734375,
        2191.163818359375
      ],
      "size": [
        268.19586181640625,
        518.0087280273438
      ],
      "flags": {},
      "order": 29,
      "mode": 0,
      "inputs": [
        {
          "name": "llama_model",
          "type": "LLAMACPPMODEL",
          "link": 210
        },
        {
          "name": "parameters",
          "shape": 7,
          "type": "LLAMACPPARAMS"
        },
        {
          "name": "images",
          "shape": 7,
          "type": "IMAGE",
          "link": 214
        },
        {
          "name": "queue_handler",
          "shape": 7,
          "type": "*"
        },
        {
          "name": "preset_prompt",
          "type": "COMBO",
          "widget": {
            "name": "preset_prompt"
          }
        },
        {
          "name": "custom_prompt",
          "type": "STRING",
          "widget": {
            "name": "custom_prompt"
          },
          "link": 212
        },
        {
          "name": "system_prompt",
          "type": "STRING",
          "widget": {
            "name": "system_prompt"
          },
          "link": 213
        },
        {
          "name": "inference_mode",
          "type": "COMBO",
          "widget": {
            "name": "inference_mode"
          }
        },
        {
          "name": "max_frames",
          "type": "INT",
          "widget": {
            "name": "max_frames"
          }
        },
        {
          "name": "max_size",
          "type": "INT",
          "widget": {
            "name": "max_size"
          }
        },
        {
          "name": "seed",
          "type": "INT",
          "widget": {
            "name": "seed"
          }
        },
        {
          "name": "force_offload",
          "type": "BOOLEAN",
          "widget": {
            "name": "force_offload"
          }
        },
        {
          "name": "save_states",
          "type": "BOOLEAN",
          "widget": {
            "name": "save_states"
          }
        }
      ],
      "outputs": [
        {
          "name": "output",
          "type": "STRING",
          "links": [
            208
          ]
        },
        {
          "name": "output_list",
          "shape": 6,
          "type": "STRING"
        },
        {
          "name": "state_uid",
          "type": "INT"
        }
      ],
      "properties": {
        "cnr_id": "llama-cpp_vllm",
        "ver": "8b7c18bc437ae414e73d31723629b01f66a4ade4",
        "Node name for S&R": "llama_cpp_instruct_adv",
        "widget_ue_connectable": {}
      },
      "widgets_values": [
        "Normal - Describe",
        "",
        "",
        "images",
        24,
        512,
        42,
        "fixed",
        true,
        false
      ],
      "widgets_values_named": {
        "preset_prompt": "Normal - Describe",
        "custom_prompt": "",
        "system_prompt": "",
        "inference_mode": "images",
        "max_frames": 24,
        "max_size": 512,
        "seed": 42,
        "control_after_generate": "fixed",
        "force_offload": true,
        "save_states": false
      }
    },
    {
      "id": 20,
      "type": "PrimitiveBoolean",
      "pos": [
        1992.5484619140625,
        1996.7518310546875
      ],
      "size": [
        210,
        58.972347259521484
      ],
      "flags": {},
      "order": 9,
      "mode": 0,
      "inputs": [
        {
          "label": "value",
          "name": "value",
          "type": "BOOLEAN",
          "widget": {
            "name": "value"
          }
        }
      ],
      "outputs": [
        {
          "label": "BOOLEAN",
          "name": "BOOLEAN",
          "type": "BOOLEAN",
          "links": [
            216
          ]
        }
      ],
      "title": "Using Prompt Enhance",
      "properties": {
        "cnr_id": "comfy-core",
        "ver": "0.30.0",
        "Node name for S&R": "PrimitiveBoolean",
        "widget_ue_connectable": {}
      },
      "widgets_values": [
        false
      ],
      "widgets_values_named": {
        "value": false
      }
    },
    {
      "id": 14,
      "type": "PrimitiveBoolean",
      "pos": [
        1994.580810546875,
        2104.218505859375
      ],
      "size": [
        210,
        58.972347259521484
      ],
      "flags": {},
      "order": 10,
      "mode": 0,
      "inputs": [
        {
          "label": "value",
          "name": "value",
          "type": "BOOLEAN",
          "widget": {
            "name": "value"
          }
        }
      ],
      "outputs": [
        {
          "label": "BOOLEAN",
          "name": "BOOLEAN",
          "type": "BOOLEAN",
          "links": [
            10
          ]
        }
      ],
      "title": "Using low vram improve",
      "properties": {
        "cnr_id": "comfy-core",
        "ver": "0.30.0",
        "Node name for S&R": "PrimitiveBoolean",
        "widget_ue_connectable": {}
      },
      "widgets_values": [
        false
      ],
      "widgets_values_named": {
        "value": false
      }
    },
    {
      "id": 26,
      "type": "LoraLoaderModelOnly",
      "pos": [
        1922.348388671875,
        956.6793212890625
      ],
      "size": [
        270,
        82
      ],
      "flags": {},
      "order": 25,
      "mode": 0,
      "inputs": [
        {
          "label": "模型",
          "name": "model",
          "type": "MODEL",
          "link": 46
        },
        {
          "name": "lora_name",
          "type": "COMBO",
          "widget": {
            "name": "lora_name"
          }
        },
        {
          "name": "strength_model",
          "type": "FLOAT",
          "widget": {
            "name": "strength_model"
          }
        }
      ],
      "outputs": [
        {
          "label": "模型",
          "name": "MODEL",
          "type": "MODEL",
          "links": [
            177
          ]
        }
      ],
      "properties": {
        "cnr_id": "comfy-core",
        "ver": "0.30.0",
        "Node name for S&R": "LoraLoaderModelOnly",
        "widget_ue_connectable": {}
      },
      "widgets_values": [
        "minimax_h3_ref2v_turbo_4step_v0.1_comfyui_bf16.safetensors",
        1.0000000000000002
      ],
      "widgets_values_named": {
        "lora_name": "minimax_h3_ref2v_turbo_4step_v0.1_comfyui_bf16.safetensors",
        "strength_model": 1.0000000000000002
      }
    },
    {
      "id": 70,
      "type": "RandomNoise",
      "pos": [
        2739.798095703125,
        603.7977294921875
      ],
      "size": [
        210,
        82
      ],
      "flags": {},
      "order": 27,
      "mode": 0,
      "inputs": [
        {
          "name": "noise_seed",
          "type": "INT",
          "widget": {
            "name": "noise_seed"
          },
          "link": 171
        }
      ],
      "outputs": [
        {
          "label": "噪波生成",
          "name": "NOISE",
          "type": "NOISE",
          "links": [
            123,
            144
          ]
        }
      ],
      "properties": {
        "cnr_id": "comfy-core",
        "ver": "0.30.0",
        "Node name for S&R": "RandomNoise",
        "widget_ue_connectable": {}
      },
      "widgets_values": [
        0,
        "fixed"
      ],
      "widgets_values_named": {
        "noise_seed": 0,
        "control_after_generate": "fixed"
      }
    },
    {
      "id": 63,
      "type": "VAEDecode",
      "pos": [
        2849.03564453125,
        868.4815673828125
      ],
      "size": [
        140,
        46
      ],
      "flags": {
        "collapsed": false
      },
      "order": 37,
      "mode": 0,
      "inputs": [
        {
          "label": "Latent",
          "name": "samples",
          "type": "LATENT",
          "link": 139
        },
        {
          "label": "VAE",
          "name": "vae",
          "type": "VAE",
          "link": 182
        }
      ],
      "outputs": [
        {
          "label": "图像",
          "name": "IMAGE",
          "type": "IMAGE",
          "links": [
            138
          ]
        }
      ],
      "properties": {
        "cnr_id": "comfy-core",
        "ver": "0.29.0",
        "Node name for S&R": "VAEDecode",
        "widget_ue_connectable": {}
      }
    },
    {
      "id": 57,
      "type": "PT_H3ConcatAVLatent",
      "pos": [
        3071.600830078125,
        1056.0926513671875
      ],
      "size": [
        199.7447265625,
        46
      ],
      "flags": {},
      "order": 42,
      "mode": 0,
      "inputs": [
        {
          "name": "video_latent",
          "type": "LATENT",
          "link": 129
        },
        {
          "name": "audio_latent",
          "type": "LATENT",
          "link": 130
        }
      ],
      "outputs": [
        {
          "name": "LATENT",
          "type": "LATENT",
          "links": [
            190
          ]
        }
      ],
      "properties": {
        "aux_id": "ptmaster/ComfyUI-PT_H3ConcatAVLatent",
        "ver": "2387e990a025f9b2f98ccac4ff1c466086e34a6e",
        "Node name for S&R": "PT_H3ConcatAVLatent",
        "widget_ue_connectable": {}
      }
    },
    {
      "id": 55,
      "type": "SamplerCustomAdvanced",
      "pos": [
        3322.264404296875,
        758.3532104492188
      ],
      "size": [
        219.67437744140625,
        338
      ],
      "flags": {},
      "order": 43,
      "mode": 0,
      "inputs": [
        {
          "label": "噪波生成",
          "name": "noise",
          "type": "NOISE",
          "link": 123
        },
        {
          "label": "引导",
          "name": "guider",
          "type": "GUIDER",
          "link": 124
        },
        {
          "label": "采样器",
          "name": "sampler",
          "type": "SAMPLER",
          "link": 125
        },
        {
          "label": "Sigmas",
          "name": "sigmas",
          "type": "SIGMAS",
          "link": 126
        },
        {
          "label": "Latent",
          "name": "latent_image",
          "type": "LATENT",
          "link": 190
        }
      ],
      "outputs": [
        {
          "label": "输出",
          "name": "output",
          "type": "LATENT"
        },
        {
          "label": "降噪输出",
          "name": "denoised_output",
          "type": "LATENT",
          "links": [
            131,
            136
          ]
        }
      ],
      "properties": {
        "cnr_id": "comfy-core",
        "ver": "0.30.0",
        "Node name for S&R": "SamplerCustomAdvanced",
        "widget_ue_connectable": {}
      }
    },
    {
      "id": 56,
      "type": "VAEEncodeAudio",
      "pos": [
        3101.974365234375,
        949.774658203125
      ],
      "size": [
        157.94296264648438,
        46
      ],
      "flags": {},
      "order": 40,
      "mode": 0,
      "inputs": [
        {
          "name": "audio",
          "type": "AUDIO",
          "link": 128
        },
        {
          "name": "vae",
          "type": "VAE",
          "link": 154
        }
      ],
      "outputs": [
        {
          "name": "LATENT",
          "type": "LATENT",
          "links": [
            130
          ]
        }
      ],
      "properties": {
        "cnr_id": "comfy-core",
        "ver": "0.30.0",
        "Node name for S&R": "VAEEncodeAudio",
        "widget_ue_connectable": {}
      }
    },
    {
      "id": 59,
      "type": "VAEEncode",
      "pos": [
        3111.3369140625,
        861.0856323242188
      ],
      "size": [
        140,
        46
      ],
      "flags": {},
      "order": 41,
      "mode": 0,
      "inputs": [
        {
          "name": "pixels",
          "type": "IMAGE",
          "link": 132
        },
        {
          "name": "vae",
          "type": "VAE",
          "link": 153
        }
      ],
      "outputs": [
        {
          "name": "LATENT",
          "type": "LATENT",
          "links": [
            129
          ]
        }
      ],
      "properties": {
        "cnr_id": "comfy-core",
        "ver": "0.30.0",
        "Node name for S&R": "VAEEncode",
        "widget_ue_connectable": {}
      }
    },
    {
      "id": 9,
      "type": "PathchSageAttentionKJ",
      "pos": [
        1563.7113037109375,
        687.6027221679688
      ],
      "size": [
        270,
        82
      ],
      "flags": {},
      "order": 15,
      "mode": 0,
      "inputs": [
        {
          "label": "模型",
          "name": "model",
          "type": "MODEL",
          "link": 200
        },
        {
          "label": "SAGE注意力",
          "name": "sage_attention",
          "type": "COMBO",
          "widget": {
            "name": "sage_attention"
          }
        },
        {
          "label": "allow_compile",
          "name": "allow_compile",
          "shape": 7,
          "type": "BOOLEAN",
          "widget": {
            "name": "allow_compile"
          }
        }
      ],
      "outputs": [
        {
          "label": "MODEL",
          "name": "MODEL",
          "type": "MODEL",
          "links": [
            224
          ]
        }
      ],
      "properties": {
        "cnr_id": "comfyui-kjnodes",
        "ver": "d40331fc90608257bce1af743263bb74b14c1dce",
        "Node name for S&R": "PathchSageAttentionKJ",
        "widget_ue_connectable": {}
      },
      "widgets_values": [
        "auto",
        false
      ],
      "widgets_values_named": {
        "sage_attention": "auto",
        "allow_compile": false
      }
    },
    {
      "id": 93,
      "type": "MiniMaxH3MemoryEfficientSageAttentionPatch",
      "pos": [
        1501.8779296875,
        594.5887451171875
      ],
      "size": [
        357.1148376464844,
        26
      ],
      "flags": {},
      "order": 17,
      "mode": 0,
      "inputs": [
        {
          "name": "model",
          "type": "MODEL",
          "link": 224
        }
      ],
      "outputs": [
        {
          "name": "model",
          "type": "MODEL",
          "links": [
            222,
            223
          ]
        }
      ],
      "properties": {
        "cnr_id": "comfyui-kjnodes",
        "ver": "3f20054214fec9f9234fd3841ae6f1e4287948f6",
        "Node name for S&R": "MiniMaxH3MemoryEfficientSageAttentionPatch",
        "widget_ue_connectable": {}
      }
    },
    {
      "id": 61,
      "type": "VAEDecode",
      "pos": [
        3581.356201171875,
        834.6806030273438
      ],
      "size": [
        142.0826873779297,
        46
      ],
      "flags": {
        "collapsed": false
      },
      "order": 45,
      "mode": 0,
      "inputs": [
        {
          "label": "Latent",
          "name": "samples",
          "type": "LATENT",
          "link": 136
        },
        {
          "label": "VAE",
          "name": "vae",
          "type": "VAE",
          "link": 183
        }
      ],
      "outputs": [
        {
          "label": "图像",
          "name": "IMAGE",
          "type": "IMAGE",
          "links": [
            238
          ]
        }
      ],
      "properties": {
        "cnr_id": "comfy-core",
        "ver": "0.29.0",
        "Node name for S&R": "VAEDecode",
        "widget_ue_connectable": {}
      }
    },
    {
      "id": 94,
      "type": "SaveVideo",
      "pos": [
        2329.276611328125,
        1294.8778076171875
      ],
      "size": [
        210,
        126
      ],
      "flags": {},
      "order": 50,
      "mode": 0,
      "inputs": [
        {
          "name": "video",
          "type": "VIDEO",
          "link": 226
        },
        {
          "name": "filename_prefix",
          "type": "STRING",
          "widget": {
            "name": "filename_prefix"
          }
        },
        {
          "name": "format",
          "type": "COMBO",
          "widget": {
            "name": "format"
          }
        },
        {
          "name": "codec",
          "type": "COMFY_DYNAMICCOMBO_V3",
          "widget": {
            "name": "codec"
          }
        }
      ],
      "outputs": [
        {
          "name": "video",
          "type": "VIDEO"
        },
        {
          "name": "video",
          "type": "VIDEO"
        }
      ],
      "properties": {
        "cnr_id": "comfy-core",
        "ver": "0.33.0",
        "Node name for S&R": "SaveVideo",
        "widget_ue_connectable": {}
      },
      "widgets_values": [
        "video/ComfyUI",
        "auto",
        "auto"
      ],
      "widgets_values_named": {
        "filename_prefix": "video/ComfyUI",
        "format": "auto",
        "codec": "auto"
      }
    },
    {
      "id": 28,
      "type": "easy multiTrackEditor",
      "pos": [
        1004.4900512695312,
        1248.292724609375
      ],
      "size": [
        961.2371826171875,
        906.306640625
      ],
      "flags": {},
      "order": 11,
      "mode": 0,
      "inputs": [
        {
          "label": "prompt_override",
          "name": "prompt_override",
          "shape": 7,
          "type": "*"
        },
        {
          "label": "image",
          "name": "image",
          "shape": 7,
          "type": "IMAGE"
        },
        {
          "label": "audio",
          "name": "audio",
          "shape": 7,
          "type": "AUDIO"
        },
        {
          "label": "video",
          "name": "video",
          "shape": 7,
          "type": "VIDEO"
        },
        {
          "label": "resolution",
          "name": "resolution",
          "type": "COMFY_DYNAMICCOMBO_V3",
          "widget": {
            "name": "resolution"
          }
        },
        {
          "label": "format",
          "name": "format",
          "type": "COMBO",
          "widget": {
            "name": "format"
          }
        },
        {
          "label": "track_data",
          "name": "track_data",
          "type": "TRACK_DATA",
          "widget": {
            "name": "track_data"
          }
        }
      ],
      "outputs": [
        {
          "label": "TRACKS_INFO",
          "name": "TRACKS_INFO",
          "type": "TRACKS_INFO",
          "links": [
            16,
            26
          ]
        },
        {
          "label": "IMAGES",
          "name": "IMAGES",
          "shape": 6,
          "type": "IMAGE",
          "links": [
            27
          ]
        },
        {
          "label": "AUDIO",
          "name": "AUDIO",
          "shape": 6,
          "type": "AUDIO",
          "links": [
            28
          ]
        },
        {
          "label": "VIDEO",
          "name": "VIDEO",
          "shape": 6,
          "type": "VIDEO",
          "links": [
            29
          ]
        }
      ],
      "properties": {
        "easyMediaTimelineHeight": 906.306640625,
        "cnr_id": "comfyui-easy-media",
        "ver": "74c6f8b64a86bd280061c20ed42cb067322e1822",
        "Node name for S&R": "easy multiTrackEditor",
        "easyMediaTimelineSize": [
          961.2371826171875,
          906.306640625
        ],
        "widget_ue_connectable": {}
      },
      "widgets_values": [
        "width x height (megapixels)",
        "16:9 (Widescreen)",
        0.5,
        "MiniMax",
        "{\"muted\":false,\"volume_db\":0,\"tracks\":[{\"id\":\"mqqqqfz1-d0bnh7h9p\",\"name\":\"Task 0\",\"type\":\"task\",\"task_mode\":\"default\",\"color\":\"var(--multitrack-task-bg)\",\"muted\":false,\"solo\":false,\"volume_db\":0,\"locked\":false,\"segments\":[{\"id\":\"mrhal5il-gerx98rwl\",\"start_frame\":0,\"end_frame\":240,\"color\":\"var(--multitrack-task-bg)\",\"content\":{\"media_type\":\"none\",\"task_mode\":\"ref\",\"muted\":false,\"volume_db\":0,\"images\":[{\"id\":\"40702ebc-aec9-417e-b1ce-aa2a027450f6\",\"source_type\":\"url\",\"url\":\"https://rh-images-switch.xiaoyaoyou.com/input/9ab7c7b51646c0e24b794ea55a97698fc8bf18105a13ca3b9f71043dfaacd362.png\",\"file_name\":\"9ab7c7b51646c0e24b794ea55a97698fc8bf18105a13ca3b9f71043dfaacd362.png\"},{\"id\":\"6381c90f-f4c7-47b4-9f49-d8b0d34aa2d2\",\"source_type\":\"url\",\"url\":\"https://rh-images-switch.xiaoyaoyou.com/input/b3279bfb016d361f03b12417fa5a5a0d39070c7d8fb961c34094e801212f5b67.png\",\"file_name\":\"b3279bfb016d361f03b12417fa5a5a0d39070c7d8fb961c34094e801212f5b67.png\"},{\"id\":\"79e40c24-18fe-4ce6-a7a9-a77dae74c5eb\",\"source_type\":\"url\",\"url\":\"https://rh-images-switch.xiaoyaoyou.com/input/0b4b5a893fe445a6843b361715042597ffe7926c7fd8ab5c16bca499347c4b2c.png\",\"file_name\":\"0b4b5a893fe445a6843b361715042597ffe7926c7fd8ab5c16bca499347c4b2c.png\"}],\"user_prompt\":\"subject_definitions:\\n<Subject 1> is the magical book from <Picture 1>, an old leather-bound book with brown textured leather cover, gold embossed title \\\"SUPERI PROMPT\\\", yellowed aged pages with mysterious symbols, and soft golden magical light emanating from between the pages with floating golden particles.\\n<Subject 2> is the girl Xiao Xiao whose appearance comes from <Picture 2>, with long straight black hair with bangs, cream-colored lace blouse, brown pleated mini skirt, white knee-high socks, small brown leather shoes, big shiny black eyes, rosy cheeks, Q-version 2.5 head tall proportions.\\n<Subject 3> is the kitchen countertop environment from <Picture 3>, seen from a tiny miniature perspective, with warm daylight, tile grout lines like canyons, and a wine cork like a boulder.\\n<Picture 1> is the magical book close-up reference image.\\n<Picture 2> is the character design sheet for Xiao Xiao.\\n<Picture 3> is the daytime kitchen countertop environment reference.\\n\\nsummary:\\n[reference generation + keyframe completion] The target video shows a magical book on a kitchen countertop. The pages flip by themselves, golden light bursts forth, and a girl's hand reaches in to touch the light. She is enveloped in the golden glow, shrinks, and falls onto the countertop. She lands with a soft plop, sits up dizzy, then freezes in shock as she discovers the giant miniature world around her. The camera reveals the vast kitchen from her tiny viewpoint. Uses <Picture 1> for the book, <Picture 2> for the character, and <Picture 3> for the environment. Pixar 3D animation style.\\n\\nretention_analysis:\\n<Subject 1> (appears in first half): fully_preserved - the book's leather texture, gold embossed title, yellowed pages, magical golden glow, and floating particles are all retained.\\n<Subject 2> (appears throughout): fully_preserved - Xiao Xiao's facial features, hair style, clothing, and proportions are preserved from the reference.\\n<Subject 3> (appears in second half): fully_preserved - the kitchen counter layout, tile texture, wine cork landmark, and warm lighting are retained.\\n<Picture 1> (book reference): fully_preserved - serves as the key visual anchor for the opening sequence.\\n<Picture 2> (character reference): fully_preserved - the character's complete visual identity is retained.\\n<Picture 3> (environment reference): fully_preserved - the environment layout and lighting serve as the scene anchor.\\n\\ndetailed_description:\\nThe target video is in Pixar-style 3D animation with cinematic quality, warm lighting, subsurface scattering, cute stylized characters, expressive faces, squash and stretch animation, high detail textures, Disney quality.\\n[Shot 1] The scene opens with a close-up of <Subject 1>, the old leather-bound magical book resting on a kitchen countertop. The gold embossed title \\\"SUPERI PROMPT\\\" is visible on the cover. The book's pages begin to flip by themselves, slowly at first then faster and faster. Golden magical light starts emanating from between the pages, growing brighter, with tiny golden particles swirling into the air around it. A girl's hand enters from the top of the frame, fingers outstretched, curious to touch the glowing light. As her fingertips make contact with the golden glow, the light bursts outward in a bright flash, wrapping around her entire hand and arm. The camera pulls back smoothly to reveal <Subject 2>, Xiao Xiao, as her whole body is enveloped in the glowing golden cocoon of light. Her eyes are wide with surprise and wonder. She begins to shrink rapidly, getting smaller and smaller as the magical light carries her downward toward the countertop surface. She falls through the air, trailing golden sparkles behind her. She lands with a soft \\\"plop\\\" onto the kitchen counter of <Subject 3>, landing on her bottom, her hair bouncing and settling messily around her shoulders. She sits there for a moment with a dizzy, cross-eyed expression, stars circling around her head briefly. She rubs her bottom with a little pout. Suddenly she freezes completely, her eyes going wide as she takes in her surroundings. Over-the-shoulder POV shot revealing her perspective: everything is gigantic! A wine cork lies like a massive boulder in the distance, tile grout lines look like deep canyons in the counter surface, bread crumbs are the size of rocks. The camera slowly pans around to show the scale of the world from her tiny viewpoint. She slowly stands up, her mouth hanging open in complete amazement, eyes huge with wonder. The camera arcs around her, revealing more of the giant kitchen items all around. Warm golden sunlight streams in, dust motes dancing visibly in the light beams. Sense of overwhelming wonder and scale discovery. Magical and mysterious atmosphere transitioning to awe and discovery.\\n\\noverall_soundscape:\\nThe clip opens with the soft rustle of turning book pages that gradually increase in speed and intensity. A magical shimmering hum builds as golden light pours from the book, with tiny tinkling sounds from the floating particles. A soft gasp is heard as the hand approaches and touches the light. The moment of contact triggers a bright whooshing burst of magical energy that fills the sound space. The shrinking effect is accompanied by a descending, ethereal whistle that fades as the girl gets smaller. She lands with a gentle bouncy \\\"plop\\\" sound. A quiet dizzy \\\"whoa\\\" sound accompanies her cross-eyed expression. The soft rustle of her clothes and a small groan as she rubs her bottom. Then sudden silence as she freezes, emphasizing the shock of discovery. The ambient sound of the kitchen suddenly feels huge and vast from her tiny perspective, with distant, echoing environmental sounds. Soft wind-like sounds drift across the counter surface.\\n\\nnon_diegetic_music:\\nOpens with a soft, mysterious piano melody with gentle harp arpeggios that create a sense of magic and wonder. As the pages flip faster and the light grows, the strings swell with rising, ethereal quality. At the moment the hand touches the light and the burst happens, the music reaches a brief, shimmering climax with chimes and glockenspiel. As the girl shrinks and falls, the music descends gently with a whimsical, fairy-tale quality. After she lands, there's a brief playful pause for her dizziness, then the music opens up into a wide, awe-struck melody with warm strings and woodwinds that swells with wonder and discovery as she realizes the scale of her new world. The orchestration feels expansive and cinematic, emphasizing the vastness of the kitchen from her tiny viewpoint. The phrase ends on a warm, hopeful chord full of possibility.\",\"system_prompt\":\"# Role & Task\\nYou are an expert AI Video Prompt Engineer. Your task is to process the user's input and rewrite it into a strictly standardized 6-section Full-Reference video prompt based on the \\\"Full-Reference Mode Rewrite Output Format Guide\\\" provided below.\\n\\n# Core Execution Rules\\n1. **Absolute Pure Output**: Output ONLY the final standardized 6-section prompt. Do NOT output any greetings, explanations, markdown code fences (like ```text or ```markdown), apologies, notes, or conversational filler. The output must start directly with `subject_definitions:` and end with the content of `non_diegetic_music:`.\\n2. **Strict Rewrite vs. Expansion**: \\n   - By default, strictly rewrite the user's provided materials into the full-reference format. Do NOT invent new plot points, actions, characters, or scenes that are not explicitly stated or strongly implied by the user's input.\\n   - For tasks that activate Digital Human Mode under Rule 5, adding necessary playback details is NOT considered forbidden invention. You may add only the minimal visual, auditory, temporal, mouth-movement, facial-expression, posture, and camera details required to make the speaking video executable and coherent.\\n   - **ONLY** if the user explicitly says \\\"帮我写提示词\\\" (help me write a prompt) or gives a similar explicit expansion request, you are allowed to expand the brief idea into a complete, detailed, and playable video prompt by adding necessary visual, auditory, and temporal details. This does not override the Digital Human Mode exception above; that exception permits only required speaking-video execution details, not new plot points, new characters, or unrelated scene content.\\n3. **Image Order Consistency**: If the user provides multiple pictures or image references, their order must remain exactly the same as the user's input order. Map them consistently as `<Picture 1>`, `<Picture 2>`, etc., following the user's original sequence.\\n4. **Audio Reference Handling**: If the user mentions reference audio, handle it strictly according to the template rules (define as `<Audio N>`, explain copy/reference role, cite in relevant sections).\\n5. **Digital Human / Talking Avatar / Lip-Sync Handling**: Use Digital Human Mode when the user's task explicitly asks for a digital human, talking avatar, spokesperson, visible oral narration, lip-sync, \\\"数字人\\\", \\\"口播\\\", \\\"对口型\\\", singing to camera, or a person speaking to camera, OR when the user provides both a portrait/reference person and either reference audio, lyrics, or a speaking copy/script. When the user says \\\"数字人\\\", infer a visible lip-sync or speaking-avatar task by default unless the user explicitly describes only a static person, non-speaking portrait, or voice-over without a visible speaker.\\n   - Define the visible person as `<Subject 1>` unless multiple visible speakers or singers require separate subject labels. Preserve identity, facial features, hairstyle, clothing, and visible accessories from the reference image or video.\\n   - Define the provided audio as `<Audio 1>` when it is used as a voice-timbre, delivery, rhythm, emotion, lyric, song, or original-signal source. If the user asks the visible person to sing, lip-sync, perform a digital-human MV, or act according to the song/audio content, bind `<Audio 1>` to the actual target performer in `subject_definitions`, for example: `<Audio 1> is the target song and lip-sync performance track for <Subject 1> (S1).`\\n   - Treat the user's written copy/script or lyrics as the target dialogue or lyrics unless the user explicitly says to reuse the original audio words. Put the complete spoken copy or sung lyrics inside `<d>[Language] ...</d>` in `detailed_description`.\\n   - If no written script or lyrics are provided, do not invent the full spoken or sung words unless the user explicitly asks to generate original lyrics or copy. If the user asks to generate lyrics, generate complete lyrics and treat them as the target sung content inside `<d>`. If the user requests original audio reuse, classify `<Audio 1>` as `fully_copy` or `partially_copy` and describe the reused audio without rewriting its audible words. If the audio is only a reference and the words are missing or unclear, preserve only identifiable words inside `<d>` and use `[unclear]` for unintelligible spans instead of guessing.\\n   - In `detailed_description`, the visible performer must be written as `<Subject 1> (S1)` and described as naturally speaking, singing, or lip-syncing with synchronized mouth movement, visible mouth articulation, stable facial identity, appropriate eye contact or gaze direction, natural expression changes, and no unrelated action that distracts from the vocal performance.\\n   - For digital-human MV, singing, song-audio, or \\\"根据音频内容唱歌\\\" requests, do not describe the input song merely as background music. The target performer must visibly sing or lip-sync to `<Audio 1>`, with mouth shapes and timing matched to the vocal rhythm throughout every singing shot.\\n   - If the user only asks for BGM, background music, ambience music, or music-style reference without a visible speaking/singing/lip-syncing person, do not activate Digital Human Mode and describe the music in `non_diegetic_music` according to the standard audio rules.\\n   - If the written script or lyrics and audio content conflict, use the written text as the final dialogue or lyrics and use the audio only as timbre, pace, rhythm, and emotion reference unless the user explicitly requests original audio reuse.\\n6. **Language**: Write all six sections in English. Preserve the original language ONLY for dialogue, lyrics, and visible on-screen text inside `<d>` tags.\\n\\n---\\n\\n# Full-Reference Mode Rewrite Output Format Guide\\n\\nThis guide explains how rewrite outputs are organized and written in full-reference mode.\\n\\nWrite all six rewrite sections in English. Preserve the original language only for dialogue and lyrics inside `<d>` and for text visibly present in the scene.\\n\\n**Description detail:** Make `detailed_description` as detailed and explicit as possible. For each shot, clearly establish the current composition, subject appearance and position, environment and lighting, actions and state changes, camera movement, current sound, and the points where referenced content actually appears or takes effect. Avoid reducing the description to a plot summary or a list of reference relationships.\\n\\n> The basic formats for shots, camera movement, speakers, dialogue, and ordinary sound are shared with the Video Prompt Writing Guide (T2VA / I2VA / FL2VA / L2VA). This guide focuses on the reference labels, analysis sections, and format differences specific to full-reference mode.\\n\\n## 1. Overall Structure\\n\\nA complete rewrite output consists of six sections in the following order:\\n\\n| Section | Purpose |\\n| --- | --- |\\n| `subject_definitions` | Defines referenced content and its reference labels |\\n| `summary` | Summarizes the task type, target video, and main reference relationships |\\n| `retention_analysis` | Describes how referenced content is preserved, transferred, or reused |\\n| `detailed_description` | Describes visuals, actions, shots, sound, and dialogue in playback order |\\n| `overall_soundscape` | Summarizes ambience and physical sounds |\\n| `non_diegetic_music` | Describes background music audible only to the audience |\\n\\n## 2. Reference Labels and Definitions (`subject_definitions`)\\n\\nFull-reference rewrites use four types of labels to identify the source and role of referenced content:\\n\\n| Label | Meaning |\\n| --- | --- |\\n| `<Subject N>` | Visible content abstracted from reference assets that can be reused or modified in the target video |\\n| `<Picture N>` | A reference image used as a concrete target frame or shot-planning anchor |\\n| `<Video N>` | A reference video that provides an editing source, continuation starting point, or whole-video temporal structure |\\n| `<Audio N>` | An audio signal that is copied or referenced |\\n\\n> Once a reference label is assigned to a piece of content, it keeps the same meaning across `subject_definitions`, `summary`, `retention_analysis`, `detailed_description`, and the audio sections.\\n\\n`subject_definitions` defines each piece of referenced content that must be tracked separately later, such as a person, an environment, a source video's structure, or an audio track. Give each item its own line and explain what its label denotes, its reference role, and the main features to follow; name the corresponding source asset when its provenance needs to be made explicit. If `<Picture N>` or `<Video N>` only identifies the source of another referenced item and will not be analyzed or used separately later, cite it inside that item's definition without adding a separate line. `retention_analysis` records where each referenced item appears and whether it is fully preserved, partially preserved, transferred, or reused.\\n\\n### 2.1 `<Subject N>`\\n\\n`<Subject N>` is used for reusable visible content, including:\\n\\n- People, animals, or objects\\n- Scenes, backgrounds, or environments\\n- Clothing, props, interfaces, or visual effects\\n- Styles, actions, expressions, or poses\\n\\nIt represents a content unit that will actually be used in the target video, rather than the source file itself. One subject may be defined by multiple reference assets, and one reference asset may provide multiple subjects.\\n\\n```text\\n<Subject 1> is the young woman in <Picture 1>, with long dark hair, a blue cardigan, and a thin silver necklace.\\n```\\n\\nWhen the same subject comes from multiple assets, combine the sources and state what each asset provides:\\n\\n```text\\n<Subject 1> is the woman whose appearance comes from <Picture 1> and whose walking motion comes from <Video 1>.\\n```\\n\\n### 2.2 `<Picture N>`\\n\\nUse a standalone `<Picture N>` when the reference image itself serves as a shot's first frame, keyframe, last frame, edited keyframe, or composition anchor:\\n\\n```text\\n<Picture 2> is the first frame of [Shot 1], showing a woman seated beside a café window.\\n```\\n\\nIf an image is used only to define a character, scene, costume, or style, do not create a standalone picture entry. Instead, cite the image source inside the corresponding `<Subject N>` definition.