[Soc-2011-dev] Weekly Status Report 1 - Mocap workflow - Pepper

[Benjy Cook]

This week:
Retargeting - Basic rotation copy is coded, for identical hierarchies. Still
needs work on creating "virtual bones" where hierarchies don't match up or
you have "skips" in the chains. Solution is straightforward: Interpolate if
user denotes match from 2 (or more) bones to 1, or extrapolate from 1 bones
to 2 or more (using IK to determine mid joint position).
Other tools - 90% done. Simplify curves is complete, using a robust
algorithm to fit beziers to a sampled fcurve, within a given error range.
Places keyframes on the same frame of all components of a bone's
rotation/location/scale.
Loop detection needs a little work, results are not totally satisfactory.
This will be revisited once basic retarget work progresses more, as it
relates to fixing footskate and stride bones.

Next Week:
Finish basic retarget algorithm, with support of varied cases such as
hierarchy mismatch, strange bone configurations (child chains that only
partially inherit rotations, etc.)

Other issues:
Keeping the communication lines open between Joshua (Aligorith) and I will
be important during coding. So far IRC and keeping a close eye on commits is
working well. He mentioned he will be working on the FSamples API that will
be relevant to parts of my project. Also, a "bake constraints" functionality
is critical for the next stage in my project, we need to decide how it
should be coded (C or Python Operator, etc.) and by whom. I want to be able
to leverage Joshua's improvements as best I can in my code. The creation of
Pepper branch really helps in this respect.

Have a good weekend!
Benjy
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.blender.org/pipermail/soc-2011-dev/attachments/20110527/a82e1ad5/attachment.htm