This Open Realtime Depth-Image-Based Renderer (OpenDIBR) takes a multi-view image/video dataset as input and renders the view from the perspective of the viewer to the desktop or VR display. The viewer has full freedom of movement (6DoF) using the keyboard and mouse or the VR controllers. Real-time performance is achieved through several optimizations, including the use of CUDA, OpenGL and the NVidia Video Codec SDK. For the VR experience, the OpenVR SDK and SteamVR are used.
- A C++11 capable compiler. The following options have been tested:
- Windows: Visual Studio 2019
- Ubuntu 18.04 LTS: gcc, g++
- CMake 3.7 or higher
- An NVidia GPU that supports hardware accelerated decoding for H265 4:2:0 (preferably up to 12-bit), check your GPU here
- CUDA Toolkit 10.2 or higher
- NVidia Video Codec SDK
If you want to use Virtual Reality:
- Steam and SteamVR
- The following HMDs have been tested:
- HTC Vive
- HTV Vive Pro
- Oculus Rift S
For Linux systems, there are several dependencies on libraries that need to be installed: FFmpeg, OpenGL, GLEW and SDL2. For example on Ubuntu:
sudo apt update
sudo apt install -y pkg-config libavcodec-dev libavformat-dev libavutil-dev libswresample-dev libglew-dev libsdl2-2.0
The following build instructions were tested on Windows 10 and Ubuntu 18.04 LTS:
Clone the repository:
git clone open-dibr
cd open-dibr
git submodule update --init --recursive
Use CMake to build the project:
cmake . -B bin
cmake --build bin --config Release
If the build was succesful, you now have an executable.
Read Running the Application for a detailed overview of how to run and use the application. It also has a section on Common Problems (scroll down) that you might run into.
Read Light field datasets to download existing light field datasets or adapt your own to use as input for the application.
Licensed datasets provided by the MPEG-I community:
Datasets provided by IDLAB:
A paper on this software tool is currently being written/evaluated. Instructions on how to replicate the results from the paper can be found here.
Our paper was published in Multimedia Tools and Applications (MTAP) by Springer. The read-only published version can be found online as part of the Springer Nature Content Sharing Initiative.
To cite the paper:
@article{opendibr2023,
author = {Artois, Julie and Courteaux, Martijn and Wallendael, Glenn and Lambert, Peter},
year = {2023},
month = {08},
pages = {25797--25815},
volume = {83},
title = {OpenDIBR: Open Real-Time Depth-Image-Based renderer of light field videos for VR},
journal = {Multimedia Tools and Applications},
publisher = {Springer},
doi = {10.1007/s11042-023-16250-8}
}
Dataset and paper by IDLab MEDIA.