search

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
808 stars 67 forks source link

Lighthouse demo doesn't seem to match teaser results. #35

Closed nanlliu closed 5 months ago

nanlliu commented 6 months ago

my environment setup:

followed everything but used pytorch 1.13.1 since my cuda driver is 11.6.

Then directly ran inference using lighthouse config.

1_sim_0 3_sim_0

Just wanted to check if i did anything wrong. Thanks!

WUMINGCHAzero commented 5 months ago

same problem... how did you solve it? thx

zekicinar commented 3 months ago

same problem... can you explain how you solved it ??