The "classic mode" is technically pretty simple. We save the position and rotation of every bones, and we interpolate between each position. (This interpolation still needs some work)
The obvious problem with that, is that the edition of an existing animation is really fastidious. Basically if you just want to modify the foot positions, you need to edit every keys. You just have to import a 30 sec. mocap animation to see how painful it could be.
The "Delta mode" tries to solve this problem. So this layer is a bit more complexe. In this layer, we want to save every bones "individually". Since the shootman skeleton has 42 bones, do we really want to have 42 different editable tracks?
The solution we opted for is to only have 5 editable tracks : you can only use the delta mode on 5 bones : both hands, both feet, and the hips. Those 5 tracks are the more important ones, they are the one we can use our Inverse Kinematics system on(
http://en.wikipedia.org/wiki/Inverse_kinematics). Basically when you move the hand, the whole arm adapts.
It is now pretty easy to use the "Delta Mode". We move the hand on top of the "classic layer" and the arms adapt itself.
In conclusion, if the delta mode only works on the hands, the feet and the hips of the shootman, it's because we thought of it that way, it's not a bug.
Having said that, we realise that the "delta mode" is still hard to understand and use efficiently, so we may modify it in the future, but it will take a lot of work and time, because we might have to rethink the whole "delta mode".
, and it helps to correct/improve animation, so we avoid in deleting several keys and remake the animation back.
