How to convert .obj to 3D TIle

3D Tiles
1

Hallo guys,
I have a layer in .obj format which I want to convert into 3D tiles. It is in ETRS89 UTM32N. I have already loaded it into cesium ion and tried to convert it with obj2tiles. But after that the layer is not in the right place but in space. How can I fix the problem and get the layer at the correct location?
cheers

2

It looks like the command line usage of Obj2Tiles includes options for specifying the longitude/latitude. If you didn’t already try that, it might be one option for achieving the right location.

For data that is uploaded to Cesium ion, you can (in many cases) use the 3D Tiles Location Editor to adjust the location and orientation of an asset.

(In doubt, it’s always possible to manually place the tileset at runtime, or manually edit the tileset JSON accordingly, but maybe these options already help to solve the issue here)

3

Thanks for the quick answer!
I have already tried that. Do you use the center of the model for this?

I have an .obj in EPSG:25832, which I want to convert into a 3D Tile. As far as I understand it, 3D Tiles uses ECEF.
I first tried to convert the .obj file using Obj2Tiles, but this resulted in the above problem. After that I tried to transform the .obj file to WGS84 first
and then converted it with Obj2Tiles. Now the model floated above the reasonably correct location, but vertically on the spot. The aim is to have the model as 3D tiles in EPSG:25832 so that it is displayed correctly where I use it.
Is this even possible?

4

I’m not an expert with the different coordinate system conventions, but know a little bit about OBJ - at least, enough to know that it does not include coordinate system information, so there are a few unknowns regarding your workflow.

But when you say

Now the model floated above the reasonably correct location, but vertically on the spot.

then it sounds like this is something that could be adjusted with the “Tileset Location Editor”. Also note that the correct vertical position may depend on factors beyond the coordinate reference system. For example, when you want to include terrain information (like the Cesium World Terrain), then the actual vertical position is somewhat independent of the theoretical height of the WGS84 ellipsoid at this point.

(If that doesn’t help: Can any of the data be shared, to get a better understanding of the issue?)