Recently upgrade to a rtx3060 from, well, nothing. Now I get to mess around a lot more (difference of minutes to seconds). I've always felt like there are a lot of characters and concepts that are criminally underrepresented here,
Warning: I have no idea what I'm really doing, yet. This was created using Anythingv5 (I think), trained for 10 epoch on 93 images. Manually tagged everything myself... terribly, I'm sure. I have a bunch of ideas already on how to properly tag and train the model for future updates. Figured I'd post this now and see if anyone wants to mess with it. It's almost to the same level as other models that I've used similar prompts for, but some clear and glaring problems.
My goal is to allow it to do various styles and simply capture Kid Buu's likeness. As of now, it's a little uncontrollable. I'm already in the process of retagging, I'd like it to be versatile with outfit. The data set includes different styles of art and some alternate coloring and outfits, but the majority is his standard look.