Heliolisk LoRA trained on NovelAI with Kohya training scripts using the DAdaptAdam optimizer at 512x512 resolution. Trained on a low quality, inconsistent dataset, expect issues. Works pretty well when nude, but breaks down once you prompt for clothing or more complex scenes. Increasing the LoRA strength and prompting for the incorrect features will help with consistency, see my example pictures to get an idea of what that looks like. Should have wide compatibility with model mixes based on NovelAI, which is most of them. Trained on a mixed male/female dataset.
Description
Initial release
FAQ
Comments (11)
Thanks a lot🦖🐉🦎
I'll post some images when I get to some extent effects.
No prob. Hopefully you don't find it too difficult to work with 😅
What a good job you do, I like your parrots, I was wondering if you could make one of Greninja?
I’m glad you’re enjoying them :). I’ve made a few attempts at Greninja, but haven’t gotten a good output yet. I’ve got a list of more difficult pokemon that I’m working on right now, and Greninja is one of them.
@KomodoStar good luck, with the dificult pokemon, I will be waiting for more of your work and for Greninja
I was wondering, will you ever do mega evolutions or gigantamax pokemon?
If they have enough images to train a good model, I'll likely be attempting them. That'll probably exclude the majority of the designs except for the more popular ones (Mega Lucario, Gardevoir, Absol, etc).
@KomodoStar understandable
when will you be back? a week without Loras is an eternity :c
I've been taking a break around the holidays, so expect that to continue through the new year. I'll be posting stuff, but a lot less frequently.