The model I think is fairly self explanatory.
Works well at a strength of 1-2 for 1.0 version. 2.0 version a strength of 1 is fine.
Previews made with novaFurryXL_illustriousV6.b
Description
Okay, that's actually the last one I'll be updating on this artist for now. I'm pretty certain with this new baking preset that all my previous issues of style lora's not applying to 'pre-existing' characters in checkpoints should finally be resolved and a non-issue going forward. I can't imagine there's much more that I really need to do with the preset, as it stands it bakes 2x faster than what I was previously doing and only needs 1500 steps to produce massively better quality than I was doing with large datasets. I guess this sort of happens every month where I spend a week losing my mind at the preset not being able to properly fulfil a request, but It's a good thing because of the jumps in quality after the short periods of experimental lora's that kind of suck.
FAQ
Comments (5)
Awesome style!
can u share that new preset 👀
Sure, just give me one minute to figure out how to compile it properly. It's a preset for LoRA Easy Trainers Scripts, but even if you don't use that specific baking tool it should still be relevant to look at the presets data in notepad to emulate on other things. I'll be back in a little bit!
https://civitai.com/articles/13714 There you go!
@Malebolgia huge, thank you !