---

What is LINO_UniPS?

LINO_UniPS is a universal photometric stereo method based on a light-invariant normal estimator. It predicts high-quality per-pixel surface normals from multi-lighting images without requiring known light directions. The method leverages diffusion model priors and handles arbitrary numbers of input images with varying illumination conditions.

---

Requirements

• Python 3.10+
• CUDA 12.x + NVIDIA GPU (ampere or newer recommended for bfloat16 support)
• Meshroom 2025+ (develop branch)

---

Quick Start

Prerequisite: a working Meshroom installation.

1. Clone the plugin

cd /path/to/your/plugins
git clone https://github.com/meshroomHub/mrLINOUniPS.git
cd mrLINOUniPS

2. Set up the virtual environment

Meshroom looks for a folder named venv at the plugin root.

python3 -m venv venv
source venv/bin/activate

pip install --upgrade pip
pip install torch torchvision
pip install -r requirements.txt

deactivate

This installs LINO_UniPS and all its dependencies automatically via pip.

3. Download pretrained weights

bash download_weights.sh

This downloads the pretrained model (~338 MB) from HuggingFace into weights/:

weights/
└── lino.pth

The plugin auto-detects this file. No config.json needed.

4. Register the plugin in Meshroom

export MESHROOM_PLUGINS_PATH=/path/to/your/plugins/mrLINOUniPS:$MESHROOM_PLUGINS_PATH

Launch Meshroom: the LINOUniPS node appears under Photometric Stereo.

---

Node Parameters

Inputs

Parameter	Label	Description
inputSfm	Input SfMData	SfMData JSON file with multi-lighting views grouped by poseId (required)
maskFolder	Mask Folder	Folder with mask PNGs named by poseId or viewId
downscale	Downscale Factor	Integer downscale factor for input images (1-8, default: 1)
nbImages	Number of Images	Number of lighting images per pose (-1 = all)
useGpu	Use GPU	Use GPU for inference (default: true)

Outputs

Parameter	Description
outputFolder	Folder containing normal map PNGs
outputSfmDataNormal	SfMData file referencing normal maps

---

Advanced: Developer Setup

If you prefer to work from a local LINO_UniPS clone instead of pip install:

1. Clone the repo: git clone -b meshroom https://github.com/meshroomHubWarehouse/LINO_UniPS.git
2. Edit meshroom/config.json:

   [
       {"key": "LINO_UNIPS_PATH", "type": "path", "value": "/path/to/LINO_UniPS"}
   ]
   

3. Place lino.pth in the LINO_UniPS directory or its weights/ subdirectory.

The node searches for weights in this order:

1. Plugin weights/ directory
2. LINO_UniPS code directory (from config.json)
3. Torch hub cache (~/.cache/torch/hub/checkpoints/lino.pth)

---

Plugin Structure

mrLINOUniPS/
├── meshroom/
│   ├── config.json                # Plugin configuration (optional for dev)
│   └── LINOUniPS/
│       ├── __init__.py
│       └── LINOUniPS.py           # Meshroom node definition
├── weights/                       # Downloaded model weights
│   └── lino.pth
├── venv/                          # Python virtual environment
├── download_weights.sh            # Weight download script
├── requirements.txt               # Python dependencies (pip install from git)
└── README.md

For more details on how Meshroom plugins work, see:

• Meshroom Plugin Install Guide
• mrHelloWorld: step-by-step tutorials for building Meshroom plugins

---

Acknowledgements

This work is supported by DOPAMIn (Diffusion Open de Photogrammetrie par AliceVision/Meshroom pour l'Industrie), selected in the 2024 cohort of the OPEN programme run by CNRS Innovation. OPEN supports the valorization of open-source scientific software by providing dedicated developer resources, governance expertise, and industry partnership support.

Lead researcher: Jean-Denis Durou, IRIT (INP-Toulouse)

---

Related Projects

Project	Description
LINO_UniPS	Light-invariant normal estimator for universal photometric stereo
mrSDMUniPS	Meshroom plugin for SDM-UniPS photometric stereo
mrOpenRNb	Meshroom plugin for neural surface reconstruction from normals

---

License

This project is licensed under the Mozilla Public License 2.0.