NESI: Shape Representation via Neural Explicit Surface Intersection

[Paper] [Project Page]

Python dependencies

Create a Python 3.11 environment:

conda create -n nesi python=3.11
conda activate nesi
pip install --upgrade pip
pip install -r ./requirements.txt

Then install Pytorch according to your CUDA version.

---

Usage

Example for running ESI view selection:

python select_views.py -i <path_to_input_mesh> -o $<output_folder> --n-jobs <number_of_threads> [--fast]

Example for generating training data:

python sample_training_data.py -i <path_to_"view_directions.txt"> -o $<output_folder> --mask

Example for training NESI:

python train_nesi.py --dataset-path <path_to_"training_samples"> --mask --normal

Example for evaluating NESI:

python eval_nesi.py --model-path <path_to_results>

Citation

If you find our method useful for your research, please cite our paper:

@article{zhang2025nesi,
author = {Zhang, Congyi and Yang, Jinfan and Hedlin, Eric and Takikawa, Suzuran and Vining, Nicholas and Yi, Kwang Moo and Wang, Wenping and Sheffer, Alla},
title = {NESI: Neural Explicit-Shape-Intersection-Based Geometry Representation},
year = {2025},
issue_date = {October 2025},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
volume = {44},
number = {5},
issn = {0730-0301},
url = {https://doi.org/10.1145/3742893},
doi = {10.1145/3742893},
journal = {ACM Trans. Graph.},
month = jul,
articleno = {166},
numpages = {23},
keywords = {Implicit Surfaces, Explicit Surfaces, Solid Modeling, CSG, Parametric Curves \& Surfaces}
}

Acknowledgement

This is partially built upon the implementation of Neural Geometric Level of Detail: Real-time Rendering with Implicit 3D Surfaces.