torcs-devel Mailing List for TORCS - The Open Racing Car Simulator

Christos Dimitrakakis wrote:
>>- If I modify the body's texture I have a transparency but I see
>>(through the window) only the other side of the body (not the floor, the
>>wheel, etc....).
>>
> 
> 
> Maybe you can take a look at the pw- wrc models. They are the only
> ones that have transparencies correct, I think. Normally one would
> like to have the view from inside looking out to be completely
> transparent.
> 
> IIRC the pw models are split in two parts: the body, and the
> floor+wheel+dashboard. Hopefully by looking at the attributes of the
> faces you can discern how to do it.
> 

The windows objects have to be named "WIxxx" so they will be rendered
in a second pass in PLIB (transparency flag set).
cars sorting is done to display the farther first and
have a correct rendering of the ttransparency.

Eric.
-- 
=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=
               TORCS - http://torcs.org
             The Open Racing Car Simulator
         AKA The Other Release Coming Soon (Skin'r)
            How soon is soon ? (RaceBlizter)
=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=

> Hi Christophe
>
> > I can't find it, but we can use the sound of an overboosted Nissan Skyline ???
> > (a V.6 3.5L bi-turbo 700hp !!!)
> >
> > Christos, I'll send you a link to download my sounds
>
> Please make sure that the sounds are compatible with our licenses
> (either Free Art License or GPL).
>
> Thank you, bye,
>

If there are copyright issues I can build a model of the sound from
the sample and then create a .wav using the model.

I imagine that a lot of samples that can be found on the web are
practically free, but: a)  few of them have an explicit permission
attached, b) most of them are not suitable for the game.

-- 
Christos Dimitrakakis
IDIAP (http://www.idiap.ch/~dimitrak/main.html)

Le vendredi 25 F=E9vrier 2005 08:55, Bernhard Wymann a =E9crit :
> Then we cannot use those sounds in TORCS, its that simple...

OK.

I think I'll find easily some people to record the peugeot V6, the alfa V6 =
and=20
the VW flat-four.

=46or the others it'll be more difficult...

Regards,
=2D-=20
Christophe BAEGERT              cba...@eu...

Hi Christophe

>>Please make sure that the sounds are compatible with our licenses
>>(either Free Art License or GPL).
> 
> 
> I'm stupid... I thought that theses files were not covered by the copyright 
> laws, since there's no real "creation". But they are...

Of course there is... its called recording;-) In fact humans never 
create something, we just reorganize matter and energy state, thats 
about it;-)

> In fact, this is almost impossible...  It let only one solution : record it 
> ourself... And it's quite hard to find a Maserati  Ghibli, an Iso Grifo or a 
> XJ220...
> 
> I'm sure I can have the agreement for a lot of sounds. But how can I make sure 
> than the perso who agrees is the real producer of the sound or have the 
> rights on the file ? Is is OK if I don't verify it (since I have absolutely 
> no way to do it) ?

Then we cannot use those sounds in TORCS, its that simple...

Bye, Bernhard.

-- 

Visit my homepage http://www.berniw.org
Official TORCS racing: The TORCS Racing Board, http://www.berniw.org/trb

Hello,

Le jeudi 24 F=E9vrier 2005 22:40, Bernhard Wymann a =E9crit :
> Please make sure that the sounds are compatible with our licenses
> (either Free Art License or GPL).

I'm stupid... I thought that theses files were not covered by the copyright=
=20
laws, since there's no real "creation". But they are...

In fact, this is almost impossible...  It let only one solution : record it=
=20
ourself... And it's quite hard to find a Maserati  Ghibli, an Iso Grifo or =
a=20
XJ220...

I'm sure I can have the agreement for a lot of sounds. But how can I make s=
ure=20
than the perso who agrees is the real producer of the sound or have the=20
rights on the file ? Is is OK if I don't verify it (since I have absolutely=
=20
no way to do it) ?

