Description:
A soft, retro cel look. Colours settle into muted greys, beige and dusty accents, shading becomes simple flat cel blocks, and outlines stay clean and slightly bold, like a quiet 90s anime frame or an old artbook page.
Known quirk: it pulls hair and clothes toward grey (in the previews, teal hair turned grey-silver). Name the colours in your prompt if you need them kept.
This isn't a trained LoRA. No images were used to make it. It comes from steering neurons inside the base model:
An automated evolutionary search mixed and mutated groups of the UNet's feed-forward neurons; this look came out of round 3 as #466.
This file is four groups of about 1,000 neurons each, across four layers (one early in the UNet, three style layers).
The LoRA stores that push as a bias change, which is why the file is only about 23 KB.
How to use
No trigger word. Just load the LoRA; it acts on the model only, not the text encoder.
Strength 1 is the look in the previews. The effect scales linearly with strength: lower it for a lighter touch; negative values push the other way.
It stacks with other LoRAs.
Preview settings: 768×1024, 20 steps, CFG 4.5, DPM++ 2M Karras.
Compatibility
Made and tested in ComfyUI (Load LoRA / LoraLoaderModelOnly) on prefectIllustriousXL v4.0.
Other UIs: the file holds only bias changes, which some UIs (A1111 / Forge) may not load. If nothing changes, that's why.
Other Illustrious checkpoints: untested, but they share the base, so it may carry over.


