difference between finetuning the unet's image layers and training motion modules

guoyww / AnimateDiff

Official implementation of AnimateDiff.

https://animatediff.github.io

Apache License 2.0

10.35k stars 847 forks source link

difference between finetuning the unet's image layers and training motion modules #343

Open song-wensong opened 4 months ago

song-wensong commented 4 months ago

What is the difference between finetuning the unet's image layers and training motion modules? Suppose I want to train animatediff on a small new dataset (about 72 minutes of video clips in total), should I use finetune?