Every version can be accurately generated, close ups and full body.
NSFW is OK.
I hope you enjoy creating it.
（Sales are prohibited. Please use for personal enjoyment）
If you like, please try "Flat Anime" version as well.
v1.0
v1.0 model has been tweaked to have a style similar to Diving-Pony Anime, while still retaining the characteristics of Illustrious.
Please check the sample image for detailed prompts.
v2.0
v2.0 model is adds realistic elements to v1.0.
v3.0
v3.0 model is adds flat coloring.
v4.0
v4.0 model is adds realistic.
v5.0
v5.0 model is adds 2.5D.
It goes well with Pony-style prompts.
v6.0
v6.0 model no longer requires the inclusion of extra LoRA and prompts.
v7.0
v7.0 model is add 3D.
v8.0
v8.0 model is more 3D.
v9.0
v9.0 model is add Real.
v Fashion and Art +VAE
This model has been adjusted to make it easier to generate images of fashion and art. And no VAE required.
v10+VAE
v10 model is add muted colors.
v11+VAE
v11 model is add noob.
v12+VAE
v12 model is add Flat Anime.
v Niji + VAE
This model adds Niji elements.
v NijiMutedColor_vol2 + VAE
This model incorporates elements of Japanese illustration.
Recommended Settings①
Sampling method：Euler a or Restart or DPM++ 2M or DPM++ SDE
Schedule type：Karras
Size：896 × 1152 (You can set it to your liking here)
Step：20〜30
CFG：4〜6
ADetailer：face_yolov8s.pt
Negative prompts
(worst quality, low quality, normal quality, caucasian:1), lowres, bad anatomy, bad hands, signature, watermarks, ugly, imperfect eyes, unnatural face, unnatural body, error, extra limb, missing limbs, Child, muscular, chinese, colored skin,Description
FAQ
Comments (11)
太棒了baby
This is awesome. This is defo in my Top3.
And for whatever reason, despite identical settings, this checkpoint generates somewhat faster than all other models I've tried so far.
How to make facial features clear? The generated image has poor facial features. Close ups of the face are good, not the whole body
I'm using ADetailer (face_yolov8s.pt)
It's no use.
@DivingSuit It's no use
V12 is absolutely a masterpiece, it's work with a lot of style but i don't like the NIJI version. Niji elements have too much impact and the model becomes less versatile than its V12.
I have to write that I accidentally generated without any LORAs while trying a new stacker and the amount of things that this Checkpoint caught and rendered with no weirdness by itself is pretty amazing. I cannot praise enough.
dpm2+exponential is better, trust me.
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.