Differentiable Geodesic Distance for Intrinsic Minimization on Triangle Meshes

Yue Li, Logan Numerow, Bernhard Thomaszewski, Stelian Coros

ACM Transactions on Graphics (Proc. SIGGRAPH 2024)

Paper Video

Project Structure

Projects/DifferentiableGeodesics/

[include & src] contains header and source files
[autodiff] contains the derivative expressions generated by a compile time autodiff library
[data] contains the meshes used in the paper
[result] is the default output folder

Compile and Run

On Linux machines, the provided docker environment should work out of the box. On MacOS, cmake can also work once all required libraries are installed using homebrew.

Install NVIDIA Docker

$ curl -fsSL https://nvidia.github.io/libnvidia-container/gpgkey | sudo gpg --dearmor -o /usr/share/keyrings/nvidia-container-toolkit-keyring.gpg
&& curl -s -L https://nvidia.github.io/libnvidia-container/stable/deb/nvidia-container-toolkit.list |
sed 's#deb https://#deb [signed-by=/usr/share/keyrings/nvidia-container-toolkit-keyring.gpg] https://#g' |
sudo tee /etc/apt/sources.list.d/nvidia-container-toolkit.list

$ sed -i -e '/experimental/ s/^#//g' /etc/apt/sources.list.d/nvidia-container-toolkit.list

$ sudo apt-get update
&& sudo apt-get install -y nvidia-container-toolkit

$ nvidia-ctk cdi generate --output=/etc/cdi/nvidia.yaml

$ sudo nvidia-ctk runtime configure --runtime=docker

Run Docker in VSCode

Download the docker image (faster option, recommended).

$ docker pull wukongsim/differentiable_geodesics_environment:latest

If you wish to build the docker image from scratch from the Dockerfile see .devcontainer/.json for instructions (simply uncommenting two lines).

Open the repository folder in VSCode. Install Docker and Dev Containers extensions in VSCode.

In VSCode, type control + p, then type >Reopen in Container (with the '>'). This option will show up in the >< tab in the bottom left corner of vscode.

The above command will open a dev container using the docker image we provided.

There you go!

Docker Image Building Time

Building this docker image can take a while, for downloading MKL libraries and compiling SuiteSparse from the source code (just to remove a single print). In case you have a powerful workstation, considering changing all the make -j8 to make -j128.

Run Comparisons

Comparisons with the Vector Heat Method can be found in the source files.

Results from Mancinelli & Puppo can be obtained using this modified code from https://github.com/liyuesolo/RCM_on_meshes

License

See the LICENSE file for license rights and limitations.

Contact

Please reach out to me (Yue Li) if you spot any bugs or having quesitons or suggestions!

If this code contributes to your research, please consider citing our work:

@article{li2024differentiable,
  title={Differentiable Geodesic Distance for Intrinsic Minimization on Triangle Meshes},
  author={Li, Yue and Numerow, Logan and Thomaszewski, Bernhard and Coros, Stelian},
  journal={ACM Transactions on Graphics (TOG)},
  volume={43},
  number={6},
  pages={1--14},
  year={2024},
  publisher={ACM New York, NY, USA}
}

liyuesolo/DifferentiableGeodesics