This one specifically is inspired by Mizuryu Kei's artwork, but I wanted to make it softer and cleaner. Mainly I like the expressive faces in Kei's artwork and wanted to try and pull that through while creating something unique that stands on its own.
That is to say if I accomplished what I was going for you wouldn't know the artist reference unless I'd told you, except maybe from a hint of a smile.
Hopefully I pulled that off!
See "About This Version" for the Sketch LoRA. It's spotty, but hey, I spent buzz. Try it out. Maybe it won't suck for you.
Description
Your mileage may very with this one. I debated even posting it, because results are so hit and miss. But I'm drunk so hey.
As I was putting together training images for Weeb02 about 1 in 10 images came out a sketchy sepia instead of the clean full color art I was going for. No rhyme or reason, just a quirk of the base model I'm using. But I kind of liked it, so I gathered a bunch of those images and trained with them. Results are... mixed. Adding 'sketch' and 'sepia' to your prompt gets it closer to the training images.