This repo is a library that provides the ability to store lidar point clouds in a GeoPackage. It includes a PDAL-based Reader and Writer.

The repo also includes a simple TMS-like tile server for use with the Rialto viewer ([https://github.com/radiantbluetechnologies/rialto-cesium]).

Rialto is an open source project by RadiantBlue Technologies.

GeoPackage does not support point clouds, so we have made a simple (and as yet unofficial) extension based on the tile-based raster model it does support.

The most recent description of the changes made for the extension are in these slides: [http://www.slideshare.net/mpgx/geo-package-pointclouds].

SCons

• to build: $ scons [debug=1] pdal_prefix=FOO install_prefix=BAR
• to install: $ scons [debug=1] pdal_prefix=FOO install_prefix=BAR install
• to clean build: $ scons [debug=1] pdal_prefix=FOO install_prefix=BAR -c
• to clean install: $ scons [debug=1] pdal_prefix=FOO install_prefix=BAR -c install
• to test: $ BAR/test/rialto_test

Make

IMPORTANT: For anything later than the 0.1 branch, the preferred build system is to use SCons, not make.

$ make all $ export DYLD_LIBRARY_PATH=...as needed... $ make test

The top-level Makefile will:

• build the librialto.so library
• build the unit tests for the library
• build a simple command-line tool

You will need to edit defs.mk to point to your preferred install location, then just do make all ; make install.

PDAL

Rialto requires the PDAL libraries ([https://github.com/PDAL/PDAL]). We normally build against a fork of PDAL at https://github.com/radiantbluetechnologies/PDAL, but the main repo should work fine as well (as of late June 2015).

Docker

Docker Hub is used to build a new container for Rialto on every push:

https://registry.hub.docker.com/u/radiantbluetechnologies/rialto-geopackage-tools/

Running

The command line app rialto_translate, can be used to create a geopackage file:

$ rialto_translate -i input.las -o output.gpkg

The app rialto_info will print information about a geopackage file:

$ rialto_info foo.gpkg

Use -h for additional options to these tools.

The public headers are in include/rialto.

No documentation exists at this point. However, see the unit tests or the command line app for example usage.

A Python script, geopackage_server.py, is provided to serve up tiles from a GeoPackage:

$ geopackage_server.py localhost 12346 somedir

where localhost is the local host, 12346 is the port to use, and somedir is the directory containing the data you wish to serve.

The server responds to these GET requests:

GET /              # returns a JSON list of the geopackages in the server's directory (e.g. `somedir`)
GET /file          # returns a JSON list of the point cloud tables in the file named `file.gpkg`.
GET /file/table    # returns a JSON object with the header info for the point cloud dataset named `table` in file `file.gpkg`
GET /file/table/$L/$X/$Y     # returns a blob with the tile at (level $L, column $X, row $Y), just like the TMS protocol

The tile blob is formatted as follows: * bytes 0..3: a uint32 with the number of points in this tile * bytes 4..7: a uint32 with a mask of which child tiles are available * bytes 8..n: the actual point data

The point data is laid out in the simplest fashion: a packed array of bytesm x0, y0, z0, t0, x1, y1, z1, t1, x2, ..., for however many dimensions are provided (e.g. x,y,z,t) and in the native datatype of each dimension. The dimensions are described in the JSON header info for the dataset.

The child mask follows the format used by the Cesium terrain server: [https://cesiumjs.org/data-and-assets/terrain/formats/heightmap-1.0.html].

With the exception of the prepended 8 bytes for the point count and the child mask, the blob format is exactly the same as how the tile is stored.

NOTE: the tile server requires (and assumes) that the dataset being served is in 4326 and has a root tile matrix of two columns by one row (again following the model of the Cesium terrain server).

Confused, baffled, or puzzled? Feel free to submit an Issue at [https://github.com/radiantbluetechnologies/rialto-geopackage/issues] or just email me at mpg@flaxen.com.

-mpg