When I was making my lora, I notice the characters eyes does not match the original style. This is a problem because it ruin consistency. So I think the problem is the checkpoint used for training. And when I switch different checkpoint, the result is different, Turns out, that illustrious 0.1 is not the only good checkpoint for training lora like how people generally suggested.I've got a hypothesis that "checkpoint trained" is more flexible than "checkpoint merge" when it comes to lora training. Because months earlier, I tried training a lora on <a rel="ugc" href="https://civarchive.com/models/827184?modelVersionId=1761560">wai illustrious 14.0 </a>and the lora was not compatitable with other checkpoints. So I think to improve my lora, I should use different training checkpoint. But it's much more complex than I thought. It has something to do with tags, the training checkpoint, and the generation checkpoint as well. The best lora i've got is trained on <a rel="ugc" href="https://civarchive.com/models/1307857/unholy-desire-mix-sinister-aesthetic-illustrious">Unholy Desire Mix - Sinister Aesthetic</a> and then the images are generated with <a rel="ugc" href="https://civarchive.com/models/1162518?modelVersionId=1714314">plant milk hemp ii.</a> Using any other checkpoint would lose some features.

Files with SHA256: 59603a6e2ef14c7f2fa69225cdd4a23cea88c970e5eab9ebb3f83b8c580d7fdc