Regards,
=2D-=20
Christophe BAEGERT              cba...@eu...

Hi Christophe

> I can't find it, but we can use the sound of an overboosted Nissan Skyline ??? 
> (a V.6 3.5L bi-turbo 700hp !!!)
> 
> Christos, I'll send you a link to download my sounds

Please make sure that the sounds are compatible with our licenses 
(either Free Art License or GPL).

Thank you, bye,

Bernhard.


-- 

Visit my homepage http://www.berniw.org
Official TORCS racing: The TORCS Racing Board, http://www.berniw.org/trb

Hello,

Le mercredi 23 F=E9vrier 2005 20:18, Christophe Baegert a =E9crit :
> Le mercredi 23 F=E9vrier 2005 18:30, Christos Dimitrakakis a =E9crit :
> > Maybe you can
> > help by posting a list of all cars without special engine sounds and
> > their associated engine type?
>
> Of course :
>

My suggestions :

> 155DTM : V6 2.5

I got it !

> 360 modena : V8 3.7

I got it !

> acura v6 3.2
OK (i got it as well)

> baja bug : flat 4  1.3
> buggy : flat 4  1.3
I've found a sound !



> nascar : v8 chevrolet 5.7
now it's a ford, but I've got some wavs of corvettes with the chevy 5.7 V8 =
!!!=20

> clk dtm : v8 4.0
i have a SL55 sound (V8 5.5).

> gt40 concept : v8 ford 5.4
we can keep our ford v8 (is it really a V8 ???)
I found the sound of a quale mangusta, it's a modern ford V8, and it sounds=
=20
really better than this one


> 2000gt : v6 2.0
I'd suggest nsx...=20

> 5300gt : v8 chevrolet 5.4
chevy...

> alpine a110 : L4 1600cm3
I'll have that one in may (currently the car is not functional...)

> alfa tz2 : L4 1600cm3
Any italian L4 sound would we ok... (fiat, lancia, alfa), their sounds are=
=20
very similar (excepted alfa boxers). But I didn't find it.

> bigh : L6 2.9
don't have that one, but I found an english L6 (the engine of the TVR Cerbe=
ra=20
4.0, but it's a modern engine...)

> ac cobra: v8 ford 4.7
ford v8 (i have the mp3 of an old v8 ford, in the "de tomaso pantera gts")

> opel conrero : L6 2.3 or L4 1.9
don't know how it sound, we could use the L6 of the TVR ?
but it's not a rare engine, a lot of big old opels use it (rekord,=20
commodore...)

> corvette : v8 chevrolet 5.7
I have it !

> ferrari daytona : v12 4.4
I have it !

> db4z: L6 3.6
we could use the tvr sound

> dbs: V8 4.3

I found a V12 Aston martin, but not a V8. We'd better use a TVR V8 (Griffit=
h=20
or Cerbera)

> dino V6 2.4
dont have that one....=20

> ghibli: V8 4.7
I only found mp3s of the new ghibli... I

> giulieta : L4 1.3
like alfa tz2

> grifo: V8 chevrolet 7L
I have it

> gt40: V8 ford 4.2
I have it

> ferrari 250 GTO: V12 3L
I have it !

> ferrari 330 P4 : V12 4L
We can use the 250 GTO engine. Better than nothing.

> lotus gt1 : V8 3.5
already OK

> mclaren f1 : V12 BMW 6L
already OK

> peugeot 406 : V6 3L
I have the sound of the xantia activa, it's the same engine

> porsche gt1 : flat 6 3.2L twin turbo
I have the gt2, it's an evolution of this engine.

> porsche gt3 rs : flat 6 3.6L
I have it !

> 206 wrc: L4 2L turbo
already OK
> 306 maxi : L4 2L turbo
we can take the 206 one

> corolla wrc : L4 2L turbo
already OK

> mitsubishi : L4 2L turbo
already OK

> focuswrc :  L4 2L turbo
already OK
> impreza : boxer 4 2L turbo
already OK

> f1 : v10 3.5L
already OK

> viper : v10 6.7L
already OK

> vm x2 : unknown ??
> vm x4 : unknown ??

