Hi,
this is my first public release. I rly dont know which checkpoints are merged because i have on my pc at least 30 personal merges. So i dont know if it contains any license etc.
All images are without post proccesing (no hires, outpaint, face rest.. etc)
All images are 768x768 with lowvram and xformers on launch.
All images are generates in only one batch with 3 images and i picked the best one. Almost every image is generated on prompts from this site from users so you can easily compare.
Description
FAQ
Comments (8)
Hi, I was unable to reproduce your results using the same prompts and same config (xformers and lowvram). Am I missing something? Thank you.
i rly dont know dude, can you upload your results? i am using automatic 1111
hello, i have same "issue", use identical settings. Good results but diferents images. Automatic1111, xformers, lowvram.
Edit: anyway, the model is really awesome. Congratulations and thank you.
@amokiller178 i dont think that you will always have same result, because its AI so some "things" are random, in prompts you didnt specific everything what (see) is in picture.
@amokiller178 i add some new screen i review.. try uniPC with cfg scale 5
@vendo19324 same, good results but diferent.
I think it's something in my installation. I tried it with Google collab and they have differences, but very small. On my pc they are very different haha (sorry im not english)
python: 3.10.9, torch: 1.13.1+cu117, xformers: 0.0.16rc425, gradio: 3.16.2, checkpoint: b0a42c3f22. Windows 10 and RTX 2060 super.
@vendo19324 Same thing, very good images but different.
I think it's something in my installation. I tried it with Google collab and they have differences, but very small. On my pc they are very different haha (sorry im not english)
python: 3.10.9, torch: 1.13.1+cu117, xformers: 0.0.16rc425, gradio: 3.16.2, Windows 10, rtx 2060 s
I've seen that this outputs well, but I am having to trim some data for space. Just wondering if any chance to get pruned? I know there is a v2 release, but I kinda like this one better. I got kohya to start messing with training models, but really no idea about it yet. Tried to convert to fp16, but got a missing module error that I will have to figure out. Let me know, thanks!