[vos-d] thought problem 1: physics

[chris]

On 2/2/07, Peter Amstutz <tetron at interreality.org> wrote:
> That's very interesting.  Which simulation engine is this?

ode - used in python-ode code. I can provide code and install
instructions if u like.

>
> Also, to reiterate my previous email, would fixed-point math improve the
> situation any?
>
see reply to last post.

> I suppose one thing to keep in mind is rigid body simulation for the
> purposes of games just needs to be "good enough" and look reasonable.
> Practical uses of rigid body physics in games that I have seen tend to
> have quite a lot of dampening to prevent the system from flipping out.
>
yeah - but my tests of the position sensitivity are quite scary for
anything u want
to be accurate or repeatable and for mission critical stuff.

> Physics is something I haven't solved in VOS.  The current ter'angreal
> uses client-side "physics" (really just trivial graviy and collision
> detection) but proper rigid body physics (so that users can push and
> pull things, stack things up and knock them over, etc) will likely need
> to be managed by the server.
>
> On Fri, Feb 02, 2007 at 11:29:35AM +0900, chris wrote:
> > On 2/2/07, Peter Amstutz <tetron at interreality.org> wrote:
> > >Well, I assume this is a trick question.  Obviously it *should* do the
> > >same thing, but because of tiny changes in floating point precision,
> > >that isn't guaranteed, and the further you go the less likely it is to
> > >produce the same results.  Do I get a cookie?
> >
> > yup - even tho it is only 10m the difference can be surprisingly large
> > and obvious just from looking at image: out of proportion to the
> > difference in resolution at 10m (which is about 1.6 x 10^-15). See
> > attached image.
> >
> > chris
> > >
> > >
>