[ODE] looks like i am moving on

Anselm Hook anselm at hook.org
Wed Jan 22 16:24:01 2003


Well, I have little to do with ODE but I will make a couple of comments:

> to work something out with Novodex or Mathengine who
> seem to be open to working a deal that is shareware
> friendly - i know that Quest3D was able to integrate

Basic dynamics technology like ODE should be be open-source from all of
the vendors.  Value should come from doing things that are hard - not from
building walls around the easy stuff.  If vendors would realize that they
could reach entirely different plateau's of revenue income by agreeing on
basics - instead of building moats around indefensible properties - then
there would likely be more success all around.  It is like trying to build
a business out of writing Linux - the real value comes from the emergence
of value added services; developer support - special plugins - etc - and
is magnified by a commonwealth of shared foundations.

> I think the other thing that is drawing me away is the
> development pace of ODE - unlike OGRE which is rapidly
> developing into a fantastic 3D engine, ODE remains on
> the slow track - it is a fantastic core engine for
> physics but it is one mans project - with the success
> of mathengine, havok, etc in todays games one would
> think that an opensource physics engine would be a
> hotbed of development activity - it seems many people

ODE is pretty authoritative - there simply isn't another open source
dynamics engine with this caliber of performance or completeness.  As well
ODE is in the enviable position of being pretty much the only reasonably
complete open source dynamics implementation - unlike say the 3D domain
where it takes months of research to distinguish between all of the
available engines.

One thing that is nice too is that at least there is one open source
option for people to get their feet wet.  Up until now it was pretty
arcane...  gatewaying to commercial tools makes sense for industrial
strength apps of course.

> on this list are writing bits and pieces completely
> separately (like Pierres soft body effects) but ODE
> itself is not moving very fast - most commercial

ODE has also just moved to source-forge and that means more people will be
able to contribute more easily.

> level - it could be that physics engines are not
> mature yet

True - but it goes both ways - I think most of the games using dymamics
today basically suck;  nobody is pushing the potential of dynamics .  The
non entertainment industry has done a lot more - for example MathEngines
Tree Cutter demo is unbelievable.  I always imagined that dynamics should
be having radicalizing effects as similar as the leap from 2d to 3d...
For example since it allows games to "go off the rails" it makes it easier
to allow multiple players in complex environments - so dynamics should be
appearing more in online games.  I'm getting a bit tired of car demos
frankly... what about a car racing game with cars that have legs?  Nate's
proj and Gene's evolving robots and a couple of others do go in these
directions... and Porrasturvat is great of course.

Perhaps as Chris Hecker argues the real problem with dynamics is the lack
of control systems.  Maybe that is a good area to build out and that might
help dynamics push into more creative areas.

Or perhaps there needs to be a declarative grammer for what dynamics "is"
- to get more synergy going - and then the engine choice itself is
secondary.  Right now there is too much emphasis on the implementation.
It is a shame that you have to choose an engine at all - at the early
stages of development.  For example there is not a similar tension between
Mozilla and IE - each is reasonably useful and has reasonable differences
and benefits - but both are conformant (more or less).

 - a