GEORIZON documentation
GEORIZON ingests geospatial data — panoramas, point clouds, aerial and oblique imagery, orthophotos — processes it, and serves it back through viewers and APIs.
Getting data in is the part that needs a contract. Every dataset has a dataset type, which decides how its delivery is read: what in your bucket is looked at, which pose data is read, and what is indexed from it. Each dataset type expects a specific folder layout and a specific metadata format. Deliver something it doesn't recognise and processing fails, usually with a message about a file it couldn't find.
These pages are that contract.
Where to start#
If you are preparing a delivery, read Concepts first — it covers the conventions every dataset type shares (coordinate systems, rotation angles, filename matching, how strict each dataset type is). Then read the page for your specific dataset type.
If you don't know which type your dataset is, look at its page in the web app (Dataset type), or ask your GEORIZON contact.
An assistant can drive the same delivery loop through the MCP server.
- Getting startedHow a GEORIZON dataset delivery works, from bucket credentials to processed output.
- ConceptsConventions shared across dataset types: coordinate systems, rotation angles, filename matching and sets.
- MCPConnect Claude Code, Claude Desktop, ChatGPT or Codex to GEORIZON over MCP.
- Dataset typesOne page per dataset type: the folder layout, metadata format and parameters each one expects.
- Data exportPackaging data GEORIZON holds into a downloadable archive, and what the formats give you.