Viewer: https://tubvsig-so2sat-vm1.srv.mwn.de/ (might already be hugged to death? When it works you get a heatmap when zoomed out, and 3d models with flat roofs when you zoom in far enough)
Looks pretty cool. Obviously only covers above-ground buildings or above-ground portions of buildings. Also seems like it tends to view buildings built wall-to-wall next to each other as the same building, but not always. So if you calculate something like average building volume you are bound to be off quite a bit
For sure not all buildings. My first query was "Costa del Este, Juan Díaz, Distrito de Panamá, Panamá Province, Panama" and no coverage. This region is packed with high rises.
So here is finally the data that I needed for my idea: walk directions from A to B while never leaving the shadows, suggesting bars and pubs where to wait while the shadows catch up and let you cross “safely” where before was sunny.
Useful for scorching weather places like south of Spain.
Currently building something like this for my city. One major challenge I've come across so far is that most APIs will give you coordinates for an address of a place within a building but none of the free, paid or crowdsourced options have reliable information on outside seating polygons. Of course, you could always display the places around you with live shadow data on a map, leaving it up to the user to zoom in and decide based on the satellite image whether the restaurant or café offers outside seating. But to plot the route and then suggest nearby sun/shadow seating options to the user, you would need this information.
You'd need stereo fotos from a low flying plane for this. Several cities are doing this for decades already. I've built such a 3D map for my city like 25 years ago.
Are you referring to the outside seating polygons? Wouldn’t these stereo images still have a lot of noise (trees, cars, trucks, smaller non-building structures) obstructing the target areas?
They've used a vision transformer to estimate building heights from monocular aerial photographs, so they're guesses at best. Calling this a map is a stretch.
What is the crowd process to upgrade/correct this? I notice massive errors in small Midwest US towns (multiple buildings being combined/ommitted, etc)?
The height measure when you click on a building is not very accurate. Clicked on two next to each other (one single floor and one with two floors) and the single story house was marked taller by the site.
Can confirm. My garage is marked double the height of my actual house. Both are incorrect. My shop building on my property, next to the garage is actually missing.
The positioning of the shapes in relation to each other are also wrong. And not in a subtle way.
Looked at the house my father built in 1985. A few hundred kilometers from my current location. The shape is wrong (as are the shapes of all neighboring buildings. As are the positions toward each other (distance between houses, rotation of shapes). The heights are also significantly wrong. The two story houses on the opposite side of the road are said to be slightly above 2 meters in height.
I've had a project idea in the back of my mind for a long time: take all of my locations that I've saved in Swarm (https://swarmapp.com) and 3D render all of the buildings that I've been to. The issue was a lack of models of the actual buildings - which is here! Maybe it is possible...
Revealing: third-world areas look different. Curiously, Australia resembles third-world building density more than it resembles the industrialized world.
Alphabet, please open your mapping data. I understand it’s your IP, but given that the value in Google Maps far exceeds the mapping data and won’t be replaced, think of all of the good that would come of it. For one thing, you wouldn’t have to worry about people posting inaccurate maps like these.
28 comments
[ 2.7 ms ] story [ 50.4 ms ] threadHow does someone screw up CSS so completely?
Viewer: https://tubvsig-so2sat-vm1.srv.mwn.de/ (might already be hugged to death? When it works you get a heatmap when zoomed out, and 3d models with flat roofs when you zoom in far enough)
Looks pretty cool. Obviously only covers above-ground buildings or above-ground portions of buildings. Also seems like it tends to view buildings built wall-to-wall next to each other as the same building, but not always. So if you calculate something like average building volume you are bound to be off quite a bit
Useful for scorching weather places like south of Spain.
https://ai.google/earth-ai/
Has anyone tried this? I’m not quite sure how to prompt it effectively.
The positioning of the shapes in relation to each other are also wrong. And not in a subtle way.
Looked at the house my father built in 1985. A few hundred kilometers from my current location. The shape is wrong (as are the shapes of all neighboring buildings. As are the positions toward each other (distance between houses, rotation of shapes). The heights are also significantly wrong. The two story houses on the opposite side of the road are said to be slightly above 2 meters in height.
I use the monthly release of Overture Maps building dataset and the last one had around 2.3 billion buildings iirc
[1] https://docs.overturemaps.org/blog/2025/11/19/release-notes/