Pixel Cube: Diffusion-based Portrait Video Relighting Through Realistic Lighting Reproduction
Researchers have developed a novel diffusion-based method called Pixel Cube for relighting portrait videos with enhanced realism and temporal consistency. This technique utilizes a hybrid dataset of real and rendered videos, coupled with an LED lighting system for accurate emulation. By integrating pre-trained video diffusion models and controlling environment maps, the system can generate temporally consistent, photorealistic relit videos that preserve subject identity and facial details. AI
IMPACT Introduces a new diffusion-based technique for realistic and temporally consistent video relighting, potentially impacting digital content creation and virtual production.