suncg_basic

Basic SUNCG scene

Renders a SUNCG scene using precomputed camera poses read from file.

Usage

Execute in the BlenderProc main directory:

python run.py examples/suncg_basic/config.yaml <path to cam_pose file> <path to house.json> examples/suncg_basic/output

• examples/suncg_basic/config.yaml: path to the configuration file with pipeline configuration.
• <path to cam_pose file>: Should point to a file which describes one camera pose per line (here the output of scn2cam from the SUNCGToolbox can be used).
• <path to house.json>: Path to the house.json file of the SUNCG scene you want to render.
• examples/suncg_basic/output: path to the output directory.

Visualization

Visualize the generated data:

python scripts/visHdf5Files.py examples/suncg_basic/output/0.hdf5

Steps

• Loads a SUNCG scene: loader.SuncgLoader module.
• Loads camera positions from a given file: camera.CameraLoader module.
• Automatically adds light sources inside each room: lighting.SuncgLighting module.
• Renders semantic segmentation map: renderer.SegMapRenderer module.
• Renders rgb, distance and normals: renderer.RgbRenderer module, by using the alpha mode.
• Merges all into an .hdf5 file: writer.Hdf5Writer module.

Config file

Global settings

There are set in the main.Initializer

"config": {
  "global": {
    "output_dir": "<args:2>",
    "pixel_aspect_x": 1.333333333
  }
}

Next to setting the output directory for all modules, we also set the pixel_aspect_x parameter for all modules. This is necessary to coincide with the aspect ratio assumed by the scn2cam script which generated the camera poses.

SuncgLoader

{
  "module": "loader.SuncgLoader",
  "config": {
    "path": "<args:1>"
  }
},

This loader automatically loads a SUNCG scene/house given the corresponding house.json file. Therefore all objects specified in the given house.json file are imported and textured. The SuncgLoader also sets the category_id of each object, such that semantic segmentation maps can be rendered in a following step.

CameraLoader

{
  "module": "camera.CameraLoader",
  "config": {
    "path": "<args:0>",
    "file_format": "location rotation/value _ _ _ fov _ _",
    "source_frame": ["X", "-Z", "Y"],
    "default_cam_param": {
      "rotation": {
        "format": "forward_vec"
      }
      "fov_is_half": true
    }
  }
},

Here the cam poses from the given file are loaded. This text based file describes one camera pose per line. The file_format describes how each line should be parsed (Here we use the format used by files created by scn2cam; _ denotes values which should be skipped).

It's also necessary here to specify a different source_frame, as scn2cam does not use the same coordinate frame as Blender.

In default_cam_param we set parameters which are the same across all cam poses: We change the rotation/format. This is necessary as rotations are specified via a forward vector in the camera file. Also fov_is_half has to be activated, as SUNCG describes the FOV as the angle between forward vector and one side of the frustum, while blender assumes the FOV describes the angle between both sides of the frustum.

SuncgLighting

{
  "module": "lighting.SuncgLighting",
},

This module automatically sets light sources inside the loaded house. Therefore each window, lamp or lampshade gets an emissive material and also the ceiling is made to slowly emit light to make sure even rooms without lights or windows are not completely dark.

More examples

• suncg_with_cam_sampling: More on rendering SUNCG scenes with dynamically sampled camera poses.
• front_3d: More on rendering 3D Front scenes with sampled camera poses.
• front_3d_with_improved_mat: More on rendering 3D Front scenes with sampled camera poses and randomized textures.