GeoMagik Technology is rock solid and built from the ground up to be as performant and compact as possible, thus suitable for both mobile and desktop uses. Our Toolkit can be used as a complete and easy all-in-one solution, or modularly for advanced uses. Our core technologies are written in C for portability and speed, but we also provide an ObjC MapView and other wrappers for ease of use when developing iOS map apps.

• Our MBT/MCTS map data file format is highly flexible and optimized for size and load performance. You can store variable precision coordinate data with accuracy up to .9cm along with rich metadata and object structure. Our comprehensive APIs and command line tools make it easy to work with, and easy to produce from OSM data. Our toolkit supports OSM XML, OSM PBF, ShapeFile, and MBT/MCTS source formats and gzip, lzma, and bzip2 compression for both input and output. Our format can hold an area of any size and density. Very large datasets can be optionally broken down into seperate files that are seamlessly tied together to represent the orginal for mobile and networked uses. Even our in-memory data representation is as compact as possible, thus suitable for mobile uses (less than 50% memory usage compared to typical toolkits). We also provide up-to-date pre-processed OSM data feeds that can be used directly by your app.
  Example: On a test machine^*, the Ile-de-france extract of OSM data (209MB osm.bz2, 144MB osm.pbf) converted to our MBT format is only 51MB (mbt.gz), and fully loaded into memory for rendering in 2.3s. Even smaller filesize can be obtained by using lzma or bz2 compression at some expense to load performance. Even faster load performance can be obtained by utilizing a MCTS dataset.
• Our MagikEngine is a high quality and high performance OpenGL (ES 2.0) based realtime map rendering engine which can be used from command line as a tile and map image rendering tool, or from our MagikMap canvas toolkit for embedding in your UI. Supports advanced stylesheets, layering, and the rendering of raster data and custom elements.
  Example: On a test machine^*, we could render tiles of dense city data using a very complex stylesheet at lvl 17 in average 20ms. At the slowest levels, we were able to get 50-100ms. With a simpler stylsheet, you can easily get 20ms per tile in the worse case.

* Our test machine is a two year old MacMini.

If you are interested in licensing our technologies, please contact us.