Snap Video: Scaled Spatiotemporal Transformers for Text-to-Video Synthesis

arXiv:2402.14797v1 Announce Type: new
Abstract: Contemporary models for generating images show remarkable quality and versatility. Swayed by these advantages, the research community repurposes them to generate videos. Since video content is highly redundant, we argue that naively bringing advances of image models to the video generation domain reduces motion fidelity, visual quality and impairs scalability. In this work, we build Snap Video, a video-first model that systematically addresses these challenges. To do that, we first extend the EDM framework to …

arxiv cs.ai cs.cv snap synthesis text text-to-video transformers type video