[Need-Community-Help] Training PIA for SDXL?

open-mmlab / PIA

[CVPR 2024] PIA, your Personalized Image Animator. Animate your images by text prompt, combing with Dreambooth, achieving stunning videos. PIA，你的个性化图像动画生成器，利用文本提示将图像变为奇妙的动画

https://pi-animator.github.io/

Apache License 2.0

888 stars 70 forks source link

[Need-Community-Help] Training PIA for SDXL? #10

Open foureyednymph opened 9 months ago

kowalgregy commented 9 months ago

johndpope commented 9 months ago

upstream animatediff has sdxl support I guess the pia checkpoint would need to be retrained for higher res?

zengyh1900 commented 3 months ago

Yes. If we would like to use PIA on SDXL, it requires re-training on SDXL. Unfortunately, we don't have any plans for the moment.
Volunteers are always welcome! :cupid: :cupid: