MR-XOTOX-NASSE-WAENDE-SDXL
More in About this version
Model 1.1 and 3.1 are not really trained models, but are provided with a Lora that contains my own photos that I once took! You can find more information about what I photograph at www.der-zerfleischer.de
At the moment I am working with Blip, Clip, Blip2 and Python to write my own local programmes for e.g. image captioning.
Kohya_ss does not run at all on my computer, neither does image captioning via ComfyUI, so I sat down to develop such a programme for myself.
You can find some videos on YouTube, but they are in German!
Description
First pseudo trained model by a trained Lora! Own real shots, I made the woman myself and provided for the decoration. You can see more on www.der-zerfleischer.de.
DPM++ 2M SDE Karras / DMP++ SDE Karras / Exponential
Steps: 30+
CFG: 2-5+
FAQ
Comments (1)
i'm not sure what you added to v3 and on, but v2 is better. That one created more unique results, where as whatever you changed in 3, now looks like all the other models, and its kind of boring because of that.