anime

Animagine XL 4.0

Anime characters and scenes, at the resolution the style was made for.

An SDXL fine-tune retrained from scratch on 8.4 million anime images, roughly 2,650 GPU hours. It runs at its full 28 steps rather than a distilled shortcut, which is why it costs about the same per image as the premium photo model. Character work is where it is strongest; it was trained on tag ordering, so tags help.

Open in StudioAPI reference
per image
20.2 s
compute
$0.0030
steps
28 · CFG 5.0
class
1024 px
runs on
GPU

What it is good at

  • Anime portraits and scenes — the best character work in the catalog
  • Ink wash, fantasy, retro cel and flat animation backgrounds
  • Holds a surface style: flat colour, thick outlines, limited palettes

Where it struggles

  • Character identity drifts to anime even when the prompt asks for something else
  • Architecture and dress drift Japanese unless the prompt fights it hard
  • Counting: ask for four figures and you often get five

Writing prompts for this model

  • A quality tag suffix is appended for you — do not write it twice
  • Spell the style out redundantly: proportions, outline weight, flat colour, no gradients
  • Naming a style loosely gets you anime; describing it concretely does not

36 looks · 108 samples · every prompt published

Its own catalog

These looks were written for this model, not shared with the others — the same prompt on a different model tells you which prompt suits which model, not which model is better. Click a look to view it full size or copy the prompt; open it in Studio from there. Every image below is the first of 3 seeds.

Licence

CreativeML Open RAIL++-M — commercial use permitted.

Other models