We have followed the step-by-step tutorials for the model-derivative apis. I can use the /manifest endpoint to get the object tree. I can also create tasks to get a .obj file with geometries of some objects. There is no way to know which geometry belongs to which object.
I could potentially create a job for each object, but that would require thousands of jobs per model which seems excessively inefficient. Are we missing something? Any pointers on how to get geometry for objects while retaining the mapping between them?
The Viewer will extract the information, but as of today, we don't have a documentation for SVF, so it may get trickier to analyze the geometry from this file. You may use JavaScript, something like described in this sample or this sample.
You can extract the OBJs for each element, but may need a big number of jobs. In this case, you need to run your code for .obj files.