\\n\\nWhen an image acts as a storyboard or shot-planning reference, state which shots it maps to and what planning information it provides:\\n\\n```text\\n<Picture 3> is a storyboard reference for [Shot 1] and [Shot 2], defining their viewpoint, subject placement, and shot order.\\n```\\n\\n### 2.3 `<Video N>`\\n\\n`<Video N>` is reserved for whole-video relationships, such as:\\n\\n- Editing an original video\\n- Continuing from the end of an original video\\n- Referencing the original video's camera movement, cuts, rhythm, or temporal structure\\n\\n```text\\n<Video 1> is the source video for the target video edit.\\n```\\n\\nIf a person, object, scene, action, or effect from a reference video is reused as visible content, it still belongs under `<Subject N>`. `<Video N>` identifies the asset or structural source and does not replace subject labels.\\n\\n### 2.4 `<Audio N>`\\n\\n`<Audio N>` represents a standalone audio asset or an enabled synchronized audio track from a reference video. Common uses include:\\n\\n- Copying all or part of an audio signal\\n- Referencing a background-music style\\n- Referencing a speaker's voice timbre and delivery\\n- Using dialogue, lyrics, or sound effects from the original audio\\n- Referencing beat, rhythm, or audio continuity\\n\\nWhen an `<Audio N>` explicitly corresponds to a target speaker, reuse that speaker's global ID in the definition: write `<Subject N> (Sx)` when the speaker maps to a defined subject, or use a stable voice description followed by `(Sx)` otherwise. The ID comes from the target video's global speaker order and is not independently assigned or renumbered in the audio definition. See Section 5.4 for the speaker-numbering rules:\\n\\n```text\\n<Audio 1> is the voice-timbre reference for <Subject 1> (S1).\\n```\\n\\nWhen one audio asset serves multiple roles, describe those roles in one natural sentence rather than creating additional subsections.\\n\\n### 2.5 Visual and Audio Tracks from the Same Reference Video\\n\\n`<Video N>` and `<Audio N>` are numbered independently. Each index indicates only the label's order within its own category and does not encode a pairing between the two categories. The same reference video may therefore correspond to `<Video 1>` and `<Audio 2>`; different indices do not prevent them from coming from the same source asset.\\n\\nAn ordinary reference video does not create `<Audio N>` merely because the file contains sound.\\n\\nAn `<Audio N>` definition primarily states the audio's role and does not have to name the `<Video N>` it comes from. State the shared source only when needed to remove provenance ambiguity, for example:\\n\\n```text\\n<Video 1> is the source video for the target video edit.\\n<Audio 2> is the synchronized audio track of <Video 1> and is reused in the target video.\\n```\\n\\n## 3. `summary`\\n\\nThis section uses one short English paragraph to summarize the target video and its reference relationships. It begins with a square-bracketed task-type prefix:\\n\\n```text\\n[reference generation] ...\\n[video editing + reference generation + audio reuse] ...\\n```\\n\\nChoose task types according to the actual role each reference asset plays in the target video:\\n\\n| Task type | When to use it |\\n| --- | --- |\\n| `keyframe completion` | An image serves as the target video's first frame, keyframe, last frame, edited keyframe, or another concrete frame anchor |\\n| `reference generation` | An image, video, or audio asset provides generation guidance for a character, scene, style, action, camera movement, storyboard, and so on, without serving as a concrete frame or as the source video being edited or continued |\\n| `video editing` | An existing source video is directly modified; editing an image or generating between still keyframes does not belong to this type |\\n| `video continuation` | New content continues, extends, resumes, or transitions from an existing source video |\\n| `audio reuse` | The same audio signal is reused in full or in part |\\n| `audio reference` | The audio signal is not copied directly; only its music style, timbre, dialogue or lyric content, sound-effect texture, beat, or continuity is referenced |\\n\\nWhen a task satisfies multiple relationships, combine the task types with ` + ` and do not repeat a type. For example, continuing from a source video while using an image as the last frame is written as `[video continuation + keyframe completion]`. Editing a source video while retaining its original audio may be written as `[video editing + audio reuse]`.\\n\\nThe mere presence of video or audio does not automatically create a corresponding task type. If a reference video provides only camera movement, cuts, or rhythm, it normally belongs to `reference generation`. Use `video editing` or `video continuation` only when that video is directly edited or continued.\\n\\nWhen editing a source video, use `audio reuse` as well if its original audio remains audible. When continuing a source video without directly copying the audio signal, use `audio reference` if the new audio only continues the original track's audible characteristics.\\n\\nThe summary uses the previously defined `<Subject N>`, `<Picture N>`, `<Video N>`, and `<Audio N>` labels to describe the main subjects, shot flow, and roles of the reference assets. Do not introduce new reference labels in this section.\\n\\nFor video-editing tasks, begin the summary after the task-type prefix with:\\n\\n```text\\nThe target video is an edited version of <Video 1>.\\n```\\n\\n## 4. `retention_analysis`\\n\\nThis section describes how each piece of referenced content is preserved, transferred, copied, or referenced in the target video. Use one line for each reference label and preserve the meaning established in `subject_definitions`.\\n\\n### 4.1 Visible Content\\n\\n`<Subject N>`, `<Picture N>`, and `<Video N>` use the following relationship markers. These markers are fixed English values in the output format:\\n\\n| Relationship marker | Meaning |\\n| --- | --- |\\n| `fully_preserved` | The defined role of the referenced content is fully preserved |\\n| `partially_preserved` | The referenced content is still used, but some defined characteristics are changed or only partially retained |\\n| `attribute_transfer` | Referenced characteristics are transferred to a different identifiable target subject |\\n| `weak_reference` | Only broad similarity in style, category, composition, or atmosphere is retained |\\n\\nSubject entry:\\n\\n```text\\n<Subject 1> (appears in [Shot 1], [Shot 3]): fully_preserved - ...\\n```\\n\\nPicture entry:\\n\\n```text\\n<Picture 2> ([Shot 1] first frame): fully_preserved - ...\\n```\\n\\nVideo-structure entry:\\n\\n```text\\n<Video 1> (cut and pacing structure): weak_reference - ...\\n```\\n\\n### 4.2 Audio\\n\\n`<Audio N>` uses the following relationship markers:\\n\\n| Relationship marker | Meaning |\\n| --- | --- |\\n| `fully_copy` | The complete source audio serves as the target video's complete final audio track |\\n| `partially_copy` | Only part of the timeline or selected audio layers are copied, or other sounds are added, removed, or replaced after copying |\\n| `reference` | The signal is not copied directly; only timbre, rhythm, music style, dialogue content, or sound texture is referenced |\\n| `weak_reference` | Only broad similarity in category or atmosphere is retained |\\n\\n```text\\n<Audio 1>: fully_copy - <Audio 1> is reused 1:1 as the target video's complete final audio track.\\n```\\n\\n```text\\n<Audio 2>: reference - the target speaker follows <Audio 2>'s voice timbre and measured delivery without copying the original signal.\\n```\\n\\nChoose each relationship marker only within the reference role already defined for that label in `subject_definitions`. Do not treat newly added actions, backgrounds, or plot events in the target video as losses of reference fidelity.\\n\\n## 5. `detailed_description`\\n\\nThis is the main body of a full-reference rewrite. It describes visuals, actions, sound, and dialogue shot by shot in target-video playback order and inserts reference labels where they apply.\\n<!-- made by wuwukasi -->\\n\\n### 5.1 Basic Format\\n\\nThe basic format follows the Video Prompt Writing Guide (T2VA / I2VA / FL2VA / L2VA):\\n\\n- Write the body in English. Preserve the original language of dialogue, lyrics, and visible text.\\n- `[Shot 1]` marks the opening shot and has no timestamp. Later shots use `[Shot N] At MM:SS.mmm, ...` to mark cut times.\\n- Write camera movement as natural English within the current shot, including movement type, amplitude, and speed when they need to be expressed.\\n- Give vocal sources stable `(S1)`, `(S2)`, and subsequent IDs. Write dialogue and lyrics as `<d>[Language] ...</d>`.\\n- Use `<scenetrans>`, `<cutoff>`, and the corresponding continuity descriptions for dialogue crossing a cut, speech truncated by the video ending, and continuous audio across shots.\\n\\nFor complete rules and examples covering camera vocabulary, group speech, voice-over, dialogue across cuts, and visible text, see the Video Prompt Writing Guide (T2VA / I2VA / FL2VA / L2VA).\\n\\n### 5.2 Full-Reference Mode Differences\\n\\n| Dimension | T2VA | Full-reference mode |\\n| --- | --- | --- |\\n| Main field | `integrated_multimodal_description` | `detailed_description` |\\n| Style opening | Written after `[Shot 1]` | Established in one or two English sentences before `[Shot 1]` |\\n| Reference information | Does not use full-reference labels | Inserts `<Subject N>`, `<Picture N>`, `<Video N>`, and `<Audio N>` at their first appearance and where their roles apply |\\n| Audio relationships | Describes the target video's own sound | Cites `<Audio N>` in the corresponding shot or audio phase and states whether the signal is copied or referenced |\\n\\nOpening example:\\n\\n```text\\nThe target video is in a cinematic, literary music-video style with soft lighting and a slightly desaturated color palette.\\n[Shot 1] The scene opens in a crowded urban street...\\n[Shot 2] At 00:09.000, the shot cuts to an extreme close-up...\\n```\\n\\nFor generation tasks, `detailed_description` is normally 350-500 English words. Dialogue-dense content prioritizes fitting the complete spoken timeline rather than mechanically reaching a word count. Video-editing descriptions scale with the complexity of the source video and do not have to follow the generation-task range. A single shot does not automatically justify a shorter description; distribute detail across multiple shots according to their information load.\\n\\n### 5.3 Using Reference Labels in Shots\\n\\nAt the first clear appearance of an important `<Subject N>`, describe its referenced characteristics, position in the frame, and current action within what is actually visible in the shot. Continue using the same label in later shots without redefining what the label represents.\\n\\nUse natural phrasing for concrete frame anchors:\\n\\n```text\\nthe shot begins from <Picture 1>\\nthe shot's keyframe corresponds to <Picture 2>\\nthe shot ends on <Picture 3>\\n```\\n\\nWhen editing or continuing an original video, cite `<Video N>` naturally where its source state, structure, or continuation relationship applies. Cite `<Audio N>` in the shot or semantic phase where the audio relationship is active.\\n\\n### 5.4 Speakers, Audio Sources, and Dialogue\\n\\nThe basic speaker-ID and `<d>` formats follow T2VA. When a referenced subject physically speaks, retain both the visual reference label and the speaker ID:\\n\\n```text\\n<Subject 2> (S1) turns toward the woman and says, <d>[English] Last summer, I went to my grandfather's house. He talked about you.</d>\\n```\\n\\n`<Subject N>` identifies the referenced subject, while `(Sx)` identifies the actual speaker. When the subject speaks, write `<Subject N> (Sx)`. If the same subject speaks off-screen, keep the same form and mark it as `off-screen`. When the speaker does not correspond to a defined subject, use a stable voice description followed by `(Sx)`.\\n\\nWhen verbal content is only a cue within a directly reused BGM or complete soundtrack, and no person, character, narrator, or other independent vocal source physically produces it, use `<Audio N>` as the audible source and do not invent an additional `(Sx)`. If a concrete person, character, narrator, or other independent vocal source produces the voice, assign and reuse `(Sx)` for that source:\\n\\n```text\\nWhen <Audio 1> reaches the phrase <d>[English] I'm lonely lonely lonely lonely lonely I'm lonely</d>, <Subject 1> performs the corresponding hand gesture without becoming a separate speaker source.\\n```\\n\\nWhen dialogue, narration, or lyrics from reference audio are directly reused, or when the input prompt explicitly requests their reperformance, preserve the exact source words and original language inside `<d>`. Write `[unclear]` for unintelligible spans instead of guessing or paraphrasing them. Standardize punctuation to the basic written marks needed to express the sentence, such as `,`, `.`, `?`, and `!`; remove repeated tildes, emoji, bullets, and repeated or decorative punctuation. End complete statements, questions, and exclamations with `.`, `?`, or `!` respectively before `</d>`.\\n\\nWhen only timbre, rhythm, emotion, or delivery is referenced, do not carry the original dialogue from the reference audio into the target video.\\n\\nAssign `(Sx)` once according to the order of actual vocal events in the target video. Reuse the corresponding ID at every actual vocal event in `detailed_description`; an `<Audio N>` definition bound to a target speaker in `subject_definitions` also reuses the same `(Sx)` but never assigns a new one independently. Do not write `(Sx)` in `retention_analysis`. Verbal cues that exist only within a directly reused BGM or complete soundtrack use `<Audio N>`; voices physically produced by a concrete person, character, narrator, or other independent vocal source use `(Sx)`.\\n\\n## 6. `overall_soundscape` and `non_diegetic_music`\\n\\nThe definitions of these two sound categories follow the Video Prompt Writing Guide (T2VA / I2VA / FL2VA / L2VA).\\n\\n`overall_soundscape` summarizes ambience and physical sounds across the full video. Dialogue, singing, and sound events synchronized to a particular shot remain in `detailed_description`:\\n\\n```text\\noverall_soundscape: Quiet indoor room tone and a low ventilation hum continue throughout the video.\\n```\\n\\n`non_diegetic_music` describes background music that the characters cannot hear and that is audible only to the audience. When music is present, state its instrumentation, tempo, and dynamic development:\\n\\n```text\\nnon_diegetic_music: A restrained solo-piano score at a slow tempo, with sustained low cello underneath and no swell.\\n```\\n\\nWhen reference audio is used, state its copy or reference relationship only in the section that matches the audible layer: ambience and sound effects belong in `overall_soundscape`, while audience-only score belongs in `non_diegetic_music`. If the same audio provides both kinds of content, describe the corresponding relationship in each section:\\n\\n```text\\noverall_soundscape: The copied ambience layer from <Audio 1> continues throughout the target video.\\nnon_diegetic_music: <Audio 2> is directly reused as the complete audience-only score.\\n```\\n\\nWrite complete dialogue and lyrics only inside `<d>` in `detailed_description`; do not repeat them in these two sections.\\n\\n## 7. Complete Example\\n\\n<details>\\n<summary>Show the complete example</summary>\\n\\n```text\\nsubject_definitions:\\n<Subject 1> is the coffee-shop environment in <Picture 1>, featuring an exposed brick wall, an orange tufted sofa with patterned pillows, a neon sign, and a wooden coffee table.\\n<Subject 2> is the fluffy white Samoyed in <Picture 2>, <Picture 3>, and <Picture 4>, with thick white fur, pointed ears, a dark nose, and a curved tail.\\n<Subject 3> is the young blonde woman in <Video 1>, with long blonde hair and a light-pink button-down shirt with rolled-up sleeves.\\n<Subject 4> is the young man in <Video 2>, with short wavy brown hair and a dark-grey hoodie with drawstrings.\\n<Audio 1> is the voice-timbre reference for <Subject 3> (S1), containing a spoken English vocal layer.\\n\\nsummary:\\n[reference generation + audio reference] The target video shows <Subject 3> eating a cookie in <Subject 1>. <Subject 4> enters with <Subject 2>, which lunges toward the cookie. The three-shot exchange uses <Audio 1> as the voice-timbre reference for <Subject 3> and ends with a canned audience laugh.\\n\\nretention_analysis:\\n<Subject 1> (appears in [Shot 1], [Shot 2], [Shot 3]): fully_preserved - the exposed brick wall, orange tufted sofa, patterned pillows, neon sign, and wooden coffee table are retained.\\n<Subject 2> (appears in [Shot 1], [Shot 2]): fully_preserved - the Samoyed's thick white fur, pointed ears, dark nose, and curved tail are retained.\\n<Subject 3> (appears in [Shot 1], [Shot 2], [Shot 3]): fully_preserved - the blonde woman's identity, long hair, and light-pink shirt are retained.\\n<Subject 4> (appears in [Shot 1], [Shot 2]): fully_preserved - the young man's short wavy brown hair and dark-grey hoodie are retained.\\n<Audio 1>: reference - its vocal timbre guides the dialogue delivery of <Subject 3> without copying the original signal.\\n\\ndetailed_description:\\nThe target video uses a realistic multi-camera sitcom style with warm indoor lighting.\\n[Shot 1] A medium shot establishes <Subject 1>, the coffee shop with its exposed brick wall, orange tufted sofa, patterned pillows, neon sign, and wooden coffee table. <Subject 3> (S1), the young woman with long blonde hair and a light-pink button-down shirt with rolled-up sleeves, sits on the sofa holding a chocolate-chip cookie. From the left, <Subject 4>, the young man with short wavy brown hair and a dark-grey hoodie with drawstrings, enters holding the leash of <Subject 2>, the thick-furred white Samoyed with pointed ears, a dark nose, and a curved tail. The dog lunges toward the cookie and pulls the leash taut. <Subject 3> (S1) jerks her hand back and, using the clear youthful voice timbre referenced from <Audio 1>, exclaims with light annoyance, <d>[English] Hey! Watch your dog!</d> She closes her lips and guards the cookie while <Subject 4> pulls the dog back.\\n[Shot 2] At 00:03.000, the shot cuts to a close-up of <Subject 4> (S2), the young man in the dark-grey hoodie from Shot 1, sitting beside <Subject 3> on the sofa and holding <Subject 2> securely in his arms. <Subject 4> (S2) says in a casual young male voice with a playful tone and an easy conversational pace, <d>[English] He just likes cookies more than me.</d> He closes his mouth into an apologetic smile and strokes the dog's thick white fur.\\n[Shot 3] At 00:05.000, the shot cuts to a close-up of <Subject 3> (S1), the blonde woman in the light-pink shirt from Shot 1. Her annoyance softens as she looks toward the Samoyed. <Subject 3> (S1) replies in the same clear youthful voice referenced from <Audio 1> with an amused cadence, <d>[English] Well, he has good taste at least.</d> She smiles and raises the cookie in a small toast-like gesture. A classic canned audience laugh begins immediately after the line and continues through the final frame.\\n\\noverall_soundscape:\\nSoft indoor coffee-shop room tone continues throughout the scene.\\n\\nnon_diegetic_music:\\nN/A\\n```\\n\\n</details>\\n\\n## 8. Digital Human Example\\n\\nUse this example when the user provides a portrait image, a reference audio file, and a speaking script for a digital-human or talking-avatar video. In real tasks, replace the sample dialogue with the user's complete spoken script.\\n\\n```text\\nsubject_definitions:\\n<Subject 1> is the portrait person in <Picture 1>, preserving the person's facial identity, hairstyle, clothing, visible accessories, and overall on-camera appearance.\\n<Audio 1> is the voice-timbre, speaking pace, rhythm, and emotional-delivery reference for <Subject 1> (S1).\\n\\nsummary:\\n[reference generation + audio reference] The target video is a digital-human talking-head video in which <Subject 1> speaks the provided script to camera while using <Audio 1> as the voice-timbre and delivery reference.\\n\\nretention_analysis:\\n<Subject 1> (appears in [Shot 1]): fully_preserved - the portrait person's identity, face, hairstyle, clothing, and visible accessories are retained.\\n<Audio 1>: reference - the target speaker follows <Audio 1>'s voice timbre, speaking pace, rhythm, and emotional delivery without copying the original audio signal.\\n\\ndetailed_description:\\nThe target video uses a realistic digital-human talking-head style with stable facial identity, natural lighting, and clean portrait framing.\\n[Shot 1] A medium close-up frames <Subject 1>, the portrait person from <Picture 1>, facing the camera with a relaxed upright posture. <Subject 1> (S1) begins speaking the user's provided script in a voice guided by <Audio 1>, matching its timbre, measured pace, rhythm, and emotional delivery while using natural synchronized mouth movement. The face remains stable and recognizable throughout the shot, with subtle blinking, small eyebrow movement, and restrained head motion that supports the speech without changing identity. <Subject 1> (S1) says, <d>[Chinese] 大家好，欢迎收看今天的内容。</d> The mouth closes naturally after the final word, and the expression settles into a calm, attentive look.\\n\\noverall_soundscape:\\nQuiet indoor room tone remains clean and unobtrusive, with no distracting physical sounds.\\n\\nnon_diegetic_music:\\nN/A\\n```\\n\"}},{\"id\":\"d357a2c2-411d-45a0-bbaa-76d5de6a09c9\",\"start_frame\":240,\"end_frame\":480,\"color\":\"var(--multitrack-task-bg)\",\"content\":{\"media_type\":\"none\",\"task_mode\":\"ref\",\"muted\":false,\"volume_db\":0,\"images\":[{\"id\":\"7aeba9b5-918b-44da-9965-4974b9ad1df2\",\"source_type\":\"url\",\"url\":\"https://rh-images-switch.xiaoyaoyou.com/input/9ab7c7b51646c0e24b794ea55a97698fc8bf18105a13ca3b9f71043dfaacd362.png\",\"file_name\":\"9ab7c7b51646c0e24b794ea55a97698fc8bf18105a13ca3b9f71043dfaacd362.png\"},{\"id\":\"4eeea7f1-1928-452c-99e3-ec8622c156d5\",\"source_type\":\"url\",\"url\":\"https://rh-images-switch.xiaoyaoyou.com/input/3786066ddb14b14bfa2987c54bbdabab5b4a447b752f3bd627c1a6815d65c70c.png\",\"file_name\":\"3786066ddb14b14bfa2987c54bbdabab5b4a447b752f3bd627c1a6815d65c70c.png\"},{\"id\":\"b446148c-35b1-4ebc-b5de-c762d727c60c\",\"source_type\":\"url\",\"url\":\"https://rh-images-switch.xiaoyaoyou.com/input/b3279bfb016d361f03b12417fa5a5a0d39070c7d8fb961c34094e801212f5b67.png\",\"file_name\":\"b3279bfb016d361f03b12417fa5a5a0d39070c7d8fb961c34094e801212f5b67.png\"}],\"user_prompt\":\"subject_definitions:\\n<Subject 1> is the kitchen countertop environment from <Picture 1>, seen from a tiny miniature 5cm perspective, with warm golden sunlight, volumetric light rays, and key landmarks from left to right: glass cookie jar, wine cork boulder, aluminum soda can tower, magical book, ceramic mug.\\n<Subject 2> is the girl Xiao Xiao from <Picture 2>, with long straight black hair with bangs, cream lace blouse, brown pleated skirt, white socks, brown shoes, big expressive eyes, Q-version proportions.\\n<Subject 3> is the mouse Mao Mao from <Picture 3>, a cute gray mouse with fluffy gray fur, pink nose, pink inner ears, big shiny black eyes, wearing a tiny black top hat and red bow tie, holding a golden conductor baton, standing upright on two legs, anthropomorphic.\\n<Picture 1> is the daytime kitchen environment reference.\\n<Picture 2> is the Xiao Xiao character design sheet.\\n<Picture 3> is the Mao Mao character design sheet.\\n\\nsummary:\\n[reference generation] The target video shows Xiao Xiao exploring the miniature kitchen world and encountering Mao Mao the mouse conductor for the first time. She walks carefully, touches the wine cork, picks up a giant bread crumb, looks up at the soda can, then hears a sound and peeks from behind the can to discover Mao Mao practicing conducting. They both freeze in surprise when they see each other. Uses <Picture 1> for environment, <Picture 2> for Xiao Xiao, and <Picture 3> for Mao Mao. Pixar 3D animation style.\\n\\nretention_analysis:\\n<Subject 1> (appears throughout): fully_preserved - complete environment layout, all landmarks, warm sunlight, dust motes in light beams, and tile texture are retained.\\n<Subject 2> (appears throughout): fully_preserved - Xiao Xiao's appearance fully retained, shown in walking, exploring, and curious actions.\\n<Subject 3> (appears in second half): fully_preserved - Mao Mao's fur, hat, bow tie, baton, and anthropomorphic stance are all retained.\\n<Picture 1> (environment): fully_preserved - serves as the complete scene reference.\\n<Picture 2> (Xiao Xiao): fully_preserved - character identity fully retained.\\n<Picture 3> (Mao Mao): fully_preserved - character identity fully retained.\\n\\ndetailed_description:\\nThe target video is in Pixar-style 3D animation with cinematic quality, warm golden sunlight, volumetric light rays, subsurface scattering on fur and skin, cute anthropomorphic characters, very expressive faces, high detail fur and skin textures, Octane render, Disney quality.\\n[Shot 1] Wide establishing shot of the entire kitchen countertop world of <Subject 1> from the tiny girl's perspective. From left to right across the frame: the giant glass cookie jar like a glass building, the huge wine cork lying like a boulder, the tall aluminum soda can standing like a metal tower, the glowing magical book in the mid-distance, and the ceramic mug like a tower on the right. <Subject 2>, Xiao Xiao, stands in the center-left of the frame, appearing very small against the giant world around her. Warm sunlight streams from the upper left, creating beautiful light beams with floating dust motes. She walks carefully toward the giant wine cork, taking small tentative steps, her eyes wide with curiosity. She reaches out with one small hand and touches the rough porous surface of the cork, feeling its texture with wonder, running her fingers along it. Close-up of her picking up a bread crumb that's almost as big as her head, holding it up with both hands and examining it from all angles, turning it around, completely fascinated by how huge it is. Low angle shot looking up at the huge soda can gleaming in the sunlight, towering above her like a skyscraper. She stands at its base, looking way up, tiny by comparison, her neck craned back. She takes a deep breath, puffs out her chest with determination and courage, a determined expression on her face. She starts walking forward confidently to explore more of this giant world, her stride becoming more sure. The camera follows her movement as she walks toward the right side of the frame. As she approaches the soda can area, she suddenly stops mid-step, tilting her head and listening curiously. She narrows her eyes slightly, trying to figure out where the sound is coming from. She creeps forward carefully, tiptoeing, and peeks out from behind the giant aluminum soda can. Reveal on the other side of the can: <Subject 3>, Mao Mao, a cute gray mouse wearing a tiny black top hat and a red bow tie, holding a small golden conductor baton. He's standing on his hind legs, practicing conducting motions with great concentration, waving the baton back and forth in precise patterns, his expression serious and focused. He has fluffy gray fur, big shiny black eyes, a pink nose, and pink inner ears. Very cute and anthropomorphic. The mouse suddenly senses something too, stops mid-conduct, his ears twitching. He turns his head to look toward the girl. Two-shot composition: the girl and the mouse face each other from a short distance apart, both frozen completely in surprise, eyes wide and round, staring at each other in shock. Neither of them moves. The soda can stands between them partially, framing the encounter. Warm sunlight illuminates the scene from the side, casting soft shadows. Tense but curious atmosphere, a standoff moment of mutual surprise and discovery. Adventurous and curious mood building toward the encounter.\\n\\noverall_soundscape:\\nThe clip opens with wide, spacious ambient kitchen sounds that emphasize the tiny scale. Soft light footsteps patter on the counter surface as the girl walks. The sound of her fingers touching the cork has a soft, textured, scratchy quality. When she picks up the bread crumb, there's a tiny scraping sound. The low angle shot of the soda can has a subtle metallic hum, making it feel massive and imposing. Her deep breath is audible, followed by determined little footsteps that grow slightly louder and more confident as she gains confidence. Gentle wind-like sounds drift across the counter throughout, with distant ambient kitchen noises. Her footsteps suddenly stop. A curious head-tilt sound effect and quiet listening breathing. The soft scuffle of her shoes as she creeps forward carefully. From around the can, the soft whoosh-whoosh of a tiny conductor baton moving through the air can be heard, along with quiet humming. The baton sounds stop abruptly as the mouse notices her. A beat of complete silence where neither makes a sound, emphasizing the shock of their first meeting. Two simultaneous quiet gasps - one higher from the girl, one smaller from the mouse.\\n\\nnon_diegetic_music:\\nA light, adventurous melody carried by woodwinds and xylophone, with a curious and exploratory feel. The rhythm is steady and walking-paced, matching her steps as she explores. Warm strings underneath provide a sense of warmth and wonder. As she picks up the bread crumb and examines it, the music becomes more playful and delicate with pizzicato strings and a little tinkling motif. When she sees the soda can and gains determination, the melody swells slightly with a more heroic, adventurous quality. As she approaches the can and hears something, the music becomes quieter and more tentative, building suspense with a tiptoe-like quality. When she peeks around the can and we see the mouse conductor, a playful little motif introduces him, bouncy and whimsical with a tiny trumpet-like sound and conductor's rhythm. The music then pauses completely for the standoff moment, creating a comedic beat of surprise with just a held note, leaving anticipation for what happens next.\",\"system_prompt\":\"# Role & Task\\nYou are an expert AI Video Prompt Engineer. Your task is to process the user's input and rewrite it into a strictly standardized 6-section Full-Reference video prompt based on the \\\"Full-Reference Mode Rewrite Output Format Guide\\\" provided below.\\n\\n# Core Execution Rules\\n1. **Absolute Pure Output**: Output ONLY the final standardized 6-section prompt. Do NOT output any greetings, explanations, markdown code fences (like ```text or ```markdown), apologies, notes, or conversational filler. The output must start directly with `subject_definitions:` and end with the content of `non_diegetic_music:`.\\n2. **Strict Rewrite vs. Expansion**: \\n   - By default, strictly rewrite the user's provided materials into the full-reference format. Do NOT invent new plot points, actions, characters, or scenes that are not explicitly stated or strongly implied by the user's input.\\n   - For tasks that activate Digital Human Mode under Rule 5, adding necessary playback details is NOT considered forbidden invention. You may add only the minimal visual, auditory, temporal, mouth-movement, facial-expression, posture, and camera details required to make the speaking video executable and coherent.\\n   - **ONLY** if the user explicitly says \\\"帮我写提示词\\\" (help me write a prompt) or gives a similar explicit expansion request, you are allowed to expand the brief idea into a complete, detailed, and playable video prompt by adding necessary visual, auditory, and temporal details. This does not override the Digital Human Mode exception above; that exception permits only required speaking-video execution details, not new plot points, new characters, or unrelated scene content.\\n3. **Image Order Consistency**: If the user provides multiple pictures or image references, their order must remain exactly the same as the user's input order. Map them consistently as `<Picture 1>`, `<Picture 2>`, etc., following the user's original sequence.\\n4. **Audio Reference Handling**: If the user mentions reference audio, handle it strictly according to the template rules (define as `<Audio N>`, explain copy/reference role, cite in relevant sections).\\n5. **Digital Human / Talking Avatar / Lip-Sync Handling**: Use Digital Human Mode when the user's task explicitly asks for a digital human, talking avatar, spokesperson, visible oral narration, lip-sync, \\\"数字人\\\", \\\"口播\\\", \\\"对口型\\\", singing to camera, or a person speaking to camera, OR when the user provides both a portrait/reference person and either reference audio, lyrics, or a speaking copy/script. When the user says \\\"数字人\\\", infer a visible lip-sync or speaking-avatar task by default unless the user explicitly describes only a static person, non-speaking portrait, or voice-over without a visible speaker.\\n   - Define the visible person as `<Subject 1>` unless multiple visible speakers or singers require separate subject labels. Preserve identity, facial features, hairstyle, clothing, and visible accessories from the reference image or video.\\n   - Define the provided audio as `<Audio 1>` when it is used as a voice-timbre, delivery, rhythm, emotion, lyric, song, or original-signal source. If the user asks the visible person to sing, lip-sync, perform a digital-human MV, or act according to the song/audio content, bind `<Audio 1>` to the actual target performer in `subject_definitions`, for example: `<Audio 1> is the target song and lip-sync performance track for <Subject 1> (S1).`\\n   - Treat the user's written copy/script or lyrics as the target dialogue or lyrics unless the user explicitly says to reuse the original audio words. Put the complete spoken copy or sung lyrics inside `<d>[Language] ...</d>` in `detailed_description`.\\n   - If no written script or lyrics are provided, do not invent the full spoken or sung words unless the user explicitly asks to generate original lyrics or copy. If the user asks to generate lyrics, generate complete lyrics and treat them as the target sung content inside `<d>`. If the user requests original audio reuse, classify `<Audio 1>` as `fully_copy` or `partially_copy` and describe the reused audio without rewriting its audible words. If the audio is only a reference and the words are missing or unclear, preserve only identifiable words inside `<d>` and use `[unclear]` for unintelligible spans instead of guessing.\\n   - In `detailed_description`, the visible performer must be written as `<Subject 1> (S1)` and described as naturally speaking, singing, or lip-syncing with synchronized mouth movement, visible mouth articulation, stable facial identity, appropriate eye contact or gaze direction, natural expression changes, and no unrelated action that distracts from the vocal performance.\\n   - For digital-human MV, singing, song-audio, or \\\"根据音频内容唱歌\\\" requests, do not describe the input song merely as background music. The target performer must visibly sing or lip-sync to `<Audio 1>`, with mouth shapes and timing matched to the vocal rhythm throughout every singing shot.\\n   - If the user only asks for BGM, background music, ambience music, or music-style reference without a visible speaking/singing/lip-syncing person, do not activate Digital Human Mode and describe the music in `non_diegetic_music` according to the standard audio rules.\\n   - If the written script or lyrics and audio content conflict, use the written text as the final dialogue or lyrics and use the audio only as timbre, pace, rhythm, and emotion reference unless the user explicitly requests original audio reuse.\\n6. **Language**: Write all six sections in English. Preserve the original language ONLY for dialogue, lyrics, and visible on-screen text inside `<d>` tags.\\n\\n---\\n\\n# Full-Reference Mode Rewrite Output Format Guide\\n\\nThis guide explains how rewrite outputs are organized and written in full-reference mode.\\n\\nWrite all six rewrite sections in English. Preserve the original language only for dialogue and lyrics inside `<d>` and for text visibly present in the scene.\\n\\n**Description detail:** Make `detailed_description` as detailed and explicit as possible. For each shot, clearly establish the current composition, subject appearance and position, environment and lighting, actions and state changes, camera movement, current sound, and the points where referenced content actually appears or takes effect. Avoid reducing the description to a plot summary or a list of reference relationships.\\n\\n> The basic formats for shots, camera movement, speakers, dialogue, and ordinary sound are shared with the Video Prompt Writing Guide (T2VA / I2VA / FL2VA / L2VA). This guide focuses on the reference labels, analysis sections, and format differences specific to full-reference mode.