> xj 220 : V6 3.5L twin turbo
I can't find it, but we can use the sound of an overboosted Nissan Skyline =
???=20
(a V.6 3.5L bi-turbo 700hp !!!)

Christos, I'll send you a link to download my sounds

Regards,
=2D-=20
Christophe Baegert

I think that the sand bug is visible at this location:

e-track-6, the sand on the left side of the track after the bridge.
If a car enters the sand, it appears that there is a force that pulls
it to the left. This happens with both simuv2 and simuv3. I am not
sure what the origin of that force is.. perhaps the normal of the
segment is wrong??

I know that there are some segments on the side whose normals and
height is not computed properly when the main track is not flat. A
case I noticed before was curbs on the right side when the track was
tilted to the left. It has been somewhat fixed by bernhard since I
discovered this problem but it seems to be still not working properly.

-- 
Christos Dimitrakakis
IDIAP (http://www.idiap.ch/~dimitrak/main.html)

> - If I modify the body's texture I have a transparency but I see
> (through the window) only the other side of the body (not the floor, the
> wheel, etc....).
>

Maybe you can take a look at the pw- wrc models. They are the only
ones that have transparencies correct, I think. Normally one would
like to have the view from inside looking out to be completely
transparent.

IIRC the pw models are split in two parts: the body, and the
floor+wheel+dashboard. Hopefully by looking at the attributes of the
faces you can discern how to do it.

-- 
Christos Dimitrakakis
IDIAP (http://www.idiap.ch/~dimitrak/main.html)

Bernhard Wymann a =E9crit :

> Hi SpeedyChonChon
>
>
> You do not need to -> Free Art License.
>
>
> Of course, see below.
>
>
> That comes from "back-face culling" I guess. You need to model the=20
> inside of the car as well... please do not use 2-sided triangles=20
> whenever another solution is possible (otherwise you get eventually=20
> z-buffer racing, depending on hardware bad performance, weird=20
> lighting, you can see the effect very well on the p206, it looks=20
> horrible sometimes if you take camera F9 and zoom out).
>
> To make transparent windows work cleanly without z-buffer artefacts=20
> you need to create geometry for the windows which are covered entirely=20
> with (mostly) transparent texels, have a look at all the other cars=20
> with windows (perhaps there are additional hints in the nfs2torcs howto=
).
>
> If you need a driver and seats I can mail them to you (created by=20
> Olaf, Free Art License).
>
> Bye, Bernhard.
>
Hi Bernhard

Thank for your fast answer. I will look at these tomorow. I want well a=20
driver. ;-)

Bye
--=20
SpeedyChonChon

----------------------------------------------------------
AIM : /SpeedyChonChon/
ICQ : /302814595/
MSN : /ab...@ho.../
http://speedy.chonchon.free.fr
----------------------------------------------------------

Hi SpeedyChonChon

> I work on the CLK-DTM in order to add an inside. (see my website on 
> Torcs pages ;-) ) (for the moment, just for me because I have not ask 
> the author). I would like use a second texture for the "cockpit" 

You do not need to -> Free Art License.

> (cockpit + chair + driver + etc). So I need to separate the objects but 
> I have problems with transparency of the body.
> - If I delete the verticices of the windows the result is correct but we 
> can't color them and it's not precise.

Of course, see below.

> - If I modify the body's texture I have a transparency but I see
> (through the window) only the other side of the body (not the floor, the 
> wheel, etc....).

That comes from "back-face culling" I guess. You need to model the 
inside of the car as well... please do not use 2-sided triangles 
whenever another solution is possible (otherwise you get eventually 
z-buffer racing, depending on hardware bad performance, weird lighting, 
you can see the effect very well on the p206, it looks horrible 
sometimes if you take camera F9 and zoom out).

