Project Icon

VADER

Optimizing Video Diffusion with Pre-trained Reward Models

Product DescriptionVADER improves video-text alignment and video creation efficiency using discriminative-based, pre-trained reward models, overcoming the need for large datasets. This method not only extends video generation to three times the sequence length but also surpasses non-gradient methods. It optimizes video models such as VideoCrafter2, Open-Sora V1.2, and ModelScope for streamlined production.
Project Details