RDBT [Anima]

Note: Experimental, but works.

A LoRA for circlestone-labs/Anima.

Finetuned + CFG distilled.

About CFG distilled:

• Use CFG scale = 1. Prefer euler and euler a.

• Because you don't need to run a forward pass for the negative prompt, you can generate 2x faster. (Almost as fast as SDXL)

Note:

This is not a style LoRA. You still need to specify styles in your prompt.

Don't expect a straightforward effect. This LoRA is not overfitted. The dataset is not small.

Versions

[base model version] [dataset version] [other flags]

(2/3/2026) preview v0.2 distilled: Speedrun attempt, mainly for testing the training script. Limited training dataset. Only covered "1 person" images plus a little bit of "furry". But it works, and way better than what I expected.