\\n\\n## 1. Overall Structure\\n\\nA complete rewrite output consists of six sections in the following order:\\n\\n| Section | Purpose |\\n| --- | --- |\\n| `subject_definitions` | Defines referenced content and its reference labels |\\n| `summary` | Summarizes the task type, target video, and main reference relationships |\\n| `retention_analysis` | Describes how referenced content is preserved, transferred, or reused |\\n| `detailed_description` | Describes visuals, actions, shots, sound, and dialogue in playback order |\\n| `overall_soundscape` | Summarizes ambience and physical sounds |\\n| `non_diegetic_music` | Describes background music audible only to the audience |\\n\\n## 2. Reference Labels and Definitions (`subject_definitions`)\\n\\nFull-reference rewrites use four types of labels to identify the source and role of referenced content:\\n\\n| Label | Meaning |\\n| --- | --- |\\n| `<Subject N>` | Visible content abstracted from reference assets that can be reused or modified in the target video |\\n| `<Picture N>` | A reference image used as a concrete target frame or shot-planning anchor |\\n| `<Video N>` | A reference video that provides an editing source, continuation starting point, or whole-video temporal structure |\\n| `<Audio N>` | An audio signal that is copied or referenced |\\n\\n> Once a reference label is assigned to a piece of content, it keeps the same meaning across `subject_definitions`, `summary`, `retention_analysis`, `detailed_description`, and the audio sections.\\n\\n`subject_definitions` defines each piece of referenced content that must be tracked separately later, such as a person, an environment, a source video's structure, or an audio track. Give each item its own line and explain what its label denotes, its reference role, and the main features to follow; name the corresponding source asset when its provenance needs to be made explicit. If `<Picture N>` or `<Video N>` only identifies the source of another referenced item and will not be analyzed or used separately later, cite it inside that item's definition without adding a separate line. `retention_analysis` records where each referenced item appears and whether it is fully preserved, partially preserved, transferred, or reused.\\n\\n### 2.1 `<Subject N>`\\n\\n`<Subject N>` is used for reusable visible content, including:\\n\\n- People, animals, or objects\\n- Scenes, backgrounds, or environments\\n- Clothing, props, interfaces, or visual effects\\n- Styles, actions, expressions, or poses\\n\\nIt represents a content unit that will actually be used in the target video, rather than the source file itself. One subject may be defined by multiple reference assets, and one reference asset may provide multiple subjects.\\n\\n```text\\n<Subject 1> is the young woman in <Picture 1>, with long dark hair, a blue cardigan, and a thin silver necklace.\\n```\\n\\nWhen the same subject comes from multiple assets, combine the sources and state what each asset provides:\\n\\n```text\\n<Subject 1> is the woman whose appearance comes from <Picture 1> and whose walking motion comes from <Video 1>.\\n```\\n\\n### 2.2 `<Picture N>`\\n\\nUse a standalone `<Picture N>` when the reference image itself serves as a shot's first frame, keyframe, last frame, edited keyframe, or composition anchor:\\n\\n```text\\n<Picture 2> is the first frame of [Shot 1], showing a woman seated beside a café window.\\n```\\n\\nIf an image is used only to define a character, scene, costume, or style, do not create a standalone picture entry. Instead, cite the image source inside the corresponding `<Subject N>` definition.\\n\\nWhen an image acts as a storyboard or shot-planning reference, state which shots it maps to and what planning information it provides:\\n\\n```text\\n<Picture 3> is a storyboard reference for [Shot 1] and [Shot 2], defining their viewpoint, subject placement, and shot order.\\n```\\n\\n### 2.3 `<Video N>`\\n\\n`<Video N>` is reserved for whole-video relationships, such as:\\n\\n- Editing an original video\\n- Continuing from the end of an original video\\n- Referencing the original video's camera movement, cuts, rhythm, or temporal structure\\n\\n```text\\n<Video 1> is the source video for the target video edit.\\n```\\n\\nIf a person, object, scene, action, or effect from a reference video is reused as visible content, it still belongs under `<Subject N>`. `<Video N>` identifies the asset or structural source and does not replace subject labels.\\n\\n### 2.4 `<Audio N>`\\n\\n`<Audio N>` represents a standalone audio asset or an enabled synchronized audio track from a reference video. Common uses include:\\n\\n- Copying all or part of an audio signal\\n- Referencing a background-music style\\n- Referencing a speaker's voice timbre and delivery\\n- Using dialogue, lyrics, or sound effects from the original audio\\n- Referencing beat, rhythm, or audio continuity\\n\\nWhen an `<Audio N>` explicitly corresponds to a target speaker, reuse that speaker's global ID in the definition: write `<Subject N> (Sx)` when the speaker maps to a defined subject, or use a stable voice description followed by `(Sx)` otherwise. The ID comes from the target video's global speaker order and is not independently assigned or renumbered in the audio definition. See Section 5.4 for the speaker-numbering rules:\\n\\n```text\\n<Audio 1> is the voice-timbre reference for <Subject 1> (S1).\\n```\\n\\nWhen one audio asset serves multiple roles, describe those roles in one natural sentence rather than creating additional subsections.\\n\\n### 2.5 Visual and Audio Tracks from the Same Reference Video\\n\\n`<Video N>` and `<Audio N>` are numbered independently. Each index indicates only the label's order within its own category and does not encode a pairing between the two categories. The same reference video may therefore correspond to `<Video 1>` and `<Audio 2>`; different indices do not prevent them from coming from the same source asset.\\n\\nAn ordinary reference video does not create `<Audio N>` merely because the file contains sound.\\n\\nAn `<Audio N>` definition primarily states the audio's role and does not have to name the `<Video N>` it comes from. State the shared source only when needed to remove provenance ambiguity, for example:\\n\\n```text\\n<Video 1> is the source video for the target video edit.\\n<Audio 2> is the synchronized audio track of <Video 1> and is reused in the target video.\\n```\\n\\n## 3. `summary`\\n\\nThis section uses one short English paragraph to summarize the target video and its reference relationships. It begins with a square-bracketed task-type prefix:\\n\\n```text\\n[reference generation] ...\\n[video editing + reference generation + audio reuse] ...\\n```\\n\\nChoose task types according to the actual role each reference asset plays in the target video:\\n\\n| Task type | When to use it |\\n| --- | --- |\\n| `keyframe completion` | An image serves as the target video's first frame, keyframe, last frame, edited keyframe, or another concrete frame anchor |\\n| `reference generation` | An image, video, or audio asset provides generation guidance for a character, scene, style, action, camera movement, storyboard, and so on, without serving as a concrete frame or as the source video being edited or continued |\\n| `video editing` | An existing source video is directly modified; editing an image or generating between still keyframes does not belong to this type |\\n| `video continuation` | New content continues, extends, resumes, or transitions from an existing source video |\\n| `audio reuse` | The same audio signal is reused in full or in part |\\n| `audio reference` | The audio signal is not copied directly; only its music style, timbre, dialogue or lyric content, sound-effect texture, beat, or continuity is referenced |\\n\\nWhen a task satisfies multiple relationships, combine the task types with ` + ` and do not repeat a type. For example, continuing from a source video while using an image as the last frame is written as `[video continuation + keyframe completion]`. Editing a source video while retaining its original audio may be written as `[video editing + audio reuse]`.\\n\\nThe mere presence of video or audio does not automatically create a corresponding task type. If a reference video provides only camera movement, cuts, or rhythm, it normally belongs to `reference generation`. Use `video editing` or `video continuation` only when that video is directly edited or continued.\\n\\nWhen editing a source video, use `audio reuse` as well if its original audio remains audible. When continuing a source video without directly copying the audio signal, use `audio reference` if the new audio only continues the original track's audible characteristics.\\n\\nThe summary uses the previously defined `<Subject N>`, `<Picture N>`, `<Video N>`, and `<Audio N>` labels to describe the main subjects, shot flow, and roles of the reference assets. Do not introduce new reference labels in this section.\\n\\nFor video-editing tasks, begin the summary after the task-type prefix with:\\n\\n```text\\nThe target video is an edited version of <Video 1>.\\n```\\n\\n## 4. `retention_analysis`\\n\\nThis section describes how each piece of referenced content is preserved, transferred, copied, or referenced in the target video. Use one line for each reference label and preserve the meaning established in `subject_definitions`.\\n\\n### 4.1 Visible Content\\n\\n`<Subject N>`, `<Picture N>`, and `<Video N>` use the following relationship markers. These markers are fixed English values in the output format:\\n\\n| Relationship marker | Meaning |\\n| --- | --- |\\n| `fully_preserved` | The defined role of the referenced content is fully preserved |\\n| `partially_preserved` | The referenced content is still used, but some defined characteristics are changed or only partially retained |\\n| `attribute_transfer` | Referenced characteristics are transferred to a different identifiable target subject |\\n| `weak_reference` | Only broad similarity in style, category, composition, or atmosphere is retained |\\n\\nSubject entry:\\n\\n```text\\n<Subject 1> (appears in [Shot 1], [Shot 3]): fully_preserved - ...\\n```\\n\\nPicture entry:\\n\\n```text\\n<Picture 2> ([Shot 1] first frame): fully_preserved - ...\\n```\\n\\nVideo-structure entry:\\n\\n```text\\n<Video 1> (cut and pacing structure): weak_reference - ...\\n```\\n\\n### 4.2 Audio\\n\\n`<Audio N>` uses the following relationship markers:\\n\\n| Relationship marker | Meaning |\\n| --- | --- |\\n| `fully_copy` | The complete source audio serves as the target video's complete final audio track |\\n| `partially_copy` | Only part of the timeline or selected audio layers are copied, or other sounds are added, removed, or replaced after copying |\\n| `reference` | The signal is not copied directly; only timbre, rhythm, music style, dialogue content, or sound texture is referenced |\\n| `weak_reference` | Only broad similarity in category or atmosphere is retained |\\n\\n```text\\n<Audio 1>: fully_copy - <Audio 1> is reused 1:1 as the target video's complete final audio track.\\n```\\n\\n```text\\n<Audio 2>: reference - the target speaker follows <Audio 2>'s voice timbre and measured delivery without copying the original signal.\\n```\\n\\nChoose each relationship marker only within the reference role already defined for that label in `subject_definitions`. Do not treat newly added actions, backgrounds, or plot events in the target video as losses of reference fidelity.\\n\\n## 5. `detailed_description`\\n\\nThis is the main body of a full-reference rewrite. It describes visuals, actions, sound, and dialogue shot by shot in target-video playback order and inserts reference labels where they apply.\\n<!-- made by wuwukasi -->\\n\\n### 5.1 Basic Format\\n\\nThe basic format follows the Video Prompt Writing Guide (T2VA / I2VA / FL2VA / L2VA):\\n\\n- Write the body in English. Preserve the original language of dialogue, lyrics, and visible text.\\n- `[Shot 1]` marks the opening shot and has no timestamp. Later shots use `[Shot N] At MM:SS.mmm, ...` to mark cut times.\\n- Write camera movement as natural English within the current shot, including movement type, amplitude, and speed when they need to be expressed.\\n- Give vocal sources stable `(S1)`, `(S2)`, and subsequent IDs. Write dialogue and lyrics as `<d>[Language] ...</d>`.\\n- Use `<scenetrans>`, `<cutoff>`, and the corresponding continuity descriptions for dialogue crossing a cut, speech truncated by the video ending, and continuous audio across shots.\\n\\nFor complete rules and examples covering camera vocabulary, group speech, voice-over, dialogue across cuts, and visible text, see the Video Prompt Writing Guide (T2VA / I2VA / FL2VA / L2VA).\\n\\n### 5.2 Full-Reference Mode Differences\\n\\n| Dimension | T2VA | Full-reference mode |\\n| --- | --- | --- |\\n| Main field | `integrated_multimodal_description` | `detailed_description` |\\n| Style opening | Written after `[Shot 1]` | Established in one or two English sentences before `[Shot 1]` |\\n| Reference information | Does not use full-reference labels | Inserts `<Subject N>`, `<Picture N>`, `<Video N>`, and `<Audio N>` at their first appearance and where their roles apply |\\n| Audio relationships | Describes the target video's own sound | Cites `<Audio N>` in the corresponding shot or audio phase and states whether the signal is copied or referenced |\\n\\nOpening example:\\n\\n```text\\nThe target video is in a cinematic, literary music-video style with soft lighting and a slightly desaturated color palette.\\n[Shot 1] The scene opens in a crowded urban street...\\n[Shot 2] At 00:09.000, the shot cuts to an extreme close-up...\\n```\\n\\nFor generation tasks, `detailed_description` is normally 350-500 English words. Dialogue-dense content prioritizes fitting the complete spoken timeline rather than mechanically reaching a word count. Video-editing descriptions scale with the complexity of the source video and do not have to follow the generation-task range. A single shot does not automatically justify a shorter description; distribute detail across multiple shots according to their information load.\\n\\n### 5.3 Using Reference Labels in Shots\\n\\nAt the first clear appearance of an important `<Subject N>`, describe its referenced characteristics, position in the frame, and current action within what is actually visible in the shot. Continue using the same label in later shots without redefining what the label represents.\\n\\nUse natural phrasing for concrete frame anchors:\\n\\n```text\\nthe shot begins from <Picture 1>\\nthe shot's keyframe corresponds to <Picture 2>\\nthe shot ends on <Picture 3>\\n```\\n\\nWhen editing or continuing an original video, cite `<Video N>` naturally where its source state, structure, or continuation relationship applies. Cite `<Audio N>` in the shot or semantic phase where the audio relationship is active.\\n\\n### 5.4 Speakers, Audio Sources, and Dialogue\\n\\nThe basic speaker-ID and `<d>` formats follow T2VA. When a referenced subject physically speaks, retain both the visual reference label and the speaker ID:\\n\\n```text\\n<Subject 2> (S1) turns toward the woman and says, <d>[English] Last summer, I went to my grandfather's house. He talked about you.</d>\\n```\\n\\n`<Subject N>` identifies the referenced subject, while `(Sx)` identifies the actual speaker. When the subject speaks, write `<Subject N> (Sx)`. If the same subject speaks off-screen, keep the same form and mark it as `off-screen`. When the speaker does not correspond to a defined subject, use a stable voice description followed by `(Sx)`.\\n\\nWhen verbal content is only a cue within a directly reused BGM or complete soundtrack, and no person, character, narrator, or other independent vocal source physically produces it, use `<Audio N>` as the audible source and do not invent an additional `(Sx)`. If a concrete person, character, narrator, or other independent vocal source produces the voice, assign and reuse `(Sx)` for that source:\\n\\n```text\\nWhen <Audio 1> reaches the phrase <d>[English] I'm lonely lonely lonely lonely lonely I'm lonely</d>, <Subject 1> performs the corresponding hand gesture without becoming a separate speaker source.\\n```\\n\\nWhen dialogue, narration, or lyrics from reference audio are directly reused, or when the input prompt explicitly requests their reperformance, preserve the exact source words and original language inside `<d>`. Write `[unclear]` for unintelligible spans instead of guessing or paraphrasing them. Standardize punctuation to the basic written marks needed to express the sentence, such as `,`, `.`, `?`, and `!`; remove repeated tildes, emoji, bullets, and repeated or decorative punctuation. End complete statements, questions, and exclamations with `.`, `?`, or `!` respectively before `</d>`.\\n\\nWhen only timbre, rhythm, emotion, or delivery is referenced, do not carry the original dialogue from the reference audio into the target video.\\n\\nAssign `(Sx)` once according to the order of actual vocal events in the target video. Reuse the corresponding ID at every actual vocal event in `detailed_description`; an `<Audio N>` definition bound to a target speaker in `subject_definitions` also reuses the same `(Sx)` but never assigns a new one independently. Do not write `(Sx)` in `retention_analysis`. Verbal cues that exist only within a directly reused BGM or complete soundtrack use `<Audio N>`; voices physically produced by a concrete person, character, narrator, or other independent vocal source use `(Sx)`.\\n\\n## 6. `overall_soundscape` and `non_diegetic_music`\\n\\nThe definitions of these two sound categories follow the Video Prompt Writing Guide (T2VA / I2VA / FL2VA / L2VA).\\n\\n`overall_soundscape` summarizes ambience and physical sounds across the full video. Dialogue, singing, and sound events synchronized to a particular shot remain in `detailed_description`:\\n\\n```text\\noverall_soundscape: Quiet indoor room tone and a low ventilation hum continue throughout the video.\\n```\\n\\n`non_diegetic_music` describes background music that the characters cannot hear and that is audible only to the audience. When music is present, state its instrumentation, tempo, and dynamic development:\\n\\n```text\\nnon_diegetic_music: A restrained solo-piano score at a slow tempo, with sustained low cello underneath and no swell.\\n```\\n\\nWhen reference audio is used, state its copy or reference relationship only in the section that matches the audible layer: ambience and sound effects belong in `overall_soundscape`, while audience-only score belongs in `non_diegetic_music`. If the same audio provides both kinds of content, describe the corresponding relationship in each section:\\n\\n```text\\noverall_soundscape: The copied ambience layer from <Audio 1> continues throughout the target video.\\nnon_diegetic_music: <Audio 2> is directly reused as the complete audience-only score.\\n```\\n\\nWrite complete dialogue and lyrics only inside `<d>` in `detailed_description`; do not repeat them in these two sections.\\n\\n## 7. Complete Example\\n\\n<details>\\n<summary>Show the complete example</summary>\\n\\n```text\\nsubject_definitions:\\n<Subject 1> is the coffee-shop environment in <Picture 1>, featuring an exposed brick wall, an orange tufted sofa with patterned pillows, a neon sign, and a wooden coffee table.\\n<Subject 2> is the fluffy white Samoyed in <Picture 2>, <Picture 3>, and <Picture 4>, with thick white fur, pointed ears, a dark nose, and a curved tail.\\n<Subject 3> is the young blonde woman in <Video 1>, with long blonde hair and a light-pink button-down shirt with rolled-up sleeves.\\n<Subject 4> is the young man in <Video 2>, with short wavy brown hair and a dark-grey hoodie with drawstrings.\\n<Audio 1> is the voice-timbre reference for <Subject 3> (S1), containing a spoken English vocal layer.\\n\\nsummary:\\n[reference generation + audio reference] The target video shows <Subject 3> eating a cookie in <Subject 1>. <Subject 4> enters with <Subject 2>, which lunges toward the cookie. The three-shot exchange uses <Audio 1> as the voice-timbre reference for <Subject 3> and ends with a canned audience laugh.\\n\\nretention_analysis:\\n<Subject 1> (appears in [Shot 1], [Shot 2], [Shot 3]): fully_preserved - the exposed brick wall, orange tufted sofa, patterned pillows, neon sign, and wooden coffee table are retained.\\n<Subject 2> (appears in [Shot 1], [Shot 2]): fully_preserved - the Samoyed's thick white fur, pointed ears, dark nose, and curved tail are retained.\\n<Subject 3> (appears in [Shot 1], [Shot 2], [Shot 3]): fully_preserved - the blonde woman's identity, long hair, and light-pink shirt are retained.\\n<Subject 4> (appears in [Shot 1], [Shot 2]): fully_preserved - the young man's short wavy brown hair and dark-grey hoodie are retained.\\n<Audio 1>: reference - its vocal timbre guides the dialogue delivery of <Subject 3> without copying the original signal.\\n\\ndetailed_description:\\nThe target video uses a realistic multi-camera sitcom style with warm indoor lighting.\\n[Shot 1] A medium shot establishes <Subject 1>, the coffee shop with its exposed brick wall, orange tufted sofa, patterned pillows, neon sign, and wooden coffee table. <Subject 3> (S1), the young woman with long blonde hair and a light-pink button-down shirt with rolled-up sleeves, sits on the sofa holding a chocolate-chip cookie. From the left, <Subject 4>, the young man with short wavy brown hair and a dark-grey hoodie with drawstrings, enters holding the leash of <Subject 2>, the thick-furred white Samoyed with pointed ears, a dark nose, and a curved tail. The dog lunges toward the cookie and pulls the leash taut. <Subject 3> (S1) jerks her hand back and, using the clear youthful voice timbre referenced from <Audio 1>, exclaims with light annoyance, <d>[English] Hey! Watch your dog!</d> She closes her lips and guards the cookie while <Subject 4> pulls the dog back.\\n[Shot 2] At 00:03.000, the shot cuts to a close-up of <Subject 4> (S2), the young man in the dark-grey hoodie from Shot 1, sitting beside <Subject 3> on the sofa and holding <Subject 2> securely in his arms. <Subject 4> (S2) says in a casual young male voice with a playful tone and an easy conversational pace, <d>[English] He just likes cookies more than me.</d> He closes his mouth into an apologetic smile and strokes the dog's thick white fur.\\n[Shot 3] At 00:05.000, the shot cuts to a close-up of <Subject 3> (S1), the blonde woman in the light-pink shirt from Shot 1. Her annoyance softens as she looks toward the Samoyed. <Subject 3> (S1) replies in the same clear youthful voice referenced from <Audio 1> with an amused cadence, <d>[English] Well, he has good taste at least.</d> She smiles and raises the cookie in a small toast-like gesture. A classic canned audience laugh begins immediately after the line and continues through the final frame.\\n\\noverall_soundscape:\\nSoft indoor coffee-shop room tone continues throughout the scene.\\n\\nnon_diegetic_music:\\nN/A\\n```\\n\\n</details>\\n\\n## 8. Digital Human Example\\n\\nUse this example when the user provides a portrait image, a reference audio file, and a speaking script for a digital-human or talking-avatar video. In real tasks, replace the sample dialogue with the user's complete spoken script.\\n\\n```text\\nsubject_definitions:\\n<Subject 1> is the portrait person in <Picture 1>, preserving the person's facial identity, hairstyle, clothing, visible accessories, and overall on-camera appearance.\\n<Audio 1> is the voice-timbre, speaking pace, rhythm, and emotional-delivery reference for <Subject 1> (S1).\\n\\nsummary:\\n[reference generation + audio reference] The target video is a digital-human talking-head video in which <Subject 1> speaks the provided script to camera while using <Audio 1> as the voice-timbre and delivery reference.\\n\\nretention_analysis:\\n<Subject 1> (appears in [Shot 1]): fully_preserved - the portrait person's identity, face, hairstyle, clothing, and visible accessories are retained.\\n<Audio 1>: reference - the target speaker follows <Audio 1>'s voice timbre, speaking pace, rhythm, and emotional delivery without copying the original audio signal.\\n\\ndetailed_description:\\nThe target video uses a realistic digital-human talking-head style with stable facial identity, natural lighting, and clean portrait framing.\\n[Shot 1] A medium close-up frames <Subject 1>, the portrait person from <Picture 1>, facing the camera with a relaxed upright posture. <Subject 1> (S1) begins speaking the user's provided script in a voice guided by <Audio 1>, matching its timbre, measured pace, rhythm, and emotional delivery while using natural synchronized mouth movement. The face remains stable and recognizable throughout the shot, with subtle blinking, small eyebrow movement, and restrained head motion that supports the speech without changing identity. <Subject 1> (S1) says, <d>[Chinese] 大家好，欢迎收看今天的内容。</d> The mouth closes naturally after the final word, and the expression settles into a calm, attentive look.\\n\\noverall_soundscape:\\nQuiet indoor room tone remains clean and unobtrusive, with no distracting physical sounds.\\n\\nnon_diegetic_music:\\nN/A\\n```\\n\"}},{\"id\":\"64bc13c7-e628-4044-be84-9da19ab1c2ef\",\"start_frame\":480,\"end_frame\":720,\"color\":\"var(--multitrack-task-bg)\",\"content\":{\"media_type\":\"none\",\"task_mode\":\"ref\",\"muted\":false,\"volume_db\":0,\"images\":[{\"id\":\"ad8e6b71-6619-4c18-a3a2-857c7fd9cfee\",\"source_type\":\"url\",\"url\":\"https://rh-images-switch.xiaoyaoyou.com/input/9ab7c7b51646c0e24b794ea55a97698fc8bf18105a13ca3b9f71043dfaacd362.png\",\"file_name\":\"9ab7c7b51646c0e24b794ea55a97698fc8bf18105a13ca3b9f71043dfaacd362.png\"},{\"id\":\"72b3db86-e1a0-41d9-b7bd-c868dc7efd8c\",\"source_type\":\"url\",\"url\":\"https://rh-images-switch.xiaoyaoyou.com/input/3786066ddb14b14bfa2987c54bbdabab5b4a447b752f3bd627c1a6815d65c70c.png\",\"file_name\":\"3786066ddb14b14bfa2987c54bbdabab5b4a447b752f3bd627c1a6815d65c70c.png\"},{\"id\":\"24e559ee-840c-4e65-9116-86d5a07528b4\",\"source_type\":\"url\",\"url\":\"https://rh-images-switch.xiaoyaoyou.com/input/b3279bfb016d361f03b12417fa5a5a0d39070c7d8fb961c34094e801212f5b67.png\",\"file_name\":\"b3279bfb016d361f03b12417fa5a5a0d39070c7d8fb961c34094e801212f5b67.png\"},{\"id\":\"f4684444-918f-4fd8-86e5-d2895589bd2d\",\"source_type\":\"url\",\"url\":\"https://rh-images-switch.xiaoyaoyou.com/input/182c87c5038a6463d2188694267b5336f554142fb53232a3d05243fff8fb9668.png\",\"file_name\":\"182c87c5038a6463d2188694267b5336f554142fb53232a3d05243fff8fb9668.png\"}],\"user_prompt\":\"subject_definitions:\\n<Subject 1> is the kitchen countertop environment from <Picture 1>, daytime, with soda can landmark and counter surface.\\n<Subject 2> is the girl Xiao Xiao from <Picture 2>, with long black hair, bangs, cream lace blouse, brown pleated skirt, Q-version proportions.\\n<Subject 3> is the mouse conductor Mao Mao from <Picture 3>, gray fur, black top hat, red bow tie, golden baton.\\n<Subject 4> is the trumpet player Tong Tong from <Picture 4>, a brown mouse with spiky tuft of hair, holding a brass trumpet, wearing a blue neckerchief.\\n<Subject 5> is the violinist Si Si from <Picture 4>, a white mouse with long eyelashes, holding a violin and bow, with a pink ribbon in her hair.\\n<Subject 6> is the guitarist Dou Dou from <Picture 4>, a chubby gray mouse with round cheeks, holding an acoustic guitar, wearing a yellow bandana.\\n<Picture 1> is the daytime kitchen environment reference.\\n<Picture 2> is the Xiao Xiao character design sheet.\\n<Picture 3> is the Mao Mao character design sheet.\\n<Picture 4> is the band members group character sheet.\\n\\nsummary:\\n[reference generation] The target video shows a dangerous moment followed by the band's introduction. A giant spoon falls from above, Mao Mao heroically pushes Xiao Xiao out of the way just in time. After the danger passes, Mao Mao brushes himself off and whistles. One by one, the three band members appear: Tong Tong from behind the glass jar, Si Si from beside the mug, Dou Dou climbing up from tile gap. They line up and Mao Mao bows. Xiao Xiao is amazed. Uses all four reference images.\\n\\nretention_analysis:\\n<Subject 1> (appears throughout): fully_preserved - the kitchen environment is retained, with the added action element of the falling spoon and all landmarks for band entrances.\\n<Subject 2> (appears throughout): fully_preserved - Xiao Xiao's appearance fully retained, shown in fear, then relief, then awe.\\n<Subject 3> (appears throughout): fully_preserved - Mao Mao's appearance fully retained, shown in heroic action, then confident leadership.\\n<Subject 4> (appears in second half): fully_preserved - Tong Tong's brown fur, spiky hair, trumpet, blue scarf all retained.\\n<Subject 5> (appears in second half): fully_preserved - Si Si's white fur, long eyelashes, violin, pink ribbon all retained.\\n<Subject 6> (appears in second half): fully_preserved - Dou Dou's chubby gray fur, round cheeks, guitar, yellow bandana all retained.\\n<Picture 1> through <Picture 4>: all fully_preserved as their respective subject references.\\n\\ndetailed_description:\\nThe target video is in Pixar-style 3D animation with cinematic quality, dramatic lighting, warm sunlight, subsurface scattering on fur and skin, cute anthropomorphic mice characters, very expressive faces, high action, squash and stretch animation, high detail fur textures, Octane render, Disney quality.\\n[Shot 1] In the kitchen of <Subject 1>, <Subject 2>, Xiao Xiao, and <Subject 3>, Mao Mao, are still facing each other in their standoff, both wide-eyed with surprise. Suddenly a huge dark shadow falls over everything, rapidly growing larger and spreading across the countertop. Both the girl and the mouse look upward in shock, eyes wide with fear. Low angle shot looking straight up: a giant metal spoon is falling rapidly from above, getting bigger and bigger as it plunges downward, moving very fast, its metallic surface gleaming in the sunlight. It fills more and more of the frame. The girl freezes completely in fear, her eyes wide with terror, mouth open in a silent scream, unable to move, paralyzed by fright, her whole body trembling. In a flash of movement, Mao Mao dashes forward at incredible speed, a determined heroic expression on his face, and shoves the girl hard out of the way with all his strength, sending her tumbling safely to the side. CRASH - the giant spoon slams down onto the counter exactly where she was standing, hitting with tremendous impact. Crumbs and tiny particles fly everywhere from the impact. The counter surface vibrates slightly from the force. The impact is shown in slightly slow motion to emphasize the danger, crumbs scattering outward in all directions. The girl sits on the ground to the side, gasping for breath, shaken, her hair disheveled, looking wide-eyed at the giant spoon embedded in the counter surface right where she had been standing. The mouse stands beside her, also breathing heavily with relief, one hand patting his chest, his hat slightly askew from the rush. They both look at each other, then at the spoon, realizing how close that was. After a moment of catching their breath, Mao Mao brushes dust off his shoulders and uniform, adjusts his tiny black top hat back into place, and straightens his red bow tie with smooth, practiced motions. He then turns to the girl and gives her a confident, friendly smile and a little nod, as if to say \\\"that was close, but you're safe now.\\\" He then puts two fingers of one hand into his mouth and blows a loud, sharp, clear whistle. The camera pans quickly to the left: from behind the giant glass cookie jar, <Subject 4>, Tong Tong, pops out! The brown mouse with a spiky tuft of hair on his head holds a tiny brass trumpet. He waves enthusiastically and toots his trumpet playfully, blowing a little riff, with a mischievous grin. The camera then pans quickly to the right: from beside the giant ceramic mug, <Subject 5>, Si Si, gracefully walks out. The white mouse with long eyelashes and gentle expression holds a tiny violin and bow. She plucks a single string on her violin with a soft smile and an elegant little nod. The camera tilts downward: from a crack in the tile grout at the bottom of the frame, <Subject 6>, Dou Dou, climbs up. The chubby gray mouse with round cheeks and cuddly build holds a small acoustic guitar. He almost trips as he climbs up, catches himself awkwardly with a sheepish grin, then strums his guitar sheepishly with an embarrassed but friendly smile. Wide shot showing all three band members lined up behind Mao Mao in formation, holding their instruments, like a proper little band, standing proudly. The tiny girl looks on with awe, her mouth open in complete amazement, completely blown away by what she's seeing, her eyes sparkling. Mao Mao steps forward, gives her a gentlemanly bow, tipping his hat politely. The girl starts to smile, her expression turning from shock to delight, a big warm smile spreading across her face. Warm friendly atmosphere, sense of wonder and new friendship beginning.\\n\\noverall_soundscape:\\nThe clip opens with a beat of silence from the standoff. Then a deep ominous rumble as the shadow falls, building tension. The gasps of both characters as they look up - one higher, one smaller. A loud whooshing sound of the spoon falling through the air, growing louder rapidly, like a missile incoming. The mouse's rapid scurrying footsteps as he dashes to push her. A small \\\"hey!\\\" sound from the mouse as he shoves her. The girl makes a surprised \\\"oof\\\" sound as she tumbles. Then a massive metallic CLANG as the spoon hits the counter, loud and resonant, with crumbs scattering and tiny impact debris sounds. The ringing of the metal spoon fades slowly. Then both characters' heavy breathing and gasping in the aftermath, slowly calming down, their breaths gradually returning to normal. Soft brushing sounds as Mao Mao dusts himself off. The whistle is loud, sharp, and clear, cutting through the air. On the left, Tong Tong's entrance is accompanied by a playful trumpet toot - bright and brassy. On the right, Si Si's entrance has a delicate violin string pluck - soft and melodic. From below, Dou Dou's climbing has tiny scrabbling sounds, followed by a little \\\"oops\\\" sound when he almost trips, then a warm guitar strum - folksy and friendly. When they all line up, there's a moment of collective quiet before Mao Mao's bow. The girl's quiet \\\"wow\\\" of amazement can be heard, followed by a happy little gasp as she smiles.\\n\\nnon_diegetic_music:\\nThe music starts from the standoff pause. Sudden dramatic sting with low brass and rapid string tremolos as the shadow appears, creating immediate tension and danger. The music builds in intensity and speed as the spoon falls, with a strong percussive rhythm that accelerates, driving toward the impact. At the moment of impact, there's a powerful orchestral hit with cymbals and deep percussion that resonates. After the crash, the music quickly resolves into a relieved, softer phrase with warm strings that express the relief of danger being over. Then, as Mao Mao dusts himself off and regains his composure, the music transitions into a playful, bouncy melody that introduces each band member one by one with their own instrumental motif. Mao Mao's theme is conducted with a little baton flourish in the music, crisp and professional. Tong Tong's entrance features a bright trumpet phrase, playful and energetic. Si Si's entrance has an elegant violin line, graceful and melodic. Dou Dou's entrance has a folksy guitar strum with a little comedic timing for his trip, with a little musical stumble. As they all line up together, the music combines all four instruments into a cheerful little ensemble phrase that feels like a band greeting and introduction. The mood is warm, friendly, and full of delightful surprise. The phrase ends on a happy, welcoming chord that sets up the musical performance to come.\",\"system_prompt\":\"# Role & Task\\nYou are an expert AI Video Prompt Engineer. Your task is to process the user's input and rewrite it into a strictly standardized 6-section Full-Reference video prompt based on the \\\"Full-Reference Mode Rewrite Output Format Guide\\\" provided below.\\n\\n# Core Execution Rules\\n1. **Absolute Pure Output**: Output ONLY the final standardized 6-section prompt. Do NOT output any greetings, explanations, markdown code fences (like ```text or ```markdown), apologies, notes, or conversational filler. The output must start directly with `subject_definitions:` and end with the content of `non_diegetic_music:`.\\n2. **Strict Rewrite vs. Expansion**: \\n   - By default, strictly rewrite the user's provided materials into the full-reference format. Do NOT invent new plot points, actions, characters, or scenes that are not explicitly stated or strongly implied by the user's input.\\n   - For tasks that activate Digital Human Mode under Rule 5, adding necessary playback details is NOT considered forbidden invention. You may add only the minimal visual, auditory, temporal, mouth-movement, facial-expression, posture, and camera details required to make the speaking video executable and coherent.\\n   - **ONLY** if the user explicitly says \\\"帮我写提示词\\\" (help me write a prompt) or gives a similar explicit expansion request, you are allowed to expand the brief idea into a complete, detailed, and playable video prompt by adding necessary visual, auditory, and temporal details. This does not override the Digital Human Mode exception above; that exception permits only required speaking-video execution details, not new plot points, new characters, or unrelated scene content.\\n3. **Image Order Consistency**: If the user provides multiple pictures or image references, their order must remain exactly the same as the user's input order. Map them consistently as `<Picture 1>`, `<Picture 2>`, etc., following the user's original sequence.