TRIGGER: kn1cstyl3
This one came out really strongly, I'm a huge fan of how it translated the shading and color. It has a bit more background variability than my usual LoRAs, so have fun with that.
MY SETTINGS FOR ILLUSTRIOUS:
Checkpoint: IllustriousXL smoothitSOLID.safetensor
VAE: fixFP16ErrorsSDXLLowerMemoryUse_v10.safetensors
Sampling Method: Euler a
Sampling Steps: 35
CFG Scale: 5
ADetailer on - FACES v1, Eyeful, Deepfashion
Enjoy what I'm doing?
Follow me and drop a few recommendations in my DMs, I do this for the love of the game, I'm not interested in monetizing this. We're lucky enough to have a free and open platform, and should all make use of it!
Description
FAQ
Comments (12)
What do you think of this one? I'm trying to notably turn up the quality with these, as my recent front page double-feature is bringing in new followers.
I just checked https://x.com/knic_toons. Man, some of this stuff from the lora seems better than what I can see on her Twitter. I can't tell if some of that is coming from any outside resources people might be using, but still. You really know your stuff.
@Diamond_Letter all of it was sourced from their public work, I do edit each of my datasets down by hand though, generally to make them more AI friendly
@Diamond_Letter specifically using IOpaint, fantastic tool and well worth supporting the dev
Strange, using WAI, Hassaku, Prefectious I don't get any of the things Lora is supposed to do. 🤨 (I got the girl in the red shirt from here)
Interesting, I might take a look at that. Try out smoothft solid and see what the results are.
@insidertrading Added the work to the gallery. I took your image and just switch the models. Doesn't really look like a sticker style.
@Twostep best thing I can suggest for you is to try out one of the suggested models I left on the page. I test my illustrious models with base, smoothft solid and SPO. Seeing as others are having better luck, and implying there aren't any weird settings in your local instance- I think it's just your choice in checkpoint
@insidertrading I'll give it a try once the generator kicks in, it's just odd that these models are as illustrious based.
@Twostep are you using the site generator? I've never used it.
@Twostep I just tried remixing the image using the on-site generator, and it looks like the anime stuff you're getting. I'll have to do a bit of research on the differences between a local stable diffusion install and the on-site generator, but I'm not sure what causes that. The style is highly consistent when used locally
@insidertrading I use SD/ Comfy, Civitai/ Tensor and similar.
When I wonder what Lora can do I try on website. If I like it, I download and try it on my PC / add it to my library to experiment later when I have free time and mood.
Hm. There used to be little difference between a website and a computer. Maybe they've changed something now. I'll have to check some of my work to see how similar it is to what I do here on the site.