So far it only happens when parent transformNode has a mesh with children + another child transformNode. Due to losing the transform matrix in the parent node, the location of these child nodes are not in correct locations.
Strange, I tried glb option it seems working in my PG…, but not the .babylon format.
My original code is using glb and babylon, but it failed for the same symptom, then I tried to reproduce it in PG.
I might forget to try glb in the PG.
If anyone can provide some details of the cause behind the .babylon exporter. It might help me to understand why it is not working with glb in my original code.
I will try to reproduce the glb case in the PG in the meaning time.
Ok this is not a bug You cannot use abstractMeshes as they are abstracts This means they do not have all the info to be used (for instance they do not support serialization)
Thanks @Deltakosh . That explains the case in my PG.
It seems it is not the same issue with my original code (although I tried my best to reproduce it in PG).
Here is some description of the original issue i met.
The logical is similar, the difference is that the hierarchy is built dynamically with the camera. I cannot figure any fundamental difference with the PG. But I realize that some of the node has “Link to Parent” value and with “None” under Parent. I tried to re-produce it, but failed.
I understand it is quite difficult to offer any help from your side. But in case you have any clue, that would be very helpful.
Please find a sample link below. As I said the hierarchy is totally lost when I open it in sandbox. Would be very helpful if you find any clue from it.
But when i export it to .babylon or .glb, the hierarchy is lost. Most of the level transformNode is removed and all child mesh nodes parents become null.
The symptom is similar as my PG example, but may with difference cause. As there is no AbstractMeshNode used in the original case.