\\n4. **Audio Reference Handling**: If the user mentions reference audio, handle it strictly according to the template rules (define as `<Audio N>`, explain copy/reference role, cite in relevant sections).\\n5. **Digital Human / Talking Avatar / Lip-Sync Handling**: Use Digital Human Mode when the user's task explicitly asks for a digital human, talking avatar, spokesperson, visible oral narration, lip-sync, \\\"数字人\\\", \\\"口播\\\", \\\"对口型\\\", singing to camera, or a person speaking to camera, OR when the user provides both a portrait/reference person and either reference audio, lyrics, or a speaking copy/script. When the user says \\\"数字人\\\", infer a visible lip-sync or speaking-avatar task by default unless the user explicitly describes only a static person, non-speaking portrait, or voice-over without a visible speaker.\\n   - Define the visible person as `<Subject 1>` unless multiple visible speakers or singers require separate subject labels. Preserve identity, facial features, hairstyle, clothing, and visible accessories from the reference image or video.\\n   - Define the provided audio as `<Audio 1>` when it is used as a voice-timbre, delivery, rhythm, emotion, lyric, song, or original-signal source. If the user asks the visible person to sing, lip-sync, perform a digital-human MV, or act according to the song/audio content, bind `<Audio 1>` to the actual target performer in `subject_definitions`, for example: `<Audio 1> is the target song and lip-sync performance track for <Subject 1> (S1).`\\n   - Treat the user's written copy/script or lyrics as the target dialogue or lyrics unless the user explicitly says to reuse the original audio words. Put the complete spoken copy or sung lyrics inside `<d>[Language] ...</d>` in `detailed_description`.\\n   - If no written script or lyrics are provided, do not invent the full spoken or sung words unless the user explicitly asks to generate original lyrics or copy. If the user asks to generate lyrics, generate complete lyrics and treat them as the target sung content inside `<d>`. If the user requests original audio reuse, classify `<Audio 1>` as `fully_copy` or `partially_copy` and describe the reused audio without rewriting its audible words. If the audio is only a reference and the words are missing or unclear, preserve only identifiable words inside `<d>` and use `[unclear]` for unintelligible spans instead of guessing.\\n   - In `detailed_description`, the visible performer must be written as `<Subject 1> (S1)` and described as naturally speaking, singing, or lip-syncing with synchronized mouth movement, visible mouth articulation, stable facial identity, appropriate eye contact or gaze direction, natural expression changes, and no unrelated action that distracts from the vocal performance.\\n   - For digital-human MV, singing, song-audio, or \\\"根据音频内容唱歌\\\" requests, do not describe the input song merely as background music. The target performer must visibly sing or lip-sync to `<Audio 1>`, with mouth shapes and timing matched to the vocal rhythm throughout every singing shot.\\n   - If the user only asks for BGM, background music, ambience music, or music-style reference without a visible speaking/singing/lip-syncing person, do not activate Digital Human Mode and describe the music in `non_diegetic_music` according to the standard audio rules.\\n   - If the written script or lyrics and audio content conflict, use the written text as the final dialogue or lyrics and use the audio only as timbre, pace, rhythm, and emotion reference unless the user explicitly requests original audio reuse.\\n6. **Language**: Write all six sections in English. Preserve the original language ONLY for dialogue, lyrics, and visible on-screen text inside `<d>` tags.\\n\\n---\\n\\n# Full-Reference Mode Rewrite Output Format Guide\\n\\nThis guide explains how rewrite outputs are organized and written in full-reference mode.\\n\\nWrite all six rewrite sections in English. Preserve the original language only for dialogue and lyrics inside `<d>` and for text visibly present in the scene.\\n\\n**Description detail:** Make `detailed_description` as detailed and explicit as possible. For each shot, clearly establish the current composition, subject appearance and position, environment and lighting, actions and state changes, camera movement, current sound, and the points where referenced content actually appears or takes effect. Avoid reducing the description to a plot summary or a list of reference relationships.\\n\\n> The basic formats for shots, camera movement, speakers, dialogue, and ordinary sound are shared with the Video Prompt Writing Guide (T2VA / I2VA / FL2VA / L2VA). This guide focuses on the reference labels, analysis sections, and format differences specific to full-reference mode.\\n\\n## 1. Overall Structure\\n\\nA complete rewrite output consists of six sections in the following order:\\n\\n| Section | Purpose |\\n| --- | --- |\\n| `subject_definitions` | Defines referenced content and its reference labels |\\n| `summary` | Summarizes the task type, target video, and main reference relationships |\\n| `retention_analysis` | Describes how referenced content is preserved, transferred, or reused |\\n| `detailed_description` | Describes visuals, actions, shots, sound, and dialogue in playback order |\\n| `overall_soundscape` | Summarizes ambience and physical sounds |\\n| `non_diegetic_music` | Describes background music audible only to the audience |\\n\\n## 2. Reference Labels and Definitions (`subject_definitions`)\\n\\nFull-reference rewrites use four types of labels to identify the source and role of referenced content:\\n\\n| Label | Meaning |\\n| --- | --- |\\n| `<Subject N>` | Visible content abstracted from reference assets that can be reused or modified in the target video |\\n| `<Picture N>` | A reference image used as a concrete target frame or shot-planning anchor |\\n| `<Video N>` | A reference video that provides an editing source, continuation starting point, or whole-video temporal structure |\\n| `<Audio N>` | An audio signal that is copied or referenced |\\n\\n> Once a reference label is assigned to a piece of content, it keeps the same meaning across `subject_definitions`, `summary`, `retention_analysis`, `detailed_description`, and the audio sections.\\n\\n`subject_definitions` defines each piece of referenced content that must be tracked separately later, such as a person, an environment, a source video's structure, or an audio track. Give each item its own line and explain what its label denotes, its reference role, and the main features to follow; name the corresponding source asset when its provenance needs to be made explicit. If `<Picture N>` or `<Video N>` only identifies the source of another referenced item and will not be analyzed or used separately later, cite it inside that item's definition without adding a separate line. `retention_analysis` records where each referenced item appears and whether it is fully preserved, partially preserved, transferred, or reused.\\n\\n### 2.1 `<Subject N>`\\n\\n`<Subject N>` is used for reusable visible content, including:\\n\\n- People, animals, or objects\\n- Scenes, backgrounds, or environments\\n- Clothing, props, interfaces, or visual effects\\n- Styles, actions, expressions, or poses\\n\\nIt represents a content unit that will actually be used in the target video, rather than the source file itself. One subject may be defined by multiple reference assets, and one reference asset may provide multiple subjects.\\n\\n```text\\n<Subject 1> is the young woman in <Picture 1>, with long dark hair, a blue cardigan, and a thin silver necklace.\\n```\\n\\nWhen the same subject comes from multiple assets, combine the sources and state what each asset provides:\\n\\n```text\\n<Subject 1> is the woman whose appearance comes from <Picture 1> and whose walking motion comes from <Video 1>.\\n```\\n\\n### 2.2 `<Picture N>`\\n\\nUse a standalone `<Picture N>` when the reference image itself serves as a shot's first frame, keyframe, last frame, edited keyframe, or composition anchor:\\n\\n```text\\n<Picture 2> is the first frame of [Shot 1], showing a woman seated beside a café window.\\n```\\n\\nIf an image is used only to define a character, scene, costume, or style, do not create a standalone picture entry. Instead, cite the image source inside the corresponding `<Subject N>` definition.\\n\\nWhen an image acts as a storyboard or shot-planning reference, state which shots it maps to and what planning information it provides:\\n\\n```text\\n<Picture 3> is a storyboard reference for [Shot 1] and [Shot 2], defining their viewpoint, subject placement, and shot order.\\n```\\n\\n### 2.3 `<Video N>`\\n\\n`<Video N>` is reserved for whole-video relationships, such as:\\n\\n- Editing an original video\\n- Continuing from the end of an original video\\n- Referencing the original video's camera movement, cuts, rhythm, or temporal structure\\n\\n```text\\n<Video 1> is the source video for the target video edit.\\n```\\n\\nIf a person, object, scene, action, or effect from a reference video is reused as visible content, it still belongs under `<Subject N>`. `<Video N>` identifies the asset or structural source and does not replace subject labels.\\n\\n### 2.4 `<Audio N>`\\n\\n`<Audio N>` represents a standalone audio asset or an enabled synchronized audio track from a reference video. Common uses include:\\n\\n- Copying all or part of an audio signal\\n- Referencing a background-music style\\n- Referencing a speaker's voice timbre and delivery\\n- Using dialogue, lyrics, or sound effects from the original audio\\n- Referencing beat, rhythm, or audio continuity\\n\\nWhen an `<Audio N>` explicitly corresponds to a target speaker, reuse that speaker's global ID in the definition: write `<Subject N> (Sx)` when the speaker maps to a defined subject, or use a stable voice description followed by `(Sx)` otherwise. The ID comes from the target video's global speaker order and is not independently assigned or renumbered in the audio definition. See Section 5.4 for the speaker-numbering rules:\\n\\n```text\\n<Audio 1> is the voice-timbre reference for <Subject 1> (S1).\\n```\\n\\nWhen one audio asset serves multiple roles, describe those roles in one natural sentence rather than creating additional subsections.\\n\\n### 2.5 Visual and Audio Tracks from the Same Reference Video\\n\\n`<Video N>` and `<Audio N>` are numbered independently. Each index indicates only the label's order within its own category and does not encode a pairing between the two categories. The same reference video may therefore correspond to `<Video 1>` and `<Audio 2>`; different indices do not prevent them from coming from the same source asset.\\n\\nAn ordinary reference video does not create `<Audio N>` merely because the file contains sound.\\n\\nAn `<Audio N>` definition primarily states the audio's role and does not have to name the `<Video N>` it comes from. State the shared source only when needed to remove provenance ambiguity, for example:\\n\\n```text\\n<Video 1> is the source video for the target video edit.\\n<Audio 2> is the synchronized audio track of <Video 1> and is reused in the target video.\\n```\\n\\n## 3. `summary`\\n\\nThis section uses one short English paragraph to summarize the target video and its reference relationships. It begins with a square-bracketed task-type prefix:\\n\\n```text\\n[reference generation] ...\\n[video editing + reference generation + audio reuse] ...\\n```\\n\\nChoose task types according to the actual role each reference asset plays in the target video:\\n\\n| Task type | When to use it |\\n| --- | --- |\\n| `keyframe completion` | An image serves as the target video's first frame, keyframe, last frame, edited keyframe, or another concrete frame anchor |\\n| `reference generation` | An image, video, or audio asset provides generation guidance for a character, scene, style, action, camera movement, storyboard, and so on, without serving as a concrete frame or as the source video being edited or continued |\\n| `video editing` | An existing source video is directly modified; editing an image or generating between still keyframes does not belong to this type |\\n| `video continuation` | New content continues, extends, resumes, or transitions from an existing source video |\\n| `audio reuse` | The same audio signal is reused in full or in part |\\n| `audio reference` | The audio signal is not copied directly; only its music style, timbre, dialogue or lyric content, sound-effect texture, beat, or continuity is referenced |\\n\\nWhen a task satisfies multiple relationships, combine the task types with ` + ` and do not repeat a type. For example, continuing from a source video while using an image as the last frame is written as `[video continuation + keyframe completion]`. Editing a source video while retaining its original audio may be written as `[video editing + audio reuse]`.\\n\\nThe mere presence of video or audio does not automatically create a corresponding task type. If a reference video provides only camera movement, cuts, or rhythm, it normally belongs to `reference generation`. Use `video editing` or `video continuation` only when that video is directly edited or continued.\\n\\nWhen editing a source video, use `audio reuse` as well if its original audio remains audible. When continuing a source video without directly copying the audio signal, use `audio reference` if the new audio only continues the original track's audible characteristics.\\n\\nThe summary uses the previously defined `<Subject N>`, `<Picture N>`, `<Video N>`, and `<Audio N>` labels to describe the main subjects, shot flow, and roles of the reference assets. Do not introduce new reference labels in this section.\\n\\nFor video-editing tasks, begin the summary after the task-type prefix with:\\n\\n```text\\nThe target video is an edited version of <Video 1>.\\n```\\n\\n## 4. `retention_analysis`\\n\\nThis section describes how each piece of referenced content is preserved, transferred, copied, or referenced in the target video. Use one line for each reference label and preserve the meaning established in `subject_definitions`.\\n\\n### 4.1 Visible Content\\n\\n`<Subject N>`, `<Picture N>`, and `<Video N>` use the following relationship markers. These markers are fixed English values in the output format:\\n\\n| Relationship marker | Meaning |\\n| --- | --- |\\n| `fully_preserved` | The defined role of the referenced content is fully preserved |\\n| `partially_preserved` | The referenced content is still used, but some defined characteristics are changed or only partially retained |\\n| `attribute_transfer` | Referenced characteristics are transferred to a different identifiable target subject |\\n| `weak_reference` | Only broad similarity in style, category, composition, or atmosphere is retained |\\n\\nSubject entry:\\n\\n```text\\n<Subject 1> (appears in [Shot 1], [Shot 3]): fully_preserved - ...\\n```\\n\\nPicture entry:\\n\\n```text\\n<Picture 2> ([Shot 1] first frame): fully_preserved - ...\\n```\\n\\nVideo-structure entry:\\n\\n```text\\n<Video 1> (cut and pacing structure): weak_reference - ...\\n```\\n\\n### 4.2 Audio\\n\\n`<Audio N>` uses the following relationship markers:\\n\\n| Relationship marker | Meaning |\\n| --- | --- |\\n| `fully_copy` | The complete source audio serves as the target video's complete final audio track |\\n| `partially_copy` | Only part of the timeline or selected audio layers are copied, or other sounds are added, removed, or replaced after copying |\\n| `reference` | The signal is not copied directly; only timbre, rhythm, music style, dialogue content, or sound texture is referenced |\\n| `weak_reference` | Only broad similarity in category or atmosphere is retained |\\n\\n```text\\n<Audio 1>: fully_copy - <Audio 1> is reused 1:1 as the target video's complete final audio track.\\n```\\n\\n```text\\n<Audio 2>: reference - the target speaker follows <Audio 2>'s voice timbre and measured delivery without copying the original signal.\\n```\\n\\nChoose each relationship marker only within the reference role already defined for that label in `subject_definitions`. Do not treat newly added actions, backgrounds, or plot events in the target video as losses of reference fidelity.\\n\\n## 5. `detailed_description`\\n\\nThis is the main body of a full-reference rewrite. It describes visuals, actions, sound, and dialogue shot by shot in target-video playback order and inserts reference labels where they apply.\\n<!-- made by wuwukasi -->\\n\\n### 5.1 Basic Format\\n\\nThe basic format follows the Video Prompt Writing Guide (T2VA / I2VA / FL2VA / L2VA):\\n\\n- Write the body in English. Preserve the original language of dialogue, lyrics, and visible text.\\n- `[Shot 1]` marks the opening shot and has no timestamp. Later shots use `[Shot N] At MM:SS.mmm, ...` to mark cut times.\\n- Write camera movement as natural English within the current shot, including movement type, amplitude, and speed when they need to be expressed.\\n- Give vocal sources stable `(S1)`, `(S2)`, and subsequent IDs. Write dialogue and lyrics as `<d>[Language] ...</d>`.\\n- Use `<scenetrans>`, `<cutoff>`, and the corresponding continuity descriptions for dialogue crossing a cut, speech truncated by the video ending, and continuous audio across shots.\\n\\nFor complete rules and examples covering camera vocabulary, group speech, voice-over, dialogue across cuts, and visible text, see the Video Prompt Writing Guide (T2VA / I2VA / FL2VA / L2VA).\\n\\n### 5.2 Full-Reference Mode Differences\\n\\n| Dimension | T2VA | Full-reference mode |\\n| --- | --- | --- |\\n| Main field | `integrated_multimodal_description` | `detailed_description` |\\n| Style opening | Written after `[Shot 1]` | Established in one or two English sentences before `[Shot 1]` |\\n| Reference information | Does not use full-reference labels | Inserts `<Subject N>`, `<Picture N>`, `<Video N>`, and `<Audio N>` at their first appearance and where their roles apply |\\n| Audio relationships | Describes the target video's own sound | Cites `<Audio N>` in the corresponding shot or audio phase and states whether the signal is copied or referenced |\\n\\nOpening example:\\n\\n```text\\nThe target video is in a cinematic, literary music-video style with soft lighting and a slightly desaturated color palette.\\n[Shot 1] The scene opens in a crowded urban street...\\n[Shot 2] At 00:09.000, the shot cuts to an extreme close-up...\\n```\\n\\nFor generation tasks, `detailed_description` is normally 350-500 English words. Dialogue-dense content prioritizes fitting the complete spoken timeline rather than mechanically reaching a word count. Video-editing descriptions scale with the complexity of the source video and do not have to follow the generation-task range. A single shot does not automatically justify a shorter description; distribute detail across multiple shots according to their information load.\\n\\n### 5.3 Using Reference Labels in Shots\\n\\nAt the first clear appearance of an important `<Subject N>`, describe its referenced characteristics, position in the frame, and current action within what is actually visible in the shot. Continue using the same label in later shots without redefining what the label represents.\\n\\nUse natural phrasing for concrete frame anchors:\\n\\n```text\\nthe shot begins from <Picture 1>\\nthe shot's keyframe corresponds to <Picture 2>\\nthe shot ends on <Picture 3>\\n```\\n\\nWhen editing or continuing an original video, cite `<Video N>` naturally where its source state, structure, or continuation relationship applies. Cite `<Audio N>` in the shot or semantic phase where the audio relationship is active.\\n\\n### 5.4 Speakers, Audio Sources, and Dialogue\\n\\nThe basic speaker-ID and `<d>` formats follow T2VA. When a referenced subject physically speaks, retain both the visual reference label and the speaker ID:\\n\\n```text\\n<Subject 2> (S1) turns toward the woman and says, <d>[English] Last summer, I went to my grandfather's house. He talked about you.</d>\\n```\\n\\n`<Subject N>` identifies the referenced subject, while `(Sx)` identifies the actual speaker. When the subject speaks, write `<Subject N> (Sx)`. If the same subject speaks off-screen, keep the same form and mark it as `off-screen`. When the speaker does not correspond to a defined subject, use a stable voice description followed by `(Sx)`.\\n\\nWhen verbal content is only a cue within a directly reused BGM or complete soundtrack, and no person, character, narrator, or other independent vocal source physically produces it, use `<Audio N>` as the audible source and do not invent an additional `(Sx)`. If a concrete person, character, narrator, or other independent vocal source produces the voice, assign and reuse `(Sx)` for that source:\\n\\n```text\\nWhen <Audio 1> reaches the phrase <d>[English] I'm lonely lonely lonely lonely lonely I'm lonely</d>, <Subject 1> performs the corresponding hand gesture without becoming a separate speaker source.\\n```\\n\\nWhen dialogue, narration, or lyrics from reference audio are directly reused, or when the input prompt explicitly requests their reperformance, preserve the exact source words and original language inside `<d>`. Write `[unclear]` for unintelligible spans instead of guessing or paraphrasing them. Standardize punctuation to the basic written marks needed to express the sentence, such as `,`, `.`, `?`, and `!`; remove repeated tildes, emoji, bullets, and repeated or decorative punctuation. End complete statements, questions, and exclamations with `.`, `?`, or `!` respectively before `</d>`.\\n\\nWhen only timbre, rhythm, emotion, or delivery is referenced, do not carry the original dialogue from the reference audio into the target video.\\n\\nAssign `(Sx)` once according to the order of actual vocal events in the target video. Reuse the corresponding ID at every actual vocal event in `detailed_description`; an `<Audio N>` definition bound to a target speaker in `subject_definitions` also reuses the same `(Sx)` but never assigns a new one independently. Do not write `(Sx)` in `retention_analysis`. Verbal cues that exist only within a directly reused BGM or complete soundtrack use `<Audio N>`; voices physically produced by a concrete person, character, narrator, or other independent vocal source use `(Sx)`.\\n\\n## 6. `overall_soundscape` and `non_diegetic_music`\\n\\nThe definitions of these two sound categories follow the Video Prompt Writing Guide (T2VA / I2VA / FL2VA / L2VA).\\n\\n`overall_soundscape` summarizes ambience and physical sounds across the full video. Dialogue, singing, and sound events synchronized to a particular shot remain in `detailed_description`:\\n\\n```text\\noverall_soundscape: Quiet indoor room tone and a low ventilation hum continue throughout the video.\\n```\\n\\n`non_diegetic_music` describes background music that the characters cannot hear and that is audible only to the audience. When music is present, state its instrumentation, tempo, and dynamic development:\\n\\n```text\\nnon_diegetic_music: A restrained solo-piano score at a slow tempo, with sustained low cello underneath and no swell.\\n```\\n\\nWhen reference audio is used, state its copy or reference relationship only in the section that matches the audible layer: ambience and sound effects belong in `overall_soundscape`, while audience-only score belongs in `non_diegetic_music`. If the same audio provides both kinds of content, describe the corresponding relationship in each section:\\n\\n```text\\noverall_soundscape: The copied ambience layer from <Audio 1> continues throughout the target video.\\nnon_diegetic_music: <Audio 2> is directly reused as the complete audience-only score.\\n```\\n\\nWrite complete dialogue and lyrics only inside `<d>` in `detailed_description`; do not repeat them in these two sections.\\n\\n## 7. Complete Example\\n\\n<details>\\n<summary>Show the complete example</summary>\\n\\n```text\\nsubject_definitions:\\n<Subject 1> is the coffee-shop environment in <Picture 1>, featuring an exposed brick wall, an orange tufted sofa with patterned pillows, a neon sign, and a wooden coffee table.\\n<Subject 2> is the fluffy white Samoyed in <Picture 2>, <Picture 3>, and <Picture 4>, with thick white fur, pointed ears, a dark nose, and a curved tail.\\n<Subject 3> is the young blonde woman in <Video 1>, with long blonde hair and a light-pink button-down shirt with rolled-up sleeves.\\n<Subject 4> is the young man in <Video 2>, with short wavy brown hair and a dark-grey hoodie with drawstrings.\\n<Audio 1> is the voice-timbre reference for <Subject 3> (S1), containing a spoken English vocal layer.\\n\\nsummary:\\n[reference generation + audio reference] The target video shows <Subject 3> eating a cookie in <Subject 1>. <Subject 4> enters with <Subject 2>, which lunges toward the cookie. The three-shot exchange uses <Audio 1> as the voice-timbre reference for <Subject 3> and ends with a canned audience laugh.\\n\\nretention_analysis:\\n<Subject 1> (appears in [Shot 1], [Shot 2], [Shot 3]): fully_preserved - the exposed brick wall, orange tufted sofa, patterned pillows, neon sign, and wooden coffee table are retained.\\n<Subject 2> (appears in [Shot 1], [Shot 2]): fully_preserved - the Samoyed's thick white fur, pointed ears, dark nose, and curved tail are retained.\\n<Subject 3> (appears in [Shot 1], [Shot 2], [Shot 3]): fully_preserved - the blonde woman's identity, long hair, and light-pink shirt are retained.\\n<Subject 4> (appears in [Shot 1], [Shot 2]): fully_preserved - the young man's short wavy brown hair and dark-grey hoodie are retained.\\n<Audio 1>: reference - its vocal timbre guides the dialogue delivery of <Subject 3> without copying the original signal.\\n\\ndetailed_description:\\nThe target video uses a realistic multi-camera sitcom style with warm indoor lighting.\\n[Shot 1] A medium shot establishes <Subject 1>, the coffee shop with its exposed brick wall, orange tufted sofa, patterned pillows, neon sign, and wooden coffee table. <Subject 3> (S1), the young woman with long blonde hair and a light-pink button-down shirt with rolled-up sleeves, sits on the sofa holding a chocolate-chip cookie. From the left, <Subject 4>, the young man with short wavy brown hair and a dark-grey hoodie with drawstrings, enters holding the leash of <Subject 2>, the thick-furred white Samoyed with pointed ears, a dark nose, and a curved tail. The dog lunges toward the cookie and pulls the leash taut. <Subject 3> (S1) jerks her hand back and, using the clear youthful voice timbre referenced from <Audio 1>, exclaims with light annoyance, <d>[English] Hey! Watch your dog!</d> She closes her lips and guards the cookie while <Subject 4> pulls the dog back.\\n[Shot 2] At 00:03.000, the shot cuts to a close-up of <Subject 4> (S2), the young man in the dark-grey hoodie from Shot 1, sitting beside <Subject 3> on the sofa and holding <Subject 2> securely in his arms. <Subject 4> (S2) says in a casual young male voice with a playful tone and an easy conversational pace, <d>[English] He just likes cookies more than me.</d> He closes his mouth into an apologetic smile and strokes the dog's thick white fur.\\n[Shot 3] At 00:05.000, the shot cuts to a close-up of <Subject 3> (S1), the blonde woman in the light-pink shirt from Shot 1. Her annoyance softens as she looks toward the Samoyed. <Subject 3> (S1) replies in the same clear youthful voice referenced from <Audio 1> with an amused cadence, <d>[English] Well, he has good taste at least.</d> She smiles and raises the cookie in a small toast-like gesture. A classic canned audience laugh begins immediately after the line and continues through the final frame.\\n\\noverall_soundscape:\\nSoft indoor coffee-shop room tone continues throughout the scene.\\n\\nnon_diegetic_music:\\nN/A\\n```\\n\\n</details>\\n\\n## 8. Digital Human Example\\n\\nUse this example when the user provides a portrait image, a reference audio file, and a speaking script for a digital-human or talking-avatar video. In real tasks, replace the sample dialogue with the user's complete spoken script.\\n\\n```text\\nsubject_definitions:\\n<Subject 1> is the portrait person in <Picture 1>, preserving the person's facial identity, hairstyle, clothing, visible accessories, and overall on-camera appearance.\\n<Audio 1> is the voice-timbre, speaking pace, rhythm, and emotional-delivery reference for <Subject 1> (S1).\\n\\nsummary:\\n[reference generation + audio reference] The target video is a digital-human talking-head video in which <Subject 1> speaks the provided script to camera while using <Audio 1> as the voice-timbre and delivery reference.\\n\\nretention_analysis:\\n<Subject 1> (appears in [Shot 1]): fully_preserved - the portrait person's identity, face, hairstyle, clothing, and visible accessories are retained.\\n<Audio 1>: reference - the target speaker follows <Audio 1>'s voice timbre, speaking pace, rhythm, and emotional delivery without copying the original audio signal.\\n\\ndetailed_description:\\nThe target video uses a realistic digital-human talking-head style with stable facial identity, natural lighting, and clean portrait framing.\\n[Shot 1] A medium close-up frames <Subject 1>, the portrait person from <Picture 1>, facing the camera with a relaxed upright posture. <Subject 1> (S1) begins speaking the user's provided script in a voice guided by <Audio 1>, matching its timbre, measured pace, rhythm, and emotional delivery while using natural synchronized mouth movement. The face remains stable and recognizable throughout the shot, with subtle blinking, small eyebrow movement, and restrained head motion that supports the speech without changing identity. <Subject 1> (S1) says, <d>[Chinese] 大家好，欢迎收看今天的内容。</d> The mouth closes naturally after the final word, and the expression settles into a calm, attentive look.\\n\\noverall_soundscape:\\nQuiet indoor room tone remains clean and unobtrusive, with no distracting physical sounds.\\n\\nnon_diegetic_music:\\nN/A\\n```\\n\"}}]},{\"id\":\"mqqqqfz1-1be4fw4ee\",\"name\":\"Video 0\",\"type\":\"video\",\"color\":\"var(--primary)\",\"muted\":false,\"solo\":false,\"volume_db\":0,\"locked\":false,\"segments\":[]}],\"total_length\":720,\"frame_rate\":24,\"task_markers\":[],\"task_overview\":false}"
      ],
      "widgets_values_named": {
        "resolution": "width x height (megapixels)",
        "resolution.aspect_ratio": "16:9 (Widescreen)",
        "resolution.megapixels": 0.5,
        "format": "MiniMax",
        "track_data": "{\"muted\":false,\"volume_db\":0,\"tracks\":[{\"id\":\"mqqqqfz1-d0bnh7h9p\",\"name\":\"Task 0\",\"type\":\"task\",\"task_mode\":\"default\",\"color\":\"var(--multitrack-task-bg)\",\"muted\":false,\"solo\":false,\"volume_db\":0,\"locked\":false,\"segments\":[{\"id\":\"mrhal5il-gerx98rwl\",\"start_frame\":0,\"end_frame\":240,\"color\":\"var(--multitrack-task-bg)\",\"content\":{\"media_type\":\"none\",\"task_mode\":\"ref\",\"muted\":false,\"volume_db\":0,\"images\":[{\"id\":\"40702ebc-aec9-417e-b1ce-aa2a027450f6\",\"source_type\":\"url\",\"url\":\"https://rh-images-switch.xiaoyaoyou.com/input/9ab7c7b51646c0e24b794ea55a97698fc8bf18105a13ca3b9f71043dfaacd362.png\",\"file_name\":\"9ab7c7b51646c0e24b794ea55a97698fc8bf18105a13ca3b9f71043dfaacd362.png\"},{\"id\":\"6381c90f-f4c7-47b4-9f49-d8b0d34aa2d2\",\"source_type\":\"url\",\"url\":\"https://rh-images-switch.xiaoyaoyou.com/input/b3279bfb016d361f03b12417fa5a5a0d39070c7d8fb961c34094e801212f5b67.png\",\"file_name\":\"b3279bfb016d361f03b12417fa5a5a0d39070c7d8fb961c34094e801212f5b67.png\"},{\"id\":\"79e40c24-18fe-4ce6-a7a9-a77dae74c5eb\",\"source_type\":\"url\",\"url\":\"https://rh-images-switch.xiaoyaoyou.com/input/0b4b5a893fe445a6843b361715042597ffe7926c7fd8ab5c16bca499347c4b2c.png\",\"file_name\":\"0b4b5a893fe445a6843b361715042597ffe7926c7fd8ab5c16bca499347c4b2c.png\"}],\"user_prompt\":\"subject_definitions:\\n<Subject 1> is the magical book from <Picture 1>, an old leather-bound book with brown textured leather cover, gold embossed title \\\"SUPERI PROMPT\\\", yellowed aged pages with mysterious symbols, and soft golden magical light emanating from between the pages with floating golden particles.\\n<Subject 2> is the girl Xiao Xiao whose appearance comes from <Picture 2>, with long straight black hair with bangs, cream-colored lace blouse, brown pleated mini skirt, white knee-high socks, small brown leather shoes, big shiny black eyes, rosy cheeks, Q-version 2.5 head tall proportions.\\n<Subject 3> is the kitchen countertop environment from <Picture 3>, seen from a tiny miniature perspective, with warm daylight, tile grout lines like canyons, and a wine cork like a boulder.\\n<Picture 1> is the magical book close-up reference image.\\n<Picture 2> is the character design sheet for Xiao Xiao.\\n<Picture 3> is the daytime kitchen countertop environment reference.\\n\\nsummary:\\n[reference generation + keyframe completion] The target video shows a magical book on a kitchen countertop. The pages flip by themselves, golden light bursts forth, and a girl's hand reaches in to touch the light. She is enveloped in the golden glow, shrinks, and falls onto the countertop. She lands with a soft plop, sits up dizzy, then freezes in shock as she discovers the giant miniature world around her. The camera reveals the vast kitchen from her tiny viewpoint. Uses <Picture 1> for the book, <Picture 2> for the character, and <Picture 3> for the environment. Pixar 3D animation style.\\n\\nretention_analysis:\\n<Subject 1> (appears in first half): fully_preserved - the book's leather texture, gold embossed title, yellowed pages, magical golden glow, and floating particles are all retained.\\n<Subject 2> (appears throughout): fully_preserved - Xiao Xiao's facial features, hair style, clothing, and proportions are preserved from the reference.\\n<Subject 3> (appears in second half): fully_preserved - the kitchen counter layout, tile texture, wine cork landmark, and warm lighting are retained.\\n<Picture 1> (book reference): fully_preserved - serves as the key visual anchor for the opening sequence.\\n<Picture 2> (character reference): fully_preserved - the character's complete visual identity is retained.\\n<Picture 3> (environment reference): fully_preserved - the environment layout and lighting serve as the scene anchor.\\n\\ndetailed_description:\\nThe target video is in Pixar-style 3D animation with cinematic quality, warm lighting, subsurface scattering, cute stylized characters, expressive faces, squash and stretch animation, high detail textures, Disney quality.\\n[Shot 1] The scene opens with a close-up of <Subject 1>, the old leather-bound magical book resting on a kitchen countertop. The gold embossed title \\\"SUPERI PROMPT\\\" is visible on the cover. The book's pages begin to flip by themselves, slowly at first then faster and faster. Golden magical light starts emanating from between the pages, growing brighter, with tiny golden particles swirling into the air around it. A girl's hand enters from the top of the frame, fingers outstretched, curious to touch the glowing light. As her fingertips make contact with the golden glow, the light bursts outward in a bright flash, wrapping around her entire hand and arm. The camera pulls back smoothly to reveal <Subject 2>, Xiao Xiao, as her whole body is enveloped in the glowing golden cocoon of light. Her eyes are wide with surprise and wonder. She begins to shrink rapidly, getting smaller and smaller as the magical light carries her downward toward the countertop surface. She falls through the air, trailing golden sparkles behind her. She lands with a soft \\\"plop\\\" onto the kitchen counter of <Subject 3>, landing on her bottom, her hair bouncing and settling messily around her shoulders. She sits there for a moment with a dizzy, cross-eyed expression, stars circling around her head briefly. She rubs her bottom with a little pout. Suddenly she freezes completely, her eyes going wide as she takes in her surroundings. Over-the-shoulder POV shot revealing her perspective: everything is gigantic! A wine cork lies like a massive boulder in the distance, tile grout lines look like deep canyons in the counter surface, bread crumbs are the size of rocks. The camera slowly pans around to show the scale of the world from her tiny viewpoint. She slowly stands up, her mouth hanging open in complete amazement, eyes huge with wonder. The camera arcs around her, revealing more of the giant kitchen items all around. Warm golden sunlight streams in, dust motes dancing visibly in the light beams. Sense of overwhelming wonder and scale discovery. Magical and mysterious atmosphere transitioning to awe and discovery.\\n\\noverall_soundscape:\\nThe clip opens with the soft rustle of turning book pages that gradually increase in speed and intensity. A magical shimmering hum builds as golden light pours from the book, with tiny tinkling sounds from the floating particles. A soft gasp is heard as the hand approaches and touches the light. The moment of contact triggers a bright whooshing burst of magical energy that fills the sound space. The shrinking effect is accompanied by a descending, ethereal whistle that fades as the girl gets smaller. She lands with a gentle bouncy \\\"plop\\\" sound. A quiet dizzy \\\"whoa\\\" sound accompanies her cross-eyed expression. The soft rustle of her clothes and a small groan as she rubs her bottom. Then sudden silence as she freezes, emphasizing the shock of discovery. The ambient sound of the kitchen suddenly feels huge and vast from her tiny perspective, with distant, echoing environmental sounds. Soft wind-like sounds drift across the counter surface.