Pudding - Grat news! I'm glad you tried my idea, and even more glad that it works :)
There's an optimisation to be found here somewhere. Let me just search my brain.......
Ah yes. Lets say you have a cube. You animate the 'top' of the cube (4 verts) so that it moves 'up', slowly elongating the object over 10 frames.
Now, if you think about it, each of the verts that make up the 'top' dont need their own bone, because they share an identical animation. They could be assigned to a single bone that moves 'up' over 10 frames. Instant optimisation!
This could be taken further with a 'threshold' parameter (although the maths would be harder). In the previous example, imagine that one of the verts on the 'top' moves 0.5 of a unit further than the rest. To have the animation export at 100% accuracy you'd need to give it its own bone. However, if you could detect verts that moved in 'similar' ways (based on a unit threshold or something) you could drastically reduce the number of bones required.
As I said, the animation wouldn't be 100% accurate, but it could be a user defined option.
Also, any verts that dont move at all could be assigned to a single, non-animated bone.
Hope this helps. I've been considering all this stuff for ages (I was going to write my own exporter eventually), but since yours is so great and I'll be using it instead you're welcome to take these ideas on :)
There's an optimisation to be found here somewhere. Let me just search my brain.......
Ah yes. Lets say you have a cube. You animate the 'top' of the cube (4 verts) so that it moves 'up', slowly elongating the object over 10 frames.
Now, if you think about it, each of the verts that make up the 'top' dont need their own bone, because they share an identical animation. They could be assigned to a single bone that moves 'up' over 10 frames. Instant optimisation!
This could be taken further with a 'threshold' parameter (although the maths would be harder). In the previous example, imagine that one of the verts on the 'top' moves 0.5 of a unit further than the rest. To have the animation export at 100% accuracy you'd need to give it its own bone. However, if you could detect verts that moved in 'similar' ways (based on a unit threshold or something) you could drastically reduce the number of bones required.
As I said, the animation wouldn't be 100% accurate, but it could be a user defined option.
Also, any verts that dont move at all could be assigned to a single, non-animated bone.
Hope this helps. I've been considering all this stuff for ages (I was going to write my own exporter eventually), but since yours is so great and I'll be using it instead you're welcome to take these ideas on :)