P-152: Virtual World Concept Update 160: Planet-sized Terrain: Part 39
The next objective is to match the 64-bit and 32-bit world coordinates together. IE the 32-bit “snapshot” that I am creating from the 64-bit world should be consistent, a player standing on the North Pole of a planet in 64-bit space, should also be standing on the North Pole in 32-bit space.
While working on this, I noticed what may be a serious issue. The alignment of the terrain in 32-bit space does not match the bounds box of the terrain. It seems to be rotated at an angle, and this is not consistent with the 64-bit space that the terrain section is translated from. There are also missing nodes in some places.
After some research, I believe that the 64-bit coordinate system is working, and the issue is with the translation itself. Lower levels of detail (longer ranges) seem to work ok, it is only at close range that the problem becomes apparent.
I am concerned that the issue might be with the fundamental subdivision algorithm. When I select a low-resolution node that is in range and subdivide it the facing of the node will be essentially inherited by the subdivided nodes. By taking a small number of high resolution nodes and subdividing them through many levels, I could be inheriting the facing of the higher level nodes, which could be causing an ever-increasing deviation with each LOD, culminating in the significant deviation seen in the final LOD. If this is the case, the issue could be very difficult to fix.

19/08/2016 , 11:48:10 PM
Torque 3D MIT – P152

19/08/2016 , 11:39:22 PM
Torque 3D MIT – P152

19/08/2016 , 11:38:39 PM
Torque 3D MIT – P152

20/08/2016 , 12:02:26 AM
Torque 3D MIT – P152

20/08/2016 , 03:57:17 AM
Torque 3D MIT – P152

20/08/2016 , 01:06:15 AM
Torque 3D MIT – P152

20/08/2016 , 12:29:48 AM
Torque 3D MIT – P152

20/08/2016 , 01:03:44 AM
Torque 3D MIT – P152

20/08/2016 , 12:05:06 AM
Torque 3D MIT – P152
