Trained with 60 images. The samples were generated with AutismMix.
Basic prompt: D_endless, 1girl, black hair, medium hair, hair between eyes, black eyes, pale skin, medium breasts, facial mark, ankh necklace, black top, bare shoulders, black pants, denim, studded belt, black lips
"anime screencap" + "source_cartoon" or "source_anime" - Use these tags for the flat cartoon style.
Recommended settings:
Sampler: DPM++ 2M Karras / Euler a - 25-30 steps
CFG: 7
Resolution: 1024*1024, 832*1280, 1280*1024
Lora weight: 0.8
HighRes: 4x_NMKD-Siax_200k, upscale by 1.5x, 20 steps, 0.35 denoising
Clip skip: 2
Description
FAQ
Comments (3)
This model is kinda funny, even that I don't have any "nude" or similar word in my prompt, I can't make it wear clothes xD maybe 1 of 10 will have some clothes on it, mostly pants with belt but to have whole set of character clothes is really hard to get :P.
That's weird, the majority of the training images are of the same outfit. Can you share your prompt so I can test it?
Edit: I've posted some pictures using the on-site generator and it generates clothed pictures just fine. There's a list of suggested keywords in the description and all of my images have the prompts embedded so you can see what keywords I used.
@bial I got over it :), I had to give a lot of "power" for clothes words, something like 1.6-1.7
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.