> 
> Can you do something for me ?

To make transparent windows work cleanly without z-buffer artefacts you 
need to create geometry for the windows which are covered entirely with 
(mostly) transparent texels, have a look at all the other cars with 
windows (perhaps there are additional hints in the nfs2torcs howto).

If you need a driver and seats I can mail them to you (created by Olaf, 
Free Art License).

Bye, Bernhard.

-- 

Visit my homepage http://www.berniw.org
Official TORCS racing: The TORCS Racing Board, http://www.berniw.org/trb

Hello.

I work on the CLK-DTM in order to add an inside. (see my website on 
Torcs pages ;-) ) (for the moment, just for me because I have not ask 
the author). I would like use a second texture for the "cockpit" 
(cockpit + chair + driver + etc). So I need to separate the objects but 
I have problems with transparency of the body.
- If I delete the verticices of the windows the result is correct but we 
can't color them and it's not precise.
- If I modify the body's texture I have a transparency but I see 
(through the window) only the other side of the body (not the floor, the 
wheel, etc....).

Can you do something for me ?

Bye
-- 
SpeedyChonChon

----------------------------------------------------------
AIM : /SpeedyChonChon/
ICQ : /302814595/
MSN : /ab...@ho.../
http://speedy.chonchon.free.fr
----------------------------------------------------------

Le mercredi 23 F=E9vrier 2005 18:30, Christos Dimitrakakis a =E9crit :
> Maybe you can
> help by posting a list of all cars without special engine sounds and
> their associated engine type?

Of course :

155DTM : V6 2.5
360 modena : V8 3.7
acura v6 3.2
baja bug : flat 4  1.3
buggy : flat 4  1.3
nascar : v8 chevrolet 5.7
clk dtm : v8 4.0
gt40 concept : v8 ford 5.4
2000gt : v6 2.0
5300gt : v8 chevrolet 5.4
alpine a110 : L4 1600cm3
alfa tz2 : L4 1600cm3
bigh : L6 2.9
ac cobra: v8 ford 4.7
opel conrero : L6 2.3 or L4 1.9
corvette : v8 chevrolet 5.7
ferrari daytona : v12 4.4
db4z: L6 3.6
dbs: V8 4.3
dino V6 2.4
ghibli: V8 4.7
giulieta : L4 1.3
grifo: V8 chevrolet 7L
gt40: V8 ford 4.2
ferrari 250 GTO: V12 3L
ferrari 330 P4 : V12 4L
lotus gt1 : V8 3.5
mclaren f1 : V12 BMW 6L
peugeot 406 : V6 3L
porsche gt1 : flat 6 3.2L twin turbo
porsche gt3 rs : flat 6 3.6L
206 wrc: L4 2L turbo
306 maxi : L4 2L turbo
corolla wrc : L4 2L turbo
mitsubishi : L4 2L turbo
focuswrc :  L4 2L turbo
impreza : boxer 4 2L turbo
f1 : v10 3.5L
viper : v10 6.7L
vm x2 : unknown ??
vm x4 : unknown ??
xj 220 : V6 3.5L twin turbo

Regards,
=2D-=20
Christophe Baegert

> sorry but ... when I asked you how to record an engine sound, I forgot to ask
> where the microphone should be located ? inside the car ? near the engine ?
> outside ?
>

For an overall balanced sound, it is best inside the car, with the
window open. That achieves a good mix between sitting inside and
outside the car. With special filters one can mapipulate the sound so
that it is different depending on position. Particularly interesting
is the ability to have a model for the exhaust.

Currently there is only one sound used for each car. If it is not a
hassle then it'd be a good idea to additionally record a)
near the engine, b) near the exhaust, and maybe even c) some distance
away from the car. This way, if we want to make the sound a bit more
sophisticated in the future, we'll already have the sounds necessary.
I am not sure they will ever be needed, though, as I might go the
filtering route.


