You just happened to cross the resident grouch.
Jan.
=---
You just happened to cross the resident grouch.
Jan.
=---
i asked for some info i remembered seeing here a while back, and ive now got
it, some people wanted to see it.
so, ill say it again.
RELAX!
we're all gplers here right?
pez
FWIW, I don't even own the game yet.
And I'll just re-iterate... it's my opinion someone casually reading this
thread could come away thinking ISI based their latest vehicle dynamics
engine on their collective imaginations. I don't know whether this is the
case or not, but indications are they, for a mainstream developer, in fact
did a very good job on the physics front, basing themselves on "real world"
data and equations. So mistakenly attributing the referred comment by the EA
Sports spokesman to them and their game is a very unfair representation,
IMO.
I think it's part of RAS' function to guide less informed netizens with
regard what is a sime an what it's not. It's therefore important we deflate
developers who exaggerate their realism claims and show support for those
who do "get it right".
The fact you, your curiosity satisfied, have neither been apologetic about
making the wrong assumption, nor have made any effort to rectify it, irks me
like hell.
I've said it before, maybe I carry a set of unusual values. So be it. Very
unfortunate for those who get on the wrong side of them and talk down to me
when challenged, but that's not to be helped.
Quit assuming things about my state of mind. It's not because I don't like
the way you go about stuff that I'm to be classified as a stressed out and
intolerant individual.
No, we're not. I still do a lot of online GPL, but I'm sure there are other
posters who either haven't touched it for years or never have at all. And
this, again, is about as irrelevant as you being Welsh and an F1 fan.
Jan.
=---
I have a hard time believing that. '
hell man, that irks me.
ill be appologetic when i do something wrong (and telling someone to shove a
post up their ass qualifies in that league i think).
pez
Simply a matter of frame of reference. Seeing as I, while I don't own the
game and, consequently, am fairly neutral about it as such, find your
assumption a serious misreprentation, your casualness about same doesn't
lend the above statement much credibility from my point of view.
Have to admit I didn't delve into the demo very deeply, but apart from the
rather curious force feedback setting, I didn't notice any of the problems
even people with the same controller as myself were reporting. So my
assumption is these are solvable by going the usual route (re-install
controller, latest drivers for everything, correct _release_ version of
directx for one's OS etc...). Ergo: someone with enough dedication to the
game can get things to work. I went through the trouble of re-wiring my
previous wheel/pedal combination after I realised only Papyrus could get
having control axis on multiple game controlllers to work properly. Wasn't
easy, but my sim horizon was broadened considerably.
So, it's ok for you to post misinformation and act like you don't care, but
it's not ok for me to tell you to check the attitude. Right, futility of
further discussion established.
Jan.
=---
I know of one small development team that started out making a racing game and
really didn't know how to go about handling the motion at all. They knew that
when you hit the brakes, the rear end should come around a little bit (or a lot
bit), so they'd come up with some code that would read the brake setting and
rotate the car according to this. Same thing would go for using a lot of
throttle. They wound up with setup screens for pre-beta testing that had
things like "braking oversteer modifier" and so on, where you'd use a slider
between -1 and 1 or something like that to get that behavior. Springs rates
had no effect on their "braking oversteer" because there weren't any springs.
Same with dampers, tires, etc.. If they wanted to have the brake bias
influence this, it would have influenced this braking oversteer modifier
directly, rather than attempt to calculate/approximate any forces acting on the
body. Basically, they'd say "if the brakes are on 20% and you're turning to
the right 50%, make it start turning to the right 60% instead."
This would be a really fudged system that isn't based on physics at all, but if
done well could be convincing...if we were still in 1982 and hadn't seen Pole
Position yet (what year was that released, anyway? I loved it when it came
out). I very much doubt that any halfway decent racing games developed over
the past five or so years have tried to use an approach like that, save for the
extremely arcadish titles that are more about shooting bad guys than driving on
the edge.
So... When using a more "proper" system where you can take an object, apply a
force to it, and it will rotate and move within 99.xxx% of how a real object
would in the real world, then start building a car simulation around this,
things are a lot better. I'm pretty sure EA's stuff is built around this very
thing. It's too complicated to get all that motion to work properly without
using a system like this, quite frankly. Why bother?
The tricky part comes in with this type of "proper" rigid body model when you
don't know what the forces are. Then you might have to make some things up and
spend time guessing how much downforce there is on a Formula 1 car, for
instance. And guessing where the centers of those forces are. I still mess
around with my Champ car and Formula 1 car in this area, doing just what you've
described. Trying to fill in the blanks not in how the physics work, but in
what the forces or inputs for certain parts of it should be. If those forces
are right, then the car will drive very close to how the real car would.
But it's very expensive to get information on how much force a certain F-1 tire
makes at 3.4 degrees slip angle, 7% slip, when the camber angle is -2.1 degrees
at that instant, the load is 1300 lb, the temperature of the tread is 200
degrees, the current air pressure is 23psi, and the tread has worn 1/8 inch.
