Sign In

JuicyFajitaMix

110
445
1
Verified:
SafeTensor
Type
Checkpoint Merge
Stats
445
Reviews
Published
Apr 25, 2023
Base Model
SD 1.5
Hash
AutoV2
64AE194DE0

Model is highly stylized and character centric. It does really well with portraits, and it works best when the width is at most equal to the height. Beyond that, characters will start to lose detail if you want both landscapes and characters.

But it has an aesthetic I really enjoy so ¯\_(ツ)_/¯ maybe I'll do more iterations and see if I can get better wide shots. We have the technology.

Model Bias

The model is very stylized because of its ingredients:

  • DalcefoOilpainting V1

  • BetterPastelMix

  • Silicon29Dark

  • MossaCannibalis_E8 (Lora)

  • Krekk0vV2 (Lora)

Dalcefo'sOilPaintingv1 model was the base and it is a portrait style model. Pastel and Dalcefoil's styles had great synergy, but both have an issue with highly detailed characters at times. So the trouble has been made double.

Mossa is a brilliant artist (that's just science don't get mad at me) whose style mixed really well with the Dalcefoil style. And I included Krekk0v because. Both are pretty lewd artists so the model trends towards risque outfits, but this bias can be controlled as per the example pics.

Final Notes

  • Occasionally the model will render blue eyes like spice eyes from dune. This should be regarded as a feature rather than a bug.

  • It is called "FajitaMix" because everyone's naming their models after food and Mossa's Aztec stuff is high culture. Me gusta mucho.

This is my first model so please let me know if you have any feedback or simply wish me bodily harm in the reviews. I wish you many healthy hands and a full set of fingers.