> What is the status on Krystal atm? How many animations are done and
> what is left to be done? I think that a fully functional demo around
> Krystal
> is still something we should try to make. Perhaps even a small game.

To summarize the status of Krystal, I would say it is just a beginning :)

What we aimed with Krystal was having a good-quality test model for every
current and future development of CS & CEL, including things like general
movements and actions (run, jump, etc), interaction with environment
(inverse kinematics, physics), facial animation, LOD management, whatever.

Korbak is working on small improvements of the mesh, then he will work on
the weight map. There will be more bones in the skeleton, mainly for the
hands. The texture should be improved, and normal + specular maps
generated. We will have to study why the FPS drops down from 500 FPS to 20
FPS (on my PC) once the animations are played.

I'm working on adding the animation of the hairs using the ragdoll plugin.
Everything is almost settle down for that.

I'd also like to animate the skirt of Krystal by adding support for soft
bodies in Bullet (ie a mesh that can be deformed procedurally), but it is
not easy to determine where we can implement that in a CS mesh, inside
animeshes? Btw, are animeshes intended to replace genmeshes at long-term?

On the animation side, there is a lot to come. I first need to manage the
transitions of the animations, it will be from the Blender's script inside
the 'Krystal.blend' file (ImportMocap.py).

Then I'd like to work on more animation nodes:
- 'movement node', ie a node updating the mesh position from the animation
data
- Inverse Kinematics. I'll start with a basic CCD algorithm, then see for
other possibilities.
- custom movements (direction and speed)
- Walking on uneven terrain, using the Inverse Kinematics.

One problem will be finding animations for all base movement we want.
There are a lot of motion capture data on the web (see
http://mocap.cs.cmu.edu/search.php?subjectnumber=%&motion=% for example)
but it is not easy to use them as is. I used the package 'Female1_bvh.zip'
from http://accad.osu.edu/research/mocap/mocap_data.htm and Krystal was
made with a skeleton close to the one of this data package. Using other
mocap data will most likely need work in order to adapt it to the skeleton
(mainly constraining from Blender the skeleton to the mocap data).

Other problems with mocap data are that they need to be cleaned, like
applying selectively a low-pass filter to attenuate the noise of the mocap
process, or adjusting eg the foot going suddenly through the roof. The
data is also not adapted to real-time rendering, as the bone's parameters
are defined for each frame (24 FPS for now) while they should be optimized
through well chosen keyframes. Blender does not have functionalities for
that. Commercial software like MotionBuider and 3DS Max's Character Studio
can work on it, but the support for that kind of functionalities is
currently rather poor.



------------------------------------------------------------------------------
SOLARIS 10 is the OS for Data Centers - provides features such as DTrace,
Predictive Self Healing and Award Winning ZFS. Get Solaris 10 NOW
http://p.sf.net/sfu/solaris-dev2dev
_______________________________________________
Crystal-main mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/crystal-main
Unsubscribe: 
mailto:[email protected]?subject=unsubscribe

Reply via email to