Bad model skeletons

[VReaperV]

I've found a rather reliable reproducer for the skeleton issue, with both 0.55.3 and current master:

1. /devmap plat23
2. Build a few buidlings. In my testing, a node, ret, and arm seemed to be enough.
3. /layoutLoad test (the layout is here: https://users.unvanquished.net/~reaper/maps/layouts/plat23/test.dat)

A couple eggs (seemingly random ones) will have some broken skeletons:

With r_showSkeleton 1:

It seems to always be 2 eggs as well.

[VReaperV]

On habitat I reproduce this by just loading the map:

[slipher]

I can reproduce the Habitat one. It works correctly with 0.55.2 gamelogic.

[slipher]

I think this is probably caused by https://github.com/DaemonEngine/Daemon/pull/1532. I will try rewriting it without QuatFromMatrix as described in https://github.com/DaemonEngine/Daemon/pull/1532#issuecomment-2620050318.

[slipher]

Yep, doing it like this fixes it.

matrix_t rotmat;
MatrixFromAngles(rotmat, bmc->modelRotation[ PITCH ], bmc->modelRotation[ YAW ],
             bmc->modelRotation[ ROLL ] );
MatrixFromVectorsFLU( axisMat, ent.axis[ 0 ], ent.axis[ 1 ], ent.axis[ 2 ] );
matrix_t out;
MatrixMultiply( axisMat, rotmat, out);
MatrixToVectorsFLU( out, ent.axis[ 0 ], ent.axis[ 1 ], ent.axis[ 2 ] );

[slipher]

One of the turrets on Habitat is also affected. It appears to point in the opposite direction of the one it's really shooting in.

[slipher]

I can't reproduce the issue with the egg spam layout on the 0.55.3 release (Windows + Nvidia).

[VReaperV]

The original issue is still there.

[VReaperV]

I can't reproduce the issue with the egg spam layout on the 0.55.3 release (Windows + Nvidia).

I can reproduce this by doing:

• /devmap plat23
• /setviewpos 0 2030 44 0 20; devteam h
• Build an arm at that viewpos and each of 0 2030 44 90 20, 0 2030 44 -90 20, and 0 2030 44 180 20

[slipher]

I don't get it with the armories either.

[VReaperV]

Might be compiler-dependent or something.

I believe @enneract was able to reproduce this issue as well.