• Stylization by illustrator H.R.Giger.

• 158 images, from different years of creativity, were used to train the model.

• the training of the LoRA was on the Civitai Trainer. (the first test. I'm not sure if it's very successful)

• we have to temporarily set early access, at the lowest possible price on the site, in order to return at least a little bit of what was spent on training the model on this site.

I see that sometimes, with the heavy weight of the model, vertical stripes are visible on the generated image. at different image resolutions, at different number of steps. apparently, the training parameters are affected, which are not available here on the site, but I'm not sure. it is quite possible that I missed some details.