Open superbagus1712 opened 3 hours ago
Thank you very much for your interest. Our model is trained on video clips of approximately 16 frames, so directly increasing the video length during testing leads to distorted content. I think we can use the last frame of the previous clip as the first frame for the next inference clip, generating long videos autoregressively.
I am very happy to use it. I want to ask how to change the video duration to 10 seconds or more?