First attempt to train a LoRA on something other than 1.5 (trained with AnyLoRA). Tested with multiple anime and realistic models using just very short prompts (subject, trigger). Works with other character LoRAs without breaking their style but of course you need to play with weights on both to find the best compromise.
Description
FAQ
Comments (6)
missing prompt in examples.
Images made with ComfyUI, they contain full workflow including prompts, you just need to copy nodes and paste them to ComfyUI canvas. But prompts were super simple here, just "Emma Watson, saiyan armor" etc.
Hello hello, just had some free time and saw that you responded to the bounty. I've played around a bit with it and have been enjoying most of my results with it. I can even play around with some of the features of the armor easily enough. Honestly all in all a good model.
A strong contender even if it is the only contender so far.
Thanks for the feedback, glad you like it! We shall see if more competition will appear to the bounty :)
Will this work with pony based checkpoints
No, this is a 1.5 model so not compatible with Pony.