This repository enhances modelscope-damo-text-to-video-synthesis by adding VTV functionality, while also optimizing memory usage. By decoding frames sequentially instead of in batches, it enables larger image resolutions and longer video durations.
This repository enhances modelscope-damo-text-to-video-synthesis by adding VTV functionality, while also optimizing memory usage. By decoding frames sequentially instead of in batches, it enables larger image resolutions and longer video durations.