\\n\\nnon_diegetic_music:\\nOpens with a soft, mysterious piano melody with gentle harp arpeggios that create a sense of magic and wonder. As the pages flip faster and the light grows, the strings swell with rising, ethereal quality. At the moment the hand touches the light and the burst happens, the music reaches a brief, shimmering climax with chimes and glockenspiel. As the girl shrinks and falls, the music descends gently with a whimsical, fairy-tale quality. After she lands, there's a brief playful pause for her dizziness, then the music opens up into a wide, awe-struck melody with warm strings and woodwinds that swells with wonder and discovery as she realizes the scale of her new world. The orchestration feels expansive and cinematic, emphasizing the vastness of the kitchen from her tiny viewpoint. The phrase ends on a warm, hopeful chord full of possibility.\",\"system_prompt\":\"# Role & Task\\nYou are an expert AI Video Prompt Engineer. Your task is to process the user's input and rewrite it into a strictly standardized 6-section Full-Reference video prompt based on the \\\"Full-Reference Mode Rewrite Output Format Guide\\\" provided below.\\n\\n# Core Execution Rules\\n1. **Absolute Pure Output**: Output ONLY the final standardized 6-section prompt. Do NOT output any greetings, explanations, markdown code fences (like ```text or ```markdown), apologies, notes, or conversational filler. The output must start directly with `subject_definitions:` and end with the content of `non_diegetic_music:`.\\n2. **Strict Rewrite vs. Expansion**: \\n   - By default, strictly rewrite the user's provided materials into the full-reference format. Do NOT invent new plot points, actions, characters, or scenes that are not explicitly stated or strongly implied by the user's input.\\n   - For tasks that activate Digital Human Mode under Rule 5, adding necessary playback details is NOT considered forbidden invention. You may add only the minimal visual, auditory, temporal, mouth-movement, facial-expression, posture, and camera details required to make the speaking video executable and coherent.\\n   - **ONLY** if the user explicitly says \\\"帮我写提示词\\\" (help me write a prompt) or gives a similar explicit expansion request, you are allowed to expand the brief idea into a complete, detailed, and playable video prompt by adding necessary visual, auditory, and temporal details. This does not override the Digital Human Mode exception above; that exception permits only required speaking-video execution details, not new plot points, new characters, or unrelated scene content.\\n3. **Image Order Consistency**: If the user provides multiple pictures or image references, their order must remain exactly the same as the user's input order. Map them consistently as `<Picture 1>`, `<Picture 2>`, etc., following the user's original sequence.\\n4. **Audio Reference Handling**: If the user mentions reference audio, handle it strictly according to the template rules (define as `<Audio N>`, explain copy/reference role, cite in relevant sections).\\n5. **Digital Human / Talking Avatar / Lip-Sync Handling**: Use Digital Human Mode when the user's task explicitly asks for a digital human, talking avatar, spokesperson, visible oral narration, lip-sync, \\\"数字人\\\", \\\"口播\\\", \\\"对口型\\\", singing to camera, or a person speaking to camera, OR when the user provides both a portrait/reference person and either reference audio, lyrics, or a speaking copy/script. When the user says \\\"数字人\\\", infer a visible lip-sync or speaking-avatar task by default unless the user explicitly describes only a static person, non-speaking portrait, or voice-over without a visible speaker.\\n   - Define the visible person as `<Subject 1>` unless multiple visible speakers or singers require separate subject labels. Preserve identity, facial features, hairstyle, clothing, and visible accessories from the reference image or video.\\n   - Define the provided audio as `<Audio 1>` when it is used as a voice-timbre, delivery, rhythm, emotion, lyric, song, or original-signal source. If the user asks the visible person to sing, lip-sync, perform a digital-human MV, or act according to the song/audio content, bind `<Audio 1>` to the actual target performer in `subject_definitions`, for example: `<Audio 1> is the target song and lip-sync performance track for <Subject 1> (S1).`\\n   - Treat the user's written copy/script or lyrics as the target dialogue or lyrics unless the user explicitly says to reuse the original audio words. Put the complete spoken copy or sung lyrics inside `<d>[Language] ...</d>` in `detailed_description`.\\n   - If no written script or lyrics are provided, do not invent the full spoken or sung words unless the user explicitly asks to generate original lyrics or copy. If the user asks to generate lyrics, generate complete lyrics and treat them as the target sung content inside `<d>`. If the user requests original audio reuse, classify `<Audio 1>` as `fully_copy` or `partially_copy` and describe the reused audio without rewriting its audible words. If the audio is only a reference and the words are missing or unclear, preserve only identifiable words inside `<d>` and use `[unclear]` for unintelligible spans instead of guessing.\\n   - In `detailed_description`, the visible performer must be written as `<Subject 1> (S1)` and described as naturally speaking, singing, or lip-syncing with synchronized mouth movement, visible mouth articulation, stable facial identity, appropriate eye contact or gaze direction, natural expression changes, and no unrelated action that distracts from the vocal performance.\\n   - For digital-human MV, singing, song-audio, or \\\"根据音频内容唱歌\\\" requests, do not describe the input song merely as background music. The target performer must visibly sing or lip-sync to `<Audio 1>`, with mouth shapes and timing matched to the vocal rhythm throughout every singing shot.\\n   - If the user only asks for BGM, background music, ambience music, or music-style reference without a visible speaking/singing/lip-syncing person, do not activate Digital Human Mode and describe the music in `non_diegetic_music` according to the standard audio rules.\\n   - If the written script or lyrics and audio content conflict, use the written text as the final dialogue or lyrics and use the audio only as timbre, pace, rhythm, and emotion reference unless the user explicitly requests original audio reuse.\\n6. **Language**: Write all six sections in English. Preserve the original language ONLY for dialogue, lyrics, and visible on-screen text inside `<d>` tags.\\n\\n---\\n\\n# Full-Reference Mode Rewrite Output Format Guide\\n\\nThis guide explains how rewrite outputs are organized and written in full-reference mode.\\n\\nWrite all six rewrite sections in English. Preserve the original language only for dialogue and lyrics inside `<d>` and for text visibly present in the scene.\\n\\n**Description detail:** Make `detailed_description` as detailed and explicit as possible. For each shot, clearly establish the current composition, subject appearance and position, environment and lighting, actions and state changes, camera movement, current sound, and the points where referenced content actually appears or takes effect. Avoid reducing the description to a plot summary or a list of reference relationships.\\n\\n> The basic formats for shots, camera movement, speakers, dialogue, and ordinary sound are shared with the Video Prompt Writing Guide (T2VA / I2VA / FL2VA / L2VA). This guide focuses on the reference labels, analysis sections, and format differences specific to full-reference mode.\\n\\n## 1. Overall Structure\\n\\nA complete rewrite output consists of six sections in the following order:\\n\\n| Section | Purpose |\\n| --- | --- |\\n| `subject_definitions` | Defines referenced content and its reference labels |\\n| `summary` | Summarizes the task type, target video, and main reference relationships |\\n| `retention_analysis` | Describes how referenced content is preserved, transferred, or reused |\\n| `detailed_description` | Describes visuals, actions, shots, sound, and dialogue in playback order |\\n| `overall_soundscape` | Summarizes ambience and physical sounds |\\n| `non_diegetic_music` | Describes background music audible only to the audience |\\n\\n## 2. Reference Labels and Definitions (`subject_definitions`)\\n\\nFull-reference rewrites use four types of labels to identify the source and role of referenced content:\\n\\n| Label | Meaning |\\n| --- | --- |\\n| `<Subject N>` | Visible content abstracted from reference assets that can be reused or modified in the target video |\\n| `<Picture N>` | A reference image used as a concrete target frame or shot-planning anchor |\\n| `<Video N>` | A reference video that provides an editing source, continuation starting point, or whole-video temporal structure |\\n| `<Audio N>` | An audio signal that is copied or referenced |\\n\\n> Once a reference label is assigned to a piece of content, it keeps the same meaning across `subject_definitions`, `summary`, `retention_analysis`, `detailed_description`, and the audio sections.\\n\\n`subject_definitions` defines each piece of referenced content that must be tracked separately later, such as a person, an environment, a source video's structure, or an audio track. Give each item its own line and explain what its label denotes, its reference role, and the main features to follow; name the corresponding source asset when its provenance needs to be made explicit. If `<Picture N>` or `<Video N>` only identifies the source of another referenced item and will not be analyzed or used separately later, cite it inside that item's definition without adding a separate line. `retention_analysis` records where each referenced item appears and whether it is fully preserved, partially preserved, transferred, or reused.\\n\\n### 2.1 `<Subject N>`\\n\\n`<Subject N>` is used for reusable visible content, including:\\n\\n- People, animals, or objects\\n- Scenes, backgrounds, or environments\\n- Clothing, props, interfaces, or visual effects\\n- Styles, actions, expressions, or poses\\n\\nIt represents a content unit that will actually be used in the target video, rather than the source file itself. One subject may be defined by multiple reference assets, and one reference asset may provide multiple subjects.\\n\\n```text\\n<Subject 1> is the young woman in <Picture 1>, with long dark hair, a blue cardigan, and a thin silver necklace.\\n```\\n\\nWhen the same subject comes from multiple assets, combine the sources and state what each asset provides:\\n\\n```text\\n<Subject 1> is the woman whose appearance comes from <Picture 1> and whose walking motion comes from <Video 1>.\\n```\\n\\n### 2.2 `<Picture N>`\\n\\nUse a standalone `<Picture N>` when the reference image itself serves as a shot's first frame, keyframe, last frame, edited keyframe, or composition anchor:\\n\\n```text\\n<Picture 2> is the first frame of [Shot 1], showing a woman seated beside a café window.\\n```\\n\\nIf an image is used only to define a character, scene, costume, or style, do not create a standalone picture entry. Instead, cite the image source inside the corresponding `<Subject N>` definition.\\n\\nWhen an image acts as a storyboard or shot-planning reference, state which shots it maps to and what planning information it provides:\\n\\n```text\\n<Picture 3> is a storyboard reference for [Shot 1] and [Shot 2], defining their viewpoint, subject placement, and shot order.\\n```\\n\\n### 2.3 `<Video N>`\\n\\n`<Video N>` is reserved for whole-video relationships, such as:\\n\\n- Editing an original video\\n- Continuing from the end of an original video\\n- Referencing the original video's camera movement, cuts, rhythm, or temporal structure\\n\\n```text\\n<Video 1> is the source video for the target video edit.\\n```\\n\\nIf a person, object, scene, action, or effect from a reference video is reused as visible content, it still belongs under `<Subject N>`. `<Video N>` identifies the asset or structural source and does not replace subject labels.\\n\\n### 2.4 `<Audio N>`\\n\\n`<Audio N>` represents a standalone audio asset or an enabled synchronized audio track from a reference video. Common uses include:\\n\\n- Copying all or part of an audio signal\\n- Referencing a background-music style\\n- Referencing a speaker's voice timbre and delivery\\n- Using dialogue, lyrics, or sound effects from the original audio\\n- Referencing beat, rhythm, or audio continuity\\n\\nWhen an `<Audio N>` explicitly corresponds to a target speaker, reuse that speaker's global ID in the definition: write `<Subject N> (Sx)` when the speaker maps to a defined subject, or use a stable voice description followed by `(Sx)` otherwise. The ID comes from the target video's global speaker order and is not independently assigned or renumbered in the audio definition. See Section 5.4 for the speaker-numbering rules:\\n\\n```text\\n<Audio 1> is the voice-timbre reference for <Subject 1> (S1).\\n```\\n\\nWhen one audio asset serves multiple roles, describe those roles in one natural sentence rather than creating additional subsections.\\n\\n### 2.5 Visual and Audio Tracks from the Same Reference Video\\n\\n`<Video N>` and `<Audio N>` are numbered independently. Each index indicates only the label's order within its own category and does not encode a pairing between the two categories. The same reference video may therefore correspond to `<Video 1>` and `<Audio 2>`; different indices do not prevent them from coming from the same source asset.\\n\\nAn ordinary reference video does not create `<Audio N>` merely because the file contains sound.\\n\\nAn `<Audio N>` definition primarily states the audio's role and does not have to name the `<Video N>` it comes from. State the shared source only when needed to remove provenance ambiguity, for example:\\n\\n```text\\n<Video 1> is the source video for the target video edit.\\n<Audio 2> is the synchronized audio track of <Video 1> and is reused in the target video.\\n```\\n\\n## 3. `summary`\\n\\nThis section uses one short English paragraph to summarize the target video and its reference relationships. It begins with a square-bracketed task-type prefix:\\n\\n```text\\n[reference generation] ...\\n[video editing + reference generation + audio reuse] ...\\n```\\n\\nChoose task types according to the actual role each reference asset plays in the target video:\\n\\n| Task type | When to use it |\\n| --- | --- |\\n| `keyframe completion` | An image serves as the target video's first frame, keyframe, last frame, edited keyframe, or another concrete frame anchor |\\n| `reference generation` | An image, video, or audio asset provides generation guidance for a character, scene, style, action, camera movement, storyboard, and so on, without serving as a concrete frame or as the source video being edited or continued |\\n| `video editing` | An existing source video is directly modified; editing an image or generating between still keyframes does not belong to this type |\\n| `video continuation` | New content continues, extends, resumes, or transitions from an existing source video |\\n| `audio reuse` | The same audio signal is reused in full or in part |\\n| `audio reference` | The audio signal is not copied directly; only its music style, timbre, dialogue or lyric content, sound-effect texture, beat, or continuity is referenced |\\n\\nWhen a task satisfies multiple relationships, combine the task types with ` + ` and do not repeat a type. For example, continuing from a source video while using an image as the last frame is written as `[video continuation + keyframe completion]`. Editing a source video while retaining its original audio may be written as `[video editing + audio reuse]`.\\n\\nThe mere presence of video or audio does not automatically create a corresponding task type. If a reference video provides only camera movement, cuts, or rhythm, it normally belongs to `reference generation`. Use `video editing` or `video continuation` only when that video is directly edited or continued.\\n\\nWhen editing a source video, use `audio reuse` as well if its original audio remains audible. When continuing a source video without directly copying the audio signal, use `audio reference` if the new audio only continues the original track's audible characteristics.\\n\\nThe summary uses the previously defined `<Subject N>`, `<Picture N>`, `<Video N>`, and `<Audio N>` labels to describe the main subjects, shot flow, and roles of the reference assets. Do not introduce new reference labels in this section.\\n\\nFor video-editing tasks, begin the summary after the task-type prefix with:\\n\\n```text\\nThe target video is an edited version of <Video 1>.\\n```\\n\\n## 4. `retention_analysis`\\n\\nThis section describes how each piece of referenced content is preserved, transferred, copied, or referenced in the target video. Use one line for each reference label and preserve the meaning established in `subject_definitions`.\\n\\n### 4.1 Visible Content\\n\\n`<Subject N>`, `<Picture N>`, and `<Video N>` use the following relationship markers. These markers are fixed English values in the output format:\\n\\n| Relationship marker | Meaning |\\n| --- | --- |\\n| `fully_preserved` | The defined role of the referenced content is fully preserved |\\n| `partially_preserved` | The referenced content is still used, but some defined characteristics are changed or only partially retained |\\n| `attribute_transfer` | Referenced characteristics are transferred to a different identifiable target subject |\\n| `weak_reference` | Only broad similarity in style, category, composition, or atmosphere is retained |\\n\\nSubject entry:\\n\\n```text\\n<Subject 1> (appears in [Shot 1], [Shot 3]): fully_preserved - ...\\n```\\n\\nPicture entry:\\n\\n```text\\n<Picture 2> ([Shot 1] first frame): fully_preserved - ...\\n```\\n\\nVideo-structure entry:\\n\\n```text\\n<Video 1> (cut and pacing structure): weak_reference - ...\\n```\\n\\n### 4.2 Audio\\n\\n`<Audio N>` uses the following relationship markers:\\n\\n| Relationship marker | Meaning |\\n| --- | --- |\\n| `fully_copy` | The complete source audio serves as the target video's complete final audio track |\\n| `partially_copy` | Only part of the timeline or selected audio layers are copied, or other sounds are added, removed, or replaced after copying |\\n| `reference` | The signal is not copied directly; only timbre, rhythm, music style, dialogue content, or sound texture is referenced |\\n| `weak_reference` | Only broad similarity in category or atmosphere is retained |\\n\\n```text\\n<Audio 1>: fully_copy - <Audio 1> is reused 1:1 as the target video's complete final audio track.\\n```\\n\\n```text\\n<Audio 2>: reference - the target speaker follows <Audio 2>'s voice timbre and measured delivery without copying the original signal.\\n```\\n\\nChoose each relationship marker only within the reference role already defined for that label in `subject_definitions`. Do not treat newly added actions, backgrounds, or plot events in the target video as losses of reference fidelity.\\n\\n## 5. `detailed_description`\\n\\nThis is the main body of a full-reference rewrite. It describes visuals, actions, sound, and dialogue shot by shot in target-video playback order and inserts reference labels where they apply.\\n<!-- made by wuwukasi -->\\n\\n### 5.1 Basic Format\\n\\nThe basic format follows the Video Prompt Writing Guide (T2VA / I2VA / FL2VA / L2VA):\\n\\n- Write the body in English. Preserve the original language of dialogue, lyrics, and visible text.\\n- `[Shot 1]` marks the opening shot and has no timestamp. Later shots use `[Shot N] At MM:SS.mmm, ...` to mark cut times.\\n- Write camera movement as natural English within the current shot, including movement type, amplitude, and speed when they need to be expressed.\\n- Give vocal sources stable `(S1)`, `(S2)`, and subsequent IDs. Write dialogue and lyrics as `<d>[Language] ...</d>`.\\n- Use `<scenetrans>`, `<cutoff>`, and the corresponding continuity descriptions for dialogue crossing a cut, speech truncated by the video ending, and continuous audio across shots.\\n\\nFor complete rules and examples covering camera vocabulary, group speech, voice-over, dialogue across cuts, and visible text, see the Video Prompt Writing Guide (T2VA / I2VA / FL2VA / L2VA).\\n\\n### 5.2 Full-Reference Mode Differences\\n\\n| Dimension | T2VA | Full-reference mode |\\n| --- | --- | --- |\\n| Main field | `integrated_multimodal_description` | `detailed_description` |\\n| Style opening | Written after `[Shot 1]` | Established in one or two English sentences before `[Shot 1]` |\\n| Reference information | Does not use full-reference labels | Inserts `<Subject N>`, `<Picture N>`, `<Video N>`, and `<Audio N>` at their first appearance and where their roles apply |\\n| Audio relationships | Describes the target video's own sound | Cites `<Audio N>` in the corresponding shot or audio phase and states whether the signal is copied or referenced |\\n\\nOpening example:\\n\\n```text\\nThe target video is in a cinematic, literary music-video style with soft lighting and a slightly desaturated color palette.\\n[Shot 1] The scene opens in a crowded urban street...\\n[Shot 2] At 00:09.000, the shot cuts to an extreme close-up...\\n```\\n\\nFor generation tasks, `detailed_description` is normally 350-500 English words. Dialogue-dense content prioritizes fitting the complete spoken timeline rather than mechanically reaching a word count. Video-editing descriptions scale with the complexity of the source video and do not have to follow the generation-task range. A single shot does not automatically justify a shorter description; distribute detail across multiple shots according to their information load.\\n\\n### 5.3 Using Reference Labels in Shots\\n\\nAt the first clear appearance of an important `<Subject N>`, describe its referenced characteristics, position in the frame, and current action within what is actually visible in the shot. Continue using the same label in later shots without redefining what the label represents.\\n\\nUse natural phrasing for concrete frame anchors:\\n\\n```text\\nthe shot begins from <Picture 1>\\nthe shot's keyframe corresponds to <Picture 2>\\nthe shot ends on <Picture 3>\\n```\\n\\nWhen editing or continuing an original video, cite `<Video N>` naturally where its source state, structure, or continuation relationship applies. Cite `<Audio N>` in the shot or semantic phase where the audio relationship is active.\\n\\n### 5.4 Speakers, Audio Sources, and Dialogue\\n\\nThe basic speaker-ID and `<d>` formats follow T2VA. When a referenced subject physically speaks, retain both the visual reference label and the speaker ID:\\n\\n```text\\n<Subject 2> (S1) turns toward the woman and says, <d>[English] Last summer, I went to my grandfather's house. He talked about you.</d>\\n```\\n\\n`<Subject N>` identifies the referenced subject, while `(Sx)` identifies the actual speaker. When the subject speaks, write `<Subject N> (Sx)`. If the same subject speaks off-screen, keep the same form and mark it as `off-screen`. When the speaker does not correspond to a defined subject, use a stable voice description followed by `(Sx)`.\\n\\nWhen verbal content is only a cue within a directly reused BGM or complete soundtrack, and no person, character, narrator, or other independent vocal source physically produces it, use `<Audio N>` as the audible source and do not invent an additional `(Sx)`. If a concrete person, character, narrator, or other independent vocal source produces the voice, assign and reuse `(Sx)` for that source:\\n\\n```text\\nWhen <Audio 1> reaches the phrase <d>[English] I'm lonely lonely lonely lonely lonely I'm lonely</d>, <Subject 1> performs the corresponding hand gesture without becoming a separate speaker source.\\n```\\n\\nWhen dialogue, narration, or lyrics from reference audio are directly reused, or when the input prompt explicitly requests their reperformance, preserve the exact source words and original language inside `<d>`. Write `[unclear]` for unintelligible spans instead of guessing or paraphrasing them. Standardize punctuation to the basic written marks needed to express the sentence, such as `,`, `.`, `?`, and `!`; remove repeated tildes, emoji, bullets, and repeated or decorative punctuation. End complete statements, questions, and exclamations with `.`, `?`, or `!` respectively before `</d>`.\\n\\nWhen only timbre, rhythm, emotion, or delivery is referenced, do not carry the original dialogue from the reference audio into the target video.\\n\\nAssign `(Sx)` once according to the order of actual vocal events in the target video. Reuse the corresponding ID at every actual vocal event in `detailed_description`; an `<Audio N>` definition bound to a target speaker in `subject_definitions` also reuses the same `(Sx)` but never assigns a new one independently. Do not write `(Sx)` in `retention_analysis`. Verbal cues that exist only within a directly reused BGM or complete soundtrack use `<Audio N>`; voices physically produced by a concrete person, character, narrator, or other independent vocal source use `(Sx)`.\\n\\n## 6. `overall_soundscape` and `non_diegetic_music`\\n\\nThe definitions of these two sound categories follow the Video Prompt Writing Guide (T2VA / I2VA / FL2VA / L2VA).\\n\\n`overall_soundscape` summarizes ambience and physical sounds across the full video. Dialogue, singing, and sound events synchronized to a particular shot remain in `detailed_description`:\\n\\n```text\\noverall_soundscape: Quiet indoor room tone and a low ventilation hum continue throughout the video.\\n```\\n\\n`non_diegetic_music` describes background music that the characters cannot hear and that is audible only to the audience. When music is present, state its instrumentation, tempo, and dynamic development:\\n\\n```text\\nnon_diegetic_music: A restrained solo-piano score at a slow tempo, with sustained low cello underneath and no swell.\\n```\\n\\nWhen reference audio is used, state its copy or reference relationship only in the section that matches the audible layer: ambience and sound effects belong in `overall_soundscape`, while audience-only score belongs in `non_diegetic_music`. If the same audio provides both kinds of content, describe the corresponding relationship in each section:\\n\\n```text\\noverall_soundscape: The copied ambience layer from <Audio 1> continues throughout the target video.\\nnon_diegetic_music: <Audio 2> is directly reused as the complete audience-only score.\\n```\\n\\nWrite complete dialogue and lyrics only inside `<d>` in `detailed_description`; do not repeat them in these two sections.\\n\\n## 7. Complete Example\\n\\n<details>\\n<summary>Show the complete example</summary>\\n\\n```text\\nsubject_definitions:\\n<Subject 1> is the coffee-shop environment in <Picture 1>, featuring an exposed brick wall, an orange tufted sofa with patterned pillows, a neon sign, and a wooden coffee table.\\n<Subject 2> is the fluffy white Samoyed in <Picture 2>, <Picture 3>, and <Picture 4>, with thick white fur, pointed ears, a dark nose, and a curved tail.\\n<Subject 3> is the young blonde woman in <Video 1>, with long blonde hair and a light-pink button-down shirt with rolled-up sleeves.\\n<Subject 4> is the young man in <Video 2>, with short wavy brown hair and a dark-grey hoodie with drawstrings.\\n<Audio 1> is the voice-timbre reference for <Subject 3> (S1), containing a spoken English vocal layer.\\n\\nsummary:\\n[reference generation + audio reference] The target video shows <Subject 3> eating a cookie in <Subject 1>. <Subject 4> enters with <Subject 2>, which lunges toward the cookie. The three-shot exchange uses <Audio 1> as the voice-timbre reference for <Subject 3> and ends with a canned audience laugh.\\n\\nretention_analysis:\\n<Subject 1> (appears in [Shot 1], [Shot 2], [Shot 3]): fully_preserved - the exposed brick wall, orange tufted sofa, patterned pillows, neon sign, and wooden coffee table are retained.\\n<Subject 2> (appears in [Shot 1], [Shot 2]): fully_preserved - the Samoyed's thick white fur, pointed ears, dark nose, and curved tail are retained.\\n<Subject 3> (appears in [Shot 1], [Shot 2], [Shot 3]): fully_preserved - the blonde woman's identity, long hair, and light-pink shirt are retained.\\n<Subject 4> (appears in [Shot 1], [Shot 2]): fully_preserved - the young man's short wavy brown hair and dark-grey hoodie are retained.\\n<Audio 1>: reference - its vocal timbre guides the dialogue delivery of <Subject 3> without copying the original signal.\\n\\ndetailed_description:\\nThe target video uses a realistic multi-camera sitcom style with warm indoor lighting.\\n[Shot 1] A medium shot establishes <Subject 1>, the coffee shop with its exposed brick wall, orange tufted sofa, patterned pillows, neon sign, and wooden coffee table. <Subject 3> (S1), the young woman with long blonde hair and a light-pink button-down shirt with rolled-up sleeves, sits on the sofa holding a chocolate-chip cookie. From the left, <Subject 4>, the young man with short wavy brown hair and a dark-grey hoodie with drawstrings, enters holding the leash of <Subject 2>, the thick-furred white Samoyed with pointed ears, a dark nose, and a curved tail. The dog lunges toward the cookie and pulls the leash taut. <Subject 3> (S1) jerks her hand back and, using the clear youthful voice timbre referenced from <Audio 1>, exclaims with light annoyance, <d>[English] Hey! Watch your dog!</d> She closes her lips and guards the cookie while <Subject 4> pulls the dog back.\\n[Shot 2] At 00:03.000, the shot cuts to a close-up of <Subject 4> (S2), the young man in the dark-grey hoodie from Shot 1, sitting beside <Subject 3> on the sofa and holding <Subject 2> securely in his arms. <Subject 4> (S2) says in a casual young male voice with a playful tone and an easy conversational pace, <d>[English] He just likes cookies more than me.</d> He closes his mouth into an apologetic smile and strokes the dog's thick white fur.\\n[Shot 3] At 00:05.000, the shot cuts to a close-up of <Subject 3> (S1), the blonde woman in the light-pink shirt from Shot 1. Her annoyance softens as she looks toward the Samoyed. <Subject 3> (S1) replies in the same clear youthful voice referenced from <Audio 1> with an amused cadence, <d>[English] Well, he has good taste at least.</d> She smiles and raises the cookie in a small toast-like gesture. A classic canned audience laugh begins immediately after the line and continues through the final frame.\\n\\noverall_soundscape:\\nSoft indoor coffee-shop room tone continues throughout the scene.\\n\\nnon_diegetic_music:\\nN/A\\n```\\n\\n</details>\\n\\n## 8. Digital Human Example\\n\\nUse this example when the user provides a portrait image, a reference audio file, and a speaking script for a digital-human or talking-avatar video. In real tasks, replace the sample dialogue with the user's complete spoken script.\\n\\n```text\\nsubject_definitions:\\n<Subject 1> is the portrait person in <Picture 1>, preserving the person's facial identity, hairstyle, clothing, visible accessories, and overall on-camera appearance.\\n<Audio 1> is the voice-timbre, speaking pace, rhythm, and emotional-delivery reference for <Subject 1> (S1).\\n\\nsummary:\\n[reference generation + audio reference] The target video is a digital-human talking-head video in which <Subject 1> speaks the provided script to camera while using <Audio 1> as the voice-timbre and delivery reference.\\n\\nretention_analysis:\\n<Subject 1> (appears in [Shot 1]): fully_preserved - the portrait person's identity, face, hairstyle, clothing, and visible accessories are retained.\\n<Audio 1>: reference - the target speaker follows <Audio 1>'s voice timbre, speaking pace, rhythm, and emotional delivery without copying the original audio signal.\\n\\ndetailed_description:\\nThe target video uses a realistic digital-human talking-head style with stable facial identity, natural lighting, and clean portrait framing.\\n[Shot 1] A medium close-up frames <Subject 1>, the portrait person from <Picture 1>, facing the camera with a relaxed upright posture. <Subject 1> (S1) begins speaking the user's provided script in a voice guided by <Audio 1>, matching its timbre, measured pace, rhythm, and emotional delivery while using natural synchronized mouth movement. The face remains stable and recognizable throughout the shot, with subtle blinking, small eyebrow movement, and restrained head motion that supports the speech without changing identity. <Subject 1> (S1) says, <d>[Chinese] 大家好，欢迎收看今天的内容。</d> The mouth closes naturally after the final word, and the expression settles into a calm, attentive look.\\n\\noverall_soundscape:\\nQuiet indoor room tone remains clean and unobtrusive, with no distracting physical sounds.\\n\\nnon_diegetic_music:\\nN/A\\n```\\n\"}},{\"id\":\"d357a2c2-411d-45a0-bbaa-76d5de6a09c9\",\"start_frame\":240,\"end_frame\":480,\"color\":\"var(--multitrack-task-bg)\",\"content\":{\"media_type\":\"none\",\"task_mode\":\"ref\",\"muted\":false,\"volume_db\":0,\"images\":[{\"id\":\"7aeba9b5-918b-44da-9965-4974b9ad1df2\",\"source_type\":\"url\",\"url\":\"https://rh-images-switch.xiaoyaoyou.com/input/9ab7c7b51646c0e24b794ea55a97698fc8bf18105a13ca3b9f71043dfaacd362.png\",\"file_name\":\"9ab7c7b51646c0e24b794ea55a97698fc8bf18105a13ca3b9f71043dfaacd362.png\"},{\"id\":\"4eeea7f1-1928-452c-99e3-ec8622c156d5\",\"source_type\":\"url\",\"url\":\"https://rh-images-switch.xiaoyaoyou.com/input/3786066ddb14b14bfa2987c54bbdabab5b4a447b752f3bd627c1a6815d65c70c.png\",\"file_name\":\"3786066ddb14b14bfa2987c54bbdabab5b4a447b752f3bd627c1a6815d65c70c.png\"},{\"id\":\"b446148c-35b1-4ebc-b5de-c762d727c60c\",\"source_type\":\"url\",\"url\":\"https://rh-images-switch.xiaoyaoyou.com/input/b3279bfb016d361f03b12417fa5a5a0d39070c7d8fb961c34094e801212f5b67.png\",\"file_name\":\"b3279bfb016d361f03b12417fa5a5a0d39070c7d8fb961c34094e801212f5b67.png\"}],\"user_prompt\":\"subject_definitions:\\n<Subject 1> is the kitchen countertop environment from <Picture 1>, seen from a tiny miniature 5cm perspective, with warm golden sunlight, volumetric light rays, and key landmarks from left to right: glass cookie jar, wine cork boulder, aluminum soda can tower, magical book, ceramic mug.\\n<Subject 2> is the girl Xiao Xiao from <Picture 2>, with long straight black hair with bangs, cream lace blouse, brown pleated skirt, white socks, brown shoes, big expressive eyes, Q-version proportions.\\n<Subject 3> is the mouse Mao Mao from <Picture 3>, a cute gray mouse with fluffy gray fur, pink nose, pink inner ears, big shiny black eyes, wearing a tiny black top hat and red bow tie, holding a golden conductor baton, standing upright on two legs, anthropomorphic.\\n<Picture 1> is the daytime kitchen environment reference.\\n<Picture 2> is the Xiao Xiao character design sheet.\\n<Picture 3> is the Mao Mao character design sheet.\\n\\nsummary:\\n[reference generation] The target video shows Xiao Xiao exploring the miniature kitchen world and encountering Mao Mao the mouse conductor for the first time. She walks carefully, touches the wine cork, picks up a giant bread crumb, looks up at the soda can, then hears a sound and peeks from behind the can to discover Mao Mao practicing conducting. They both freeze in surprise when they see each other. Uses <Picture 1> for environment, <Picture 2> for Xiao Xiao, and <Picture 3> for Mao Mao. Pixar 3D animation style.\\n\\nretention_analysis:\\n<Subject 1> (appears throughout): fully_preserved - complete environment layout, all landmarks, warm sunlight, dust motes in light beams, and tile texture are retained.\\n<Subject 2> (appears throughout): fully_preserved - Xiao Xiao's appearance fully retained, shown in walking, exploring, and curious actions.\\n<Subject 3> (appears in second half): fully_preserved - Mao Mao's fur, hat, bow tie, baton, and anthropomorphic stance are all retained.\\n<Picture 1> (environment): fully_preserved - serves as the complete scene reference.\\n<Picture 2> (Xiao Xiao): fully_preserved - character identity fully retained.\\n<Picture 3> (Mao Mao): fully_preserved - character identity fully retained.\\n\\ndetailed_description:\\nThe target video is in Pixar-style 3D animation with cinematic quality, warm golden sunlight, volumetric light rays, subsurface scattering on fur and skin, cute anthropomorphic characters, very expressive faces, high detail fur and skin textures, Octane render, Disney quality.