-- 
Christos Dimitrakakis
IDIAP (http://www.idiap.ch/~dimitrak/main.html)

Le mercredi 23 F=E9vrier 2005 18:53, Christos Dimitrakakis a =E9crit :
> The F360, F1, CLKDTM, 155-DTM, Viper cars have been udpated with new
> sounds. The CLK and Viper sounds are dominated by the exhaust,
> so they sound better on driveby cams than in internal cams.

sorry but ... when I asked you how to record an engine sound, I forgot to a=
sk=20
where the microphone should be located ? inside the car ? near the engine ?=
=20
outside ?

Regards,
=2D-=20
Christophe Baegert

The F360, F1, CLKDTM, 155-DTM, Viper cars have been udpated with new
sounds. The CLK and Viper sounds are dominated by the exhaust,
so they sound better on driveby cams than in internal cams.

-- 
Christos Dimitrakakis
IDIAP (http://www.idiap.ch/~dimitrak/main.html)

> > - Added a distance-based filter for the engine sound. This makes it
> > slightly more realistic.
>

> About realistic sound... I saw that the Porsche GT1 uses the sound
> 944.wav....
>
> I think we should better use the motor sound of the same type than the sound
> of the same maker (generally, not only for this model) : the GT1 has a flat-6
> and the 944 is a big L4...
>

I am afraid I have no idea how motors with different n of cylinders
sound :(

> the turbo). Maybe it's possible to add a generic turbo sound to a naturally
> aspired motor noise ? (by example porsche_engine.wav + turbo.wav would
> certainly be better for the GT1 than 944.wav..

(Funnily enough, porsche_engine.wav is an electric motor)
If you think you can help that, it'll be cool. The V4/6/8/10/12
sounds, we can make if we find an appropriate sound. Maybe you can
help by posting a list of all cars without special engine sounds and
their associated engine type?

With respect to the turbo, we need to know at least which engines have
turbo. Note that currently the simulation does not include turbo
(which would be probably implemented as 'turbo lag').

About the turbo sound itself.. The turbo is usually audible only
inside the car, right? In cars passing at high speed, you only hear
the engine and the exhaust usually right? (particularly the exhaust I
think). I was thinking of making it at a level similar to the axle
noise. (which might already be a bit too loud, I think).



-- 
Christos Dimitrakakis
IDIAP (http://www.idiap.ch/~dimitrak/main.html)

Hello,

Le mercredi 23 F=E9vrier 2005 17:17, Christos Dimitrakakis a =E9crit :
> - Added a distance-based filter for the engine sound. This makes it
> slightly more realistic.

About realistic sound... I saw that the Porsche GT1 uses the sound 944.wav.=
=2E..=20

I think we should better use the motor sound of the same type than the soun=
d=20
of the same maker (generally, not only for this model) : the GT1 has a flat=
=2D6=20
and the 944 is a big L4...

=46or each V6, if we don't have the good sound, we should use the sound of =
an=20
another V6 (or a L6), even for historical models.

Idem for L4, V8, V10, V12.

The flat engines are particular, so we should distinguish between V&Ln and=
=20
=46lat-n engines... Turbocharged and naturally aspirated motors are differe=
nt=20
as well (the first is a bit muffled and normally we can hear the whistle of=
=20
the turbo). Maybe it's possible to add a generic turbo sound to a naturally=
=20
aspired motor noise ? (by example porsche_engine.wav + turbo.wav would=20
certainly be better for the GT1 than 944.wav...)

Regards,
=2D-=20
Christophe BAEGERT              cba...@eu...

- Fixed a tiny bug with the gear changing sound (some times that an
opponent passed you, a gear changing sound was heard even if it should
not).

- Added a distance-based filter for the engine sound. This makes it
slightly more realistic.



