/OITAlgorithm

OIT

Primary LanguageC++OtherNOASSERTION

NOTE

This program has been discontinued in favor of https://github.com/chrismile/LineVis. LineVis offers more rendering modes, has cleaner code and can be compiled on a larger range of compilers (e.g., using Microsoft Visual Studio).

OIT Rendering Tool (PixelSyncOIT)

DOI

A visualization tool for rendering triangle, line and point data sets using order independent transparency (OIT).

This tool uses OpenGL 4.5 together with pixel synchronization (GL_ARB_fragment_shader_interlock) to render these data sets. It was created for the paper "A Comparison of Rendering Techniques for 3D Line Sets with Transparency" (submitted to TVCG).

IMPORTANT NOTICE: This tool was created as a testing tool for comparing the performance, quality and memory consumption of different OIT rendering algorithms. It is not meant to be a stable and full-fledged visualization tool like ParaView. Use at your own risk!

The followings rendering algorithms are currently supported.

  • Per-pixel linked lists (LL)
  • Moment-based Order-Independent Transparency (MBOIT)
  • Multi-Layer Alpha Blending (MLAB)
  • Multi-Layer Alpha Blending using depth buckets (MLABDB)
  • K-Buffer (KB)
  • Hybrid Transparency (HT)
  • Weighted Blended Order-Independent Transparency (WBOIT)
  • Depth Peeling (DP)
  • Voxel Ray Casting (VRC)
  • Ray tracing using OSPRay (triangles, lines using Embree, generalized tube primitives, see https://github.com/MengjiaoH/ospray-module-tubes)

The .glsl shaders of all algorithms can be found in 'Data/Shaders'.

Prerequisites (build currently only supported on Linux):

  • sgl: https://github.com/chrismile/sgl (recommended to use sudo make install to install this library on your system)
  • The datasets are not supplied with the repository. Currently, to render all datasets, the following files are needed:
    • Rings: Data/Rings/rings.obj
    • Aneurysm: Data/Trajectories/9213_streamlines.obj
    • Convection Rolls: Data/ConvectionRolls/output.obj
    • Turbulence: Data/ConvectionRolls/turbulence80000.obj
    • Convection Rolls (Small): Data/ConvectionRolls/turbulence20000.obj
    • ...

The data sets can be downloaded in the supplemental material section. For internal use, these data sets are converted to .binmesh files (.binmesh_lines files for line data sets not converted to triangle hulls).

How to add new data sets

In src/Performance/InternalState.hpp, add the file name of the data set to MODEL_FILENAMES and the display name to MODEL_DISPLAYNAMES. In src/MainApp.cpp, the name of the data set that should be loaded at start-up can be set in startupModelName.

Currently, the program supports line and triangle data sets stored in .obj files and triangle data sets stored in .bobj files. Additionally, it has loaders for data set specific NetCDF .nc formats for lines and .xml and .bin formats for point data sets. Internally, these data sets are converted to .binmesh files (.binmesh_lines for line data sets not converted to triangle hulls).

Building and running the programm

On Ubuntu 18.04 for example, you can install all necessary packages with this command (additionally to the prerequisites required by sgl):

sudo apt-get install libnetcdf-dev netcdf-bin

After installing sgl (see above) execute in the repository directory:

mkdir build
cd build
cmake ..
make -j 4
ln -s ../Data .

(Alternatively, use 'cp -R ../Data .' to copy the Data directory instead of creating a soft link to it).

The build process was also tested on Windows 10 64-bit using MSYS2 and Mingw-w64 (http://www.msys2.org/). Using MSYS2 and Pacman, the following packages need to be installed additionally to the prerequisites required by sgl.

pacman -S mingw64/mingw-w64-x86_64-netcdf

On Windows, using MSYS2 and Mingw-w64 (http://www.msys2.org/), it is best to use the following CMake command:

cmake .. -G"MSYS Makefiles"

To run the program, execute:

export LD_LIBRARY_PATH=/usr/local/lib
./PixelSyncOIT

Ray tracing with OSPRay

If the user wants to build the program with support for ray tracing with OSPRay, USE_RAYTRACING must be set to ON when using cmake. Additonally, the user also needs to compile OSPRay with support for generalized tube primitives (see https://github.com/MengjiaoH/ospray-module-tubes).

Ray tracing with RTX

Click on the DOI link to obtain the RTX program:

DOI

RTX ray tracing is build upon a test environment separated from the main program. Users need to compile our RTX program with CMake and Visual Studio 2017, first. Note that we currently support Windows 10 only for RTX since SIMD operations are Microsoft-specific. In the future, we plan to adapt the code to support Linux systems, as well.

To set-up RTX, make sure you have the latest NVidia driver (415 or later / 440 recommended) installed and a ray-tracing capable NVidia graphics card (GTX 10xx / RTX 20xx). Load the source code (see link below) and build the project file using CMake. Vulkan headers are provided in the source code. Build the project "RTX" with Visual Studio 2017 in Release Mode and execute the program.

Supplemental Material

Data Sets

Click on the DOI link to obtain all line data sets:

DOI

Benchmark Results

https://www.in.tum.de/fileadmin/w00bws/cg/Research/Publications/2019/3D_Line_Sets_Transparency/benchmark_results.tar

This tar archive includes all performance measurements (rendering times), image quality measurements (SSIM / PSNR), depth complexity, and the used transfer functions.

Videos

https://youtu.be/bxV71bkJM4k