Skip to content

ShaderPointCloud

Renders ShaderPointClouds in the format of an RGB-D image, where the image is split into the rgb information and the depth encoded information. Works best with Space-Stream, a python based app that takes live feed from the most common 3d cameras and transfers it into the rgb-d format. Space-Stream can stream the video directly via SpoutReceiver or SyphonReceiver to a SPARCK project.

RGB-D

ShaderPointCloud Node

Live 3D Camera Integration

ShaderPointCloud renders real-time point clouds from depth cameras (Intel RealSense, Azure Kinect, etc.) directly in SPARCK. Combined with Space-Stream, you can capture live 3D environments and integrate them into your projection mapping or Spatial Augmented Reality installations.

Reference

The following properties can be configured for this node:

Property Type Description
drawto - set the render group. Capture/Beamer/3DViewer have an equivalent in which you can choose which group to render.
resolution [px] resolution of the rgb-image (half the width of the received stream)
encoded - recomended format is 'uniform hue'. choose from 'linear 8bit', 'linear 16bit', 'uniform hue', 'inverse hue', 'hsv'
point size [px] size of a point in the cloud
hole detect - tries to fix holes in the point cloud
principal point - intrinsic lense setting
focal point - intrinsic lense setting
distance min - min distance from where the depth encoding starts
distance max - max distance to where the depth encoding stops
cull near - min distance from where the points are rendered
cull far - max distance to where the points are rendered
frontface draw..
  • polygons - only faces
  • lines - only lines
  • points - only vertices
    		•
    backface draw..
  • polygons - only faces
  • lines - only lines
  • points - only vertices
    		•
    cullface - choose cullface:
  • 0=Off
  • 1=Back
  • 2=Front
    		•
    parent - parent transformation node
    position (local transformation) position x y z
    rotation (local transformation) rotation x y z
    scale (local transformation) scale x y z
    Inlet Type Description
    properties properties properties | use message [set <propertyPath> <value(s)>] (without node/<nodeName> at the beginning) to set internal properties
    texture texture texture one (also used for shader)
    Outlet Type Description

    RGB-D Format

    Understanding the Input Format

    The RGB-D format combines color and depth information in a single texture:

    • Right half: RGB color image
    • Left half: Depth encoded as color values

    The resolution property should be set to half the width of the incoming stream (the width of just the RGB portion).

    Depth Encoding Modes

    Choosing an Encoding Format

    The encoded property determines how depth values are decoded from the color data:

    Mode Description Best For
    uniform hue Recommended default — depth encoded as hue values with uniform distribution Most use cases
    linear 8bit Simple linear mapping, 8-bit precision Low depth range
    linear 16bit Linear mapping with 16-bit precision High precision needs
    inverse hue Inverted hue encoding Special cases
    hsv Full HSV color space encoding Maximum precision

    The encoding must match what your depth camera or Space-Stream is outputting.

    Space-Stream Setup

    Complete Workflow

    To stream point clouds from a depth camera to SPARCK:

    1. Install Space-Stream (Python-based application)
    2. Connect your depth camera (RealSense, Azure Kinect, etc.)
    3. In Space-Stream, select uniform hue encoding
    4. Disable 'normalized intrinsics' in Space-Stream UI to get absolute values
    5. Start streaming via Spout (Windows) or Syphon (macOS)
    6. In SPARCK, create a SpoutReceiver or SyphonReceiver
    7. Connect the receiver output to ShaderPointCloud's texture inlet
    8. Match all intrinsic parameters (see Parameter Matching below)

    Depth Culling

    Controlling Visible Range

    Two sets of distance controls allow fine-tuning:

    Encoding Range (must match source):

    • distance min: Where depth encoding starts (near plane of encoding)
    • distance max: Where depth encoding ends (far plane of encoding)

    Render Culling (can be adjusted freely):

    • cull near: Minimum distance to render points (hides close objects)
    • cull far: Maximum distance to render points (hides distant objects)

    Use culling to remove unwanted parts of the point cloud (floors, ceilings, background walls).

    Point Rendering

    Visual Quality Settings

    • point size: Larger values fill gaps but reduce detail; smaller values show more detail but may have visible gaps
    • hole detect: Attempts to fill holes in the point cloud — enable for cleaner surfaces
    • frontface/backface: Control how points are rendered (polygons, lines, or points)
    • cullface: Hide front or back faces to improve performance or visual clarity

    Important Notes

    Parameter Matching

    To have the best results receiving point clouds encoded in the RGB-D format, you must ensure these properties match between Space-Stream and ShaderPointCloud:

    • Resolution
    • Principal Point
    • Focal Point
    • Min Distance
    • Max Distance

    Space-Stream stores some properties normalized, while this shader takes them absolute. Get absolute values by disabling 'normalized intrinsics' in the Space-Stream UI.

    Reference

    The algorithm of this shader is based on this Intel whitepaper. A bug was detected in their math during implementation — this has been corrected in SPARCK's implementation.

    Need help or want to suggest improvements?

    Report an issue Improve the Docs

    Last updated: 2025-12-01 | Edit this page on GitHub