Top 3 Recommended Stable Diffusion Models: Outstanding Quality & Stability with AingDiffusion and More

This site uses affiliate links.

【Special Offer at Article’s End】Choose from 9 Colors! Free Commercial Use Illustration Card Set

Hello, this is kojirom.
Today, I’ll introduce three recommended models for image generation using Stable Diffusion.
After testing several models, I’ve identified three top-performing models in terms of efficiency and track record.
These models should provide high quality and stable performance.
The recommended models are as follows:


The popular model ‘AingDiffusion’ has been updated and powered up.
By incorporating the ‘Aissist – Negative Embedding – v2‘ embedding created by the author, it seems that the best performance can be achieved simply by setting negative prompts.

The installation method is simple.
Just put the downloaded file in the folder
and enter ‘AissistV2-neg’ in the negative prompt field.
Give it a try!
Oh, and please be careful with the spelling. It’s ‘Aissist,’ not ‘Assist.'”


I just found out that three types of Mix by the popular modeler, Mr. Butaman, have been updated at once.
Since only one type was displayed on Civitai, it seems I missed the other two.
Not only the model with the No. 1 download count, but the new two types also generate quite powerful and energetic illustrations.
Be sure to check them out!


‘SoteMix’ generates illustrations of beautiful girls who look great in pure white sweaters.
Detailed settings for this model can be found on the author’s site.
It’s a must-check for modelers, of course, and for those who want to generate more beautiful girl illustrations.

Please generate many illustrations and find your favorites.
Of course, all are free for personal use!”


note: Site Summary

  • “AingDiffusion” is a model specialized in anime image generation, producing high-quality images tailored to the creator’s preferences.
  • Model updates can be checked under “About this version,” with the option to use older versions.
  • Available on HuggingFace and, but note that the author is not active on HuggingFace.
  • Significant resources are required for model maintenance, and support through Ko-fi is requested.
  • The model is always free, with updates for the EXP model at
  • The guide for image generation includes using VAE, negative text inversion (※ embeddings), recommended sampler selection, necessity for Hi-res fix, and settings for Clip skip and ENSD.
  • Frequent updates optimize the model and share improvements.
  • If struggling with image generation, adjusting prompts or changing sampling methods is recommended.

“This is an advertisement.”

FANZA Illustration Collection (R18)


note: Site Summary

Greetings to HENTAI gentlemen all over the world!
The purpose of ButamanMix is simple: to easily create HENTAIART!
If there’s a significant response, I’ll release the next ButamanMix!



note: Site Summary
“SoteMix V3” is a model trained specifically for high-resolution image generation at 1024×1536.
It’s available on Hugging Face and supports OpenVINO and ONNX.
The model has been tested with SD.Next’s Diffusers backend and HyperTile size set to 0 (auto).
For positive prompts, it’s recommended to use terms like masterpiece, best quality, highres, 1girl.
Recommended negative prompts include (worst quality, low quality, lowres), zombie, interlocked fingers, water.
Do not use embeddings.
The suggested sampler is “Euler a” with 30-40 steps, clip skip at 1, and CFG between 4-7.

All Illustrations Are Commercially Usable

Some illustrations may have incomplete details such as hands and fingers. We kindly ask for your understanding.

When using these images, please feel free to crop or modify them as needed.

We encourage you to fully express your creativity and use these illustrations with a free and imaginative approach.

The illustrations featured on this site are available for commercial use.

Beautiful Illustration Cards in Your Hands!

We’ve prepared illustration cards with plenty of space for your messages. A stunningly beautiful girl will convey your words. Download now for free!

Prompts and Settings for Image Generation

Note: The prompts are generally aligned with the parameters set on the CIVITAI author page.


Positive prompt: {1 girl|wearing a swimwear}
Negative prompt: AissistV2-neg,

Steps: 50
Sampler: DPM++ 2M SDE Karras
CFG scale: 7
Seed: 2382849811
Size: 512×512
Model hash: 4fb1719268
Model: aingdiffusion_v15
Denoising strength: 0.4
Hires upscale: 1.5
Hires steps: 10
Hires upscaler: ultramix_balanced
Version: v1.7.0


Positive prompt: {1 girl|wearing a swimwear}
Negative prompt: (worst quality, low quality:1.4),(negative_hand-neg:1.1),(greyscale:1.1) ,(monochrome:1.1),censored,bad proportions,jpeg artifacts,(signature, watermark,logo, text:1.2),

Steps: 60
Sampler: DPM++ 2M Karras
CFG scale: 4
Seed: 1212332176
Size: 512×512
Model hash: 981e83de21
Model: butamanmix_80
VAE hash: df3c506e51
Denoising strength: 0.3
Clip skip: 2
Hires upscale: 1.5
Hires steps: 10
Hires upscaler: R-ESRGAN 4x+ Anime6B
Version: v1.7.0


Positive prompt: {1 girl|wearing a swimwear}
Negative prompt: (worst quality, low quality, lowres), zombie, interlocked fingers, water, realistic,

Steps: 40
Sampler: Euler a
CFG scale: 4
Seed: 2760101277
Size: 512×512
Model hash: dcc16969a0
Model: sotemix_v30
Denoising strength: 0.3
ENSD: 31337
Hires upscale: 1.5
Hires steps: 20
Hires upscaler: 4x-AnimeSharp
Version: v1.7.0