Awesome

nvdiffrecmc (NeurIPS 2022)

Official PyTorch implementation

Teaser image

Joint optimization of topology, materials and lighting from multi-view image observations as described in the paper Shape, Light, and Material Decomposition from Images using Monte Carlo Rendering and Denoising (NeurIPS 2022).

Licenses

This work is made available under the Nvidia Source Code License.

For business inquiries, please visit our website and submit the form: NVIDIA Research Licensing.

Installation

Requires Python 3.6+, VS2019+, Cuda 11.3+ and PyTorch 1.10+, and an NVIDIA GPU with a modern driver supporting OptiX 7.3 or newer.

Tested in Anaconda3 with Python 3.9 and PyTorch 1.12 on the following GPUs: V100, RTX3090, and A6000.

One time setup

Install the Cuda toolkit (required to build the PyTorch extensions). We support Cuda 11.3 and above. Pick the appropriate version of PyTorch compatible with the installed Cuda toolkit. Below is an example with Cuda 11.6

conda create -n dmodel python=3.9
activate dmodel
conda install pytorch torchvision torchaudio cudatoolkit=11.6 -c pytorch -c conda-forge
pip install ninja imageio PyOpenGL glfw xatlas gdown
pip install git+https://github.com/NVlabs/nvdiffrast/
pip install --global-option="--no-networks" git+https://github.com/NVlabs/tiny-cuda-nn#subdirectory=bindings/torch
imageio_download_bin freeimage

Every new command prompt

activate dmodel

Examples

Our approach is designed for high-end NVIDIA GPUs with large amounts of memory. To run on mid-range GPU's, reduce the batch size parameter in the .json files.

Simple genus 1 reconstruction example:

python train.py --config configs/bob.json

Visualize training progress (only supported on Windows):

python train.py --config configs/bob.json --display-interval 20

Below, we show the starting point, our extracted diffuse albedo texture, our final result and the reference.

Our result

The results will be stored in the out folder. The Spot and Bob models were created and released into the public domain by Keenan Crane.

Included examples

spot.json - Extracting a 3D model of the spot model. Geometry, materials, and lighting from image observations.
spot_metal.json - Example of joint learning of materials and high frequency environment lighting
bob.json - Simple example of a genus 1 model.

Datasets

We additionally include configs (nerf_*.json, nerd_*.json) to reproduce the main results of the paper. We rely on third party datasets, which are courtesy of their respective authors. Please note that individual licenses apply to each dataset. To automatically download and pre-process the NeRF and NeRD datasets, run the download_datasets.py script (note that the NeRFactor dataset is large (~10GB)):

activate dmodel
cd data
python download_datasets.py

Below follows more information and instructions on how to manually install the datasets (in case the automated script fails). Note that the NeRFactor dataset is large (~10GB).

NeRF synthetic dataset Our view interpolation results use the synthetic dataset from the original NeRF paper. To manually install it, download the NeRF synthetic dataset archive and unzip it into the nvdiffrec/data folder. This is required for running any of the nerf_*.json configs.

NeRD dataset We use a dataset from the NeRD paper, which features real-world photogrammetry and inaccurate (manually annotated) segmentation masks. Clone the NeRD datasets using git and rescale them to 512 x 512 pixels resolution using the script scale_images.py. This is required for running the nerd_gold.json config.

activate dmodel
cd nvdiffrec/data/nerd
git clone https://github.com/vork/moldGoldCape.git
python scale_images.py

NeRFactor dataset We use datasets from the NeRFactor paper, which features a simplified version of a subset of the NeRF Synthetic dataset (low frequency lighting). Download the NeRFactor datasets from their Google Drive. This is required for running any of the nerfactor_*.json configs. The four datasets should be placed in the folders data\nerfactor\hotdog_2163, data\nerfactor\drums_3072, data\nerfactor\ficus_2188, and data\nerfactor\lego_3072.

Use the extracted 3D models in Blender

We include a script blender.py that can be used to load our geometry, materials and environment lighting in Blender. An example of the shading graph is included below Teaser image

Server usage (through Docker)

Build docker image.

cd docker
./make_image.sh nvdiffrec:v1

Start an interactive docker container: docker run --gpus device=0 -it --rm -v /raid:/raid -it nvdiffrec:v1 bash
Detached docker: docker run --gpus device=1 -d -v /raid:/raid -w=[path to the code] nvdiffrec:v1 python train.py --config configs/bob.json