Auratons / renderer_ray_marching

OpenGL-based ray marching pointcloud renderer

Pointcloud Renderer

OpenGL-based ray marching pointcloud renderer.

The renderer models points of the point cloud as spheres with diameters defined through the nearest neighbor. For the k-NN problem solution, used also for view frustum culling, a fixed GPU implementation from the FLANN project is used. The radii get stored in a special file with .radii suffix saved alongside of the input PLY file so that it can be caches and later reused to shorten startup time. The renderer features headless rendering mode, an option to load a general point cloud, render specific views stored in a file and also a windows FPS camera mode.

The repository contains git submodules, so either clone the repository with --recurse-submodules option or inside of the folder run git submodule init && git subbmodule update --recursive.

Pointcloud Renderer Usage: marcher.bin [OPTIONS]

Options: -h,--help Print this help message and exit -f,--file TEXT Path to PLY pointcloud to render -m,--matrices TEXT Path to view matrices json for which to render pointcloud in case of headless rendering -o,--output_path TEXT Path where to store renders in case of headless rendering -s,--max_points INT Take exact number of points from the PLY file -r,--max_radius FLOAT Filter possible outliers in radii file by settings max radius -d,--headless Run headlessly without a window -p,--precompute-radii Precompute radii even if already precomputed -i,--ignore_existing Ignore existing renders and forcefully rewrite them

For headless rendering on a multi-gpu machine, NVIDIA drivers may prevent running the application on other than GPU0 with a cryptic EGL error. It is a bug of the driver, not this application.

The format of camera paramater matrices json file:

{
  "train": {
    "<relative path against --output_path or an absolute path to the output png>": {
      "calibration_mat": [
        [
          803.802316909057,
          0.0,
          533.5,
          0.0
        ],
        [
          0.0,
          803.802316909057,
          345.0,
          0.0
        ],
        [
          0.0,
          0.0,
          1.0,
          0.0
        ],
        [
          0.0,
          0.0,
          0.0,
          1.0
        ]
      ],
      "camera_pose": [
        [
          0.999810651632215,
          0.005047574637627222,
          0.01879316027290225,
          0.16393394238855838
        ],
        [
          0.006089153405678126,
          -0.9984242270151509,
          -0.05578516935530967,
          0.25632920943896725
        ],
        [
          0.018481966712650667,
          0.055889040960424845,
          -0.9982659124737038,
          1.0211781008610987
        ],
        [
          0.0,
          0.0,
          0.0,
          1.0
        ]
      ]
    }
  },
  "val": {}
}