Novus Convert
Convert files
Vector & CAD format

PCD converter

A .pcd file stores a point cloud: a plain-text header declaring which fields each point carries and how they are laid out, followed by the points themselves in ASCII or in a packed binary block. There are no faces and no surfaces — a PCD is a measurement, not a model.

Private for supported formats — processed in your browser

Convert supported files

Runs on your device

Drop PCD files here

Batch files can each use a different output. Nothing uploads for local conversions.

Working inputs include camera RAW, browser-local audio/video, PDF, CBZ/CBR comics, office documents, ebooks, markup, 3D models, structured text, images, and archives.

Where PCD comes from

PCD was created for the Point Cloud Library, which Radu Rusu began at Willow Garage around 2010 while working on robot perception. Existing formats could not express the per-point attributes LiDAR and depth cameras produce — intensity, return number, RGB, surface normals — so PCL defined its own. Version 0.7 added the VIEWPOINT field and has been stable since; PCD remains the working format of most robotics and SLAM codebases.

How PCD works

  • The header is always ASCII and declares FIELDS, SIZE, TYPE, COUNT, WIDTH, HEIGHT, VIEWPOINT, POINTS and DATA.
  • TYPE is I for signed integer, U for unsigned and F for float, paired with SIZE in bytes — so F/4 is a 32-bit float.
  • DATA may be ascii, binary or binary_compressed; the compressed variant uses LZF and is not interchangeable with the others.
  • WIDTH and HEIGHT describe organised clouds from depth cameras, where the point order encodes the sensor grid; an unorganised cloud has HEIGHT 1.
  • VIEWPOINT records the sensor pose as a translation and a quaternion, which is what makes multi-scan registration possible.

When to use PCD

  • Robotics perception, SLAM and autonomous-vehicle datasets
  • LiDAR and depth-camera captures before surface reconstruction
  • 3D scanning workflows where raw measurements must be kept separate from any fitted mesh
  • Converting a scan into PLY or glTF so it can be viewed without PCL

Strengths and limitations

Strengths

  • Designed for point attributes, not retrofitted from a mesh format
  • Records sensor pose, which most point formats simply drop
  • Human-readable header even when the payload is binary, so a file can always be identified

Limitations

  • Closely tied to the PCL ecosystem; general 3D tools often cannot open it
  • The binary_compressed variant is poorly supported outside PCL
  • ASCII clouds are enormous — tens of bytes per point

Compatibility

PCL, Open3D, CloudCompare, MeshLab and ROS read PCD. No operating system previews it and no browser renders it. Novus Convert reads ASCII and uncompressed binary PCD locally and converts to PLY, GLB and glTF as genuine point primitives — not as a mesh with zero faces, which is what a naive converter produces. Colour, intensity and custom fields are not carried across in this release; positions are.

VIEWPOINT sensor pose and field declarations are read, but the point targets have no place to record them.

Convert PCD to another format