| { | |
| "metadata": { | |
| "task_id": "TXT_INTER_00445", | |
| "task_level": "Level1", | |
| "task_type": "TXT_INTER", | |
| "original_video": "Data/Raw/00445/video.mp4", | |
| "original_motion": "Data/Raw/00445/human_motion.pkl", | |
| "duration": 30.002 | |
| }, | |
| "input": { | |
| "prompts": { | |
| "general_instruction": "You are an advanced 3D motion generation model designed to synthesize physically plausible human movements. I will provide you with semantic text inputs detailing the starting segment and the ending segment of a human action, with a temporal gap in between. These descriptions detail the kinematics of the person's whole body, hands, and legs for those specific segments. Your objective is to predict and generate the missing intermediate movements to bridge this temporal gap. You must ensure the generated motion logically connects the starting posture to the ending posture and maintains physical balance. Your output MUST be a continuous sequence of ACTION TOKENS representing the missing middle sequence.", | |
| "modifier_instruction": "" | |
| }, | |
| "modalities": { | |
| "text": "Background: The person is sitting on a stone ledge in front of a fountain while performing a variety of expressive gestures and poses before adjusting an accessory. Ends details: The person is sitting on a stone ledge in front of a fountain, maintaining a stable posture. She starts with her hands near her chest, forms a heart shape with her fingers, and then spreads her hands apart. The middle sequence is missing, but she remains seated on the ledge and uses her right hand to adjust the watch on her left wrist.", | |
| "audios": [], | |
| "videos_processed": [], | |
| "spatial_coordinates": [], | |
| "input_images": [] | |
| } | |
| }, | |
| "ground_truth": { | |
| "motion_parameters": "Data/Processed/Text/TXT_INTER/00445_inter_gt.pkl", | |
| "rendered_video": "Data/Raw/00445/rendered.mp4" | |
| } | |
| } |