This is the area where you have to fudge things. It's not that the code isn't
letting the car follow the laws of physics, it's that the forces are unknown to
a large degree. Of course, some models might include forces that other models
don't. GPL doesn't let the roll centers move with suspension travel, for
instance, which changes the forces on the chassis when the tire discribed above
exerts whatever force it exerts. But when playing, who would know the
difference if they hadn't driven a model that had dynamic roll centers before?
A pretty good way of finding this missing force data is by taking known
performance data and working backwards, just as you described, Dave. For
instance, I knew (or really just decided because I didn't really know) that the
top speed of my F-1 car with full downforce would be 200mph (or whatever), and
knew it made xxx power at 15,000 rpm. I'd set up the downforce and tires so it
would corner at 4.5g at 170 mph, and turn at low speed at 1.4g without the
downforce (or whatever the performance data from Car Craft or wherever said).
Then, I'd mess with the drag coefficient so the top speed would be right
(starting with approximate numbers found online and in books). After setting
up the gears, it wound up spinning the tires to about 80mph in either first or
midway through second gear, which matched up with driver's descriptions I'd
read. And so on... basically.
So yeah, there's a lot of playing around and throwing in basically random info
(in some areas) involved if you don't have access to the real life data (which
is really fun, BTW!) An 80,000 Hp Corvette is very difficult to drive, and
with enough grip, can do a backflip and land on its wheels again if you time it
right and ignore ground impacts, especially at the rear bumper ;-)
One thing about the Formula 1 cars. I haven't driven a F1 or Indycar sim since
ICR2 and World Circuit (wasn't that GP1?), but after throwing in the best
numbers I both found and "invented" where necessary as you described to get
performance data to match up to the info I found as closely as possible, I
think it's safe to say that traction control in a commercially released sim
marketed to the general public would be mandatory to a large extent!
My sim version of the Champ/F-1 car (both are the same thing except for
wheelbase, weight, and track width in my stuff, not super accurate), is not
impossible to drive, nor impossible to drive very quickly and aggressively....
when going over 90-120+ mph or so. Below this, it's tough. Not impossible,
mind you, but the car can spin very quickly. Too quickly for most people, but
probably right on par with a diehard GPLer's "make it as real as it can be"
taste. No, the car doesn't go flying off the track if gravity is not increased
100 times.
I've read comments by people (recently here, but in the past in other places as
well), that when a simulated car is done using all the math stuff out there in
the vehicle dynamics books, that they're impossible to drive. I can see where
this could happen. However, when a model is taken to that level of complexity
you really have to engineer the car at that point to get it to handle well. I
have a feeling that people tend to "chince" out on that once the model is done
(for school or whatever), and don't follow through by spending a lot of time
designing a car that handles well using the math model they've written. They
can get the equations, and they may even implement them correctly, but do they
know how to set up the car so it's driveable afterwards by Mom? Can they set
it up so Dave Henrie can rip some fast laps with the tail wagging about a bit
leaving turns 3,4 and 7?
That's two different things, and I suspect that folks that claim a full blown
model is undriveable because they've "used all the equations and tried it",
didn't know/learn how the parts were interacting with each other well enough to
design a good handling car after the model was set up. Honestly, if most of us
went in our garages and could build a car from scratch, but we had to do
EVERYTHING, suspension, engine, select tires, etc., how many of us would wind
up with even a driveable, much less competitive, vehicle at the end? No, the
physics aren't wrong, the design was lousy. Call Doug Arnao and get him to set
up your suspension for you.
Anyway, in the model itself, there are sweet spots in there where suddenly the
car seems to come alive and really start driving the way you'd expect, and yes,
with a real crazy type of model that does the suspension, RC's, and all that.
If you take any inputs too far outside these areas, suddenly the handling goes
to crap and you think it's a problem with the vehicle model. Nope.
Gregor Veble had mentioned some time ago how he'd found it interesting that
after having written his model, he suddenly found himself in the position of
being a race engineer. Indeed, just to make a car that's driveable, a good
amount of attention has to be paid to a lot of different areas.
This reminds me (this is getting long and no one is reading by now, but what
the hell, I'm enjoying myself). F-1 brakes were brought up in another
discussion. I got information that an F-1 car (probably Champ cars too) would
decelerate at over 5g at 200 mph when full brakes were applied. This is right
at about the limit of traction, meaning it might very well be possible to lock
the brakes on an F-1 car at full speed. Makes sense, doesn't it? Why not have
brakes that will take the tires to their very limit? This 5g figure seems to
indicate that this is really what's done. My simulated F-1 car (in
...
read more »
"J. Todd Wasson
Whew! At least I read the whole....thing....now why does my head
hurt????
dave henrie
Cheers,
SB
PS - thanks again to Dave P. for help and guidance on that project.
<great post snipped>