[ODE] Documentation about physics simulation libraries

[Kenneth Bodin]

Hi,
I do not agree that quality of engines is a subjective thing. However, 
the metric is
multi-dimensional and the relevance is context dependent, and therefore it
is hard to summarize. The main reason for subjective measures of 
"plausibility" is
that noone is allocting enough time to do some serious and systematic 
testing.

We have done some comparisons of various physics engines. The ranking 
was done
from a specific perspective (choosing an engine for user integration 
with Virtools, to be
used in VR projects at the Interactive Institute), but some of the tests 
were carefully
choosen to measure quality of the simulation engines, their models, 
solvers and overall behaviour.

http://www.cs.umu.se/education/examina/Rapporter/SeuglingRolin.pdf

The Msc thesis is just a first step, but we plan to do more work on this 
during 2007, though
not really at the engine level, but rather at the algorithmic/model 
level. However, it should be fairly easy
to test further also at the engine level since e.g. Collada Physics 
makes it possible to port simulation
scenes between engines without too much work.

/Kenneth


Kenneth Bodin Holmlund
VRlab/HPC2N, Umeå University, Sweden



ode-request at q12.org skrev:
> There is some physics engine list here:
> http://www.diku.dk/hjemmesider/ansatte/kenny/other.html 
>
> It would be interesting to have some feature and performance comparison, but 
> that tends to be subjective. In other words, there are always ways to make 
> one Physics SDK look better then another. 
>
>   


-- 
Kenneth Bodin, VRlab/HPC2N
Umeå University, Sweden
bodin at hpc2n.umu.se

Philosophy of science is about as useful
to scientists as ornithology is to birds.
-- Richard P. Feynman