Viktor Rudnev, Mohamed Elgharib, William Smith, Lingjie Liu, Vladislav Golyanik, Christian Theobalt
Codebase for ECCV 2022 paper "NeRF for Outdoor Scene Relighting".
Based on NeRF++ codebase and inherits the same training data preprocessing and format.
Our datasets and preprocessed Trevi dataset from PhotoTourism can be found here. Put the downloaded folders into data/
sub-folder in the code directory.
See NeRF++ sections on data and COLMAP on how to create adapt a new dataset for training. In addition, we also support masking via adding mask
directory with monochrome masks alongside rgb
directory in the dataset folder. For more details, refer to the provided datasets.
We provide pre-trained models here. Put the folders into logs/
sub-directory. Use the scripts from scripts/
subfolder for testing.
conda env create --file environment.yml
conda activate nerfosr
Use the scripts from scripts/
subfolder for training and testing.
Please find precompiled binaries, source code, and the extracted Site 1 mesh from here.
To run the demo, make sure you have an OpenVR runtime such as SteamVR and launch run.bat
in hellovr_opengl
directory.
To extract the mesh from another model, run
ddp_mesh_nerf.py --config lk2/final.txt
The list of folder name correspondences can be found in the README of the dataset.
Note that in the VR demo executable, we also clip the model to keep only the main building on ll. 1446-1449. The bounds are hard-coded for the Site 1.
To recompile the code, refer to OpenVR instructions, as the demo is based on one of the samples.
Please cite our work if you use the code.
@InProceedings{rudnev2022nerfosr,
title={NeRF for Outdoor Scene Relighting},
author={Viktor Rudnev and Mohamed Elgharib and William Smith and Lingjie Liu and Vladislav Golyanik and Christian Theobalt},
booktitle={European Conference on Computer Vision (ECCV)},
year={2022}
}
Permission is hereby granted, free of charge, to any person or company obtaining a copy of this software and associated documentation files (the "Software") from the copyright holders to use the Software for any non-commercial purpose. Publication, redistribution and (re)selling of the software, of modifications, extensions, and derivates of it, and of other software containing portions of the licensed Software, are not permitted. The Copyright holder is permitted to publically disclose and advertise the use of the software by any licensee.
Packaging or distributing parts or whole of the provided software (including code, models and data) as is or as part of other software is prohibited. Commercial use of parts or whole of the provided software (including code, models and data) is strictly prohibited. Using the provided software for promotion of a commercial entity or product, or in any other manner which directly or indirectly results in commercial gains is strictly prohibited.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.