Hi! I’m the author of ValTerra, a plugin for Autodesk 3ds Max used in architectural visualization. One of its features builds schematic site context from OpenStreetMap: the user draws a rectangle on a globe inside 3ds Max, and the plugin brings in building volumes (extruded by building:height or building:levels where tagged), roads as splines, and open elevation terrain, all draped over the relief. Measured on about 2 km² of downtown Montreal: 2,368 volumes and 1,865 road ways.
Question to people who know ODbL better than I do: is that attribution sufficient for this kind of use, and is there anything you would expect us to do differently? Site with a free trial: https://getvalterra.com
Please have a look at the thread on Overpass performance issues and look into using a commercial offering of Overpass services for the plugin if you haven’t yet done so.
Not sure that I am one of those, but I suspect, the final visualization (the grey picture above) must mention openstreetmap contributors to satisfy the licence, as the extract is beyond the lower bound. C.f. Licence/Attribution Guidelines - OpenStreetMap Foundation
Static images
Static images must be generally attributed the same way as interactive maps. […]
In accordance with the Substantial Guideline, static images of fewer than 100 features do not require attribution. In addition, static images of areas less than 10,000 m2 do not require attribution. Small thumbnails/icons do not require attribution.
Not sure how the overall function of your addin will be, though you need to keep in mind that the data (buildings) you download from OSM is ODbL licensed. If you add locally another building and publish it, the whole stuff will need to be ODbL licensed.
Thank you for the pointer, I have read the thread. The plugin will identify itself with a full User-Agent including a contact, stop retrying across mirrors and back off on errors; a re-import of the same area will come from a local cache. I am also looking at taking the building and road data from a bulk open source instead of live Overpass queries, so the public instances are not carrying a paid tool.
You are right, that still is well above the threshold. I have replaced both images with versions that carry the credit in the picture. For users, the plugin writes the credit into every imported object and into the import report, and the site’s attribution page says where it has to appear in a published render: https://getvalterra.com/legal/attribution
Agreed on the licence. The way I read ODbL, a render or animation made with that geometry is a Produced Work: it needs the OSM credit, but it does not put the artist’s own scene under share-alike. Handing the imported geometry itself on to someone as data is a different matter, and the attribution page on the site says exactly that and points people to their own legal advice for it. If I have that wrong, I would rather hear it here than from a client.
Let me be more specific. Let’s say an architect wants to use your plugin to render his building and surround it with our buildings and then publish the result. ODbL license would require, that his building data also needs to be released under ODbL as this is together with the OSM buildings a derivative database. The rendering result (produced work) can be made available under a different license.
Not a lawyer here. One at hand certainly nice to have. As far as I know, ODbL is viral and affects mixed databases.
I do not think that the original poster asked for advice on that. As long as processing the data into a static image is done in-house, there should not be a problem?
Say the architect also modifies a building in the background (from OpenStreetMap) by shaping the roof in his proprietary 3D-modeling software, what exactly must they share? The modified 3D-model (which is already an interpretation of our data), in some proprietary format that you can’t read without the software? An export into an open 3D-format? A reconversion into the OpenStreetMap data format?
it is likely they will give the results to their client.
Not a lawyer, but ODbL is not limited to the data format. Like just because you take the location and area and height=* and store that information as a 3DS-cube doesn’t change anything on the fact that data is ODbL. For sure you can render graphics, videos, … from this cube and share it as produced work and this just requires crediting OSM, but then you also need to share the derivative database (or a diff) as ODbL.
I’d rather see the 3D model as a produced work, similar to a 2D map made from the data. If you modify a produced work (made from an unmodified OpenStreetMap database), you do not have to share your modifications, or do you?
I would think it’s somehow similar to Garmin maps. If the planned usage is just view it like a picture, that might be a produced work, if you plan to use it as a dataset, it’s not.
So like if the architect would create one such produced work only from OSM data and a totally in-depended model of his building and then overlays the two, that might be possible.
But if you load OSM data into another software, where you can for example change the building height from 4m to 5m, I would still see this file as a derived database.
Sorry for the long silence, I was reading everything here and trying to actually understand it xD. Thank you all, this really helped me a lot. I did what you suggested on the Overpass side: one server, a proper User-Agent with a contact, it waits when the server says so, results are cached for a week and requests are capped at 4 km². On share-alike I went with where @aighes and @dieterdreist ended up: a render needs the OSM credit, and if someone edits the OSM buildings and publishes renders of that, they should offer those edits. All imported objects are tagged so they’re easy to pull out of a scene. One thing I still don’t know: if an architect wanted to give their edits back, what would you actually want to get? OSM XML, GeoJSON, something else?
Realistically, the only contribution that will find the way back to our database would be edits to our data. Integrating any export from the 3D-model back into some other format would likely mean more work than modeling it from scratch in OSM. Also in OSM we want to represent the real world, an architect in his presentation could also modify the real world (on purpose) to make it more pleasing in their rendering (completely valid, but not what we want in our database). Behing our license is the idea the data user produces something similar to what we have, i.e. a map or geo database that tries to represent the real world, e.g. if someone drew some fantasy structures into an OpenStreetMap-map, we would not want those for inclusion, although they would have to be provided according to the license text, if they were mixed in and not just layered on top.
Makes sense, thanks Martin. So no export then, that saves me building something nobody would use xD. The OSM objects stay tagged and separate from whatever the architect models on top, so nothing gets mixed in. And if someone spots a wrong height or a missing building while working, the right move is to go fix it in OSM itself, I’ll point people there.
well, it is just my interpretation of the situation. Legally you have to follow the license and its provisions, but the license does not require you to proactively deliver the modified data, only on request. What is essential is the attribution requirement, share alike of the modified data is less likely to become relevant in your context (but again, just my own personal opinion).
Strictly speaking, the derived database has to be ODbL (if the part taken from OpenStretMap is substancial), no specific format is required. Data should be accessible.
If the shape of the building is refined, then the best workflow is to refine in OpenStreetMap, then to use your stuff.
ODbL is not exactly viral. The French federation of French profesionals around OSM made a document to clarify that:
The ODbL does not only serves OSM. Even if OSM is not interested in the back-feed, others might be interested. For example the competitor of our imagined architect.
