Fizz (LoL) LoRA版本v3 (ID: 245116)

I've decided I should upload most of the LoRAs or other content I make to CivitAI. I did not expect to make much, but I am uploading now to expand to a broader audience. Maybe something I create is what one other is looking for, but cannot find due to my current choices of upload sites.

This LoRA, while labelled V3, should still be considered my first LoRA, so it is rough and maybe overbaked due to the excessive image count in the dataset. If I ever make a V4, it will be with a much smaller dataset. Preview images are from a mix of V3 versions.

Trained using FluffyRock at 576px with 525 images using Prodigy optimizer, around 2100 steps. Tested with many other furry-based models, and generally performs well as long as the model uses e621 tagging.

Trigger/Activation keyword is lolfizz, best results with a weight of 0.8 ~ 1.

Recommended positive prompts: tentacle hair, big head, big eyes, green eyes, webbed feet, arm markings, leg markings (If desired or needed, add fangs when prompting for open mouths.)

Recommened negative prompts: countershade face, webbed hands, eyebrows, lagomorph (You may or may not want to add white markings to the negatives as well if the markings/tattoos on the arms and legs are showing up white instead of blue)

描述:

训练词语: lolfizz