💡 Login to apply to projects
Project Overview Objective: Naturally transform a person in a challenge/dance video into a desired person (or avatar) Replace the person in the challenge video with a specified person while maintaining the original dance (poses/timing/rhythm) but displaying the target person's face/style (hair, accessories, clothing tone). Input/Output Input: Challenge video (input video), 1 or more frontal images of target person/character (input image) Output: Video of target person performing the same challenge dance (output video mp4, 5+ seconds) Quality Requirements: - Movement Preservation: Maintain original dance movements (poses/timing/rhythm) as much as possible - Facial Consistency: Ensure consistent appearance throughout the video - Temporal Stability: Minimize flickering, boundary shaking, and other artifacts - Synthesis/Adjustment: Ensure natural blending of colors, brightness, and background - Standardization: Ensure robust performance across different input videos and target persons Current Development Stage: Currently developing using WAN 2.2 model in COMFYUI (details in attached materials) Reference workflow provided Not limited to Comfy: Can be produced using other commercial/alternative tools Deliverables: - Input video and input person/character images - Output video (mp4), 5+ seconds minimum - 3 different examples using different videos and persons Production method guide will be provided once videos are selected
