https://www.lumiere-video.github.io/ | United States | 1998 |
Lumiere is a text-to-video diffusion model developed by a team of researchers from Google Research, a division of the multinational technology company Google (NASDAQ: GOOGL). The model is designed to address the challenges in video synthesis, particularly in achieving realistic and coherent motion. Lumiere’s key innovation is its Space-Time U-Net architecture, which allows the model to generate the entire temporal duration of a video in a single pass, rather than relying on the traditional approach of synthesizing distant keyframes and then using temporal super-resolution. This approach is aimed at improving the global temporal consistency of the generated videos.
