Experiences with current setup

[rduivenvoorde]

Hi,

Eager to test this as geo-people in Nieuwegein, we thought to start with a small part of the city: "galecop". Below our experiences: We downloaded gml, and tried to run the commands in BGT_conversion_full_gpkg.bat one by one.

The two replacement commands in powershell ended ok, but the resulting GML was not valid anymore... in a parts of the gml, srsDimension="2"> became srsDimension=" 2\>. Not sure if this particular gml was the problem or if Windows just cannot handle such big textfiles ;-) Fixed by hand, they all looked good.

Then we thought to be clever and adjust the ogr2ogr commands to ALSO clip our gml to the exact part 'galecop' by adding a galecop.shp file and putting -clipsrc galecop.shp in the ogr2ogr commands... That resulted in a lot of warnings and errors, probably because the clipping results in non-valid or split geometries/objects?

Is it maybe possible to do this 'clip/extract' step in your software? In that case at least the input is still valid.

The input geopackages look OK in QGIS:

But we also see some holes in the data:

Then running the final 3dfier command results in an obj and json file, but trying to open the obj file in Meshlab, it opens (with material file next to it), but only on a very large scale:

But impossible to zoom in further, as then the area disappears...

If needed, I'm happy to provide the input or output material I used. Just let me know. I'm aware that this are mostly input/data issues but thought to let you know.

[rduivenvoorde]

@hugoledoux project is still being worked on yes?

Ok, little further. So many parameters.... Now we have something like this:

Any idea what the giant spikes can be? They appear not to come from the las file??

[balazsdukai]

Hi @rduivenvoorde , thank you for your detailed description.

• Its possible that the GML conversion scripts are somewhat outdated, so the best approach if you read what the script does and evaluate if it can be applied to your case, evtl. adapt it to newer tooling. Just a note that 3dfier can also convert curves to segments if you use the option

• I'm not sure what can the be problem with the clipping, quite a lot of things. 3dfier is complex enough already and to be honest I don't see why to increase the complexity by adding clipping too, since it can be done with other tools.

• Such spikes can appear when there is not enough points within a vertex' buffer to calculate a good height. You could try increasing the search radius for the vertices (see the options at http://tudelft3d.github.io/3dfier/output_options.html )