Home Artists Posts Import Register

Content

Sorry everyone, didn't had much time to answer everyone in the last two days. Tomorrow morning will sit and answer everyone.

Been doing some crazy experimentations on SD and already managed to get some stuff working.


Outpaiting (as show in the images) is working a little better now, it sometimes manage to make a outpainting that make sense, it also apply to inpainting.


Got negative prompt working. I think?

Fixed the bug if the prompt is too big will crash.

Some .ckpt files was trained using the lighting library, i think now those are working with the importer.


Been also messing with some adicional importers for other models, but still very experimental.


Would like to add prompt weight modificator and a few more schedulers before the next release.


Files

Comments

Anonymous

Yeah, that's awesome. Specify things like Negative Prompt: deformed body, unusual face, blurry image, etc...

Anonymous

I wanna use SD for somthing like coloring frames in animation.,. is that possible?

Anonymous

I’m stoked for negative prompts! Looking forward to it.

Anonymous

Schedulers are important, some of them can do double it/s than LMS, also ancestral ones give a very different results

Anonymous

Dreambooth will be nice to be integrated, it needs less than 10Gb

Anonymous

I am personally looking forward to getting the fundamental stuff working perfectly (all schedulers, support for different models, negative prompts, prompt weight modificators, etc) more than the more exotic stuff (inpainting, outpainting, animation, etc).

Anonymous

Looking forward to outpainting, thank you for your hard work

DAINAPP

I did some tests with that, its not great for now, perhaps with some creative code it could work better. Once I get some time will try to focus some energy to see if I make it work. My test was very simple, just make the white color transparent and use inpatining.

3DCinetv

Is stable Diffusion GUI 0.53 the latest one?

Jables

Is there a hard limit on the number of tags? Sometimes I will enter a prompt and if I add one more term and get: File "diffusers\pipelines\stable_diffusion\pipeline_stable_diffusion.py", line 196, in Txt2Img NameError: name 'logger' is not defined Sometimes if there is a queue it will blast through the queue without producing anything.

Anonymous

Awesome news keen for it

Anonymous

I think the number it is 77 tokens. Here you can count how many tokens has a prompt https://beta.openai.com/tokenizer