25launches

The first launch of LiliPixel V1.

LiliPixel V1 page

LiliPixel V1

Launching today

Open-weight pixel-art model trained from scratch

Hunted by IlyasHunter
Visit huggingface.co (opens in a new tab)

LiliPixel V1 is a text-to-pixel-art model trained from scratch: a 191M-parameter diffusion transformer that draws sprites at a native 16x16 or 32x32 with a transparent background, or 64x64 scenes with an optional realistic mode. It needs about 1 GB of memory and runs with Diffusers or a ComfyUI node, with a free demo on Hugging Face. Trained on four public pixel-art datasets, two of them downscaled from FLUX images. Weights under OpenMDW-1.1. Complex scenes and text are weak.

Meet the team

Hunted by Ilyas. The makers are not on 25launches yet.

H (Hunter)
  • LiliPixel V1 screenshot 1
  • LiliPixel V1 screenshot 2
  • LiliPixel V1 screenshot 3

Discussion 1

Ask questions, give feedback. Comments don't count in the ranking.

Sign in to join the discussion.

  • IlyasHunter Pinned

    Most AI pixel art is a big image model told to look pixelated, then cleaned up by hand. LiliPixel works at the real sprite size, so a 32x32 chest comes out as 32x32 pixels on a transparent background, from a model that fits in about 1 GB. I tried the demo: small game items look clean, while 64x64 scenes are much rougher, as the model card itself warns.