zhangwei

@Artiprocher Thanks first for your excellent open-source work on DiffSynth-Studio — it’s a fantastic project. After analyzing the Wan-Animate training flow, I believe the “drop the last 4 frames” operation...

@gugite We faced the same issue initially. However, after addressing the problems mentioned above, our training seems to be proceeding normally.

> [@zwplus](https://github.com/zwplus) thank you for your reply. I'm a bit confused — after this change, does it still end up with a latent of 81-frame input video matching against a...