OrionAI now knows a forest is an area
Named forests, lakes and parks arrive as outlines, one county's wards load on their own, and country-wide datasets are clipped to the place you ask about. Small changes that make many more maps finish.
Ask OrionAI to "track vegetation month by month in Karura Forest" and it has to know where the forest is. Until now, looking up a named place gave a single point on the map: fine for a dam or a town, but useless for measuring a forest, a lake or a national park. OrionAI now brings in the outline of places that are areas, so the analysis runs over the place itself.
The second change is about size. Many of the most useful open datasets, such as OpenStreetMap roads, schools and markets, are published one country at a time, and a whole country can be hundreds of megabytes. "Map the roads of Nairobi" used to stop with "too large to copy". Orion now prepares these large datasets once so that it can cut out any area in seconds. The first time a dataset is used this takes a few minutes, and OrionAI tells you when that happens.
Boundaries got the same treatment. "Show the wards of Nairobi county" now brings in Nairobi's wards and nothing else, each labelled with its county, so a map of one county's sub-areas no longer starts with the whole country.
We also made OrionAI more forgiving of its own small slips in step settings, so it spends its steps on your map rather than on retries.
Read the full OrionAI 2.16.0 release notes.
Release notes: OrionAI 2.16.0