---
license: other
license_name: nscl-v1
license_link: LICENSE
tags:
- image-generation
- class-conditional
- diffusion
- pixel-space
- dit
- imagenet
library_name: pytorch
pipeline_tag: unconditional-image-generation
---

<p align="center">
  <img src="https://raw.githubusercontent.com/pixeldit/pixeldit.github.io/main/static/images/pixeldit-logo.png" height="120" />
</p>

<h2 align="center">PixelDiT: Pixel Diffusion Transformers for Image Generation</h2>

<p align="center">
  <a href="https://www.yongshengyu.com/">Yongsheng Yu</a><sup>1,2</sup> &nbsp;
  <a href="https://wxiong.me/">Wei Xiong</a><sup>1†</sup> &nbsp;
  <a href="https://weilinie.github.io/">Weili Nie</a><sup>1</sup> &nbsp;
  <a href="https://shengcn.github.io/">Yichen Sheng</a><sup>1</sup> &nbsp;
  <a href="http://behindthepixels.io/">Shiqiu Liu</a><sup>1</sup> &nbsp;
  <a href="https://www.cs.rochester.edu/u/jluo/">Jiebo Luo</a><sup>2</sup>
</p>
<p align="center">
  <sup>1</sup>NVIDIA &nbsp; <sup>2</sup>University of Rochester
  <br>
  <sup>†</sup>Project Lead and Main Advising
</p>

<p align="center">
  <a href="https://pixeldit.github.io/"><img src="https://img.shields.io/badge/Website-Project_Page-2ea44f" /></a>
  &nbsp;
  <a href="https://arxiv.org/abs/2511.20645"><img src="https://img.shields.io/badge/arXiv-2511.20645-b31b1b.svg" /></a>
  &nbsp;
  <a href="https://github.com/NVlabs/PixelDiT"><img src="https://img.shields.io/badge/GitHub-Code-blue" /></a>
</p>

## Model Overview

**PixelDiT-XL** (797M parameters) is a class-conditional image generation model trained on ImageNet, operating directly in **pixel space** — no VAE, no latent space. It uses a dual-level architecture combining a patch-level DiT for global semantics with a pixel-level DiT for fine texture details.

## Pre-trained Checkpoints

| Checkpoint | Resolution | Epochs | gFID | CFG Scale | Time Shift | CFG Interval |
|:---|:---:|:---:|:---:|:---:|:---:|:---:|
| `imagenet256_pixeldit_xl_epoch80.ckpt` | 256x256 | 80 | **2.36** | 3.25 | 1.0 | [0.1, 1.0] |
| `imagenet256_pixeldit_xl_epoch160.ckpt` | 256x256 | 160 | **1.97** | 3.25 | 1.0 | [0.1, 1.0] |
| `imagenet256_pixeldit_xl_epoch320.ckpt` | 256x256 | 320 | **1.61** | 2.75 | 1.0 | [0.1, 0.9] |
| `imagenet512_pixeldit_xl.ckpt` | 512x512 | 850 | **1.78** | 3.5 | 2.0 | [0.1, 1.0] |

All evaluations use **FlowDPMSolver** with **100 steps**. 50K samples. Metrics follow the ADM evaluation protocol.

## Usage

### Installation

```bash
pip install torch torchvision lightning omegaconf timm wandb h5py
```

### Evaluation (Generate 50K Samples)

```bash
cd c2i/

# ImageNet 256x256 (epoch 320, best FID)
torchrun --nproc_per_node=8 main.py predict \
  -c configs/pix256_xl.yaml \
  --ckpt_path=imagenet256_pixeldit_xl_epoch320.ckpt \
  --model.diffusion_sampler.class_path=src.diffusion.FlowDPMSolverSampler \
  --model.diffusion_sampler.init_args.num_steps=100 \
  --model.diffusion_sampler.init_args.guidance=2.75 \
  --model.diffusion_sampler.init_args.timeshift=1.0 \
  --model.diffusion_sampler.init_args.guidance_interval_min=0.1 \
  --model.diffusion_sampler.init_args.guidance_interval_max=0.9 \
  --per_run_seed=false --seed_everything=1000

# ImageNet 512x512
torchrun --nproc_per_node=8 main.py predict \
  -c configs/pix512_xl.yaml \
  --ckpt_path=imagenet512_pixeldit_xl.ckpt \
  --model.diffusion_sampler.class_path=src.diffusion.FlowDPMSolverSampler \
  --model.diffusion_sampler.init_args.num_steps=100 \
  --model.diffusion_sampler.init_args.guidance=3.5 \
  --model.diffusion_sampler.init_args.timeshift=2.0 \
  --model.diffusion_sampler.init_args.guidance_interval_min=0.1 \
  --model.diffusion_sampler.init_args.guidance_interval_max=1.0 \
  --per_run_seed=false --seed_everything=10000
```

After generating samples, compute FID with the [ADM evaluation toolkit](https://github.com/openai/guided-diffusion/tree/main/evaluations).

## Model Architecture

| Component | Value |
|-----------|-------|
| Parameters | 797M |
| Input channels | 3 (RGB) |
| Patch size | 16 |
| Hidden size | 1152 |
| Attention heads | 16 |
| Patch-level depth | 26 |
| Pixel-level depth | 4 |
| Pixel hidden size | 16 |
| Classes | 1000 (ImageNet) |

## Citation

```bibtex
@misc{yu2025pixeldit,
      title={PixelDiT: Pixel Diffusion Transformers for Image Generation},
      author={Yongsheng Yu and Wei Xiong and Weili Nie and Yichen Sheng and Shiqiu Liu and Jiebo Luo},
      year={2025},
      eprint={2511.20645},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2511.20645},
}
```

## License

This model is released under the [NVIDIA OneWay Non-Commercial License](LICENSE). The work and any derivative works may only be used for non-commercial (research or evaluation) purposes.