ComfyUI-Pixal3D
ComfyUI nodes for Pixal3D (SIGGRAPH 2026, TencentARC) - pixel-aligned image-to-3D. Academic-only model license.
[!WARNING] Warning, uses experimental package
comfy-envto attempt a one click isolated install. Will download and use pixi package manager.
ComfyUI-Pixal3D
Installation
Three options, in order of speed → reliability:
- ComfyUI Manager (recommended) — search for
Pixal3Din the Manager and click Install from the highest version displayed. If that doesn't work, try nightly. - Manager via Git URL — in ComfyUI Manager: "Install via Git URL" with
https://github.com/PozzettiAndrea/ComfyUI-Pixal3D.git. - Manual (most reliable):
cd ComfyUI/custom_nodes git clone https://github.com/PozzettiAndrea/ComfyUI-Pixal3D.git cd ComfyUI-Pixal3D pip install -r requirements.txt --upgrade python install.py
Please report any problems you hit during installation or use of my nodes — open a Discussion or Issue. Very grateful for your help! 🙏
<div align="center"> <a href="https://pozzettiandrea.github.io/ComfyUI-Pixal3D/"> <img src="https://pozzettiandrea.github.io/ComfyUI-Pixal3D/gallery-preview.png" alt="Workflow Test Gallery" width="800"> </a> <br> <b><a href="https://pozzettiandrea.github.io/ComfyUI-Pixal3D/">View Live Test Gallery →</a></b> </div>
ComfyUI nodes for Pixal3D (SIGGRAPH 2026, TencentARC) — pixel-aligned image-to-3D generation. Single image in, textured GLB out.
- Project page: https://ldyang694.github.io/projects/pixal3d/
- Paper: https://arxiv.org/abs/2605.10922
- Upstream code: https://github.com/TencentARC/Pixal3D
- Model weights: https://huggingface.co/TencentARC/Pixal3D
Nodes (MVP)
| Node | Purpose |
|------|---------|
| Pixal3DLoadPipeline | Loads the cascade pipeline + four DinoV3 cond models. Auto-downloads Pixal3D weights. |
| Pixal3DLoadMoGe | Loads MoGe-2 for camera-intrinsic estimation. |
| Pixal3DPreprocessImage | Background removal + alpha bbox crop + 1024-max resize. |
| Pixal3DEstimateCamera | Runs MoGe-2 to estimate camera_angle_x and distance from the input image. |
| Pixal3DGenerate | Runs the four-stage cascade (SS → shape LR 512 → shape HR 1024 → texture 1024). |
| Pixal3DExtractGLB | Extracts a textured GLB via o_voxel.postprocess.to_glb. Saves to output/. |
Hardware
- NVIDIA GPU with SM ≥ 8.0 (Ampere/Ada/Hopper/Blackwell). flash-attn-3 has no fallback for older GPUs.
- ≥24 GB VRAM recommended for
1024_cascadewithlow_vram=True. More for1536_cascade. - ~30 GB free disk for model weights.
Community
Questions or feature requests? Open a Discussion on GitHub.
Join the Comfy3D Discord for help, updates, and chat about 3D workflows in ComfyUI.
Credits
Built on the work of the Pixal3D authors (Li, Zhao, Chen, Hu, Guo, Zhang, Shan, Hu — Tsinghua / Tencent ARC / Victoria University of Wellington), Microsoft TRELLIS.2, Direct3D-S2, MoGe (Microsoft), and DINOv3 (Meta).
Built with DINOv3.
Wrapper authored by Andrea Pozzetti.
Contributing
Contributions are welcome! Please feel free to submit issues and pull requests.