[ODE] Natural motion

[Kenneth Holmlund]

I think the papers of van de Panne et al. might be useful.
See e.g. http://www.cs.ucla.edu/~pfal/papers/siggraph2001.pdf
They used SDFast for the physics, and their other software can be
downloaded (Dance + one of the free SVM libraries around, try Google).
The papers are really useful since they list lots of parameters taken
from the biomehanics literature.
To take this to the level of NaturalMotion (they use different methods)
requires a good bit of work though, and probably also some work
on specialized constraints in ODE.

Cheers,
/Kenneth

VRlab, Umeå University, Sweden
holmlund@hpc2n.umu.se

Russ Smith wrote:

>>I would like to know if any of you know a bit about NaturalMotion and
>>their Active Character Technology. It sounds rather convincing,
>>impressive, promising and a whole range of wonderful things.
>>    
>>
>
>i just took a look at their demo movies. i think it's mostly kinematically
>specified poses that are used to set target positions for joints in a
>dynamical skeleton. in other words, this is only slightly different from a
>floppy rag doll. in particular, i see no evidence that the virtual
>characters react to their environment (i.e. no "AI" at all) - it's telling
>that most of the demos consist of the characters moving to pre-specified
>poses and then falling down.
>
>the graphics are very nice though, and i think it's easy to mistake good
>graphics for good AI (because the character already looks "more real").
>the stuff on their web site about how this is "new" don't really hold up
>from what i have seen. my conclusion: nothing too interesting here yet.
>
>it would be *quite* easy to do the same thing in ODE. will someone take up
>the challenge?
>
>russ.
>
>  
>