[Bf-committers] blender2.5's new IK

[Benoit Bolsee]

On Tue, 01 Sep 2009 11:06:40 +1000, Matt Ebb <matt at mke3.net> wrote:
[...] 
> I hear in this thread about problems in the existing IK system,
> but in my experience (animation) I've never seen problems in
> practice, nor have I heard our animators/rigger complain about
> it either. 

Here is an example of the 'flipping' and performance problem of the old
solver:
http://graphicall.org/ftp/ben2610/iksolver_justin.avi
This example is a bit extreme since the entire armature is controlled by
IK but you will see that the robot is moving slowly (CPU 100%) and
eratically.

Here is now the exact same animation running with itasc (nothing is
changed except the solver):
http://graphicall.org/ftp/ben2610/itasc_justin.avi
It plays in real time (CPU %5) and the movements are smooth and natural.

To achieve the same result with the old solver, you would have to add a
lot of keyframes or secondary targets, or control a large part of the
armature manually.

This proves the superiority of the simulation approach in certain
situations. 

> I understand it may not work well for robotics,
> and it's good to make life easier for people working in this area.
> However robotics is not blender's primary target, animation is,
> so if there are to be any changes they should result in something
> at least as usable for animators as the previous system.

This project is targetting at both simulation and animation. The last
thing we want to say is that it's only for robotics applications. If a
new mode is needed for animators, then we'll just add it.  

/benoit