stable-diffusion How do you do unconditional image generation?

How do you do unconditional image generation?

Open sgbaird opened this issue 2 years ago • 5 comments

Similar to what imagen-pytorch supports, i.e. train on images and then generate new images from noise.

Sep 21 '22 21:09 sgbaird

The same demand. Do you have any idea? @sgbaird

Nov 11 '22 09:11 ziyeZzz

I saw this https://github.com/CompVis/stable-diffusion/commit/171cf29fb54afe048b03ec73da8abb9d102d0614. Might just be undocumented?

Nov 11 '22 22:11 sgbaird

Hi @sgbaird, did you figure out how to train an unconditional LDM? Is the undocumented readme file helpful?

Mar 07 '23 14:03 forever208

for unconditional training and sampling, it is better to look at their LDM repo

Mar 07 '23 16:03 forever208

@forever208 I haven't figured it out. Thanks for the link!

Mar 08 '23 05:03 sgbaird