Description:A semi-real, grainy 2.5D look. Colours go cool and greyish, surfaces get a rough painted grain, and faces and bodies are shaded with more realistic volume.Known quirk: it frames shots tighter (it can crop the top of the head) and makes body proportions curvier.This isn't a trained LoRA. No images were used to make it. It comes from steering neurons inside the base model:<ul><li>I mapped the UNet's feed-forward neurons into groups. This file is five groups across three layers (two early in the UNet, one style layer), mixed by a random combination search.</li><li>The LoRA stores that push as a bias change, which is why the file is only about 15 KB.</li></ul>How to use<ul><li>No trigger word. Just load the LoRA; it acts on the model only, not the text encoder.</li><li>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.</li><li>It stacks with other LoRAs.</li><li>Preview settings: 768×1024, 20 steps, CFG 4.5, DPM++ 2M Karras.</li></ul>Compatibility<ul><li>Made and tested in ComfyUI (Load LoRA / LoraLoaderModelOnly) on prefectIllustriousXL v4.0.</li><li>Other UIs: the file holds only bias changes, which some UIs (A1111 / Forge) may not load. If nothing changes, that's why.</li><li>Other Illustrious checkpoints: untested, but they share the base, so it may carry over.</li></ul>

Files with SHA256: eb5a0e720055b317927062f9201634f76609211edebc4eb60e3deed0aebd28ff