BudgetPix-PixelDiT

🌐 Project page  |  💻 Code  |  📄 Paper  |  🧩 BudgetAny

One checkpoint, any token budget. Text-to-image generation at 1024 x 1024 in pixel space with PixelDiT-T2I 1.3B, fine-tuned with BudgetPix compute-adaptive tokenization. At inference you choose how many tokens the diffusion transformer runs on: an entropy-based detail map merges the flattest regions into coarse tokens first and gives the fine tokens to the most detailed ones. The same weights serve every budget.

Unofficial, open-source implementation, taken from budgetpix/BudgetPix-public: a public reimplementation of BudgetPix: compute-adaptive tokenization for pixel-space image diffusion (JiT, MiniT2I, PixelDiT, pMF). This model is part of BudgetAny, a community reimplementation of BudgetPix (Kara et al., 2026). It is not an official release. Thanks to the BudgetPix authors for the method.

Files

file	base model	dense tokens	parameters	size	SHA-256
budgetpix_pixeldit_1024.pt	PixelDiT-T2I 1.3B, 1024 px	4,096	1,307 M	5.23 GB	0c3d82e42cfbe229…

SHA256SUMS lists the full checksums. The text encoder (Efficient-Large-Model/gemma-2-2b-it) is downloaded from the Hub on first use.

Usage

git clone https://github.com/BudgetAny/BudgetPix && cd BudgetPix/pixeldit    # setup: see README.md
hf download BudgetAny/BudgetPix-PixelDiT budgetpix_pixeldit_1024.pt --local-dir weights
python sample.py --ckpt weights/budgetpix_pixeldit_1024.pt --prompts "a red fox in a snowy forest" --budget 2048

--budget is the number of tokens per image (4,096 is the dense grid). Default guidance 2.75; about 4.5 at 25 % of the grid.

Citation

If you use this model or the code, please cite the BudgetPix paper:

@article{kara2026budgetpix,
  title={BudgetPix: Compute-Adaptive Tokenization for Pixel-Space Image Diffusion},
  author={Ozgur Kara and Yujia Chen and Daniel Watson and David Forsyth and James Matthew Rehg and Wen-Sheng Chu and Du Tran},
  journal={arXiv preprint arXiv:2610.12307},
  year={2026}
}

Built on PixelDiT and APT.

License

PixelDiT and its released weights are distributed by NVIDIA under the NVIDIA License (NSCLv1), which permits non-commercial use only; this checkpoint is a derivative of those weights and is released under the same terms.