Note: Cover images are raw output at 1024px. No upscale, no face/hand detailer, no negative prompt, no CFG (CFG scale =1). For demonstration purposes only.

RDBT [Anima]

It was trained as a LoRA natively, main model page: https://civitai.com/models/2364703.

This page contains merged fp16 base model. Which = bf16 pretrained anima base model + fp32 LoRA.

If you have already downloaded the pretrained anima base model, you can just download the LoRA.

p1 [fp8]:

fp8 anima preview. I uploaded it here for convenience.

About fp8 models:

Quantized fp8 models in ComfyUI format. If you GPU supports it (aka. rtx 4000 and later), ComfyUI will use hardware fp8 automatically, which might (blame Nvidia) be faster. More about ComfyUI fp8, see ComfyUI docs TensorCoreFP8Layout.