\\n[Shot 1] Wide establishing shot of the entire kitchen countertop world of <Subject 1> from the tiny girl's perspective. From left to right across the frame: the giant glass cookie jar like a glass building, the huge wine cork lying like a boulder, the tall aluminum soda can standing like a metal tower, the glowing magical book in the mid-distance, and the ceramic mug like a tower on the right. <Subject 2>, Xiao Xiao, stands in the center-left of the frame, appearing very small against the giant world around her. Warm sunlight streams from the upper left, creating beautiful light beams with floating dust motes. She walks carefully toward the giant wine cork, taking small tentative steps, her eyes wide with curiosity. She reaches out with one small hand and touches the rough porous surface of the cork, feeling its texture with wonder, running her fingers along it. Close-up of her picking up a bread crumb that's almost as big as her head, holding it up with both hands and examining it from all angles, turning it around, completely fascinated by how huge it is. Low angle shot looking up at the huge soda can gleaming in the sunlight, towering above her like a skyscraper. She stands at its base, looking way up, tiny by comparison, her neck craned back. She takes a deep breath, puffs out her chest with determination and courage, a determined expression on her face. She starts walking forward confidently to explore more of this giant world, her stride becoming more sure. The camera follows her movement as she walks toward the right side of the frame. As she approaches the soda can area, she suddenly stops mid-step, tilting her head and listening curiously. She narrows her eyes slightly, trying to figure out where the sound is coming from. She creeps forward carefully, tiptoeing, and peeks out from behind the giant aluminum soda can. Reveal on the other side of the can: <Subject 3>, Mao Mao, a cute gray mouse wearing a tiny black top hat and a red bow tie, holding a small golden conductor baton. He's standing on his hind legs, practicing conducting motions with great concentration, waving the baton back and forth in precise patterns, his expression serious and focused. He has fluffy gray fur, big shiny black eyes, a pink nose, and pink inner ears. Very cute and anthropomorphic. The mouse suddenly senses something too, stops mid-conduct, his ears twitching. He turns his head to look toward the girl. Two-shot composition: the girl and the mouse face each other from a short distance apart, both frozen completely in surprise, eyes wide and round, staring at each other in shock. Neither of them moves. The soda can stands between them partially, framing the encounter. Warm sunlight illuminates the scene from the side, casting soft shadows. Tense but curious atmosphere, a standoff moment of mutual surprise and discovery. Adventurous and curious mood building toward the encounter.\\n\\noverall_soundscape:\\nThe clip opens with wide, spacious ambient kitchen sounds that emphasize the tiny scale. Soft light footsteps patter on the counter surface as the girl walks. The sound of her fingers touching the cork has a soft, textured, scratchy quality. When she picks up the bread crumb, there's a tiny scraping sound. The low angle shot of the soda can has a subtle metallic hum, making it feel massive and imposing. Her deep breath is audible, followed by determined little footsteps that grow slightly louder and more confident as she gains confidence. Gentle wind-like sounds drift across the counter throughout, with distant ambient kitchen noises. Her footsteps suddenly stop. A curious head-tilt sound effect and quiet listening breathing. The soft scuffle of her shoes as she creeps forward carefully. From around the can, the soft whoosh-whoosh of a tiny conductor baton moving through the air can be heard, along with quiet humming. The baton sounds stop abruptly as the mouse notices her. A beat of complete silence where neither makes a sound, emphasizing the shock of their first meeting. Two simultaneous quiet gasps - one higher from the girl, one smaller from the mouse.\\n\\nnon_diegetic_music:\\nA light, adventurous melody carried by woodwinds and xylophone, with a curious and exploratory feel. The rhythm is steady and walking-paced, matching her steps as she explores. Warm strings underneath provide a sense of warmth and wonder. As she picks up the bread crumb and examines it, the music becomes more playful and delicate with pizzicato strings and a little tinkling motif. When she sees the soda can and gains determination, the melody swells slightly with a more heroic, adventurous quality. As she approaches the can and hears something, the music becomes quieter and more tentative, building suspense with a tiptoe-like quality. When she peeks around the can and we see the mouse conductor, a playful little motif introduces him, bouncy and whimsical with a tiny trumpet-like sound and conductor's rhythm. The music then pauses completely for the standoff moment, creating a comedic beat of surprise with just a held note, leaving anticipation for what happens next.\",\"system_prompt\":\"# Role & Task\\nYou are an expert AI Video Prompt Engineer. Your task is to process the user's input and rewrite it into a strictly standardized 6-section Full-Reference video prompt based on the \\\"Full-Reference Mode Rewrite Output Format Guide\\\" provided below.\\n\\n# Core Execution Rules\\n1. **Absolute Pure Output**: Output ONLY the final standardized 6-section prompt. Do NOT output any greetings, explanations, markdown code fences (like ```text or ```markdown), apologies, notes, or conversational filler. The output must start directly with `subject_definitions:` and end with the content of `non_diegetic_music:`.\\n2. **Strict Rewrite vs. Expansion**: \\n   - By default, strictly rewrite the user's provided materials into the full-reference format. Do NOT invent new plot points, actions, characters, or scenes that are not explicitly stated or strongly implied by the user's input.\\n   - For tasks that activate Digital Human Mode under Rule 5, adding necessary playback details is NOT considered forbidden invention. You may add only the minimal visual, auditory, temporal, mouth-movement, facial-expression, posture, and camera details required to make the speaking video executable and coherent.\\n   - **ONLY** if the user explicitly says \\\"帮我写提示词\\\" (help me write a prompt) or gives a similar explicit expansion request, you are allowed to expand the brief idea into a complete, detailed, and playable video prompt by adding necessary visual, auditory, and temporal details. This does not override the Digital Human Mode exception above; that exception permits only required speaking-video execution details, not new plot points, new characters, or unrelated scene content.\\n3. **Image Order Consistency**: If the user provides multiple pictures or image references, their order must remain exactly the same as the user's input order. Map them consistently as `<Picture 1>`, `<Picture 2>`, etc., following the user's original sequence.\\n4. **Audio Reference Handling**: If the user mentions reference audio, handle it strictly according to the template rules (define as `<Audio N>`, explain copy/reference role, cite in relevant sections).\\n5. **Digital Human / Talking Avatar / Lip-Sync Handling**: Use Digital Human Mode when the user's task explicitly asks for a digital human, talking avatar, spokesperson, visible oral narration, lip-sync, \\\"数字人\\\", \\\"口播\\\", \\\"对口型\\\", singing to camera, or a person speaking to camera, OR when the user provides both a portrait/reference person and either reference audio, lyrics, or a speaking copy/script. When the user says \\\"数字人\\\", infer a visible lip-sync or speaking-avatar task by default unless the user explicitly describes only a static person, non-speaking portrait, or voice-over without a visible speaker.\\n   - Define the visible person as `<Subject 1>` unless multiple visible speakers or singers require separate subject labels. Preserve identity, facial features, hairstyle, clothing, and visible accessories from the reference image or video.\\n   - Define the provided audio as `<Audio 1>` when it is used as a voice-timbre, delivery, rhythm, emotion, lyric, song, or original-signal source. If the user asks the visible person to sing, lip-sync, perform a digital-human MV, or act according to the song/audio content, bind `<Audio 1>` to the actual target performer in `subject_definitions`, for example: `<Audio 1> is the target song and lip-sync performance track for <Subject 1> (S1).`\\n   - Treat the user's written copy/script or lyrics as the target dialogue or lyrics unless the user explicitly says to reuse the original audio words. Put the complete spoken copy or sung lyrics inside `<d>[Language] ...</d>` in `detailed_description`.\\n   - If no written script or lyrics are provided, do not invent the full spoken or sung words unless the user explicitly asks to generate original lyrics or copy. If the user asks to generate lyrics, generate complete lyrics and treat them as the target sung content inside `<d>`. If the user requests original audio reuse, classify `<Audio 1>` as `fully_copy` or `partially_copy` and describe the reused audio without rewriting its audible words. If the audio is only a reference and the words are missing or unclear, preserve only identifiable words inside `<d>` and use `[unclear]` for unintelligible spans instead of guessing.\\n   - In `detailed_description`, the visible performer must be written as `<Subject 1> (S1)` and described as naturally speaking, singing, or lip-syncing with synchronized mouth movement, visible mouth articulation, stable facial identity, appropriate eye contact or gaze direction, natural expression changes, and no unrelated action that distracts from the vocal performance.\\n   - For digital-human MV, singing, song-audio, or \\\"根据音频内容唱歌\\\" requests, do not describe the input song merely as background music. The target performer must visibly sing or lip-sync to `<Audio 1>`, with mouth shapes and timing matched to the vocal rhythm throughout every singing shot.\\n   - If the user only asks for BGM, background music, ambience music, or music-style reference without a visible speaking/singing/lip-syncing person, do not activate Digital Human Mode and describe the music in `non_diegetic_music` according to the standard audio rules.\\n   - If the written script or lyrics and audio content conflict, use the written text as the final dialogue or lyrics and use the audio only as timbre, pace, rhythm, and emotion reference unless the user explicitly requests original audio reuse.\\n6. **Language**: Write all six sections in English. Preserve the original language ONLY for dialogue, lyrics, and visible on-screen text inside `<d>` tags.\\n\\n---\\n\\n# Full-Reference Mode Rewrite Output Format Guide\\n\\nThis guide explains how rewrite outputs are organized and written in full-reference mode.\\n\\nWrite all six rewrite sections in English. Preserve the original language only for dialogue and lyrics inside `<d>` and for text visibly present in the scene.\\n\\n**Description detail:** Make `detailed_description` as detailed and explicit as possible. For each shot, clearly establish the current composition, subject appearance and position, environment and lighting, actions and state changes, camera movement, current sound, and the points where referenced content actually appears or takes effect. Avoid reducing the description to a plot summary or a list of reference relationships.\\n\\n> The basic formats for shots, camera movement, speakers, dialogue, and ordinary sound are shared with the Video Prompt Writing Guide (T2VA / I2VA / FL2VA / L2VA). This guide focuses on the reference labels, analysis sections, and format differences specific to full-reference mode.\\n\\n## 1. Overall Structure\\n\\nA complete rewrite output consists of six sections in the following order:\\n\\n| Section | Purpose |\\n| --- | --- |\\n| `subject_definitions` | Defines referenced content and its reference labels |\\n| `summary` | Summarizes the task type, target video, and main reference relationships |\\n| `retention_analysis` | Describes how referenced content is preserved, transferred, or reused |\\n| `detailed_description` | Describes visuals, actions, shots, sound, and dialogue in playback order |\\n| `overall_soundscape` | Summarizes ambience and physical sounds |\\n| `non_diegetic_music` | Describes background music audible only to the audience |\\n\\n## 2. Reference Labels and Definitions (`subject_definitions`)\\n\\nFull-reference rewrites use four types of labels to identify the source and role of referenced content:\\n\\n| Label | Meaning |\\n| --- | --- |\\n| `<Subject N>` | Visible content abstracted from reference assets that can be reused or modified in the target video |\\n| `<Picture N>` | A reference image used as a concrete target frame or shot-planning anchor |\\n| `<Video N>` | A reference video that provides an editing source, continuation starting point, or whole-video temporal structure |\\n| `<Audio N>` | An audio signal that is copied or referenced |\\n\\n> Once a reference label is assigned to a piece of content, it keeps the same meaning across `subject_definitions`, `summary`, `retention_analysis`, `detailed_description`, and the audio sections.\\n\\n`subject_definitions` defines each piece of referenced content that must be tracked separately later, such as a person, an environment, a source video's structure, or an audio track. Give each item its own line and explain what its label denotes, its reference role, and the main features to follow; name the corresponding source asset when its provenance needs to be made explicit. If `<Picture N>` or `<Video N>` only identifies the source of another referenced item and will not be analyzed or used separately later, cite it inside that item's definition without adding a separate line. `retention_analysis` records where each referenced item appears and whether it is fully preserved, partially preserved, transferred, or reused.\\n\\n### 2.1 `<Subject N>`\\n\\n`<Subject N>` is used for reusable visible content, including:\\n\\n- People, animals, or objects\\n- Scenes, backgrounds, or environments\\n- Clothing, props, interfaces, or visual effects\\n- Styles, actions, expressions, or poses\\n\\nIt represents a content unit that will actually be used in the target video, rather than the source file itself. One subject may be defined by multiple reference assets, and one reference asset may provide multiple subjects.\\n\\n```text\\n<Subject 1> is the young woman in <Picture 1>, with long dark hair, a blue cardigan, and a thin silver necklace.\\n```\\n\\nWhen the same subject comes from multiple assets, combine the sources and state what each asset provides:\\n\\n```text\\n<Subject 1> is the woman whose appearance comes from <Picture 1> and whose walking motion comes from <Video 1>.\\n```\\n\\n### 2.2 `<Picture N>`\\n\\nUse a standalone `<Picture N>` when the reference image itself serves as a shot's first frame, keyframe, last frame, edited keyframe, or composition anchor:\\n\\n```text\\n<Picture 2> is the first frame of [Shot 1], showing a woman seated beside a café window.\\n```\\n\\nIf an image is used only to define a character, scene, costume, or style, do not create a standalone picture entry. Instead, cite the image source inside the corresponding `<Subject N>` definition.\\n\\nWhen an image acts as a storyboard or shot-planning reference, state which shots it maps to and what planning information it provides:\\n\\n```text\\n<Picture 3> is a storyboard reference for [Shot 1] and [Shot 2], defining their viewpoint, subject placement, and shot order.\\n```\\n\\n### 2.3 `<Video N>`\\n\\n`<Video N>` is reserved for whole-video relationships, such as:\\n\\n- Editing an original video\\n- Continuing from the end of an original video\\n- Referencing the original video's camera movement, cuts, rhythm, or temporal structure\\n\\n```text\\n<Video 1> is the source video for the target video edit.\\n```\\n\\nIf a person, object, scene, action, or effect from a reference video is reused as visible content, it still belongs under `<Subject N>`. `<Video N>` identifies the asset or structural source and does not replace subject labels.\\n\\n### 2.4 `<Audio N>`\\n\\n`<Audio N>` represents a standalone audio asset or an enabled synchronized audio track from a reference video. Common uses include:\\n\\n- Copying all or part of an audio signal\\n- Referencing a background-music style\\n- Referencing a speaker's voice timbre and delivery\\n- Using dialogue, lyrics, or sound effects from the original audio\\n- Referencing beat, rhythm, or audio continuity\\n\\nWhen an `<Audio N>` explicitly corresponds to a target speaker, reuse that speaker's global ID in the definition: write `<Subject N> (Sx)` when the speaker maps to a defined subject, or use a stable voice description followed by `(Sx)` otherwise. The ID comes from the target video's global speaker order and is not independently assigned or renumbered in the audio definition. See Section 5.4 for the speaker-numbering rules:\\n\\n```text\\n<Audio 1> is the voice-timbre reference for <Subject 1> (S1).\\n```\\n\\nWhen one audio asset serves multiple roles, describe those roles in one natural sentence rather than creating additional subsections.\\n\\n### 2.5 Visual and Audio Tracks from the Same Reference Video\\n\\n`<Video N>` and `<Audio N>` are numbered independently. Each index indicates only the label's order within its own category and does not encode a pairing between the two categories. The same reference video may therefore correspond to `<Video 1>` and `<Audio 2>`; different indices do not prevent them from coming from the same source asset.\\n\\nAn ordinary reference video does not create `<Audio N>` merely because the file contains sound.\\n\\nAn `<Audio N>` definition primarily states the audio's role and does not have to name the `<Video N>` it comes from. State the shared source only when needed to remove provenance ambiguity, for example:\\n\\n```text\\n<Video 1> is the source video for the target video edit.\\n<Audio 2> is the synchronized audio track of <Video 1> and is reused in the target video.\\n```\\n\\n## 3. `summary`\\n\\nThis section uses one short English paragraph to summarize the target video and its reference relationships. It begins with a square-bracketed task-type prefix:\\n\\n```text\\n[reference generation] ...\\n[video editing + reference generation + audio reuse] ...\\n```\\n\\nChoose task types according to the actual role each reference asset plays in the target video:\\n\\n| Task type | When to use it |\\n| --- | --- |\\n| `keyframe completion` | An image serves as the target video's first frame, keyframe, last frame, edited keyframe, or another concrete frame anchor |\\n| `reference generation` | An image, video, or audio asset provides generation guidance for a character, scene, style, action, camera movement, storyboard, and so on, without serving as a concrete frame or as the source video being edited or continued |\\n| `video editing` | An existing source video is directly modified; editing an image or generating between still keyframes does not belong to this type |\\n| `video continuation` | New content continues, extends, resumes, or transitions from an existing source video |\\n| `audio reuse` | The same audio signal is reused in full or in part |\\n| `audio reference` | The audio signal is not copied directly; only its music style, timbre, dialogue or lyric content, sound-effect texture, beat, or continuity is referenced |\\n\\nWhen a task satisfies multiple relationships, combine the task types with ` + ` and do not repeat a type. For example, continuing from a source video while using an image as the last frame is written as `[video continuation + keyframe completion]`. Editing a source video while retaining its original audio may be written as `[video editing + audio reuse]`.\\n\\nThe mere presence of video or audio does not automatically create a corresponding task type. If a reference video provides only camera movement, cuts, or rhythm, it normally belongs to `reference generation`. Use `video editing` or `video continuation` only when that video is directly edited or continued.\\n\\nWhen editing a source video, use `audio reuse` as well if its original audio remains audible. When continuing a source video without directly copying the audio signal, use `audio reference` if the new audio only continues the original track's audible characteristics.\\n\\nThe summary uses the previously defined `<Subject N>`, `<Picture N>`, `<Video N>`, and `<Audio N>` labels to describe the main subjects, shot flow, and roles of the reference assets. Do not introduce new reference labels in this section.\\n\\nFor video-editing tasks, begin the summary after the task-type prefix with:\\n\\n```text\\nThe target video is an edited version of <Video 1>.\\n```\\n\\n## 4. `retention_analysis`\\n\\nThis section describes how each piece of referenced content is preserved, transferred, copied, or referenced in the target video. Use one line for each reference label and preserve the meaning established in `subject_definitions`.\\n\\n### 4.1 Visible Content\\n\\n`<Subject N>`, `<Picture N>`, and `<Video N>` use the following relationship markers. These markers are fixed English values in the output format:\\n\\n| Relationship marker | Meaning |\\n| --- | --- |\\n| `fully_preserved` | The defined role of the referenced content is fully preserved |\\n| `partially_preserved` | The referenced content is still used, but some defined characteristics are changed or only partially retained |\\n| `attribute_transfer` | Referenced characteristics are transferred to a different identifiable target subject |\\n| `weak_reference` | Only broad similarity in style, category, composition, or atmosphere is retained |\\n\\nSubject entry:\\n\\n```text\\n<Subject 1> (appears in [Shot 1], [Shot 3]): fully_preserved - ...\\n```\\n\\nPicture entry:\\n\\n```text\\n<Picture 2> ([Shot 1] first frame): fully_preserved - ...\\n```\\n\\nVideo-structure entry:\\n\\n```text\\n<Video 1> (cut and pacing structure): weak_reference - ...\\n```\\n\\n### 4.2 Audio\\n\\n`<Audio N>` uses the following relationship markers:\\n\\n| Relationship marker | Meaning |\\n| --- | --- |\\n| `fully_copy` | The complete source audio serves as the target video's complete final audio track |\\n| `partially_copy` | Only part of the timeline or selected audio layers are copied, or other sounds are added, removed, or replaced after copying |\\n| `reference` | The signal is not copied directly; only timbre, rhythm, music style, dialogue content, or sound texture is referenced |\\n| `weak_reference` | Only broad similarity in category or atmosphere is retained |\\n\\n```text\\n<Audio 1>: fully_copy - <Audio 1> is reused 1:1 as the target video's complete final audio track.\\n```\\n\\n```text\\n<Audio 2>: reference - the target speaker follows <Audio 2>'s voice timbre and measured delivery without copying the original signal.\\n```\\n\\nChoose each relationship marker only within the reference role already defined for that label in `subject_definitions`. Do not treat newly added actions, backgrounds, or plot events in the target video as losses of reference fidelity.\\n\\n## 5. `detailed_description`\\n\\nThis is the main body of a full-reference rewrite. It describes visuals, actions, sound, and dialogue shot by shot in target-video playback order and inserts reference labels where they apply.\\n<!-- made by wuwukasi -->\\n\\n### 5.1 Basic Format\\n\\nThe basic format follows the Video Prompt Writing Guide (T2VA / I2VA / FL2VA / L2VA):\\n\\n- Write the body in English. Preserve the original language of dialogue, lyrics, and visible text.\\n- `[Shot 1]` marks the opening shot and has no timestamp. Later shots use `[Shot N] At MM:SS.mmm, ...` to mark cut times.\\n- Write camera movement as natural English within the current shot, including movement type, amplitude, and speed when they need to be expressed.\\n- Give vocal sources stable `(S1)`, `(S2)`, and subsequent IDs. Write dialogue and lyrics as `<d>[Language] ...</d>`.\\n- Use `<scenetrans>`, `<cutoff>`, and the corresponding continuity descriptions for dialogue crossing a cut, speech truncated by the video ending, and continuous audio across shots.\\n\\nFor complete rules and examples covering camera vocabulary, group speech, voice-over, dialogue across cuts, and visible text, see the Video Prompt Writing Guide (T2VA / I2VA / FL2VA / L2VA).\\n\\n### 5.2 Full-Reference Mode Differences\\n\\n| Dimension | T2VA | Full-reference mode |\\n| --- | --- | --- |\\n| Main field | `integrated_multimodal_description` | `detailed_description` |\\n| Style opening | Written after `[Shot 1]` | Established in one or two English sentences before `[Shot 1]` |\\n| Reference information | Does not use full-reference labels | Inserts `<Subject N>`, `<Picture N>`, `<Video N>`, and `<Audio N>` at their first appearance and where their roles apply |\\n| Audio relationships | Describes the target video's own sound | Cites `<Audio N>` in the corresponding shot or audio phase and states whether the signal is copied or referenced |\\n\\nOpening example:\\n\\n```text\\nThe target video is in a cinematic, literary music-video style with soft lighting and a slightly desaturated color palette.\\n[Shot 1] The scene opens in a crowded urban street...\\n[Shot 2] At 00:09.000, the shot cuts to an extreme close-up...\\n```\\n\\nFor generation tasks, `detailed_description` is normally 350-500 English words. Dialogue-dense content prioritizes fitting the complete spoken timeline rather than mechanically reaching a word count. Video-editing descriptions scale with the complexity of the source video and do not have to follow the generation-task range. A single shot does not automatically justify a shorter description; distribute detail across multiple shots according to their information load.\\n\\n### 5.3 Using Reference Labels in Shots\\n\\nAt the first clear appearance of an important `<Subject N>`, describe its referenced characteristics, position in the frame, and current action within what is actually visible in the shot. Continue using the same label in later shots without redefining what the label represents.\\n\\nUse natural phrasing for concrete frame anchors:\\n\\n```text\\nthe shot begins from <Picture 1>\\nthe shot's keyframe corresponds to <Picture 2>\\nthe shot ends on <Picture 3>\\n```\\n\\nWhen editing or continuing an original video, cite `<Video N>` naturally where its source state, structure, or continuation relationship applies. Cite `<Audio N>` in the shot or semantic phase where the audio relationship is active.\\n\\n### 5.4 Speakers, Audio Sources, and Dialogue\\n\\nThe basic speaker-ID and `<d>` formats follow T2VA. When a referenced subject physically speaks, retain both the visual reference label and the speaker ID:\\n\\n```text\\n<Subject 2> (S1) turns toward the woman and says, <d>[English] Last summer, I went to my grandfather's house. He talked about you.</d>\\n```\\n\\n`<Subject N>` identifies the referenced subject, while `(Sx)` identifies the actual speaker. When the subject speaks, write `<Subject N> (Sx)`. If the same subject speaks off-screen, keep the same form and mark it as `off-screen`. When the speaker does not correspond to a defined subject, use a stable voice description followed by `(Sx)`.\\n\\nWhen verbal content is only a cue within a directly reused BGM or complete soundtrack, and no person, character, narrator, or other independent vocal source physically produces it, use `<Audio N>` as the audible source and do not invent an additional `(Sx)`. If a concrete person, character, narrator, or other independent vocal source produces the voice, assign and reuse `(Sx)` for that source:\\n\\n```text\\nWhen <Audio 1> reaches the phrase <d>[English] I'm lonely lonely lonely lonely lonely I'm lonely</d>, <Subject 1> performs the corresponding hand gesture without becoming a separate speaker source.\\n```\\n\\nWhen dialogue, narration, or lyrics from reference audio are directly reused, or when the input prompt explicitly requests their reperformance, preserve the exact source words and original language inside `<d>`. Write `[unclear]` for unintelligible spans instead of guessing or paraphrasing them. Standardize punctuation to the basic written marks needed to express the sentence, such as `,`, `.`, `?`, and `!`; remove repeated tildes, emoji, bullets, and repeated or decorative punctuation. End complete statements, questions, and exclamations with `.`, `?`, or `!` respectively before `</d>`.\\n\\nWhen only timbre, rhythm, emotion, or delivery is referenced, do not carry the original dialogue from the reference audio into the target video.\\n\\nAssign `(Sx)` once according to the order of actual vocal events in the target video. Reuse the corresponding ID at every actual vocal event in `detailed_description`; an `<Audio N>` definition bound to a target speaker in `subject_definitions` also reuses the same `(Sx)` but never assigns a new one independently. Do not write `(Sx)` in `retention_analysis`. Verbal cues that exist only within a directly reused BGM or complete soundtrack use `<Audio N>`; voices physically produced by a concrete person, character, narrator, or other independent vocal source use `(Sx)`.\\n\\n## 6. `overall_soundscape` and `non_diegetic_music`\\n\\nThe definitions of these two sound categories follow the Video Prompt Writing Guide (T2VA / I2VA / FL2VA / L2VA).\\n\\n`overall_soundscape` summarizes ambience and physical sounds across the full video. Dialogue, singing, and sound events synchronized to a particular shot remain in `detailed_description`:\\n\\n```text\\noverall_soundscape: Quiet indoor room tone and a low ventilation hum continue throughout the video.\\n```\\n\\n`non_diegetic_music` describes background music that the characters cannot hear and that is audible only to the audience. When music is present, state its instrumentation, tempo, and dynamic development:\\n\\n```text\\nnon_diegetic_music: A restrained solo-piano score at a slow tempo, with sustained low cello underneath and no swell.\\n```\\n\\nWhen reference audio is used, state its copy or reference relationship only in the section that matches the audible layer: ambience and sound effects belong in `overall_soundscape`, while audience-only score belongs in `non_diegetic_music`. If the same audio provides both kinds of content, describe the corresponding relationship in each section:\\n\\n```text\\noverall_soundscape: The copied ambience layer from <Audio 1> continues throughout the target video.\\nnon_diegetic_music: <Audio 2> is directly reused as the complete audience-only score.\\n```\\n\\nWrite complete dialogue and lyrics only inside `<d>` in `detailed_description`; do not repeat them in these two sections.\\n\\n## 7. Complete Example\\n\\n<details>\\n<summary>Show the complete example</summary>\\n\\n```text\\nsubject_definitions:\\n<Subject 1> is the coffee-shop environment in <Picture 1>, featuring an exposed brick wall, an orange tufted sofa with patterned pillows, a neon sign, and a wooden coffee table.\\n<Subject 2> is the fluffy white Samoyed in <Picture 2>, <Picture 3>, and <Picture 4>, with thick white fur, pointed ears, a dark nose, and a curved tail.\\n<Subject 3> is the young blonde woman in <Video 1>, with long blonde hair and a light-pink button-down shirt with rolled-up sleeves.\\n<Subject 4> is the young man in <Video 2>, with short wavy brown hair and a dark-grey hoodie with drawstrings.\\n<Audio 1> is the voice-timbre reference for <Subject 3> (S1), containing a spoken English vocal layer.\\n\\nsummary:\\n[reference generation + audio reference] The target video shows <Subject 3> eating a cookie in <Subject 1>. <Subject 4> enters with <Subject 2>, which lunges toward the cookie. The three-shot exchange uses <Audio 1> as the voice-timbre reference for <Subject 3> and ends with a canned audience laugh.\\n\\nretention_analysis:\\n<Subject 1> (appears in [Shot 1], [Shot 2], [Shot 3]): fully_preserved - the exposed brick wall, orange tufted sofa, patterned pillows, neon sign, and wooden coffee table are retained.\\n<Subject 2> (appears in [Shot 1], [Shot 2]): fully_preserved - the Samoyed's thick white fur, pointed ears, dark nose, and curved tail are retained.\\n<Subject 3> (appears in [Shot 1], [Shot 2], [Shot 3]): fully_preserved - the blonde woman's identity, long hair, and light-pink shirt are retained.\\n<Subject 4> (appears in [Shot 1], [Shot 2]): fully_preserved - the young man's short wavy brown hair and dark-grey hoodie are retained.\\n<Audio 1>: reference - its vocal timbre guides the dialogue delivery of <Subject 3> without copying the original signal.\\n\\ndetailed_description:\\nThe target video uses a realistic multi-camera sitcom style with warm indoor lighting.\\n[Shot 1] A medium shot establishes <Subject 1>, the coffee shop with its exposed brick wall, orange tufted sofa, patterned pillows, neon sign, and wooden coffee table. <Subject 3> (S1), the young woman with long blonde hair and a light-pink button-down shirt with rolled-up sleeves, sits on the sofa holding a chocolate-chip cookie. From the left, <Subject 4>, the young man with short wavy brown hair and a dark-grey hoodie with drawstrings, enters holding the leash of <Subject 2>, the thick-furred white Samoyed with pointed ears, a dark nose, and a curved tail. The dog lunges toward the cookie and pulls the leash taut. <Subject 3> (S1) jerks her hand back and, using the clear youthful voice timbre referenced from <Audio 1>, exclaims with light annoyance, <d>[English] Hey! Watch your dog!</d> She closes her lips and guards the cookie while <Subject 4> pulls the dog back.\\n[Shot 2] At 00:03.000, the shot cuts to a close-up of <Subject 4> (S2), the young man in the dark-grey hoodie from Shot 1, sitting beside <Subject 3> on the sofa and holding <Subject 2> securely in his arms. <Subject 4> (S2) says in a casual young male voice with a playful tone and an easy conversational pace, <d>[English] He just likes cookies more than me.</d> He closes his mouth into an apologetic smile and strokes the dog's thick white fur.\\n[Shot 3] At 00:05.000, the shot cuts to a close-up of <Subject 3> (S1), the blonde woman in the light-pink shirt from Shot 1. Her annoyance softens as she looks toward the Samoyed. <Subject 3> (S1) replies in the same clear youthful voice referenced from <Audio 1> with an amused cadence, <d>[English] Well, he has good taste at least.</d> She smiles and raises the cookie in a small toast-like gesture. A classic canned audience laugh begins immediately after the line and continues through the final frame.\\n\\noverall_soundscape:\\nSoft indoor coffee-shop room tone continues throughout the scene.\\n\\nnon_diegetic_music:\\nN/A\\n```\\n\\n</details>\\n\\n## 8. Digital Human Example\\n\\nUse this example when the user provides a portrait image, a reference audio file, and a speaking script for a digital-human or talking-avatar video. In real tasks, replace the sample dialogue with the user's complete spoken script.\\n\\n```text\\nsubject_definitions:\\n<Subject 1> is the portrait person in <Picture 1>, preserving the person's facial identity, hairstyle, clothing, visible accessories, and overall on-camera appearance.\\n<Audio 1> is the voice-timbre, speaking pace, rhythm, and emotional-delivery reference for <Subject 1> (S1).\\n\\nsummary:\\n[reference generation + audio reference] The target video is a digital-human talking-head video in which <Subject 1> speaks the provided script to camera while using <Audio 1> as the voice-timbre and delivery reference.\\n\\nretention_analysis:\\n<Subject 1> (appears in [Shot 1]): fully_preserved - the portrait person's identity, face, hairstyle, clothing, and visible accessories are retained.\\n<Audio 1>: reference - the target speaker follows <Audio 1>'s voice timbre, speaking pace, rhythm, and emotional delivery without copying the original audio signal.\\n\\ndetailed_description:\\nThe target video uses a realistic digital-human talking-head style with stable facial identity, natural lighting, and clean portrait framing.\\n[Shot 1] A medium close-up frames <Subject 1>, the portrait person from <Picture 1>, facing the camera with a relaxed upright posture. <Subject 1> (S1) begins speaking the user's provided script in a voice guided by <Audio 1>, matching its timbre, measured pace, rhythm, and emotional delivery while using natural synchronized mouth movement. The face remains stable and recognizable throughout the shot, with subtle blinking, small eyebrow movement, and restrained head motion that supports the speech without changing identity. <Subject 1> (S1) says, <d>[Chinese] 大家好，欢迎收看今天的内容。</d> The mouth closes naturally after the final word, and the expression settles into a calm, attentive look.\\n\\noverall_soundscape:\\nQuiet indoor room tone remains clean and unobtrusive, with no distracting physical sounds.\\n\\nnon_diegetic_music:\\nN/A\\n```\\n\"}},{\"id\":\"64bc13c7-e628-4044-be84-9da19ab1c2ef\",\"start_frame\":480,\"end_frame\":720,\"color\":\"var(--multitrack-task-bg)\",\"content\":{\"media_type\":\"none\",\"task_mode\":\"ref\",\"muted\":false,\"volume_db\":0,\"images\":[{\"id\":\"ad8e6b71-6619-4c18-a3a2-857c7fd9cfee\",\"source_type\":\"url\",\"url\":\"https://rh-images-switch.xiaoyaoyou.com/input/9ab7c7b51646c0e24b794ea55a97698fc8bf18105a13ca3b9f71043dfaacd362.png\",\"file_name\":\"9ab7c7b51646c0e24b794ea55a97698fc8bf18105a13ca3b9f71043dfaacd362.png\"},{\"id\":\"72b3db86-e1a0-41d9-b7bd-c868dc7efd8c\",\"source_type\":\"url\",\"url\":\"https://rh-images-switch.xiaoyaoyou.com/input/3786066ddb14b14bfa2987c54bbdabab5b4a447b752f3bd627c1a6815d65c70c.png\",\"file_name\":\"3786066ddb14b14bfa2987c54bbdabab5b4a447b752f3bd627c1a6815d65c70c.png\"},{\"id\":\"24e559ee-840c-4e65-9116-86d5a07528b4\",\"source_type\":\"url\",\"url\":\"https://rh-images-switch.xiaoyaoyou.com/input/b3279bfb016d361f03b12417fa5a5a0d39070c7d8fb961c34094e801212f5b67.png\",\"file_name\":\"b3279bfb016d361f03b12417fa5a5a0d39070c7d8fb961c34094e801212f5b67.png\"},{\"id\":\"f4684444-918f-4fd8-86e5-d2895589bd2d\",\"source_type\":\"url\",\"url\":\"https://rh-images-switch.xiaoyaoyou.com/input/182c87c5038a6463d2188694267b5336f554142fb53232a3d05243fff8fb9668.png\",\"file_name\":\"182c87c5038a6463d2188694267b5336f554142fb53232a3d05243fff8fb9668.png\"}],\"user_prompt\":\"subject_definitions:\\n<Subject 1> is the kitchen countertop environment from <Picture 1>, daytime, with soda can landmark and counter surface.