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

Hi Mart

> For the first solution, a test "car->_pit" have to be added in line 139, or as 
> first test in line 141 (in raceengine.cpp).

Fixed, thank you for the report.

Bye, Bernhard.

PS: the spam filter of my provider currently swallows the torcs mailing list
postings...


-- 

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

Am Mittwoch, den 22.02.2006, 18:10 +0100 schrieb Christos Dimitrakakis:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>=20
> >=20
> > Thanks. Just one thing: in grcam.cpp the matching of the field-of-view
> > of the background camera is commented out. Was that on purpose?
> >=20
> No, I didn't change your patch at all - it seems it had already been
> commented out when you sent it.

Oops, the first version of my patch had it commented out. The second
version had it active, though. Sorry about that.

Regards,
  Felix

>=20
> Note that Bernhard objects to zooming too much in the background because
> of extreme pixelation - so I was thinking of limiting the fov for the
> background just to avoid that - for me it's more distracting to have the
> wrong foc than to have pixelation.  This is quite a minor issue to fight
> over, but .. well, any opinions?
>=20
>=20
>=20
> - --
> Christos Dimitrakakis
> Homepage: http://www.idiap.ch/~dimitrak/main.html
> Music: http://olethros.dmusic.com
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.1 (GNU/Linux)
> Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org
>=20
> iD8DBQFD/Jr7BsQIVGq8N4ERAkB2AJ47BzXrsYfCRT1leKOnq4LNJ/horACfVrdv
> bly3iaxUMcFK8qHrk6uphSc=3D
> =3DlVRb
> -----END PGP SIGNATURE-----

--=20
| Felix K=C3=BChling <fx...@gm...>                     http://fxk.de.vu =
|
| PGP Fingerprint: 6A3C 9566 5B30 DDED 73C3  B152 151C 5CC1 D888 E595 |

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

> 
> Thanks. Just one thing: in grcam.cpp the matching of the field-of-view
> of the background camera is commented out. Was that on purpose?
> 
No, I didn't change your patch at all - it seems it had already been
commented out when you sent it.

Note that Bernhard objects to zooming too much in the background because
of extreme pixelation - so I was thinking of limiting the fov for the
background just to avoid that - for me it's more distracting to have the
wrong foc than to have pixelation.  This is quite a minor issue to fight
over, but .. well, any opinions?



- --
Christos Dimitrakakis
Homepage: http://www.idiap.ch/~dimitrak/main.html
Music: http://olethros.dmusic.com
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFD/Jr7BsQIVGq8N4ERAkB2AJ47BzXrsYfCRT1leKOnq4LNJ/horACfVrdv
bly3iaxUMcFK8qHrk6uphSc=
=lVRb
-----END PGP SIGNATURE-----

Am Mittwoch, den 22.02.2006, 11:10 +0100 schrieb Christos Dimitrakakis:
> Incorporated Felix's PoV patches.

Thanks. Just one thing: in grcam.cpp the matching of the field-of-view
of the background camera is commented out. Was that on purpose?

Regards,
  Felix

--=20
| Felix K=C3=BChling <fx...@gm...>                     http://fxk.de.vu =
|
| PGP Fingerprint: 6A3C 9566 5B30 DDED 73C3  B152 151C 5CC1 D888 E595 |

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Mart Kelder wrote:
> Hi,
> 
> I think the problem is caused by a car which ask for a pitstop and doesn't 
> have a pit box. The human driver does this when passing the starting line.
> 
> I think there are at least two solutions possible: check if the driver have a 
> pit before checking if it is empty, or not allowing bots to ask for a pitstop 
> when they don't have a pit box.
> 
> For the first solution, a test "car->_pit" have to be added in line 139, or as 
> first test in line 141 (in raceengine.cpp).
> 

I think it'd be preferrable to do the fix in raceengine.cpp since the
bots can may also ask for a pit-stop later on.  Will try it later, thanks.


- --
Christos Dimitrakakis
Homepage: http://www.idiap.ch/~dimitrak/main.html
Music: http://olethros.dmusic.com
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFD/G3nBsQIVGq8N4ERAkANAJ0Zlq445tUYgkTt5v+PcWdSNWF+9gCfXzuU
lGWiEm3qR4ejAZ/II1dUODA=
=kUuS
-----END PGP SIGNATURE-----

Hi,

I think the problem is caused by a car which ask for a pitstop and doesn't 
have a pit box. The human driver does this when passing the starting line.

