[Bf-committers] Bf-committers Digest, Vol 61, Issue 44

[Benoit Bolsee]

The new IK is definately designed for simulation: it estimates the
targets velocity, it operates in a 'true time' context, it ignores
rotation from keyframes (since it is in full charge of the bones), etc.
However, I understand that this is not what animators are looking for,
the predictibility and immediate result is much more important for them.

It's really not a problem to add a stateless operational mode to the new
solver that will provide predictable result on a frame by frame basis.
>From our conversation on IRC, I can summarize the principle of FK
keyframes as follow: the solver itself is stateless but some kind of
state is provided through bone rotation interpolation between keyframes
that are automatically inserted when the animator moves a target. I'm
not sure how to insert keyframes, but making the solver evaluate the
bone rotation and start converging from there is something I can do
easily. 

For those who are afraid of loosing things after the merge, I'm not
planning to remove the old solver. I have implemented a plugin system
that allows to choose between the two solvers on a per armature basis.
So all the existing animations will work identically since they will use
the old solver by default. Animators can continue to use the old solver
if they prefer. 

/benoit 

On Mon, 31 Aug 2009 14:53:42 +0200, Brecht Van Lommel
<brecht at blender.org> wrote:
> 
> The current system definitely has a disadvantage that it does not
> ensure consistency between frames, which may lead to flips. However,
> riggers make IK setups to avoid this problem, and if we wanted to
> solve the consistency problem there are other ways to do it without a
> cache, based on inserting FK keyframes after moving an IK targert.
> 
> In robotics of course you approach things more like a simulation, then
> you need the cache. In animation the animator needs to be in complete
> control, and will spend time anyway to get each frame looking right.
> So automatically ensuring consistency then is less important, and the
> drawbacks of such a cache system become a bigger problem.
>