Extensions/ComfyUI-UniDepth
ComfyUI Extension

ComfyUI-UniDepth

ComfyUI nodes for UniDepthV2 (universal monocular metric depth estimation).

By PozzettiAndrea·Created 3 months ago·Updated a day ago· 1
PozzettiAndrea/ComfyUI-UniDepth
Nodes
On cloudLocal install
Stars1
Updateda day ago
Readme

[!WARNING] Warning, uses experimental package comfy-env to attempt a one click isolated install. Will download and use pixi package manager.

ComfyUI-UniDepth

ComfyUI nodes for UniDepthV2 — universal monocular metric depth estimation by Piccinelli et al. (ETH Zürich, CVPR 2024 + 2025 follow-up).

From a single RGB image, UniDepth predicts:

  • Metric depth (meters)
  • Camera intrinsics (or honours yours if you provide them)
  • Per-pixel 3D point map (XYZ in camera space)
  • Per-pixel ray directions
  • Confidence map

Nodes

| Node | Purpose | |---|---| | UniDepthLoader | Load a UniDepth model (V2 ViT-S/B/L or V1 ViT-L / ConvNeXt-L). Auto-downloads from HuggingFace into ComfyUI/models/unidepth/. | | UniDepthInfer | Run inference. RGB image in → metric depth, points, intrinsics, rays, confidence. | | UniDepthCameraIntrinsics | Build a Pinhole 3×3 K from fx/fy/cx/cy (or fov + image size). | | UniDepthDepthPreview | Colorize raw metric depth to an IMAGE for visual inspection. | | UniDepthSavePointCloud | Save the per-pixel point map as a colored PLY. |

License

  • Wrapper code (everything outside nodes/unidepth/): MIT — see LICENSE.
  • Vendored upstream source (nodes/unidepth/) and pretrained weights downloaded at runtime: CC-BY-NC 4.0 — see LICENSE-UPSTREAM and NOTICE.

Because the model weights and the vendored upstream code are non-commercial, end-user use of this node pack is non-commercial only.

Citation

@inproceedings{piccinelli2024unidepth,
  title     = {UniDepth: Universal Monocular Metric Depth Estimation},
  author    = {Piccinelli, Luigi and Yang, Yung-Hsu and Sakaridis, Christos and Segu, Mattia and Li, Siyuan and Van Gool, Luc and Yu, Fisher},
  booktitle = {CVPR},
  year      = {2024}
}

@misc{piccinelli2025unidepthv2,
  title = {UniDepthV2: Universal Monocular Metric Depth Estimation Made Simpler},
  author = {Piccinelli, Luigi and Sakaridis, Christos and Yang, Yung-Hsu and Segu, Mattia and Li, Siyuan and Abbeloos, Wim and Van Gool, Luc},
  year = {2025},
  eprint = {2502.20110},
  archivePrefix = {arXiv}
}