Stable Diffusion 2.0
Sorry if it's an obvious question, but is it possible / are there any plans, to train a new model based on SD 2.0? It seems to have some impressive improvements that might benefit AI-generated anime art as well!
I cannot find any information online on SD 2.0 based models :D
SD 2.0 is unfortunately a step back since StabilityAI wants to go commercial and they stripped artists, nudity and most copyrighted references from the model.
Can't even get SD 2.0 to download after launching the webui-user bat. It skips to loading WD weights instead. Anyone know how to fix this? The only way for me to not have it break things has been to pull out the SD 2.0 model files from my model folder.
SD 2.0 is unfortunately a step back since StabilityAI wants to go commercial and they stripped artists, nudity and most copyrighted references from the model.
Yeah, I heard that too. I don't think they stripped nudity (erotic still works), so maybe only full nudity? Point is, do you think it will be hard to train/condition it for the missing aspects?
Part which might suck, is that because of lack of nudity, it might also have an even worse understanding of anatomy than it already does!
Can't even get SD 2.0 to download after launching the webui-user bat. It skips to loading WD weights instead. Anyone know how to fix this? The only way for me to not have it break things has been to pull out the SD 2.0 model files from my model folder.
Make sure to follow the instructions on webui! Worked for me! And remember to correctly place the .yaml file!
Also, for me, I had to completely reinstall webui cause the cached dependency somewhere doesn't work.