lat/lon origin question

[wolfsnipes]

Do the lat/lon arguments describe the origin (0,0 point) in the obj file, or somehow the interpolated center coordinate? If it is the center, I would appreciate knowing how the 'center' is defined so I can provide the lat/lon values appropriately.
Thanks!

[HeDo88TH]

--lat/--lon represent the origin (0,0,0) of the OBJ file, not the computed center of the model.

In other words, they describe which point on the globe corresponds to the point (0, 0, 0) in the OBJ coordinate system.

1. The GPS coordinates are converted into an ENU-to-ECEF transformation matrix which is applied as the transform of the root node in tileset.json.
2. The bounding volumes of each tile use the OBJ vertex coordinates directly, without any re-centering.
3. There is no centroid computation or model offset: the OBJ geometry is placed "as-is" in the ENU frame anchored at the given (lat, lon, alt) point.

if the geometry in your OBJ is not centered at the origin but, for example, around (100, 50, 0), the model will appear shifted ~100 m East and ~50 m North from the specified GPS position. To get correct placement you should either:

• Provide the lat/lon corresponding to the origin (0,0,0) of the coordinate system used to generate the OBJ (typically the local reference frame), or
• Pre-process the OBJ to translate it so its content is centered at the origin before conversion.

[wolfsnipes]

Ok that's great thanks for the response! For an alternate workflow, I want to represent data in a projected/planar coordinate system (not ECEF). Is it safe to just use lat/lon = 0/0 and then use the transform matrix to manually offset the model, in a way that makes sense with my coordinate system?

[HeDo88TH]

Ok that's great thanks for the response! For an alternate workflow, I want to represent data in a projected/planar coordinate system (not ECEF). Is it safe to just use lat/lon = 0/0 and then use the transform matrix to manually offset the model, in a way that makes sense with my coordinate system?

Yes