Sign In

Cunetta Mix

Updated: Aug 1, 2025

base model

Download

1 variant available

fp16 SafeTensor

CunettaMix_illustrious_VAE.safetensors

Half precision, best balance (pruned) • 6.62 GB

Verified:

Type
Checkpoint Merge
Stats

507

Reviews
Published

Aug 1, 2025

Base Model

Illustrious

Usage Tips
Clip Skip: 2
Hash
AutoV2
BD0DA7F744
default creator card background decoration
Reactions - 30518

30.5K

Followers - 4129

4.1K

Likes - 5388

5.4K

Operates at lower CFG than most IL models so increase weight of negatives and quality positives or use that feature to get more gritty retro style and western styles.

I had a NAI and Mistoon Sapphire SD1.5 model mix that I liked and wanted to recreate that feel. Mixed from a bunch of different models and distilled to remove even more base XL and Pony remnants by add-difference merges onto itself. Didn't solve the problems with eye pupil detail or the random six fingers, but the aesthetics and colors stuck out. With an insanely narrow prompt it will line up with most IL models, but at the lower CFGs it differs a lot. With the right Lora mix and prompting you can probably reach styles in this model that others can't do as well, but it should do anything that any IL model can just in a different way.

Lacks finer detail at distance and tends to have higher noise so adjusting ETA can do a lot. Illustrious is all about the right prompt lining things up though so that's your main tool for better results.