TwistedReality-XL
Version 4 (full SDXL finetune)
This version is a full finetune from the base SDXL checkpoint, and not base on Cyberrealistic XL.
https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0
These are the hyperparameters used for the training:
8 A40s
5000 steps
lr = 1e-4
batch size = 8
gradient accumulation = 4
total effective batch size = 8 x 8 x 4 = 256
The dataset has been captioned with either xcomposer or qwen vlm, and compressed into comma separated prompts. I've let out the pronouns, verb, so use nouns and adjectives to describe the scene.
It is not very good with hardcore NSFW stuff (this is not pony, illustrious or noobai), so you need to use a Lora for this.
I might look into data balancing techniques to promote rarer concepts...
Versions 3 and below
This model is a Dora trained model on top of the excellent CyberrealisticXL checkpoint. This model enhances the realism of the photos and force the output to be in the style of the adult movie studios Playboy and ScoreGroup. I strongly suggest using Perturbed Attention Guidance (PAG) to fix some of the anatomy problems that can come up with this model. CFG can go from 1.0 to 7.0,
Description
This version uses the fixed code for the PEFT FourierFT training technique, with more gradient accumulation steps.
FAQ
Comments (5)
I cannot make this work no matter what i try...still trying to avoid comfy though
This is a regular checkpoint, this should work in Automatic1111 also.
@WhyNaNWhenNyan I am trying it there, but my settings only give me garbled outputs. Are your test images using img2img? I set up PAG to try it out and i just can't make it look good. It gets close sometimes, but nothing like what you achieve. Thank you!
@vbob Make sure to use DDIM scheduler or DDPM, there are some schedulers that oversaturate the outputs or crank too much the details.
You should try the new version, this should work out of the box since this is a finetune from the base SDXL checkpoint.