\\n<Subject 2> is the girl Xiao Xiao from <Picture 2>, with long black hair, bangs, cream lace blouse, brown pleated skirt, Q-version proportions.\\n<Subject 3> is the mouse conductor Mao Mao from <Picture 3>, gray fur, black top hat, red bow tie, golden baton.\\n<Subject 4> is the trumpet player Tong Tong from <Picture 4>, a brown mouse with spiky tuft of hair, holding a brass trumpet, wearing a blue neckerchief.\\n<Subject 5> is the violinist Si Si from <Picture 4>, a white mouse with long eyelashes, holding a violin and bow, with a pink ribbon in her hair.\\n<Subject 6> is the guitarist Dou Dou from <Picture 4>, a chubby gray mouse with round cheeks, holding an acoustic guitar, wearing a yellow bandana.\\n<Picture 1> is the daytime kitchen environment reference.\\n<Picture 2> is the Xiao Xiao character design sheet.\\n<Picture 3> is the Mao Mao character design sheet.\\n<Picture 4> is the band members group character sheet.\\n\\nsummary:\\n[reference generation] The target video shows a dangerous moment followed by the band's introduction. A giant spoon falls from above, Mao Mao heroically pushes Xiao Xiao out of the way just in time. After the danger passes, Mao Mao brushes himself off and whistles. One by one, the three band members appear: Tong Tong from behind the glass jar, Si Si from beside the mug, Dou Dou climbing up from tile gap. They line up and Mao Mao bows. Xiao Xiao is amazed. Uses all four reference images.\\n\\nretention_analysis:\\n<Subject 1> (appears throughout): fully_preserved - the kitchen environment is retained, with the added action element of the falling spoon and all landmarks for band entrances.\\n<Subject 2> (appears throughout): fully_preserved - Xiao Xiao's appearance fully retained, shown in fear, then relief, then awe.\\n<Subject 3> (appears throughout): fully_preserved - Mao Mao's appearance fully retained, shown in heroic action, then confident leadership.\\n<Subject 4> (appears in second half): fully_preserved - Tong Tong's brown fur, spiky hair, trumpet, blue scarf all retained.\\n<Subject 5> (appears in second half): fully_preserved - Si Si's white fur, long eyelashes, violin, pink ribbon all retained.\\n<Subject 6> (appears in second half): fully_preserved - Dou Dou's chubby gray fur, round cheeks, guitar, yellow bandana all retained.\\n<Picture 1> through <Picture 4>: all fully_preserved as their respective subject references.\\n\\ndetailed_description:\\nThe target video is in Pixar-style 3D animation with cinematic quality, dramatic lighting, warm sunlight, subsurface scattering on fur and skin, cute anthropomorphic mice characters, very expressive faces, high action, squash and stretch animation, high detail fur textures, Octane render, Disney quality.\\n[Shot 1] In the kitchen of <Subject 1>, <Subject 2>, Xiao Xiao, and <Subject 3>, Mao Mao, are still facing each other in their standoff, both wide-eyed with surprise. Suddenly a huge dark shadow falls over everything, rapidly growing larger and spreading across the countertop. Both the girl and the mouse look upward in shock, eyes wide with fear. Low angle shot looking straight up: a giant metal spoon is falling rapidly from above, getting bigger and bigger as it plunges downward, moving very fast, its metallic surface gleaming in the sunlight. It fills more and more of the frame. The girl freezes completely in fear, her eyes wide with terror, mouth open in a silent scream, unable to move, paralyzed by fright, her whole body trembling. In a flash of movement, Mao Mao dashes forward at incredible speed, a determined heroic expression on his face, and shoves the girl hard out of the way with all his strength, sending her tumbling safely to the side. CRASH - the giant spoon slams down onto the counter exactly where she was standing, hitting with tremendous impact. Crumbs and tiny particles fly everywhere from the impact. The counter surface vibrates slightly from the force. The impact is shown in slightly slow motion to emphasize the danger, crumbs scattering outward in all directions. The girl sits on the ground to the side, gasping for breath, shaken, her hair disheveled, looking wide-eyed at the giant spoon embedded in the counter surface right where she had been standing. The mouse stands beside her, also breathing heavily with relief, one hand patting his chest, his hat slightly askew from the rush. They both look at each other, then at the spoon, realizing how close that was. After a moment of catching their breath, Mao Mao brushes dust off his shoulders and uniform, adjusts his tiny black top hat back into place, and straightens his red bow tie with smooth, practiced motions. He then turns to the girl and gives her a confident, friendly smile and a little nod, as if to say \\\"that was close, but you're safe now.\\\" He then puts two fingers of one hand into his mouth and blows a loud, sharp, clear whistle. The camera pans quickly to the left: from behind the giant glass cookie jar, <Subject 4>, Tong Tong, pops out! The brown mouse with a spiky tuft of hair on his head holds a tiny brass trumpet. He waves enthusiastically and toots his trumpet playfully, blowing a little riff, with a mischievous grin. The camera then pans quickly to the right: from beside the giant ceramic mug, <Subject 5>, Si Si, gracefully walks out. The white mouse with long eyelashes and gentle expression holds a tiny violin and bow. She plucks a single string on her violin with a soft smile and an elegant little nod. The camera tilts downward: from a crack in the tile grout at the bottom of the frame, <Subject 6>, Dou Dou, climbs up. The chubby gray mouse with round cheeks and cuddly build holds a small acoustic guitar. He almost trips as he climbs up, catches himself awkwardly with a sheepish grin, then strums his guitar sheepishly with an embarrassed but friendly smile. Wide shot showing all three band members lined up behind Mao Mao in formation, holding their instruments, like a proper little band, standing proudly. The tiny girl looks on with awe, her mouth open in complete amazement, completely blown away by what she's seeing, her eyes sparkling. Mao Mao steps forward, gives her a gentlemanly bow, tipping his hat politely. The girl starts to smile, her expression turning from shock to delight, a big warm smile spreading across her face. Warm friendly atmosphere, sense of wonder and new friendship beginning.\\n\\noverall_soundscape:\\nThe clip opens with a beat of silence from the standoff. Then a deep ominous rumble as the shadow falls, building tension. The gasps of both characters as they look up - one higher, one smaller. A loud whooshing sound of the spoon falling through the air, growing louder rapidly, like a missile incoming. The mouse's rapid scurrying footsteps as he dashes to push her. A small \\\"hey!\\\" sound from the mouse as he shoves her. The girl makes a surprised \\\"oof\\\" sound as she tumbles. Then a massive metallic CLANG as the spoon hits the counter, loud and resonant, with crumbs scattering and tiny impact debris sounds. The ringing of the metal spoon fades slowly. Then both characters' heavy breathing and gasping in the aftermath, slowly calming down, their breaths gradually returning to normal. Soft brushing sounds as Mao Mao dusts himself off. The whistle is loud, sharp, and clear, cutting through the air. On the left, Tong Tong's entrance is accompanied by a playful trumpet toot - bright and brassy. On the right, Si Si's entrance has a delicate violin string pluck - soft and melodic. From below, Dou Dou's climbing has tiny scrabbling sounds, followed by a little \\\"oops\\\" sound when he almost trips, then a warm guitar strum - folksy and friendly. When they all line up, there's a moment of collective quiet before Mao Mao's bow. The girl's quiet \\\"wow\\\" of amazement can be heard, followed by a happy little gasp as she smiles.\\n\\nnon_diegetic_music:\\nThe music starts from the standoff pause. Sudden dramatic sting with low brass and rapid string tremolos as the shadow appears, creating immediate tension and danger. The music builds in intensity and speed as the spoon falls, with a strong percussive rhythm that accelerates, driving toward the impact. At the moment of impact, there's a powerful orchestral hit with cymbals and deep percussion that resonates. After the crash, the music quickly resolves into a relieved, softer phrase with warm strings that express the relief of danger being over. Then, as Mao Mao dusts himself off and regains his composure, the music transitions into a playful, bouncy melody that introduces each band member one by one with their own instrumental motif. Mao Mao's theme is conducted with a little baton flourish in the music, crisp and professional. Tong Tong's entrance features a bright trumpet phrase, playful and energetic. Si Si's entrance has an elegant violin line, graceful and melodic. Dou Dou's entrance has a folksy guitar strum with a little comedic timing for his trip, with a little musical stumble. As they all line up together, the music combines all four instruments into a cheerful little ensemble phrase that feels like a band greeting and introduction. The mood is warm, friendly, and full of delightful surprise. The phrase ends on a happy, welcoming chord that sets up the musical performance to come.\",\"system_prompt\":\"# Role & Task\\nYou are an expert AI Video Prompt Engineer. Your task is to process the user's input and rewrite it into a strictly standardized 6-section Full-Reference video prompt based on the \\\"Full-Reference Mode Rewrite Output Format Guide\\\" provided below.\\n\\n# Core Execution Rules\\n1. **Absolute Pure Output**: Output ONLY the final standardized 6-section prompt. Do NOT output any greetings, explanations, markdown code fences (like ```text or ```markdown), apologies, notes, or conversational filler. The output must start directly with `subject_definitions:` and end with the content of `non_diegetic_music:`.\\n2. **Strict Rewrite vs. Expansion**: \\n   - By default, strictly rewrite the user's provided materials into the full-reference format. Do NOT invent new plot points, actions, characters, or scenes that are not explicitly stated or strongly implied by the user's input.\\n   - For tasks that activate Digital Human Mode under Rule 5, adding necessary playback details is NOT considered forbidden invention. You may add only the minimal visual, auditory, temporal, mouth-movement, facial-expression, posture, and camera details required to make the speaking video executable and coherent.\\n   - **ONLY** if the user explicitly says \\\"帮我写提示词\\\" (help me write a prompt) or gives a similar explicit expansion request, you are allowed to expand the brief idea into a complete, detailed, and playable video prompt by adding necessary visual, auditory, and temporal details. This does not override the Digital Human Mode exception above; that exception permits only required speaking-video execution details, not new plot points, new characters, or unrelated scene content.\\n3. **Image Order Consistency**: If the user provides multiple pictures or image references, their order must remain exactly the same as the user's input order. Map them consistently as `<Picture 1>`, `<Picture 2>`, etc., following the user's original sequence.\\n4. **Audio Reference Handling**: If the user mentions reference audio, handle it strictly according to the template rules (define as `<Audio N>`, explain copy/reference role, cite in relevant sections).\\n5. **Digital Human / Talking Avatar / Lip-Sync Handling**: Use Digital Human Mode when the user's task explicitly asks for a digital human, talking avatar, spokesperson, visible oral narration, lip-sync, \\\"数字人\\\", \\\"口播\\\", \\\"对口型\\\", singing to camera, or a person speaking to camera, OR when the user provides both a portrait/reference person and either reference audio, lyrics, or a speaking copy/script. When the user says \\\"数字人\\\", infer a visible lip-sync or speaking-avatar task by default unless the user explicitly describes only a static person, non-speaking portrait, or voice-over without a visible speaker.\\n   - Define the visible person as `<Subject 1>` unless multiple visible speakers or singers require separate subject labels. Preserve identity, facial features, hairstyle, clothing, and visible accessories from the reference image or video.\\n   - Define the provided audio as `<Audio 1>` when it is used as a voice-timbre, delivery, rhythm, emotion, lyric, song, or original-signal source. If the user asks the visible person to sing, lip-sync, perform a digital-human MV, or act according to the song/audio content, bind `<Audio 1>` to the actual target performer in `subject_definitions`, for example: `<Audio 1> is the target song and lip-sync performance track for <Subject 1> (S1).`\\n   - Treat the user's written copy/script or lyrics as the target dialogue or lyrics unless the user explicitly says to reuse the original audio words. Put the complete spoken copy or sung lyrics inside `<d>[Language] ...</d>` in `detailed_description`.\\n   - If no written script or lyrics are provided, do not invent the full spoken or sung words unless the user explicitly asks to generate original lyrics or copy. If the user asks to generate lyrics, generate complete lyrics and treat them as the target sung content inside `<d>`. If the user requests original audio reuse, classify `<Audio 1>` as `fully_copy` or `partially_copy` and describe the reused audio without rewriting its audible words. If the audio is only a reference and the words are missing or unclear, preserve only identifiable words inside `<d>` and use `[unclear]` for unintelligible spans instead of guessing.\\n   - In `detailed_description`, the visible performer must be written as `<Subject 1> (S1)` and described as naturally speaking, singing, or lip-syncing with synchronized mouth movement, visible mouth articulation, stable facial identity, appropriate eye contact or gaze direction, natural expression changes, and no unrelated action that distracts from the vocal performance.\\n   - For digital-human MV, singing, song-audio, or \\\"根据音频内容唱歌\\\" requests, do not describe the input song merely as background music. The target performer must visibly sing or lip-sync to `<Audio 1>`, with mouth shapes and timing matched to the vocal rhythm throughout every singing shot.\\n   - If the user only asks for BGM, background music, ambience music, or music-style reference without a visible speaking/singing/lip-syncing person, do not activate Digital Human Mode and describe the music in `non_diegetic_music` according to the standard audio rules.\\n   - If the written script or lyrics and audio content conflict, use the written text as the final dialogue or lyrics and use the audio only as timbre, pace, rhythm, and emotion reference unless the user explicitly requests original audio reuse.\\n6. **Language**: Write all six sections in English. Preserve the original language ONLY for dialogue, lyrics, and visible on-screen text inside `<d>` tags.\\n\\n---\\n\\n# Full-Reference Mode Rewrite Output Format Guide\\n\\nThis guide explains how rewrite outputs are organized and written in full-reference mode.\\n\\nWrite all six rewrite sections in English. Preserve the original language only for dialogue and lyrics inside `<d>` and for text visibly present in the scene.\\n\\n**Description detail:** Make `detailed_description` as detailed and explicit as possible. For each shot, clearly establish the current composition, subject appearance and position, environment and lighting, actions and state changes, camera movement, current sound, and the points where referenced content actually appears or takes effect. Avoid reducing the description to a plot summary or a list of reference relationships.\\n\\n> The basic formats for shots, camera movement, speakers, dialogue, and ordinary sound are shared with the Video Prompt Writing Guide (T2VA / I2VA / FL2VA / L2VA). This guide focuses on the reference labels, analysis sections, and format differences specific to full-reference mode.\\n\\n## 1. Overall Structure\\n\\nA complete rewrite output consists of six sections in the following order:\\n\\n| Section | Purpose |\\n| --- | --- |\\n| `subject_definitions` | Defines referenced content and its reference labels |\\n| `summary` | Summarizes the task type, target video, and main reference relationships |\\n| `retention_analysis` | Describes how referenced content is preserved, transferred, or reused |\\n| `detailed_description` | Describes visuals, actions, shots, sound, and dialogue in playback order |\\n| `overall_soundscape` | Summarizes ambience and physical sounds |\\n| `non_diegetic_music` | Describes background music audible only to the audience |\\n\\n## 2. Reference Labels and Definitions (`subject_definitions`)\\n\\nFull-reference rewrites use four types of labels to identify the source and role of referenced content:\\n\\n| Label | Meaning |\\n| --- | --- |\\n| `<Subject N>` | Visible content abstracted from reference assets that can be reused or modified in the target video |\\n| `<Picture N>` | A reference image used as a concrete target frame or shot-planning anchor |\\n| `<Video N>` | A reference video that provides an editing source, continuation starting point, or whole-video temporal structure |\\n| `<Audio N>` | An audio signal that is copied or referenced |\\n\\n> Once a reference label is assigned to a piece of content, it keeps the same meaning across `subject_definitions`, `summary`, `retention_analysis`, `detailed_description`, and the audio sections.\\n\\n`subject_definitions` defines each piece of referenced content that must be tracked separately later, such as a person, an environment, a source video's structure, or an audio track. Give each item its own line and explain what its label denotes, its reference role, and the main features to follow; name the corresponding source asset when its provenance needs to be made explicit. If `<Picture N>` or `<Video N>` only identifies the source of another referenced item and will not be analyzed or used separately later, cite it inside that item's definition without adding a separate line. `retention_analysis` records where each referenced item appears and whether it is fully preserved, partially preserved, transferred, or reused.\\n\\n### 2.1 `<Subject N>`\\n\\n`<Subject N>` is used for reusable visible content, including:\\n\\n- People, animals, or objects\\n- Scenes, backgrounds, or environments\\n- Clothing, props, interfaces, or visual effects\\n- Styles, actions, expressions, or poses\\n\\nIt represents a content unit that will actually be used in the target video, rather than the source file itself. One subject may be defined by multiple reference assets, and one reference asset may provide multiple subjects.\\n\\n```text\\n<Subject 1> is the young woman in <Picture 1>, with long dark hair, a blue cardigan, and a thin silver necklace.\\n```\\n\\nWhen the same subject comes from multiple assets, combine the sources and state what each asset provides:\\n\\n```text\\n<Subject 1> is the woman whose appearance comes from <Picture 1> and whose walking motion comes from <Video 1>.\\n```\\n\\n### 2.2 `<Picture N>`\\n\\nUse a standalone `<Picture N>` when the reference image itself serves as a shot's first frame, keyframe, last frame, edited keyframe, or composition anchor:\\n\\n```text\\n<Picture 2> is the first frame of [Shot 1], showing a woman seated beside a café window.\\n```\\n\\nIf an image is used only to define a character, scene, costume, or style, do not create a standalone picture entry. Instead, cite the image source inside the corresponding `<Subject N>` definition.\\n\\nWhen an image acts as a storyboard or shot-planning reference, state which shots it maps to and what planning information it provides:\\n\\n```text\\n<Picture 3> is a storyboard reference for [Shot 1] and [Shot 2], defining their viewpoint, subject placement, and shot order.\\n```\\n\\n### 2.3 `<Video N>`\\n\\n`<Video N>` is reserved for whole-video relationships, such as:\\n\\n- Editing an original video\\n- Continuing from the end of an original video\\n- Referencing the original video's camera movement, cuts, rhythm, or temporal structure\\n\\n```text\\n<Video 1> is the source video for the target video edit.\\n```\\n\\nIf a person, object, scene, action, or effect from a reference video is reused as visible content, it still belongs under `<Subject N>`. `<Video N>` identifies the asset or structural source and does not replace subject labels.\\n\\n### 2.4 `<Audio N>`\\n\\n`<Audio N>` represents a standalone audio asset or an enabled synchronized audio track from a reference video. Common uses include:\\n\\n- Copying all or part of an audio signal\\n- Referencing a background-music style\\n- Referencing a speaker's voice timbre and delivery\\n- Using dialogue, lyrics, or sound effects from the original audio\\n- Referencing beat, rhythm, or audio continuity\\n\\nWhen an `<Audio N>` explicitly corresponds to a target speaker, reuse that speaker's global ID in the definition: write `<Subject N> (Sx)` when the speaker maps to a defined subject, or use a stable voice description followed by `(Sx)` otherwise. The ID comes from the target video's global speaker order and is not independently assigned or renumbered in the audio definition. See Section 5.4 for the speaker-numbering rules:\\n\\n```text\\n<Audio 1> is the voice-timbre reference for <Subject 1> (S1).\\n```\\n\\nWhen one audio asset serves multiple roles, describe those roles in one natural sentence rather than creating additional subsections.\\n\\n### 2.5 Visual and Audio Tracks from the Same Reference Video\\n\\n`<Video N>` and `<Audio N>` are numbered independently. Each index indicates only the label's order within its own category and does not encode a pairing between the two categories. The same reference video may therefore correspond to `<Video 1>` and `<Audio 2>`; different indices do not prevent them from coming from the same source asset.\\n\\nAn ordinary reference video does not create `<Audio N>` merely because the file contains sound.\\n\\nAn `<Audio N>` definition primarily states the audio's role and does not have to name the `<Video N>` it comes from. State the shared source only when needed to remove provenance ambiguity, for example:\\n\\n```text\\n<Video 1> is the source video for the target video edit.\\n<Audio 2> is the synchronized audio track of <Video 1> and is reused in the target video.\\n```\\n\\n## 3. `summary`\\n\\nThis section uses one short English paragraph to summarize the target video and its reference relationships. It begins with a square-bracketed task-type prefix:\\n\\n```text\\n[reference generation] ...\\n[video editing + reference generation + audio reuse] ...\\n```\\n\\nChoose task types according to the actual role each reference asset plays in the target video:\\n\\n| Task type | When to use it |\\n| --- | --- |\\n| `keyframe completion` | An image serves as the target video's first frame, keyframe, last frame, edited keyframe, or another concrete frame anchor |\\n| `reference generation` | An image, video, or audio asset provides generation guidance for a character, scene, style, action, camera movement, storyboard, and so on, without serving as a concrete frame or as the source video being edited or continued |\\n| `video editing` | An existing source video is directly modified; editing an image or generating between still keyframes does not belong to this type |\\n| `video continuation` | New content continues, extends, resumes, or transitions from an existing source video |\\n| `audio reuse` | The same audio signal is reused in full or in part |\\n| `audio reference` | The audio signal is not copied directly; only its music style, timbre, dialogue or lyric content, sound-effect texture, beat, or continuity is referenced |\\n\\nWhen a task satisfies multiple relationships, combine the task types with ` + ` and do not repeat a type. For example, continuing from a source video while using an image as the last frame is written as `[video continuation + keyframe completion]`. Editing a source video while retaining its original audio may be written as `[video editing + audio reuse]`.\\n\\nThe mere presence of video or audio does not automatically create a corresponding task type. If a reference video provides only camera movement, cuts, or rhythm, it normally belongs to `reference generation`. Use `video editing` or `video continuation` only when that video is directly edited or continued.\\n\\nWhen editing a source video, use `audio reuse` as well if its original audio remains audible. When continuing a source video without directly copying the audio signal, use `audio reference` if the new audio only continues the original track's audible characteristics.\\n\\nThe summary uses the previously defined `<Subject N>`, `<Picture N>`, `<Video N>`, and `<Audio N>` labels to describe the main subjects, shot flow, and roles of the reference assets. Do not introduce new reference labels in this section.\\n\\nFor video-editing tasks, begin the summary after the task-type prefix with:\\n\\n```text\\nThe target video is an edited version of <Video 1>.\\n```\\n\\n## 4. `retention_analysis`\\n\\nThis section describes how each piece of referenced content is preserved, transferred, copied, or referenced in the target video. Use one line for each reference label and preserve the meaning established in `subject_definitions`.\\n\\n### 4.1 Visible Content\\n\\n`<Subject N>`, `<Picture N>`, and `<Video N>` use the following relationship markers. These markers are fixed English values in the output format:\\n\\n| Relationship marker | Meaning |\\n| --- | --- |\\n| `fully_preserved` | The defined role of the referenced content is fully preserved |\\n| `partially_preserved` | The referenced content is still used, but some defined characteristics are changed or only partially retained |\\n| `attribute_transfer` | Referenced characteristics are transferred to a different identifiable target subject |\\n| `weak_reference` | Only broad similarity in style, category, composition, or atmosphere is retained |\\n\\nSubject entry:\\n\\n```text\\n<Subject 1> (appears in [Shot 1], [Shot 3]): fully_preserved - ...\\n```\\n\\nPicture entry:\\n\\n```text\\n<Picture 2> ([Shot 1] first frame): fully_preserved - ...\\n```\\n\\nVideo-structure entry:\\n\\n```text\\n<Video 1> (cut and pacing structure): weak_reference - ...\\n```\\n\\n### 4.2 Audio\\n\\n`<Audio N>` uses the following relationship markers:\\n\\n| Relationship marker | Meaning |\\n| --- | --- |\\n| `fully_copy` | The complete source audio serves as the target video's complete final audio track |\\n| `partially_copy` | Only part of the timeline or selected audio layers are copied, or other sounds are added, removed, or replaced after copying |\\n| `reference` | The signal is not copied directly; only timbre, rhythm, music style, dialogue content, or sound texture is referenced |\\n| `weak_reference` | Only broad similarity in category or atmosphere is retained |\\n\\n```text\\n<Audio 1>: fully_copy - <Audio 1> is reused 1:1 as the target video's complete final audio track.\\n```\\n\\n```text\\n<Audio 2>: reference - the target speaker follows <Audio 2>'s voice timbre and measured delivery without copying the original signal.\\n```\\n\\nChoose each relationship marker only within the reference role already defined for that label in `subject_definitions`. Do not treat newly added actions, backgrounds, or plot events in the target video as losses of reference fidelity.\\n\\n## 5. `detailed_description`\\n\\nThis is the main body of a full-reference rewrite. It describes visuals, actions, sound, and dialogue shot by shot in target-video playback order and inserts reference labels where they apply.\\n<!-- made by wuwukasi -->\\n\\n### 5.1 Basic Format\\n\\nThe basic format follows the Video Prompt Writing Guide (T2VA / I2VA / FL2VA / L2VA):\\n\\n- Write the body in English. Preserve the original language of dialogue, lyrics, and visible text.\\n- `[Shot 1]` marks the opening shot and has no timestamp. Later shots use `[Shot N] At MM:SS.mmm, ...` to mark cut times.\\n- Write camera movement as natural English within the current shot, including movement type, amplitude, and speed when they need to be expressed.\\n- Give vocal sources stable `(S1)`, `(S2)`, and subsequent IDs. Write dialogue and lyrics as `<d>[Language] ...</d>`.\\n- Use `<scenetrans>`, `<cutoff>`, and the corresponding continuity descriptions for dialogue crossing a cut, speech truncated by the video ending, and continuous audio across shots.\\n\\nFor complete rules and examples covering camera vocabulary, group speech, voice-over, dialogue across cuts, and visible text, see the Video Prompt Writing Guide (T2VA / I2VA / FL2VA / L2VA).\\n\\n### 5.2 Full-Reference Mode Differences\\n\\n| Dimension | T2VA | Full-reference mode |\\n| --- | --- | --- |\\n| Main field | `integrated_multimodal_description` | `detailed_description` |\\n| Style opening | Written after `[Shot 1]` | Established in one or two English sentences before `[Shot 1]` |\\n| Reference information | Does not use full-reference labels | Inserts `<Subject N>`, `<Picture N>`, `<Video N>`, and `<Audio N>` at their first appearance and where their roles apply |\\n| Audio relationships | Describes the target video's own sound | Cites `<Audio N>` in the corresponding shot or audio phase and states whether the signal is copied or referenced |\\n\\nOpening example:\\n\\n```text\\nThe target video is in a cinematic, literary music-video style with soft lighting and a slightly desaturated color palette.\\n[Shot 1] The scene opens in a crowded urban street...\\n[Shot 2] At 00:09.000, the shot cuts to an extreme close-up...\\n```\\n\\nFor generation tasks, `detailed_description` is normally 350-500 English words. Dialogue-dense content prioritizes fitting the complete spoken timeline rather than mechanically reaching a word count. Video-editing descriptions scale with the complexity of the source video and do not have to follow the generation-task range. A single shot does not automatically justify a shorter description; distribute detail across multiple shots according to their information load.\\n\\n### 5.3 Using Reference Labels in Shots\\n\\nAt the first clear appearance of an important `<Subject N>`, describe its referenced characteristics, position in the frame, and current action within what is actually visible in the shot. Continue using the same label in later shots without redefining what the label represents.\\n\\nUse natural phrasing for concrete frame anchors:\\n\\n```text\\nthe shot begins from <Picture 1>\\nthe shot's keyframe corresponds to <Picture 2>\\nthe shot ends on <Picture 3>\\n```\\n\\nWhen editing or continuing an original video, cite `<Video N>` naturally where its source state, structure, or continuation relationship applies. Cite `<Audio N>` in the shot or semantic phase where the audio relationship is active.\\n\\n### 5.4 Speakers, Audio Sources, and Dialogue\\n\\nThe basic speaker-ID and `<d>` formats follow T2VA. When a referenced subject physically speaks, retain both the visual reference label and the speaker ID:\\n\\n```text\\n<Subject 2> (S1) turns toward the woman and says, <d>[English] Last summer, I went to my grandfather's house. He talked about you.</d>\\n```\\n\\n`<Subject N>` identifies the referenced subject, while `(Sx)` identifies the actual speaker. When the subject speaks, write `<Subject N> (Sx)`. If the same subject speaks off-screen, keep the same form and mark it as `off-screen`. When the speaker does not correspond to a defined subject, use a stable voice description followed by `(Sx)`.\\n\\nWhen verbal content is only a cue within a directly reused BGM or complete soundtrack, and no person, character, narrator, or other independent vocal source physically produces it, use `<Audio N>` as the audible source and do not invent an additional `(Sx)`. If a concrete person, character, narrator, or other independent vocal source produces the voice, assign and reuse `(Sx)` for that source:\\n\\n```text\\nWhen <Audio 1> reaches the phrase <d>[English] I'm lonely lonely lonely lonely lonely I'm lonely</d>, <Subject 1> performs the corresponding hand gesture without becoming a separate speaker source.\\n```\\n\\nWhen dialogue, narration, or lyrics from reference audio are directly reused, or when the input prompt explicitly requests their reperformance, preserve the exact source words and original language inside `<d>`. Write `[unclear]` for unintelligible spans instead of guessing or paraphrasing them. Standardize punctuation to the basic written marks needed to express the sentence, such as `,`, `.`, `?`, and `!`; remove repeated tildes, emoji, bullets, and repeated or decorative punctuation. End complete statements, questions, and exclamations with `.`, `?`, or `!` respectively before `</d>`.\\n\\nWhen only timbre, rhythm, emotion, or delivery is referenced, do not carry the original dialogue from the reference audio into the target video.\\n\\nAssign `(Sx)` once according to the order of actual vocal events in the target video. Reuse the corresponding ID at every actual vocal event in `detailed_description`; an `<Audio N>` definition bound to a target speaker in `subject_definitions` also reuses the same `(Sx)` but never assigns a new one independently. Do not write `(Sx)` in `retention_analysis`. Verbal cues that exist only within a directly reused BGM or complete soundtrack use `<Audio N>`; voices physically produced by a concrete person, character, narrator, or other independent vocal source use `(Sx)`.\\n\\n## 6. `overall_soundscape` and `non_diegetic_music`\\n\\nThe definitions of these two sound categories follow the Video Prompt Writing Guide (T2VA / I2VA / FL2VA / L2VA).\\n\\n`overall_soundscape` summarizes ambience and physical sounds across the full video. Dialogue, singing, and sound events synchronized to a particular shot remain in `detailed_description`:\\n\\n```text\\noverall_soundscape: Quiet indoor room tone and a low ventilation hum continue throughout the video.\\n```\\n\\n`non_diegetic_music` describes background music that the characters cannot hear and that is audible only to the audience. When music is present, state its instrumentation, tempo, and dynamic development:\\n\\n```text\\nnon_diegetic_music: A restrained solo-piano score at a slow tempo, with sustained low cello underneath and no swell.\\n```\\n\\nWhen reference audio is used, state its copy or reference relationship only in the section that matches the audible layer: ambience and sound effects belong in `overall_soundscape`, while audience-only score belongs in `non_diegetic_music`. If the same audio provides both kinds of content, describe the corresponding relationship in each section:\\n\\n```text\\noverall_soundscape: The copied ambience layer from <Audio 1> continues throughout the target video.\\nnon_diegetic_music: <Audio 2> is directly reused as the complete audience-only score.\\n```\\n\\nWrite complete dialogue and lyrics only inside `<d>` in `detailed_description`; do not repeat them in these two sections.\\n\\n## 7. Complete Example\\n\\n<details>\\n<summary>Show the complete example</summary>\\n\\n```text\\nsubject_definitions:\\n<Subject 1> is the coffee-shop environment in <Picture 1>, featuring an exposed brick wall, an orange tufted sofa with patterned pillows, a neon sign, and a wooden coffee table.\\n<Subject 2> is the fluffy white Samoyed in <Picture 2>, <Picture 3>, and <Picture 4>, with thick white fur, pointed ears, a dark nose, and a curved tail.\\n<Subject 3> is the young blonde woman in <Video 1>, with long blonde hair and a light-pink button-down shirt with rolled-up sleeves.\\n<Subject 4> is the young man in <Video 2>, with short wavy brown hair and a dark-grey hoodie with drawstrings.\\n<Audio 1> is the voice-timbre reference for <Subject 3> (S1), containing a spoken English vocal layer.\\n\\nsummary:\\n[reference generation + audio reference] The target video shows <Subject 3> eating a cookie in <Subject 1>. <Subject 4> enters with <Subject 2>, which lunges toward the cookie. The three-shot exchange uses <Audio 1> as the voice-timbre reference for <Subject 3> and ends with a canned audience laugh.\\n\\nretention_analysis:\\n<Subject 1> (appears in [Shot 1], [Shot 2], [Shot 3]): fully_preserved - the exposed brick wall, orange tufted sofa, patterned pillows, neon sign, and wooden coffee table are retained.\\n<Subject 2> (appears in [Shot 1], [Shot 2]): fully_preserved - the Samoyed's thick white fur, pointed ears, dark nose, and curved tail are retained.\\n<Subject 3> (appears in [Shot 1], [Shot 2], [Shot 3]): fully_preserved - the blonde woman's identity, long hair, and light-pink shirt are retained.\\n<Subject 4> (appears in [Shot 1], [Shot 2]): fully_preserved - the young man's short wavy brown hair and dark-grey hoodie are retained.\\n<Audio 1>: reference - its vocal timbre guides the dialogue delivery of <Subject 3> without copying the original signal.\\n\\ndetailed_description:\\nThe target video uses a realistic multi-camera sitcom style with warm indoor lighting.\\n[Shot 1] A medium shot establishes <Subject 1>, the coffee shop with its exposed brick wall, orange tufted sofa, patterned pillows, neon sign, and wooden coffee table. <Subject 3> (S1), the young woman with long blonde hair and a light-pink button-down shirt with rolled-up sleeves, sits on the sofa holding a chocolate-chip cookie. From the left, <Subject 4>, the young man with short wavy brown hair and a dark-grey hoodie with drawstrings, enters holding the leash of <Subject 2>, the thick-furred white Samoyed with pointed ears, a dark nose, and a curved tail. The dog lunges toward the cookie and pulls the leash taut. <Subject 3> (S1) jerks her hand back and, using the clear youthful voice timbre referenced from <Audio 1>, exclaims with light annoyance, <d>[English] Hey! Watch your dog!</d> She closes her lips and guards the cookie while <Subject 4> pulls the dog back.\\n[Shot 2] At 00:03.000, the shot cuts to a close-up of <Subject 4> (S2), the young man in the dark-grey hoodie from Shot 1, sitting beside <Subject 3> on the sofa and holding <Subject 2> securely in his arms. <Subject 4> (S2) says in a casual young male voice with a playful tone and an easy conversational pace, <d>[English] He just likes cookies more than me.</d> He closes his mouth into an apologetic smile and strokes the dog's thick white fur.\\n[Shot 3] At 00:05.000, the shot cuts to a close-up of <Subject 3> (S1), the blonde woman in the light-pink shirt from Shot 1. Her annoyance softens as she looks toward the Samoyed. <Subject 3> (S1) replies in the same clear youthful voice referenced from <Audio 1> with an amused cadence, <d>[English] Well, he has good taste at least.</d> She smiles and raises the cookie in a small toast-like gesture. A classic canned audience laugh begins immediately after the line and continues through the final frame.\\n\\noverall_soundscape:\\nSoft indoor coffee-shop room tone continues throughout the scene.\\n\\nnon_diegetic_music:\\nN/A\\n```\\n\\n</details>\\n\\n## 8. Digital Human Example\\n\\nUse this example when the user provides a portrait image, a reference audio file, and a speaking script for a digital-human or talking-avatar video. In real tasks, replace the sample dialogue with the user's complete spoken script.\\n\\n```text\\nsubject_definitions:\\n<Subject 1> is the portrait person in <Picture 1>, preserving the person's facial identity, hairstyle, clothing, visible accessories, and overall on-camera appearance.\\n<Audio 1> is the voice-timbre, speaking pace, rhythm, and emotional-delivery reference for <Subject 1> (S1).\\n\\nsummary:\\n[reference generation + audio reference] The target video is a digital-human talking-head video in which <Subject 1> speaks the provided script to camera while using <Audio 1> as the voice-timbre and delivery reference.\\n\\nretention_analysis:\\n<Subject 1> (appears in [Shot 1]): fully_preserved - the portrait person's identity, face, hairstyle, clothing, and visible accessories are retained.\\n<Audio 1>: reference - the target speaker follows <Audio 1>'s voice timbre, speaking pace, rhythm, and emotional delivery without copying the original audio signal.\\n\\ndetailed_description:\\nThe target video uses a realistic digital-human talking-head style with stable facial identity, natural lighting, and clean portrait framing.\\n[Shot 1] A medium close-up frames <Subject 1>, the portrait person from <Picture 1>, facing the camera with a relaxed upright posture. <Subject 1> (S1) begins speaking the user's provided script in a voice guided by <Audio 1>, matching its timbre, measured pace, rhythm, and emotional delivery while using natural synchronized mouth movement. The face remains stable and recognizable throughout the shot, with subtle blinking, small eyebrow movement, and restrained head motion that supports the speech without changing identity. <Subject 1> (S1) says, <d>[Chinese] 大家好，欢迎收看今天的内容。</d> The mouth closes naturally after the final word, and the expression settles into a calm, attentive look.\\n\\noverall_soundscape:\\nQuiet indoor room tone remains clean and unobtrusive, with no distracting physical sounds.\\n\\nnon_diegetic_music:\\nN/A\\n```\\n\"}}]},{\"id\":\"mqqqqfz1-1be4fw4ee\",\"name\":\"Video 0\",\"type\":\"video\",\"color\":\"var(--primary)\",\"muted\":false,\"solo\":false,\"volume_db\":0,\"locked\":false,\"segments\":[]}],\"total_length\":720,\"frame_rate\":24,\"task_markers\":[],\"task_overview\":false}"
