The model I think is fairly self explanatory.
Works fine at a strength of 1. Previews made with novaFurryXL_illustriousV20.
Description
FAQ
Comments (2)
Yo, bit curious about something if you'd indulge it.
When you train these LoRAs, do you use all of the doujin versions?
Like this one has the B/W, and two different colorizations.
Do you vet the colorizations to ensure they aren't using bad colors and just use the best or not use them at all if the colors are bad or do you straight just use any and all versions?
I tend to often use a colorized version if there's one available, I vaguely recall doing this one's colorization but I couldn't really tell you which one of the two since I don't backup my datasets once I'm done with them I delete them to save harddrive space. I usually just run one whole doujin, pruning pages that don't matter, such as pages that are largely blank or a 'joke page' of chibi art and such. I put a detailed explanation of my process at https://civitai.com/models/1169975?dialog=commentThread&commentId=671577&highlight=671620 that has some information you might find relevant such as my baking preset and an example dataset of what my datasets typically look like when I bake them.