This is a short outline on how to convert point cloud data into 3D PDF, where the 3D point cloud is shown directly in a viewport. Point clouds may be imported from PTS, CSV, XYZ, LAS and several other types. Any point cloud input should have a minimum of 3 channels (columns) for X, Y, Z coordinates per point. Additional channels (columns) may be optionally used for attributes such as intensity (i) or color (RGB).
Initially, to get a good result in a short time for review, it is recommended to set the subsample rate to achieve 1M or 500K point count after decimation. The Subsample parameter selects every Nth value in the point list for processing, ignoring the points between those selected. Once a good result is achieved, then the Subsample value can be lowered, to create higher fidelity outputs. In broad terms, for 1M points, a file size of 12 - 15MiB should be expected. The Subsample parameter is on the Gridding tab.
The initial assumption is that all X,Y,Z values are in meters. If XY are in Km or degrees, and Z is in meters, or some other set of units, this can be specified on the Geospatial tab by selecting the "Input Units" section.
On the Gridding tab, the default value for Interpolation is "IDW". This should be removed and switched to "Point Cloud". If left in IDW mode, a low-resolution terrain surface reconstruction step will be generated. In "Point Cloud" mode, this is skipped so the points are shown directly without any surface reconstruction.
Use the Interface Settings Color Mapping and Elevation tab to browse to and specify a colormap file. Then adjust the "Binding Mode" to "Along Z", "With Attribute". If the point cloud already contains full color, then don't select any colormap file and set "Color RGB".
By using just "Subsample", the size of the point cloud can be quickly reduced, however this ignores shapes and attributes. A more accurate method is to enable Advanced tab Simplification, and set a value for the number of points desired in the output. This uses a careful mathematical statistical method to select the most important subset of points to represent the shape and features within the point cloud. This can be used in combination, for instance to use Subsample to reduce the considered points in half (value=2), then reduce further by enabling simplification.