I think there are at least two solutions possible: check if the driver have a 
pit before checking if it is empty, or not allowing bots to ask for a pitstop 
when they don't have a pit box.

For the first solution, a test "car->_pit" have to be added in line 139, or as 
first test in line 141 (in raceengine.cpp).

For the second solution, the same test can be added in "human.cpp" in the 
human driver. The test could be added in line 488.

Regards,

Mart

There seems to be a bug related to the new pits code.

Whenever the human player (not the bots, apparently)
a) passes the start line at the beginning of the race
b) in tracks which have no pits

A segfault happens at raceengine.cpp:141

Can you verify?
-- 
Christos Dimitrakakis
Homepage: http://www.idiap.ch/~dimitrak/main.html
Music: http://olethros.dmusic.com

Incorporated Felix's PoV patches.
-- 
Christos Dimitrakakis
Homepage: http://www.idiap.ch/~dimitrak/main.html
Music: http://olethros.dmusic.com

OK, I have updated the FoV... it works great. Thanks!

Note that the same thing must be done for the 2D graphics, which become
stretched.

I will put the other patch on CVS as soon as I have tested it, which
should be any day now

-- 
Christos Dimitrakakis
Homepage: http://www.idiap.ch/~dimitrak/main.html
Music: http://olethros.dmusic.com

Hi everyone,
I am new to Torcs.  I think your game is quite
impressive.

I would like to offer some help on your project in the
graphics area.  I am fairly experienced with OpenGL. 
I would like to add cube map reflection to the cars
and render a glass material for the car windows.  Of
course I would do it in a way that people with older
machines would have it work like it does currently.

I have a couple of questions.  I couldn't get the CVS
version to run.  I am using windows and MSVC 6.  It
compiles but it doesn't show the opening text doesn't
appear.    

Have you guys thought about using the GLEW library to
handle OpenGL extensions?

Brian

__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com 

Am Montag, den 20.02.2006, 11:24 +0100 schrieb Christos Dimitrakakis:
> OK, tested the FOV patch. Here are my results:
>=20
> 1) OK, fine, apart from one thing:
> - On some extra-wide screens, such as for example the screens that you
> get when pressing [ or ], everything is stretched. Does this patch take
> into account the aspect ratio of the screen??? Can you check? It might
> be that it does and that the aspect ratio is not set correctly for these
> screens or that it does not check and it assumes a 1:1 ratio ..

An updated patch is attached. It removes the limitation of the aspect
ratio to 2.3, which fixes the distortion in split screen mode.

What about the other patch for the mode list generation. Can you commit
that to CVS too?

Regards,
  Felix

>=20
>=20
> 2) I guess OK, but not tested.
>=20
> 3) Should be OK, we must change grBackgroundCam() so that it gets the
> FOV of the camera that is using it.
>=20
> If you can check my problem with point 1) I could do 3) and apply the
> whole thing on CVS
>=20
--=20
| Felix K=C3=BChling <fx...@gm...>                     http://fxk.de.vu =
|
| PGP Fingerprint: 6A3C 9566 5B30 DDED 73C3  B152 151C 5CC1 D888 E595 |

Hi all

There is a little update in the CVS.
- Pit sharing support (for TRB, the default is one car per pit because the
robots in the CVS do not yet support it). Have a look at the readme.
- Fixed wrong car corners for cars with nonmatching center of gravity and
geometric center.

Bye, Bernhard.

-- 

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

Am Montag, den 20.02.2006, 11:24 +0100 schrieb Christos Dimitrakakis:
> OK, tested the FOV patch. Here are my results:
>=20
> 1) OK, fine, apart from one thing:
> - On some extra-wide screens, such as for example the screens that you
> get when pressing [ or ], everything is stretched. Does this patch take
> into account the aspect ratio of the screen??? Can you check? It might
> be that it does and that the aspect ratio is not set correctly for these
> screens or that it does not check and it assumes a 1:1 ratio ..

Maybe on split screen it still assumes the aspect ratio of the full
screen and squeezes it to fit on half the screen. Or I believe I saw
some code that limits the aspect ratio to 2.3. I guess that would have
to be disabled for horizontal screen splitting to work reasonably,
especially on wide-screen displays. I suspect this issue was previously
masked by the wrong aspect ration computation. I'll look into it.

>=20
>=20
> 2) I guess OK, but not tested.
>=20
> 3) Should be OK, we must change grBackgroundCam() so that it gets the
> FOV of the camera that is using it.
>=20
> If you can check my problem with point 1) I could do 3) and apply the
> whole thing on CVS

Sounds good.

