Support edges input as constraints

[kannes]

It would be great if the triangulation could support user-supplied vector lines as constraints for the triangulation. These are sometimes available as high quality lines gathered in surveying.

For example for Hamburg, Germany there is a DEM as regular raster: http://suche.transparenz.hamburg.de/dataset/digitales-hohenmodell-hamburg-dgm-15

And an accompanying "Bruchkanten" file where specific known "hard" edges in the terrain were drawn: http://suche.transparenz.hamburg.de/dataset/bruchkanten-hamburg8?forceWeb=true (eww, dxf and lots of layers, I mean the B_Boeschungen_curve_ here).

Those edges are more realistic than what any algorithm could ever dream of from a regular raster.

[delfrrr]

Sounds interesting to me 👍 Thanks!

@kannes can you, please, explain or send a link explaining how these "structural lines" are created (mapped)? Do you think vector map (buildings for example) can be combined with DSM to produce more useful datasets?

Also, I'd appreciate if you could explain your use case of using tin-terrain

[kannes]

Use case is mostly having hobbyist fun visualising geodata but also dabbling with web and VR products where this tool would give much nicer results than "dumb two tris per raster cell" approaches. ;)

From what I know those lines usually come from traditional land surveying. They basically say, regardless what the raster says, there is a edge going through the world exactly here. Commonly a part of "Topografisches Aufmaß", see e.g. https://www.google.de/search?q=topografie+aufmaß+bruchkanten . Also just spotted this interested related work https://www.tuwien.ac.at/aktuelles/news_detail/article/3370/
Sorry about the all German stuff, just ask some colleagues in doubt. :)