If you enjoyed this model, why don't you check out my other works? I put a lot of effort into correctly recreating artist styles and making the occasional character LoRA.😊
You can also buy me a Coffee ☕ at https://ko-fi.com/stickyricky
Drop some suggestions for other artists/styles in the comments. Otherwise I'm gonna end up reworking everything I've uploaded this far again and again.
These models were trained using the collab/guide by holostrawberry [Guide] Make your own Loras, easy and free - colabs | Stable Diffusion Other | Civitai
About the artist:
Twitter: Nyantcha (@ThiccWithaQ) / Twitter
Patreon: ThiccWithaQ | Creating illustrated art | Patreon
Expect nsfw content with a tendency towards shiny skin,thick thighs and large breasts.
Weights & Prompts
v2.0:
New goto version. It is so much better at getting the detailed style mostly used for very nsfw images.
Use a weight of 0.8-1.2, adjust to taste and depending on your checkpoint.
Check the XYZ plot for reference.
Many of the examples are the exact same prompts with the same seed. Swap between this and the previous version to see the difference.
LoCon (NAI):
Use a weight of 1.0, adjust to taste. This works even with other LoRA. But in return it isn't quite as intense as the AOM version.
Version 1.5is a slightly improved variant. I'm currently testing out quite a few training options. This was the best one while testing file size and learning rate.
LoCon (AOM):
Use a weight of 1.0. Go a bit lower (I used 0.8) for character loras. As usual I have included an XYZ plot for reference.
Please leave a review if you enjoyed this model or have problems with it. I'd also be interested in how the model behaves with other base checkpoints, so why don't you give it a try and throw some images down below! 👇
Description
Pruned dataset removing images with less saturation or deviating styles
FAQ
Comments (6)
question, did you use fullanime or AOM3 as the checkpoint? I tried using this Lora with a different character Lora but the resulting character wasn't even close to acceptable
Version 2.0 was trained and tested on fullanime.
What character LoRA did you use? In my experience like 95% have an unacceptable impact on the style (part of the reasons I keep coming back to using the Jinx LoRA)
@StickyRicky i'm using Warspite (Azur Lane) - ウォースパイト, I'm trying to replicate those Nyantcha style pics I've been seeing of Pixiv by Henrik N, the art is clean and very good
@shadowrider57Â Haven't tested the model itself but looking at the examples I'm pretty sure it struggles when combined with more involved style models. Image posted by pensive_penguin (civitai.com)
Henrik N used novelAI i.e. fullanime but I doubt that they took the character LoRA (or this one) from civit. The style is probably also a combination of multiple styles (with a strong influence from nyantcha). They almost surely did a lot of additional editing to get such clean output while generating on a lot more than 512*512 pixels.
@StickyRicky hmmm, thanks for the info, I'll see what I can do from here
Good good
Details
Available On (2 platforms)
Same model published on other platforms. May have additional downloads or version variants.