An uncensored model mainly capable of photorealism.

v2.1

Small weight adjustment of v2. Very similar but better imo.

v2

Clamped add difference v1 with cyberrealistic 3.1. More contrast.

v1

Two models are used (+ itercomp).

• bigASP 2.0 (training details)

• Lustify 4.0 DMD2

I chose these models because they are trained on large comprehensive datasets of SFW and NSFW images, with tags and natural language captions.

The following steps were performed with sd_mecha/comfy-mecha:

1. Clamped add difference with the two models against sdxl base 1.0.

2. Merge "rotate" (orthogonal procrustes) the result model with bigASP 2.0.

3. Clamped add difference with the result of step 2 against IterComp.

4. Add the original CLIP and VAE from step 2 into the model of step 3.