LoRA is generated entirely from anime images.
Should work well with any model.
Check samples for prompts :)
For anime clothes just be specific in describing them and it should generate well.
Leave a like, comment, review but first of all, ENJOY ;)
If you wish to support me in my work,
you can do that here: Ko-Fi.com/LittleJelly
Description
lowered file size
fixed clothes tags
removed bad images from dataset
FAQ
Comments (2)
test these prompts in exact the way i post them and you will see that it breaks your lora like crazy with a weight of 1:
(((insanely high quality, insanely high resolution, insanely high details)))
her face looks totally out of place in every checkpoint with these prompts
And? I don't use these prompts in my samples.
Also here's guide how I generate my images:
Step #1: txt-2-img 768x768 or 512x768 with hiresfix, Upscaler: 4x-AnimeSharp, denoising 0.65
Step #2: Send to img2img, default resize (should be 1024x1024, or 1024x1536), same sampling as in step #1, denoising 0.1
Step #3: Send to extras, Resize x2
Upscaler, 4x-AnimeSharp
That's it.
Step #2 overwrites metadata sadly. Idk if that can be turned off.