Thanks,
  Felix

--=20
| Felix K=C3=BChling <fx...@gm...>                     http://fxk.de.vu =
|
| PGP Fingerprint: 6A3C 9566 5B30 DDED 73C3  B152 151C 5CC1 D888 E595 |

Hi all.

Can you please have a look at my forum <http://speedy.chonchon.free.fr>?
A new Torcs addict needs some help. So far I could help him. But now he 
need to compile stripe on Windows and I don't know anything about that.
The subject is : "Export a car from blender into torcs. Help is needed. 
<http://speedy.chonchon.free.fr/phpBB2/viewtopic.php?t=37&start=15&postdays=0&postorder=asc&highlight=>"

Thank

Speedy
<http://speedy.chonchon.free.fr/phpBB2/viewtopic.php?t=37&start=15&postdays=0&postorder=asc&highlight=>

OK, tested the FOV patch. Here are my results:

1) OK, fine, apart from one thing:
- On some extra-wide screens, such as for example the screens that you
get when pressing [ or ], everything is stretched. Does this patch take
into account the aspect ratio of the screen??? Can you check? It might
be that it does and that the aspect ratio is not set correctly for these
screens or that it does not check and it assumes a 1:1 ratio ..


2) I guess OK, but not tested.

3) Should be OK, we must change grBackgroundCam() so that it gets the
FOV of the camera that is using it.

If you can check my problem with point 1) I could do 3) and apply the
whole thing on CVS

-- 
Christos Dimitrakakis
Homepage: http://www.idiap.ch/~dimitrak/main.html
Music: http://olethros.dmusic.com

Hi,

I attached two patches. The first one fixes three issues concerning the
field of view.

     1. When computing the fovx base on fovy and the screen aspect
        ratio, you can't use angles. Instead the computation must be
        done with lengths in the projection plane. The old way of doing
        the computation with angles is approximately correct for small
        angles but gets more and more distorted with larger angles. Now
        the aspect ratio looks correct, even with extreme zoom levels
        and on wide-screen displays.
     2. Now the limitFov methods that limit the fovy such that fovx
        never gets bigger than 90=C2=B0 doesn't make sense any more. With t=
he
        aspect ratio computation in the projection plane you always get
        a valid fovx angle for any fovy angle < 180=C2=B0. Therefore I
        removed all those limitFov methods. The respective camera
        classes inherit the empty implementation from cGrPerspCamera.
     3. The field of view of the background did not match the field of
        view of the rest of the scene. In particular the background did
        not change when zooming. Just enable the zoomed track-view
        camera (F9) to see what I mean. :) I added a getFovY method to
        the camera base class and use it in cGrBackgroundCam::update to
        match the scene's fovy. On orthographic cameras it returns a
        constant 67.5=C2=B0 (the default value of the background camera), o=
n
        perspective cameras it returns the real fovy.

The second one fixes a glitch in the display mode list generation. On my
system it would result in the highest resolution mode missing from the
list. With this fix I can finally play Torcs in fullscreen mode on my
wide-screen display. :)

Please consider applying these to CVS.

Regards,
  Felix

--=20
| Felix K=C3=BChling <fx...@gm...>                     http://fxk.de.vu =
|
| PGP Fingerprint: 6A3C 9566 5B30 DDED 73C3  B152 151C 5CC1 D888 E595 |

Hi,

by default the installation directories of TORCS are world-writable.
This is a pretty bad thing from a security point-of-view. This small
patch to the mkinstalldirs script in CVS fixes the problem:

Index: mkinstalldirs
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
RCS file: /cvsroot/torcs/torcs/torcs/mkinstalldirs,v
retrieving revision 1.1.1.1
diff -u -r1.1.1.1 mkinstalldirs
--- mkinstalldirs       24 Jun 2001 17:43:16 -0000      1.1.1.1
+++ mkinstalldirs       19 Feb 2006 18:00:35 -0000
@@ -8,8 +8,6 @@

 errstatus=3D0

-umask 000
-
 for file
 do
    set fnord `echo ":$file" | sed -ne 's/^:\//#/;s/^://;s/\// /g;s/^#/\//;=
p'`



Please consider applying this to CVS.

Regards,
  Felix

--=20
| Felix K=C3=BChling <fx...@gm...>                     http://fxk.de.vu =
|
| PGP Fingerprint: 6A3C 9566 5B30 DDED 73C3  B152 151C 5CC1 D888 E595 |