Sustainability of Digital Formats: Planning for Library of Congress Collections

The XYZ Point Cloud (XYZ) format is a simple, text-based data dictionary format for recording points in 3D space, or point cloud data. Stored values are aligned with the Cartesian coordinate system ('x','y', and 'z'). These values can be transformed and visualized as 3D objects, as described by the FME software documentation.

The FME documentation defines XYZ files as ASCII database files in which each column in a row is delimited by a separator character (like a space, comma, or tab), each row represents a point within a point cloud, and each column represents a component for that point. The Automapki software documentation notes that files may optionally include 3D normal vector coordinates for each point.

XYZ files have simple but flexible guidelines, which can create compatibility issues. From a post from Vercator, "There are no unit standardisations for XYZ files. Although there is wide compatibility across programs for this type of file, the lack of standardisation surrounding units and specifications makes it a fundamentally faulty method of data transfer unless additional information is supplied."

While some see this flexibility as a hindrance, others see it as a value. From Wisconsin Institute for Discovery, "We wanted a file format that would be maximally accessible, both now and in the future. A ASCII text-based system should enable users to understand the data structure with limited documentation, and not leave the data bound to any one program."

Outlined by computational scientist and professor John Burkardt, XYZ files have the following structural restrictions:

• Comment lines begin with a "#" in column 1 and may appear anywhere.
• The coordinates of a point are written on a single line, separated by spaces.
• Blank lines may appear anywhere in the file.

Burkardt's site offers several sample XYZ files.

ROCKrobotic's FAQ page on XYZ asserts that the structure can include commas or not include commas. It doesn't include the # marks or the space between the title and the data. This variation would look like, for example:

X,Y,Z

289814.15,4320978.61,170.76

289814.64,4320978.84,170.76

XYZ files may also have additional lines that represent color (RGB values), as outlined in "PointXYZRGB Struct Reference" (link via Internet Archive). This format is used by Wisconsin Institute for Discovery, which documents the pattern like so:

PositionX PositionY PositionZ ColorR ColorG ColorB

Wisconsin adds, "[PositionX], [PositionY] and [PositionZ] are all described as floating point numbers, while the colors [ColorR], [ColorG] and [ColorB] are unsigned bytes described as values from 0-255."

These files would look like this:

0.792 9.978 12.769 234 220 209

0.792 9.978 12.768 242 228 217

0.794 9.978 12.771 241 227 214

The above example is taken from this StackOverflow question.

All of these above are valid structures for this format.

Software commonly used with these files are CloudCompare and MeshLab, as noted by Wisconsin Institute for Discovery. Wisconsin also includes sample C++ for working with XYZ files with the XYZRGB structure.

Wisconsin Institute for Discovery recommends that the format be transferred to another type for being able to read/write faster: "While you could use these directly, you would likely want to convert these files into another format. While text files require the computer to parse these files into points, binary files enable the possibility for these files to be read directly."