Style LoRA trained on qiandaiyiyu (前代未夕) illustration look for Anima.

Trigger: qiandaiyiyu

Recommended: strength ~0.8–1.0 (accept used 0.9), 30 steps, CFG ~4.5, on official anima-base.

Train: anima-base-v1.0, lora_anima unet-only, dim/alpha 32, 12 epochs (~3036 steps), 506 images.

All sample images on this page were generated with this LoRA + anima-base locally.

Tags: anime, style, anima, qiandaiyiyu