This is my ninth style lora so go easy with me on this one. V1: More 2d and toony. Good as a "cutify" lora or combined with others. V2: More heavily trained to match Dacad's style better, which includes more depth and better matching character design.As always, which one you prefer to use is up to you. I've remade V1 images in V2 for easy comparison. Aiming for the look this artist uses in their more high-quality animations and drawings. Very happy with the outcome. Unfortunately, a lot of the artist's work doesn't pass the rules of Civitai so I had to leave out some stuff, but there was still plenty of data to work with (V2 uses 400 images, believe it or not). Like with other style loras, feel free to use both the custom d4c4d tag, along with e621 tag of dacad that your checkpoint might already be trained on.

Dacad Style

Complete rework. Using a much, much larger dataset (my largest used yet). Trained on a higher step count and tweaked settings.

Files with SHA256: f5dc260b122412d4699c253b80d3702c832c6dc38eaa154afd082a1e4e590c28