Field-defined records
Each PointCloud2 message declares PointField entries, including name, byte offset, datatype, and count. A reader uses that layout to find x, y, z and optional dimensions.
ROS 2 point clouds
ROS 2 PointCloud2 Viewer
A ROS 2 PointCloud2 message is a binary, field-described point cloud. WolkeViz decodes supported PointCloud2 streams from indexed MCAP recordings in the browser, helping you verify fields, geometry, timestamps, and coordinate-frame identifiers before writing or debugging a ROS 2 processing node.
At a glance
What to know before opening the file
Organized or unorganized
Width and height describe whether a cloud is arranged like an image grid or a flat collection of points. Both forms can be valid and need different assumptions in downstream processing.
Frames and time
The message header identifies the coordinate frame and timestamp. Those values matter when comparing sensor data or transforming points into a vehicle, map, or camera frame.
Practical workflow
Validate a PointCloud2 stream before coding
- Open a supported MCAP file and select the PointCloud2 topic you want to inspect.
- Confirm the reported fields, frame count, point count, and bounds match the expected sensor output.
- Move through the timeline and compare frames or accumulated views to understand motion and density.
- Use the viewer for validation, then retain the topic, frame ID, and timestamps in any ROS 2 processing pipeline.
Technical notes
Reading PointCloud2 correctly
PointCloud2 is not a fixed XYZ structure. The fields array and point_step specify the binary record layout, while is_bigendian identifies byte order. Coordinates are often float32 fields, but a producer may include additional types and names.
The header's frame_id does not transform the points by itself; it tells ROS consumers which coordinate frame the data belongs to. WolkeViz does not apply TF, odometry, SLAM, or motion compensation, so correct fusion or measurement across topics still requires the corresponding transforms and a deliberate time policy.
The is_dense flag is a useful hint about invalid values, but viewers and processing nodes should still be prepared for NaNs or missing returns where the producer permits them. Inspect the declared fields and several frames rather than assuming one visible frame represents the complete stream.
See it in use
Watch the workflow.
See WolkeViz open and inspect a ROS 2 MCAP point-cloud recording in the browser.
Questions and answers
Common questions
What is a ROS 2 PointCloud2 message?
It is the standard sensor_msgs/msg/PointCloud2 message type for binary point-cloud data. It carries a timestamped header, field layout, dimensions, and the raw point payload.
Can PointCloud2 contain RGB and intensity together?
Yes. The fields list can include x, y, z, rgb or rgba, intensity, ring, time, and other producer-defined attributes. The exact fields depend on the driver and sensor.
Why is the frame ID important?
It identifies the coordinate frame used by the points. You need the correct transform and time alignment before comparing a point cloud with other vehicle, map, or camera data.
Keep exploring