-- 
Christos Dimitrakakis
IDIAP (http://www.idiap.ch/~dimitrak/main.html)

(A set of fixes backported from simuv3)

wheel.cpp: Properly updating wheel slip export (for sound)

transmission.cpp: Moved differential->in.I and
differential->outAxis[]->I updates out of if() clauses in
SimGearboxUpdate(). This fixes the 'drivertrain-inertia-smaller-with-
gear-engaged-than-without' bug.

engine.cpp|h : Added engine exhaust pressure calculation and exporting
carElt->smoke variable. This enables the backfiring sound.

sim.h : Added inline urandom() which is used occasionally in simu.
Perhaps urandom() should become globally available?

I tested transmission change in particular, for seeing whether it
makes any difference with robots. It does not seem to. I don't know if
this would also fix the bug that bernhard mentioned with getting stuck
in the sand or not.

-- 
Christos Dimitrakakis
IDIAP (http://www.idiap.ch/~dimitrak/main.html)

I'm kicking around the idea of adding an RPC interface for robot control
of the cars, with the intent of distributing the race across a LAN of 
participating
computers.

There are obvious issues with this, primarily network latencies, but
I think even discussing it could open up new insights.  I havent yet 
properly studied
the code, so this is all somewhat premature, Im hoping to find out 
whether this is
just madness before I invest lots of time. (or conversely if anyone else 
is thinking quietly about it)

. registration interface, protocol.

cars connect to race-manager socket, identify themselves, send various
checksums of track, cars, etc, so that race-manager can verify that all
competitors are racing under same assumptions, track, data, etc...

. simulation, event loop

vastly oversimplifying (probably), the simulator invokes robot
callbacks, supplies situation data, and expects control vector back.
This is qualitatively the only coupling between racers (the situation
data carries "full" knowledge of competitors positions, so no other
data is needed - crossing process boundaries should prove this).


. rendering, video

obviously want each competitor/team-owner to be able to watch the
race.  This suggests that all participating computers are executing
the 'same' simulation, but Im not sure such a viewpoint can clarify
whats needed.  Nonetheless, it would be cool to have 6 displays side
by side, each dedicated to its driver's view.  Would be entertaining
to watch, even without a steering wheel.


. should work on single box.

using shared memory for the RPC, should have nearly no slowdown.  Im
assuming that 1 shmem segment will hold 2 arrays; 1 of situation
vectors, 1 of robot responses, and each robot (via proper RPC
abstraction) will stay out of others data.

. round-robin simulation.

presumably the current approach is to invoke each racer-robot in a
loop, then update everones situation vectors and repeat, until race is
finished..  IIUC, RARS stipulated that robots should not chew up all
the cpu, but did not enforce this.

do any robots currently build up a 'memory' of how theyve driven the
previous laps ?  It seems this would be unwelcome currently, but would
be ok/interesting when run on a separate racer-box.

. decoupling sim-loop from rendering loop

(caveat, WAG follows. (wild-ass speculation)) Having just successfully
installed ATI's fglrx video drivers on my Radeon 9700 laptop, Ive just
seen my FPS jump from 0.6 to 70.  Needless to say, it was formerly
impossible to drive, now it looks really good.


Well, there are many other tangents possible, Im hoping you all are willing
to brainstorm down some of those possibilities.

tia
Jim Cromie

>
> > However it might be OK to put it as the default option on pro
> > mode.
>
> I would not change it at all in simuv2 because the jump is simply too
> big then.

This is only for simuv2.

> Another issue is that the data you have found on the web might
> not be reliable at all, so who knows... (btw. I have as well the feeling
> that the coefficients are too big, but I am against a change in simuv2).
>

I dunno, is it reasonable for 1000Kg + cars to be able to turn at 3g?
The Ferrari data is taken from the Ferrari website. They seem to be
very proud of the 280Kg downforce they achieved for their latest model
:) I changed the coefficients so I get 400Kg on the 360, and it was
quite a bit more difficult.

