Looking at your code snippet again, I noticed that you set the glow's rotation matrix to the one of the object. I'm not sure if that's a good idea, because it sets the exact some instance. I might be better to make one the child of the other or do a cloneMatrix() on that rotation matrix before setting it. Maybe that helps.

Maybe because the transformation as a whole is different (some additional rotation involved)? Try to offset the glow by a scaled version of the results from get?Axis() instead.

Must be something else then. The controller works in object space. I don't see how this code snippet alone can cause wrong scaling unless the object is somehow rotated around Z and your Z-scaling should actually be Y.

I think that I don't understand your actual problem. The idea is: You create a new sub sequence, add all the keyframes, continue with the next one. If you animate a sequence, it's limited to that sequence's animations unless you animate 0, which includes all keyframes. What's the actual issue with that?

That's just some viewer application that displays an existing texture. I haven't done anything to it in terms of merging.

Then something isn't identical between the meshes no matter what. If the vertex count would match, they would at least make it all into the animation. However, even then, the usage of these vertices in the triangles could be different as well, which is something that a keyframe animation can't deal with. The assigments of vertices to triangles have to stay the same in all files.

No, they aren't. RL_Face002 for example differs from frame to frame by +-1 vertex in that log file.

Try to set http://www.jpct.net/doc/com/threed/jpct/Loader.html#setVertexOptimization(boolean) to false before loading the files.