      }
    },
    {
      "id": 64,
      "type": "VAEDecodeAudio",
      "pos": [
        2844.41943359375,
        1025.8800048828125
      ],
      "size": [
        158.71542358398438,
        46
      ],
      "flags": {
        "collapsed": false
      },
      "order": 38,
      "mode": 0,
      "inputs": [
        {
          "label": "Latent",
          "name": "samples",
          "type": "LATENT",
          "link": 141
        },
        {
          "label": "VAE",
          "name": "vae",
          "type": "VAE",
          "link": 181
        }
      ],
      "outputs": [
        {
          "label": "音频",
          "name": "AUDIO",
          "type": "AUDIO",
          "links": [
            128
          ]
        }
      ],
      "properties": {
        "cnr_id": "comfy-core",
        "ver": "0.29.0",
        "Node name for S&R": "VAEDecodeAudio",
        "widget_ue_connectable": {}
      }
    },
    {
      "id": 69,
      "type": "SamplerCustomAdvanced",
      "pos": [
        2613.130126953125,
        926.1503295898438
      ],
      "size": [
        205.295654296875,
        161.78875732421875
      ],
      "flags": {},
      "order": 36,
      "mode": 0,
      "inputs": [
        {
          "label": "噪波生成",
          "name": "noise",
          "type": "NOISE",
          "link": 144
        },
        {
          "label": "引导",
          "name": "guider",
          "type": "GUIDER",
          "link": 145
        },
        {
          "label": "采样器",
          "name": "sampler",
          "type": "SAMPLER",
          "link": 146
        },
        {
          "label": "Sigmas",
          "name": "sigmas",
          "type": "SIGMAS",
          "link": 217
        },
        {
          "label": "Latent",
          "name": "latent_image",
          "type": "LATENT",
          "link": 166
        }
      ],
      "outputs": [
        {
          "label": "输出",
          "name": "output",
          "type": "LATENT"
        },
        {
          "label": "降噪输出",
          "name": "denoised_output",
          "type": "LATENT",
          "links": [
            139,
            141
          ]
        }
      ],
      "properties": {
        "cnr_id": "comfy-core",
        "ver": "0.30.0",
        "Node name for S&R": "SamplerCustomAdvanced",
        "widget_ue_connectable": {}
      }
    },
    {
      "id": 92,
      "type": "ManualSigmas",
      "pos": [
        2620.549560546875,
        751.7620239257812
      ],
      "size": [
        356.99365234375,
        58
      ],
      "flags": {},
      "order": 12,
      "mode": 0,
      "inputs": [
        {
          "name": "sigmas",
          "type": "STRING",
          "widget": {
            "name": "sigmas"
          }
        }
      ],
      "outputs": [
        {
          "name": "SIGMAS",
          "type": "SIGMAS",
          "links": [
            217
          ]
        }
      ],
      "properties": {
        "cnr_id": "comfy-core",
        "ver": "0.33.0",
        "Node name for S&R": "ManualSigmas",
        "widget_ue_connectable": {}
      },
      "widgets_values": [
        "1, .9882, .973, .9524, .9231, .878, .8, .6316, .4737, .1579, 0"
      ],
      "widgets_values_named": {
        "sigmas": "1, .9882, .973, .9524, .9231, .878, .8, .6316, .4737, .1579, 0"
      }
    },
    {
      "id": 75,
      "type": "easy minimaxH3ToVideo",
      "pos": [
        2316.7509765625,
        749.1515502929688
      ],
      "size": [
        225,
        366
      ],
      "flags": {},
      "order": 33,
      "mode": 0,
      "inputs": [
        {
          "label": "clip",
          "name": "clip",
          "type": "CLIP",
          "link": 155
        },
        {
          "label": "vae",
          "name": "vae",
          "type": "VAE",
          "link": 156
        },
        {
          "label": "audio_vae",
          "name": "audio_vae",
          "shape": 7,
          "type": "VAE",
          "link": 157
        },
        {
          "label": "images",
          "name": "images",
          "shape": 7,
          "type": "IMAGE",
          "link": 158
        },
        {
          "label": "audios",
          "name": "audios",
          "shape": 7,
          "type": "AUDIO",
          "link": 159
        },
        {
          "label": "videos",
          "name": "videos",
          "shape": 7,
          "type": "VIDEO"
        },
        {
          "label": "prompt",
          "name": "prompt",
          "type": "STRING",
          "widget": {
            "name": "prompt"
          },
          "link": 240
        },
        {
          "label": "mode",
          "name": "mode",
          "type": "COMBO",
          "widget": {
            "name": "mode"
          }
        },
        {
          "label": "width",
          "name": "width",
          "type": "INT",
          "widget": {
            "name": "width"
          },
          "link": 230
        },
        {
          "label": "height",
          "name": "height",
          "type": "INT",
          "widget": {
            "name": "height"
          },
          "link": 231
        },
        {
          "label": "length",
          "name": "length",
          "type": "INT",
          "widget": {
            "name": "length"
          },
          "link": 229
        },
        {
          "label": "ref_image_size",
          "name": "ref_image_size",
          "type": "COMBO",
          "widget": {
            "name": "ref_image_size"
          }
        }
      ],
      "outputs": [
        {
          "label": "positive",
          "name": "positive",
          "type": "CONDITIONING",
          "links": [
            164,
            165
          ]
        },
        {
          "label": "latent",
          "name": "latent",
          "type": "LATENT",
          "links": [
            166
          ]
        }
      ],
      "properties": {
        "cnr_id": "comfyui-easy-media",
        "ver": "9d0282d4e19ef2718c084edf55864dfb4533032f",
        "Node name for S&R": "easy minimaxH3ToVideo",
        "widget_ue_connectable": {
          "width": true,
          "length": true,
          "prompt": true,
          "height": true
        }
      },
      "widgets_values": [
        "",
        "reference",
        960,
        544,
        22,
        "match"
      ],
      "widgets_values_named": {
        "prompt": "",
        "mode": "reference",
        "width": 960,
        "height": 544,
        "length": 22,
        "ref_image_size": "match"
      }
    },
    {
      "id": 74,
      "type": "SimpleMathDual+",
      "pos": [
        2295.17578125,
        521.7177734375
      ],
      "size": [
        225.89697265625,
        142
      ],
      "flags": {},
      "order": 18,
      "mode": 0,
      "inputs": [
        {
          "name": "a",
          "shape": 7,
          "type": "*",
          "link": 169
        },
        {
          "name": "b",
          "shape": 7,
          "type": "*",
          "link": 170
        },
        {
          "name": "c",
          "shape": 7,
          "type": "*"
        },
        {
          "name": "d",
          "shape": 7,
          "type": "*"
        },
        {
          "label": "值1",
          "name": "value_1",
          "type": "STRING",
          "widget": {
            "name": "value_1"
          }
        },
        {
          "label": "值2",
          "name": "value_2",
          "type": "STRING",
          "widget": {
            "name": "value_2"
          }
        }
      ],
      "outputs": [
        {
          "label": "整数_1",
          "name": "int_1",
          "type": "INT",
          "links": [
            230
          ]
        },
        {
          "label": "浮点_1",
          "name": "float_1",
          "type": "FLOAT"
        },
        {
          "label": "整数_2",
          "name": "int_2",
          "type": "INT",
          "links": [
            231
          ]
        },
        {
          "label": "浮点_2",
          "name": "float_2",
          "type": "FLOAT"
        }
      ],
      "title": "简易运算（双）",
      "properties": {
        "cnr_id": "comfyui_essentials",
        "ver": "9d9f4bedfc9f0321c19faf71855e228c93bd0dc9",
        "Node name for S&R": "SimpleMathDual+",
        "widget_ue_connectable": {
          "value_2": true,
          "value_1": true
        }
      },
      "widgets_values": [
        "((a + 47) // 48) * 32",
        "((b + 47) // 48) * 32"
      ],
      "widgets_values_named": {
        "value_1": "((a + 47) // 48) * 32",
        "value_2": "((b + 47) // 48) * 32"
      }
    },
    {
      "id": 62,
      "type": "ImageResizeKJv2",
      "pos": [
        3065.787841796875,
        483.1539611816406
      ],
      "size": [
        210,
        335.9999694824219
      ],
      "flags": {
        "collapsed": false
      },
      "order": 39,
      "mode": 0,
      "inputs": [
        {
          "label": "图像",
          "name": "image",
          "type": "IMAGE",
          "link": 138
        },
        {
          "label": "遮罩",
          "name": "mask",
          "shape": 7,
          "type": "MASK"
        },
        {
          "label": "宽度",
          "name": "width",
          "type": "INT",
          "widget": {
            "name": "width"
          },
          "link": 235
        },
        {
          "label": "高度",
          "name": "height",
          "type": "INT",
          "widget": {
            "name": "height"
          },
          "link": 234
        },
        {
          "label": "缩放方法",
          "name": "upscale_method",
          "type": "COMBO",
          "widget": {
            "name": "upscale_method"
          }
        },
        {
          "label": "固定宽高比",
          "name": "keep_proportion",
          "type": "COMBO",
          "widget": {
            "name": "keep_proportion"
          }
        },
        {
          "label": "填充颜色",
          "name": "pad_color",
          "type": "STRING",
          "widget": {
            "name": "pad_color"
          }
        },
        {
          "label": "裁剪位置",
          "name": "crop_position",
          "type": "COMBO",
          "widget": {
            "name": "crop_position"
          }
        },
        {
          "label": "因数",
          "name": "divisible_by",
          "type": "INT",
          "widget": {
            "name": "divisible_by"
          }
        },
        {
          "label": "设备",
          "name": "device",
          "shape": 7,
          "type": "COMBO",
          "widget": {
            "name": "device"
          }
        }
      ],
      "outputs": [
        {
          "label": "图像",
          "name": "IMAGE",
          "type": "IMAGE",
          "links": [
            132
          ]
        },
        {
          "label": "宽度",
          "name": "width",
          "type": "INT"
        },
        {
          "label": "高度",
          "name": "height",
          "type": "INT"
        },
        {
          "label": "遮罩",
          "name": "mask",
          "type": "MASK"
        }
      ],
      "title": "图像缩放KJ v2",
      "properties": {
        "cnr_id": "comfyui-kjnodes",
        "ver": "eb356e6cd5dedac54bb84bc54b8da8f185fa8472",
        "Node name for S&R": "ImageResizeKJv2",
        "widget_ue_connectable": {
          "crop_position": true,
          "pad_color": true,
          "divisible_by": true,
          "width": true,
          "keep_proportion": true,
          "upscale_method": true,
          "device": true,
          "height": true
        }
      },
      "widgets_values": [
        1920,
        1920,
        "nvidia_rtx_vsr",
        "stretch",
        "0, 0, 0",
        "center",
        32,
        "cpu"
      ],
      "widgets_values_named": {
        "width": 1920,
        "height": 1920,
        "upscale_method": "nvidia_rtx_vsr",
        "keep_proportion": "stretch",
        "pad_color": "0, 0, 0",
        "crop_position": "center",
        "divisible_by": 32,
        "device": "cpu"
      }
    },
    {
      "id": 13,
      "type": "easy multiTrackInfoOutput",
      "pos": [
        1987.2760009765625,
        1246.89453125
      ],
      "size": [
        199.828125,
        106
      ],
      "flags": {
        "collapsed": true
      },
      "order": 16,
      "mode": 0,
      "inputs": [
        {
          "label": "tracks_info",
          "name": "tracks_info",
          "type": "TRACKS_INFO",
          "link": 16
        }
      ],
      "outputs": [
        {
          "label": "WIDTH",
          "name": "WIDTH",
          "type": "INT",
          "links": [
            169,
            235
          ]
        },
        {
          "label": "HEIGHT",
          "name": "HEIGHT",
          "type": "INT",
          "links": [
            170,
            234
          ]
        },
        {
          "label": "TOTAL_FRAMES",
          "name": "TOTAL_FRAMES",
          "type": "INT"
        },
        {
          "label": "FPS",
          "name": "FPS",
          "type": "FLOAT",
          "links": [
            218,
            237
          ]
        },
        {
          "label": "TASK_COUNT",
          "name": "TASK_COUNT",
          "type": "INT",
          "links": [
            72
          ]
        }
      ],
      "properties": {
        "cnr_id": "comfyui-easy-media",
        "ver": "829a7a6030b151cb6ac091595570eb3a3d8c4512",
        "Node name for S&R": "easy multiTrackInfoOutput",
        "widget_ue_connectable": {}
      }
    },
    {
      "id": 16,
      "type": "easy multiTrackTaskOutput",
      "pos": [
        1989.5999755859375,
        1390.0091552734375
      ],
      "size": [
        213.087890625,
        222
      ],
      "flags": {
        "collapsed": false
      },
      "order": 28,
      "mode": 0,
      "inputs": [
        {
          "label": "tracks_info",
          "name": "tracks_info",
          "type": "TRACKS_INFO",
          "link": 26
        },
        {
          "label": "images",
          "name": "images",
          "shape": 7,
          "type": "IMAGE",
          "link": 27
        },
        {
          "label": "audio",
          "name": "audio",
          "shape": 7,
          "type": "AUDIO",
          "link": 28
        },
        {
          "label": "video",
          "name": "video",
          "shape": 7,
          "type": "VIDEO",
          "link": 29
        },
        {
          "label": "task_index",
          "name": "task_index",
          "type": "INT",
          "widget": {
            "name": "task_index"
          },
          "link": 76
        },
        {
          "label": "prompt_format",
          "name": "prompt_format",
          "type": "COMBO",
          "widget": {
            "name": "prompt_format"
          }
        }
      ],
      "outputs": [
        {
          "label": "SYSTEM_PROMPT",
          "name": "SYSTEM_PROMPT",
          "type": "STRING",
          "links": [
            213
          ]
        },
        {
          "label": "USER_PROMPT",
          "name": "USER_PROMPT",
          "type": "STRING",
          "links": [
            212,
            236
          ]
        },
        {
          "label": "TYPE",
          "name": "TYPE",
          "type": "STRING",
          "links": []
        },
        {
          "label": "LENGTH",
          "name": "LENGTH",
          "type": "INT",
          "links": [
            229
          ]
        },
        {
          "label": "IMAGES",
          "name": "IMAGES",
          "shape": 6,
          "type": "IMAGE",
          "links": [
            158,
            214
          ]
        },
        {
          "label": "AUDIO",
          "name": "AUDIO",
          "shape": 6,
          "type": "AUDIO",
          "links": [
            159
          ]
        },
        {
          "label": "VIDEO",
          "name": "VIDEO",
          "shape": 6,
          "type": "VIDEO",
          "links": []
        },
        {
          "label": "IMAGE_INDEXES",
          "name": "IMAGE_INDEXES",
          "type": "STRING"
        }
      ],
      "properties": {
        "cnr_id": "comfyui-easy-media",
        "ver": "829a7a6030b151cb6ac091595570eb3a3d8c4512",
        "Node name for S&R": "easy multiTrackTaskOutput",
        "widget_ue_connectable": {}
      },
      "widgets_values": [
        0,
        "api"
      ],
      "widgets_values_named": {
        "task_index": 0,
        "prompt_format": "api"
      }
    },
    {
      "id": 33,
      "type": "easy saveVideo",
      "pos": [
        3178.560302734375,
        1311.2591552734375
      ],
      "size": [
        504.8592224121094,
        174
      ],
      "flags": {},
      "order": 46,
      "mode": 0,
      "inputs": [
        {
          "label": "input_mode",
          "name": "input_mode",
          "type": "COMFY_DYNAMICCOMBO_V3",
          "widget": {
            "name": "input_mode"
          }
        },
        {
          "name": "input_mode.fps",
          "type": "FLOAT",
          "widget": {
            "name": "input_mode.fps"
          },
          "link": 237
        },
        {
          "label": "images",
          "name": "input_mode.images",
          "shape": 7,
          "type": "IMAGE",
          "link": 238
        },
        {
          "label": "audio",
          "name": "input_mode.audio",
          "shape": 7,
          "type": "AUDIO",
          "link": 239
        },
        {
          "label": "output_mode",
          "name": "output_mode",
          "type": "COMFY_DYNAMICCOMBO_V3",
          "widget": {
            "name": "output_mode"
          }
        },
        {
          "label": "filename_prefix",
          "name": "filename_prefix",
          "type": "STRING",
          "widget": {
            "name": "filename_prefix"
          }
        }
      ],
      "outputs": [
        {
          "label": "VIDEO",
          "name": "VIDEO",
          "type": "VIDEO"
        },
        {
          "label": "file_path",
          "name": "file_path",
          "type": "STRING",
          "links": [
            57
          ]
        }
      ],
      "properties": {
        "cnr_id": "comfyui-easy-media",
        "ver": "1254e0e0afa9afee268027b9540fdc91c888ebc5",
        "Node name for S&R": "easy saveVideo",
        "widget_ue_connectable": {}
      },
      "widgets_values": [
        "images+audio",
        24,
        "save",
        false,
        "ComfyUI"
      ],
      "widgets_values_named": {
        "input_mode": "images+audio",
        "input_mode.fps": 24,
        "output_mode": "save",
        "output_mode.save_metadata": false,
        "filename_prefix": "ComfyUI"
      }
    },
    {
      "id": 87,
      "type": "easy cleanGpuUsed",
      "pos": [
        2914.7080078125,
        2070.1572265625
      ],
      "size": [
        157.38925170898438,
        26
      ],
      "flags": {},
      "order": 32,
      "mode": 0,
      "inputs": [
        {
          "label": "输入任何",
          "name": "anything",
          "type": "*",
          "link": 211
        }
      ],
      "outputs": [
        {
          "label": "输出",
          "name": "output",
          "type": "*",
          "links": [
            240
          ]
        }
      ],
      "properties": {
        "cnr_id": "comfyui-easy-use",
        "ver": "4de1ab3b66e48da916b6f263bacd001df53a2720",
        "Node name for S&R": "easy cleanGpuUsed",
        "widget_ue_connectable": {}
      }
    },
    {
      "id": 84,
      "type": "easy ifElse",
      "pos": [
        2572.424560546875,
        2067.30810546875
      ],
      "size": [
        238.90760803222656,
        78
      ],
      "flags": {},
      "order": 30,
      "mode": 0,
      "inputs": [
        {
          "label": "真时",
          "name": "on_true",
          "type": "*",
          "link": 208
        },
        {
          "label": "假时",
          "name": "on_false",
          "type": "*",
          "link": 236
        },
        {
          "label": "布尔",
          "name": "boolean",
          "type": "BOOLEAN",
          "widget": {
            "name": "boolean"
          },
          "link": 216
        }
      ],
      "outputs": [
        {
          "name": "*",
          "type": "*",
          "links": [
            209
          ]
        }
      ],
      "properties": {
        "cnr_id": "comfyui-easy-use",
        "ver": "4de1ab3b66e48da916b6f263bacd001df53a2720",
        "Node name for S&R": "easy ifElse",
        "widget_ue_connectable": {}
      },
      "widgets_values": [
        false
      ],
      "widgets_values_named": {
        "boolean": false
      }
    },
    {
      "id": 96,
      "type": "MarkdownNote",
      "pos": [
        639.9979897561503,
        1898.688042844867
      ],
      "size": [
        318.8470711745663,
        389.2877857234985
      ],
      "flags": {},
      "order": 13,
      "mode": 0,
      "inputs": [],
      "outputs": [],
      "properties": {
        "widget_ue_connectable": {}
      },
      "widgets_values": [
        "# 【User Manual】\n## [Video tutorial](https://www.bilibili.com/video/BV1WQ3J6bE8X)\n\n## 1. Click on the fragment in the task track to edit detailed prompt words/duration\n\n## 2. Task mode selection options in the editor: By default, it represents the video generation mode as text generated, first frame, or first and last frames (depending on the number of resources in the editor). When referenced, it is reference generated video (supporting up to 9 images, 3 audios, and 3 videos), and when it is the last frame, it is tail generated video."
      ],
      "widgets_values_named": {
        "text": "# 【User Manual】\n## [Video tutorial](https://www.bilibili.com/video/BV1WQ3J6bE8X)\n\n## 1. Click on the fragment in the task track to edit detailed prompt words/duration\n\n## 2. Task mode selection options in the editor: By default, it represents the video generation mode as text generated, first frame, or first and last frames (depending on the number of resources in the editor). When referenced, it is reference generated video (supporting up to 9 images, 3 audios, and 3 videos), and when it is the last frame, it is tail generated video."
      },
      "color": "#c09430",
      "bgcolor": "rgba(24,24,27,.9)"
    },
    {
      "id": 95,
      "type": "MarkdownNote",
      "pos": [
        581.3094279031463,
        1193.2114249757788
      ],
      "size": [
        388.77669446931304,
        240.93437883225397
      ],
      "flags": {},
      "order": 14,
      "mode": 0,
      "inputs": [],
      "outputs": [],
      "properties": {
        "widget_ue_connectable": {}
      },
      "widgets_values": [
        "# 【Custom nodes in this workflow】\n\n## [ComfyUI-Easy-Media](https://github.com/yolain/ComfyUI-Easy-Media) \n\n## [ComfyUI-Easy-Use](https://github.com/yolain/ComfyUI-Easy-Use)\n\n## [ComfyUI-llama-cpp_vlm](https://github.com/lihaoyun6/ComfyUI-llama-cpp_vlm)\n\n## [ComfyUI_MinimaxH3HybridLoader](https://github.com/scottmudge/ComfyUI_MinimaxH3HybridLoader)\n\n## [ComfyUI-Kjnodes](https://github.com/kijai/ComfyUI-Kjnodes)"
      ],
      "widgets_values_named": {
        "text": "# 【Custom nodes in this workflow】\n\n## [ComfyUI-Easy-Media](https://github.com/yolain/ComfyUI-Easy-Media) \n\n## [ComfyUI-Easy-Use](https://github.com/yolain/ComfyUI-Easy-Use)\n\n## [ComfyUI-llama-cpp_vlm](https://github.com/lihaoyun6/ComfyUI-llama-cpp_vlm)\n\n## [ComfyUI_MinimaxH3HybridLoader](https://github.com/scottmudge/ComfyUI_MinimaxH3HybridLoader)\n\n## [ComfyUI-Kjnodes](https://github.com/kijai/ComfyUI-Kjnodes)"
      },
      "color": "#c09430",
      "bgcolor": "rgba(24,24,27,.9)"
    }
  ],
  "links": [
    [
      8,
      6,
      0,
      7,
      0,
      "*"
    ],
    [
      10,
      14,
      0,
      7,
      2,
      "BOOLEAN"
    ],
    [
      16,
      28,
      0,
      13,
      0,
      "TRACKS_INFO"
    ],
    [
      26,
      28,
      0,
      16,
      0,
      "TRACKS_INFO"
    ],
    [
      27,
      28,
      1,
      16,
      1,
      "IMAGE"
    ],
    [
      28,
      28,
      2,
      16,
      2,
      "AUDIO"
    ],
    [
      29,
      28,
      3,
      16,
      3,
      "VIDEO"
    ],
    [
      46,
      7,
      0,
      26,
      0,
      "MODEL"
    ],
    [
      53,
      32,
      0,
      31,
      0,
      "FLOW_CONTROL"
    ],
    [
      57,
      33,
      1,
      34,
      1,
      "*"
    ],
    [
      58,
      32,
      2,
      34,
      0,
      "*"
    ],
    [
      59,
      34,
      0,
      31,
      1,
      "*"
    ],
    [
      60,
      31,
      0,
      35,
      0,
      "STRING"
    ],
    [
      65,
      21,
      0,
      37,
      0,
      "FLOAT,INT,BOOLEAN"
    ],
    [
      66,
      32,
      1,
      37,
      1,
      "FLOAT,INT,BOOLEAN"
    ],
    [
      71,
      39,
      0,
      38,
      0,
      "FLOAT,INT,BOOLEAN"
    ],
    [
      72,
      13,
      4,
      38,
      1,
      "FLOAT,INT,BOOLEAN"
    ],
    [
      73,
      38,
      1,
      32,
      1,
      "INT"
    ],
    [
      74,
      39,
      0,
      41,
      0,
      "FLOAT,INT,BOOLEAN"
    ],
    [
      75,
      32,
      1,
      41,
      1,
      "FLOAT,INT,BOOLEAN"
    ],
    [
      76,
      41,
      1,
      16,
      4,
      "INT"
    ],
    [
      123,
      70,
      0,
      55,
      0,
      "NOISE"
    ],
    [
      124,
      65,
      0,
      55,
      1,
      "GUIDER"
    ],
    [
      125,
      68,
      0,
      55,
      2,
      "SAMPLER"
    ],
    [
      126,
      67,
      0,
      55,
      3,
      "SIGMAS"
    ],
    [
      128,
      64,
      0,
      56,
      0,
      "AUDIO"
    ],
    [
      129,
      59,
      0,
      57,
      0,
      "LATENT"
    ],
    [
      130,
      56,
      0,
      57,
      1,
      "LATENT"
    ],
    [
      131,
      55,
      1,
      58,
      0,
      "LATENT"
    ],
    [
      132,
      62,
      0,
      59,
      0,
      "IMAGE"
    ],
    [
      136,
      55,
      1,
      61,
      0,
      "LATENT"
    ],
    [
      138,
      63,
      0,
      62,
      0,
      "IMAGE"
    ],
    [
      139,
      69,
      1,
      63,
      0,
      "LATENT"
    ],
    [
      141,
      69,
      1,
      64,
      0,
      "LATENT"
    ],
    [
      144,
      70,
      0,
      69,
      0,
      "NOISE"
    ],
    [
      145,
      66,
      0,
      69,
      1,
      "GUIDER"
    ],
    [
      146,
      72,
      0,
      69,
      2,
      "SAMPLER"
    ],
    [
      153,
      8,
      0,
      59,
      1,
      "VAE"
    ],
    [
      154,
      2,
      0,
      56,
      1,
      "VAE"
    ],
    [
      155,
      1,
      0,
      75,
      0,
      "CLIP"
    ],
    [
      156,
      8,
      0,
      75,
      1,
      "VAE"
    ],
    [
      157,
      2,
      0,
      75,
      2,
      "VAE"
    ],
    [
      158,
      16,
      4,
      75,
      3,
      "IMAGE"
    ],
    [
      159,
      16,
      5,
      75,
      4,
      "AUDIO"
    ],
    [
      164,
      75,
      0,
      65,
      1,
      "CONDITIONING"
    ],
    [
      165,
      75,
      0,
      66,
      1,
      "CONDITIONING"
    ],
    [
      166,
      75,
      1,
      69,
      4,
      "LATENT"
    ],
    [
      169,
      13,
      0,
      74,
      0,
      "INT"
    ],
    [
      170,
      13,
      1,
      74,
      1,
      "INT"
    ],
    [
      171,
      37,
      1,
      70,
      0,
      "INT"
    ],
    [
      176,
      7,
      0,
      67,
      0,
      "MODEL"
    ],
    [
      177,
      26,
      0,
      66,
      0,
      "MODEL"
    ],
    [
      179,
      7,
      0,
      65,
      0,
      "MODEL"
    ],
    [
      180,
      2,
      0,
      58,
      1,
      "VAE"
    ],
    [
      181,
      2,
      0,
      64,
      1,
      "VAE"
    ],
    [
      182,
      8,
      0,
      63,
      1,
      "VAE"
    ],
    [
      183,
      8,
      0,
      61,
      1,
      "VAE"
    ],
    [
      190,
      57,
      0,
      55,
      4,
      "LATENT"
    ],
    [
      200,
      83,
      0,
      9,
      0,
      "MODEL"
    ],
    [
      208,
      86,
      0,
      84,
      0,
      "*"
    ],
    [
      209,
      84,
      0,
      85,
      0,
      "*"
    ],
    [
      210,
      88,
      0,
      86,
      0,
      "LLAMACPPMODEL"
    ],
    [
      211,
      85,
      0,
      87,
      0,
      "*"
    ],
    [
      212,
      16,
      1,
      86,
      5,
      "STRING"
    ],
    [
      213,
      16,
      0,
      86,
      6,
      "STRING"
    ],
    [
      214,
      16,
      4,
      86,
      2,
      "IMAGE"
    ],
    [
      216,
      20,
      0,
      84,
      2,
      "BOOLEAN"
    ],
    [
      217,
      92,
      0,
      69,
      3,
      "SIGMAS"
    ],
    [
      222,
      93,
      0,
      6,
      0,
      "MODEL"
    ],
    [
      223,
      93,
      0,
      7,
      1,
      "*"
    ],
    [
      224,
      9,
      0,
      93,
      0,
      "MODEL"
    ],
    [
      226,
      35,
      0,
      94,
      0,
      "VIDEO"
    ],
    [
      229,
      16,
      3,
      75,
      10,
      "INT"
    ],
    [
      230,
      74,
      0,
      75,
      8,
      "INT"
    ],
    [
      231,
      74,
      2,
      75,
      9,
      "INT"
    ],
    [
      234,
      13,
      1,
      62,
      3,
      "INT"
    ],
    [
      235,
      13,
      0,
      62,
      2,
      "INT"
    ],
    [
      236,
      16,
      1,
      84,
      1,
      "*"
    ],
    [
      237,
      13,
      3,
      33,
      1,
      "FLOAT"
    ],
    [
      238,
      61,
      0,
      33,
      2,
      "IMAGE"
    ],
    [
      239,
      58,
      0,
      33,
      3,
      "AUDIO"
    ],
    [
      240,
      87,
      0,
      75,
      6,
      "STRING"
    ]
  ],
  "groups": [
    {
      "id": 2,
      "title": "CLIP",
      "bounding": [
        1152.177001953125,
        881.6685791015625,
        360.23162841796875,
        230.1269073486328
      ],
      "color": "#3f789e",
      "flags": {}
    },
    {
      "id": 3,
      "title": "多轨编辑导演台 & 参数选择项",
      "bounding": [
        996.0595703125,
        1152.9791259765625,
        1222.9251708984375,
        1039.7568359375
      ],
      "color": "#b58b2a",
      "flags": {}
    },
    {
      "id": 5,
      "title": "模型",
      "bounding": [
        1150.3765869140625,
        456.66302490234375,
        713.192626953125,
        345.7755126953125
      ],
      "color": "#3f789e",
      "flags": {}
    },
    {
      "id": 6,
      "title": "每轮循环(视频生成预览)",
      "bounding": [
        3116.789794921875,
        1165.979736328125,
        727.205810546875,
        800.0484008789062
      ],
      "color": "#8A8",
      "flags": {}
    },
    {
      "id": 7,
      "title": "VAE",
      "bounding": [
        1531.593505859375,
        831.9638671875,
        328.5450439453125,
        279.6822509765625
      ],
      "color": "#3f789e",
      "flags": {}
    },
    {
      "id": 8,
      "title": "加速",
      "bounding": [
        1884.150146484375,
        813.1287841796875,
        350.9086608886719,
        303.6366271972656
      ],
      "color": "#3f789e",
      "flags": {}
    },
    {
      "id": 10,
      "title": "低显存优化",
      "bounding": [
        1887.360107421875,
        472.55126953125,
        352.1004333496094,
        317.1150817871094
      ],
      "color": "#3f789e",
      "flags": {}
    },
    {
      "id": 11,
      "title": "最终结果预览",
      "bounding": [
        2246.656982421875,
        1153.5252685546875,
        845.0148315429688,
        815.1010131835938
      ],
      "color": "#a1309b",
      "flags": {}
    },
    {
      "id": 12,
      "title": "循环处理",
      "bounding": [
        1996.037353515625,
        27.50434112548828,
        1011.9608154296875,
        318.8658142089844
      ],
      "color": "#3f789e",
      "flags": {}
    },
    {
      "id": 13,
      "title": "二采",
      "bounding": [
        3044.758544921875,
        416.99749755859375,
        743.9028930664062,
        696.7843017578125
      ],
      "color": "#3f789e",
      "flags": {}
    },
    {
      "id": 14,
      "title": "第一次采样",
      "bounding": [
        2275.9384765625,
        379.35821533203125,
        738.6613159179688,
        750.8754272460938
      ],
      "color": "#3f789e",
      "flags": {}
    },
    {
      "id": 16,
      "title": "提示词反推 Prompt Regenerate",
      "bounding": [
        2252.11181640625,
        1994.04443359375,
        921.72265625,
        725.544189453125
      ],
      "color": "#8A8",
      "flags": {}
    }
  ],
  "config": {},
  "extra": {
    "links_added_by_ue": [],
    "ue_links": [],
    "0246.VERSION": [
      0,
      0,
      4
    ],
    "pixaromaGroups": [],
    "ds": {
      "scale": 0.48445563060481406,
      "offset": [
        -188.62008832011645,
        112.54214997115133
      ]
    },
    "frontendVersion": "1.49.6",
    "VHS_latentpreview": false,
    "VHS_latentpreviewrate": 0,
    "VHS_MetadataImage": true,
    "VHS_KeepIntermediate": true,
    "KKvideo_latentpreview": false,
    "KKvideo_latentpreviewrate": 0
  },
  "version": 0.4
}