On the plus side, the traction becomes a bit more linear so it can be
more easily modelled by robots (and humans, I guess) and oversteering
is smoother. On the minus side, RWD cars become more prone to power
spins and some cars almost lift off on the hill in the long straight
of e-track-4 :)

I have just done an options system for simuv3, so I'll add this as an
option for realism junkies. [It can certainly be intimidating to
novices. Some friends of mine that tried to play found even
simuv2/rookie hard.] There is currently neither GUI nor .XML file for
controlling the simuv3 options, of course.

-- 
Christos Dimitrakakis
IDIAP (http://www.idiap.ch/~dimitrak/main.html)

Hi Christos

> I fiddled a bit with the aero settings and lowered it
> down so that with the 360-modena you get around double the downforce
> of the real car. While there is a huge difference in speed that you
> can use for curves, it seems that if you slow done sufficiently the
> car handling is slightly easier.
> 
> Unfortunately, no matter how I implement this in the code, it means
> that all robots that try to duplicate the functionality of the
> simulation are broken, because they read the parameters directly from
> the file with GfParmGetNum().
> 
> As it stands, altering the aero to reduce downforce to more reasonable
> levels seems to not be worth it, considering the amount of things that
> break. (Car wing settings will need to be changed as well).

Yes:-)

> However it might be OK to put it as the default option on pro
> mode.

I would not change it at all in simuv2 because the jump is simply too 
big then. Another issue is that the data you have found on the web might 
not be reliable at all, so who knows... (btw. I have as well the feeling 
that the coefficients are too big, but I am against a change in simuv2).

Bye, Bernhard.


-- 

Visit my homepage http://www.berniw.org
Official TORCS racing: The TORCS Racing Board, http://www.berniw.org/trb

I fiddled a bit with the aero settings and lowered it
down so that with the 360-modena you get around double the downforce
of the real car. While there is a huge difference in speed that you
can use for curves, it seems that if you slow done sufficiently the
car handling is slightly easier.

Unfortunately, no matter how I implement this in the code, it means
that all robots that try to duplicate the functionality of the
simulation are broken, because they read the parameters directly from
the file with GfParmGetNum().

As it stands, altering the aero to reduce downforce to more reasonable
levels seems to not be worth it, considering the amount of things that
break. (Car wing settings will need to be changed as well).

However it might be OK to put it as the default option on pro
mode.

-- 
Christos Dimitrakakis
IDIAP (http://www.idiap.ch/~dimitrak/main.html)

I was just going through the aero code again. It is hard to find
figures for real cars, but I chanced across something in ferrari's
site which said that the ferrari 360 modena generates 200Kg at
300Km/h. In proper units, that is 2000N. Now, checking aero.cpp,
we have

total_lift = C * v^2 * hm, where C = 0.7+0.8 for the ferrari.
hm=2*exp(-3.0*(1.5*0.4)^4), [the 0.4 figure comes from the ride
height set up, though it could be even lower) really.

At a speed of 300Km/h this gives 8000 N, four times as much as the
original figure.

In order to bring it in line with the real figure, we need to either
change the C coefficients, or make it so that the car's front area and
Cx is taken into account.

We could change the line:
  hm = 2 * exp(-3.0*hm);
into:
  hm = FrntArea * Cx * exp(-3.0*hm);

I guess the 3.0 is arbitrary and could have been any other value.
Values near zero would indicate that the downforce is mostly due to
ground effect's decrease of pressure in the bottom and lower values
that it is mostly due the body's increase of pressure in the top.

I am sure that if we change this cars will be much more difficult to
handle. I don't know if the 360's value is representative of that of
other cars' downforce without any spoilers, but people on various
websites seem to value its downforce very highly. But perhaps they are
just echoing Ferrari's press release.


-- 
Christos Dimitrakakis
IDIAP (http://www.idiap.ch/~dimitrak/main.html)