Well, after a year or two, I'm finally learning how to train LoRAs offline. Hopefully I can revisit my first models, trained here on CivitAI, and make them actually functional, haha.

This lil fella is the first of the new gen, and he works pretty well. Trained on Pony Diffusion. Funnily enough, without any additional tags, he gens as some interesting black dragon serpent kind of thing. With the tags, he gens pretty consistently at strength 1. I haven't played around with the different strengths yet.

Main trigger is em3raldtr3boa. To get the fully true gen, add green scales, white stripes, to the positive prompt, and horns and dragon to the negative prompt.

Unfortunately, this LoRA doesn't work with Easy Diffusion. I don't have the OneTrainer settings right, so it keeps on saying there's no Unet linear embedding. Works fine on Comfy UI, and I expect, most others.