This is a merge between my custom model and Realistic Vision 3.0 which can be found here

Generated images are quite similar to Realistic Vision but the faces are a bit aesthetically pleasing (in my humble opinion)

Model is EXTREMELY biased towards NSFW generations. Heavy de-emphasis using prompts may be required for SFW images.

I suggest using the following as a base prompt and remove what you dont need.

Positive: subject, (RAW photo, photorealistic:1.3), candid, 16mm, color graded, remarkable color, ultra realisitic, textured skin, remarkable detailed pupils, (imperfect skin:1.1), realistic dull skin noise, visible skin detail, skin fuzz, dry skin, film still

Negative: doll, (deformed, distorted, disfigured:1.3), poorly drawn, bad anatomy, wrong anatomy, extra limb, missing limb, floating limbs, (mutated hands and fingers:1.4), disconnected limbs, mutation, mutated, ugly, disgusting, blurry, amputation


What this model does well:

1. Good portraits

2. Realistic skin textures

What this model doesnt do well:

1. Backgrounds (servicable but not very reliable)

2. SFW generation requires some work

3. Ethnicities are a little whack

I have tried it with multiple lora's and it worked well for me but your mileage may vary.