- MSIAfterbu...png
- - 3.02 MB
- (3369x1068)
Last year I did a bunch of experiments with Stable Diffusion image generation stuff and of course one of the first things I did was create Kuon and Aquaplus LORAs for my own personal use. Well, now for SDXL (specifically Pony which is a finetune of SDXL so that it's actually usable for this kind of thing) is now something I wish to try, I have begun the process of getting Kuon in higher quality. By that I mean, higher quality AI generations! Kuon could not be any higher quality herself!
The first step of the SDXL Kuon LORA process has begun after a week of careful image cropping, editing and tagging! Yes, you don't need to be this specific about it, but it's worth doing. I've started the style training thing before going to bed, but I still need to create a curated set of Kuon-specific images to train the concept of her appearance for the second part! It will take me another week or so to gather all those and carefully curate them, but since I'll be looking at images of Kuon the entire time it won't feel like work.
Well, assuming this first style thing works. Who knows what I'll wake up to? It could be a huge mess since I'm not quite accustomed to this SDXL stuff.
KUUUUUUUUUUUUUUUUOOOOO