Pete
Pete
> Bill Bollinger
> www.gsxn.com
:)
Regards,
Mark Davison
www.v8thunder.com
Then they need to be re-nailed, if what happens in gpl, etc. is the
alternative.
You see, if you have a prediction code like in Nascar Heat, you have very
smooth cars that don't seem much warpy. Unfortunately, most of that
smoothness is due to the more heavy prediction code. When it predicts, it's
not always telling the truth. In NH and in I think ISI multiplayer, the
server relies too much on predicting, giving the whole experience a
ghost-like feeling of not really driving against other human beings, but
predicting of what they will do in the next split second (that action being
replicated each next split second after). Now that gives you the advantage
that since it predicts more, the solid mass of cars doesn't give you
impossible physics as much as in those Papy multiplayer crashes.
Because we need to remember that the graphic engine of those cars, the
physical appearance on screen has NO EFFECT on physics and crashes. It's
all done in "another reality" within the virtual reality of the sim.
Because of the more rough prediction code of Papyrus, you get a more real
sense of driving against other humans, because it predicts less. So you are
given closer to reality data of the other opponents. That unfortunately
makes it also more possible to have physics that go bezerk.
Until we start seeing physics that are done with the polygons themselves in
real-time (and not a reality within the virtual reality), things like that
will continue to happen. Only by putting "brackets" so that let's say, a
speed of a stock car cannot exceed 245mph or that it's velocity cannot
exceed X, or go below Y in a Z lapse of time, can we diminish it.
Of course when (and we already have in certain simulations) real-time
physics are done within the game engine itself, and with the advancing of
computer power, you won't have that anymore. Cars will be real entities on
the tracks, so if a collision arrives, it will be because of those polygons
hitting each other. They will not pass through each other, and the physics
won't go bezerk anymore. But those technologies are really crude compared
to the potential in the future, with real-time physics in the game engine
acting with all the entities on screen. From the wheels to even the simple
needle in the gauge or the pitroad wall, will be an entity with a mass and a
volume, not lifeless polygons that have specific hidden information like it
is nowadays. I've said a long time ago that we will have a regression of
the simulation with that transition, as it's so complicated that sims that
start using those techniques, will be seen as more crude compared to those
who predict.
I hope somebody still reads this ;)
--
-- Fran?ois Mnard <ymenard>
-- This announcement is brought to you by the Shimago-Dominguez
Corporation - helping America into the New World...
> Prediction code is, unfortunately, a two-sided knife. It cannot be
perfect,
> it can only predict. If it's too soft, it can be as worse as if it was
too
> tight. Papyrus perhaps went a little too tight in their earlier
multiplayer
> release, and have loosened up to a level in NR2003 that is very
satisfying.
> You see, if you have a prediction code like in Nascar Heat, you have very
> smooth cars that don't seem much warpy. Unfortunately, most of that
> smoothness is due to the more heavy prediction code. When it predicts,
it's
> not always telling the truth. In NH and in I think ISI multiplayer, the
> server relies too much on predicting, giving the whole experience a
> ghost-like feeling of not really driving against other human beings, but
> predicting of what they will do in the next split second (that action
being
> replicated each next split second after). Now that gives you the
advantage
> that since it predicts more, the solid mass of cars doesn't give you
> impossible physics as much as in those Papy multiplayer crashes.
> Because we need to remember that the graphic engine of those cars, the
> physical appearance on screen has NO EFFECT on physics and crashes. It's
> all done in "another reality" within the virtual reality of the sim.
> Because of the more rough prediction code of Papyrus, you get a more real
> sense of driving against other humans, because it predicts less. So you
are
> given closer to reality data of the other opponents. That unfortunately
> makes it also more possible to have physics that go bezerk.
> Until we start seeing physics that are done with the polygons themselves
in
> real-time (and not a reality within the virtual reality), things like that
> will continue to happen. Only by putting "brackets" so that let's say, a
> speed of a stock car cannot exceed 245mph or that it's velocity cannot
> exceed X, or go below Y in a Z lapse of time, can we diminish it.
> Of course when (and we already have in certain simulations) real-time
> physics are done within the game engine itself, and with the advancing of
> computer power, you won't have that anymore. Cars will be real entities
on
> the tracks, so if a collision arrives, it will be because of those
polygons
> hitting each other. They will not pass through each other, and the
physics
> won't go bezerk anymore. But those technologies are really crude compared
> to the potential in the future, with real-time physics in the game engine
> acting with all the entities on screen. From the wheels to even the
simple
> needle in the gauge or the pitroad wall, will be an entity with a mass and
a
> volume, not lifeless polygons that have specific hidden information like
it
> is nowadays. I've said a long time ago that we will have a regression of
> the simulation with that transition, as it's so complicated that sims that
> start using those techniques, will be seen as more crude compared to those
> who predict.
> I hope somebody still reads this ;)
> --
> -- Fran?ois Mnard <ymenard>
> -- This announcement is brought to you by the Shimago-Dominguez
> Corporation - helping America into the New World...
Having said that - yes I too think they can improve it... just look at LFS.
--
Ed_
Bill Bollinger
www.gsxn.com
> Having said that - yes I too think they can improve it... just look at
LFS.
> --
> Ed_
> > Then why didn't N3 have the same type of wrecks then? My guess is
because
> > Charlie Heath did the net code for N3 and someone else did GPL/N4.
> > Bill Bollinger
> > www.gsxn.com