Microsoft debuts VASA neural network for lifelike talking face videos up to 512x512 resolution at 40fps.
Microsoft has launched the neural network VASA, capable of bringing photos and virtual characters to life by creating videos of talking faces with a wide range of emotions and natural movements. The technology supports online video generation with resolutions up to 512x512 at 40fps. VASA is praised for its realism, controllable generation, and ability to process diverse data. Microsoft acknowledges the risks and currently has no plans to release a product, API, or online demo.
Source: https://t.me/allhypes