Re: [Flightgear-devel] Re: [Flightgear-cvslogs] CVS: FlightGear/src/Main main.cxx,1.245,1.246

2002-03-05 Thread D Luff

On 5 Mar 2002, at 8:40, Curtis L. Olson wrote:

 David,
 
 This is going to mess up rendering of distant mountains for people
 with voodoo cards (i.e. 16 bit buffers).
 

And not just 16 bit cards - I get flashing polygons in the sky with 
two different 32 bit cards at 0.5, progressively cured by pushing out 
to 1.2ish, although no-one else seems to be afflicted by this.

Cheers - Dave

--
[EMAIL PROTECTED]

___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel



Re: [Flightgear-devel] Re: [Flightgear-cvslogs] CVS: FlightGear/src/Main main.cxx,1.245,1.246

2002-03-05 Thread D Luff

Alex Perry writes:

 and for the latter you're in ground haze in any case.

You Californians speak for yourselves! 

:-)

Cheers - Dave

 

--
[EMAIL PROTECTED]

___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel



Re: [Flightgear-devel] Re: [Flightgear-cvslogs] CVS: FlightGear/src/Main main.cxx,1.245,1.246

2002-03-05 Thread Andy Ross

Curtis L. Olson wrote:
  I'm really nervous about forcing this down to 0.1 for all cases.  This
  will mess things up for people with voodoo cards.
 
  Can we change this back?
 
  Perhaps we need to do a separate pass for rendering the 3d model and
  change the near clip plane just for that portion of the rendering.

I'm with Curt.  Coming up with a single Z range for mountains and
gauges isn't likely to work well.  A better idea would be to
investigate glDepthRange() to isolate the two regimes (this trades ~1
bit in depth buffer precision and allows you two or more completely
separated clip plane sets).

Or just turn the depth buffer off for the cockpit and sort the
geometry; for most cockpit layouts, this should be pretty feasible.
It won't work if the cockpit has moving parts (yokes or whatnot) that
obscure other pieces.

Or just eat the cost and clear the depth buffer between the scenery
and aircraft/cockpit.  On modern hardware, it's actually not that
expensive.

Andy

-- 
Andrew J. RossNextBus Information Systems
Senior Software Engineer  Emeryville, CA
[EMAIL PROTECTED]  http://www.nextbus.com
Men go crazy in conflagrations.  They only get better one by one.
  - Sting (misquoted)


___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel



Re: [Flightgear-devel] Re: [Flightgear-cvslogs] CVS: FlightGear/src/Main main.cxx,1.245,1.246

2002-03-05 Thread Wolfram Kuss

Curt wrote:

  Perhaps we need to do a separate pass for rendering the 3d model and
  change the near clip plane just for that portion of the rendering.

Yes. That should be easy to do. BoB does it as well, they even have 3
or 4 different parts of the geometry that they render with 3 or 4
different near values. 

Bye bye,
Wolfram.

___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel



Re: [Flightgear-devel] Re: [Flightgear-cvslogs] CVS: FlightGear/src/Main main.cxx,1.245,1.246

2002-03-05 Thread David Megginson

Andy Ross writes:

  Or just turn the depth buffer off for the cockpit and sort the
  geometry; for most cockpit layouts, this should be pretty feasible.
  It won't work if the cockpit has moving parts (yokes or whatnot) that
  obscure other pieces.

I don't think this is an easy option, at least not with a true 3D
model wrapped around the viewer.  We'll have to find a more robust
solution.  To start, I can make the depth buffer 0.1 only when the
interior model view is enabled, so no one loses without it; old
hardware probably won't do well with the interior view anyway.


All the best,


David

-- 
David Megginson
[EMAIL PROTECTED]


___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel



Re: [Flightgear-devel] Re: [Flightgear-cvslogs] CVS: FlightGear/src/Main main.cxx,1.245,1.246

2002-03-05 Thread Jim Wilson

David Megginson [EMAIL PROTECTED] said:

 
 I don't think this is an easy option, at least not with a true 3D
 model wrapped around the viewer.  We'll have to find a more robust
 solution.  To start, I can make the depth buffer 0.1 only when the
 interior model view is enabled, so no one loses without it; old
 hardware probably won't do well with the interior view anyway.
 

Ummm...why not clear the z-buffer and use a seperate scene graph?  Note that
there are *lots* of people using pre-geforce generation hardware.  Probably
more than not.

Best,

Jim

___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel



Re: [Flightgear-devel] Re: [Flightgear-cvslogs] CVS: FlightGear/src/Main main.cxx,1.245,1.246

2002-03-05 Thread Jim Wilson

Curtis L. Olson [EMAIL PROTECTED] said:

 
 David,
 
 Assuming all of this is being drawn via plib/ssg then you could put
 all the geometry in a separate ssgRoot node and call ssgCullandDraw()
 on this root after everything else has been rendered.  We have a
 couple ssgRoot's already so we can have some high level control over
 sorting geometry ... sky, terrain, lights, cockpit ...
 
 Regards,
 
 Curt.


Please, yes.

___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel



Re: [Flightgear-devel] Re: [Flightgear-cvslogs] CVS: FlightGear/src/Main main.cxx,1.245,1.246

2002-03-05 Thread Andy Ross

Jim Wilson wrote:
  Ummm...why not clear the z-buffer and use a seperate scene graph?
  Note that there are *lots* of people using pre-geforce generation
  hardware.  Probably more than not.

Actually, it's those folks who are at risk.  Clearing the whole z
buffer soaks up big chunks of framebuffer bandwidth, and is a frame
rate killer on early hardware.  But I agree, it's the simplest and
most robust solution.

Folks who haven't looked at it before should definitely read the man
pages for glDepthRange().  For this situation (the cockpit doesn't
need a lot of z precision), it's a good fit.  The only worry is that
it's a relatively obscure feature and thus more susceptible to driver
bugs and incompatibilities.

Andy

-- 
Andrew J. RossNextBus Information Systems
Senior Software Engineer  Emeryville, CA
[EMAIL PROTECTED]  http://www.nextbus.com
Men go crazy in conflagrations.  They only get better one by one.
  - Sting (misquoted)


___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel



Re: [Flightgear-devel] Re: [Flightgear-cvslogs] CVS: FlightGear/src/Main main.cxx,1.245,1.246

2002-03-05 Thread David Megginson

Jim Wilson writes:

  Ummm...why not clear the z-buffer and use a seperate scene graph?  Note that
  there are *lots* of people using pre-geforce generation hardware.  Probably
  more than not.

Time -- I don't know if I'll have time to figure all that out this
week.


All the best,


David

-- 
David Megginson
[EMAIL PROTECTED]


___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel