pre6 - crashing at Saturn's rings

Report bugs, bug fixes and workarounds here.
Avatar
Topic author
fsgregs
Posts: 1307
Joined: 07.10.2002
With us: 22 years 1 month
Location: Manassas, VA

pre6 - crashing at Saturn's rings

Post #1by fsgregs » 14.03.2004, 03:18

Hi. I have tried out 1.3.2pre 6 and am having some crash problems. I go to Saturn to view the new rings and encounter two problems. First, the chopped edges of the planet shadow on the ring that DJ Night complained about in a separate thread last week is really bad with pre6 on my system, even through I have an Nvidia GeForce 4 TI 4200 with 128 MB video RAM. It is present with 1.3.1 final, but nowhere near as bad. I've set my screen with ctrl V to the best vortex shading I can get but no difference at all.

Secondly, Celestia locks up and crashes when I'm looking at Saturn's rings about 2 in every 3 tries. Has happened 9 or 10 times so far. I don't know why. I'm using Win XP on a pentium IV with Saturn's default texture and the new ring textures that come with pre6. I don't get a crash using version 1.3.1 and the same texture and solarsys.ssc file. It seems to be something in the pre6 program execution.

Third, I noticed that with the new ring dimensions in the ssc file, the moon Pan is no longer in the gap. It is now inside its former path.

Thanks in advance, Chris.

Frank

granthutchison
Developer
Posts: 1863
Joined: 21.11.2002
With us: 22 years

Re: pre6 - crashing at Saturn's rings

Post #2by granthutchison » 14.03.2004, 13:13

fsgregs wrote:Third, I noticed that with the new ring dimensions in the ssc file, the moon Pan is no longer in the gap. It is now inside its former path.
You've picked up the wrong rings or the wrong dimensions somewhere.
The correct ring definition for the new (1.3.2pre6) rings appears in the corresponding solarsys.ssc:

Code: Select all

   Rings {
      Inner   74500  # includes some ringlets inside edge of C ring at 74660
      Outer  140220
      Texture "saturn-rings.png"
   }

Grant

Guest

Post #3by Guest » 14.03.2004, 14:27

Grant:

I'm using the default solarsys file so it must be my Pan ssc file that is wrong. Could you give me the figures for Pan?

Thanks

Frank

granthutchison
Developer
Posts: 1863
Joined: 21.11.2002
With us: 22 years

Post #4by granthutchison » 14.03.2004, 15:15

Code: Select all

   EllipticalOrbit
   {
   Period      0.5750
   SemiMajorAxis   133583
   Eccentricity   0.0000
   Inclination   0.000
   MeanLongitude   188.905 #J2000.0
   }

Grant

Avatar
selden
Developer
Posts: 10192
Joined: 04.09.2002
With us: 22 years 2 months
Location: NY, USA

Post #5by selden » 14.03.2004, 18:04

Frank,

I have a 128MB Ti4200 also, and haven't been able to get it to crash. Which drivers are you using? v53.03 is available on Nvidia's Web site, although v56.55 seems to be available on some vendors' sites.

Also, if you can provide the program counter of the crash, that might help Chris to figure out where it's happening. It's available in one of the informational links in the Windows crash popup.

I suspect the ragged edges of the shadows on the rings won't be fixed for Ti cards. Apparently they're due to the resolution of the image of the circle used to draw them. To get the best shadows with v1.3.2pre6 you need an Nvidia FX card. They require an OpenGL routine that is not available in Nvidia's other cards. An equivalent routine is available in the high end ATI cards, so presumably the new shadow features will be avaliable for them RealSoonNow.
Selden

Guest

Post #6by Guest » 14.03.2004, 20:10

Selden:

I'll get back to you on your questions about Saturn crashing. I have to research things in my system. Thanks for your consideration. I'll post a response soon.

Grant, I may be wrong, but I thought Pan is supposed to be centered in the gap to the right of the image below, having swept out that empty gap in the rings. If true, then here is a screen shot of what I am getting, using the new default ring.png texture that comes with 1.3.2pre 6, plus your ssc file values for both Saturn's rings and Pan.

Image

As you can see, Pan is not in a gap any longer. Maybe I'm wrong and it never was supposed to be in a gap, but .... if I'm correct, then something is screwy.

Frank

Avatar
selden
Developer
Posts: 10192
Joined: 04.09.2002
With us: 22 years 2 months
Location: NY, USA

Post #7by selden » 14.03.2004, 20:49

Frank,

I'm sure you'll be glad to know that I've found a way to crash Celestia every time. Right now, at any rate -- I add that caveat because sometimes it depends on the phases of the moons, well, their exact locations, anyhow. :)

cel://Follow/Sol:Saturn:Pan/2002-06-10T23:56:24.87548?x=Re0R1H4MGS3aDA&y=JrG5e94XyG38/////////w&z=hcF4FcvvHfBz/////////w&ow=0.930567&ox=0.290933&oy=0.208392&oz=0.077307&select=Sol:Saturn:Pan&fov=45.000000&ts=0.001000<d=0&rf=71431&lm=2

Address: 0x0...09d6546d

It does not crash Celestia v1.3.1 final.

System:
256MB 500MHz P3, Win XP Pro
128MB GF4 Ti 4200, drivers v56.55
Celestia v1.3.2pre6
Selden

Avatar
selden
Developer
Posts: 10192
Joined: 04.09.2002
With us: 22 years 2 months
Location: NY, USA

Post #8by selden » 14.03.2004, 21:00

Frank,

It would seem that you may be using a version of solarsys.ssc that does not contain the current values for Saturn's rings. Here's a picture taken of Pan using an unmodified copy of Celestia v1.3.2pre6:

Image
(this links to a larger image)

Pan is right in the middle of the gap.
Selden

chris
Site Admin
Posts: 4211
Joined: 28.01.2002
With us: 22 years 9 months
Location: Seattle, Washington, USA

Re: pre6 - crashing at Saturn's rings

Post #9by chris » 15.03.2004, 08:13

fsgregs wrote:Hi. I have tried out 1.3.2pre 6 and am having some crash problems. I go to Saturn to view the new rings and encounter two problems. First, the chopped edges of the planet shadow on the ring that DJ Night complained about in a separate thread last week is really bad with pre6 on my system, even through I have an Nvidia GeForce 4 TI 4200 with 128 MB video RAM. It is present with 1.3.1 final, but nowhere near as bad.
Really? Can you should me some screen shots? I compared 1.3.1 and it doesn't look any better to me.

I've set my screen with ctrl V to the best vortex shading I can get but no difference at all.
The new ring shadow code only works on GeForce FX cards. GeForce 4 cards don't have advanced enough pixel shaders to render the improved ring shadows. I will be adding celestia.cfg parameters to set the shadow texture resolution--this will give you smoother shadows, though won't look as good as the GeForce FX pixel shader shadows.

Secondly, Celestia locks up and crashes when I'm looking at Saturn's rings about 2 in every 3 tries. Has happened 9 or 10 times so far. I don't know why. I'm using Win XP on a pentium IV with Saturn's default texture and the new ring textures that come with pre6. I don't get a crash using version 1.3.1 and the same texture and solarsys.ssc file. It seems to be something in the pre6 program execution.

Thanks for this report Frank and thanks to Selden for the cel:// URL to reproduce the problem. I was finally able to track the bug down and fix it. I think this bug was the source of most of the recent crashing problems reported in the forum, including Rassilon's troubles with emissive planets. The fix is now in CVS and I'll produce another prerelease soon.

--Chris

Avatar
Topic author
fsgregs
Posts: 1307
Joined: 07.10.2002
With us: 22 years 1 month
Location: Manassas, VA

Post #10by fsgregs » 16.03.2004, 03:59

Selden:

Chris:

Here are the screenshots you requested of the rings with 1.3.2pre6, and 1.3.1., taken from exactly the same point. The first shot is 1.3.1. However, like Andre's post elsewhere in the forum, I am using the KCelestia version of 1.3.1. As you can see, it results in very little of the ugly hatch marks on the rings. The second shot is 1.3.2.



Image


Here is the Celestia 1.3.2 version from the same viewpoint.

Image

I have antialiasing set to 4Xs (the highest on my GeForce 4 TI 4200 card), and anisotrophic filtering set to 4X. I've tried other settings, with no improvement on 1.3.2.


Selden:

OK, I solved my problem with Pan. I was using an older version of the ring.png texture, which I had overwritten when I moved my existing textures back into the Celestia folder after installing 1.3.2pre6. When I reloaded the default texture, Pan fell into place. Thanks for working on it for me.

Frank

Frank

chris
Site Admin
Posts: 4211
Joined: 28.01.2002
With us: 22 years 9 months
Location: Seattle, Washington, USA

Post #11by chris » 16.03.2004, 09:57

fsgregs wrote:Chris:

Here are the screenshots you requested of the rings with 1.3.2pre6, and 1.3.1., taken from exactly the same point. The first shot is 1.3.1. However, like Andre's post elsewhere in the forum, I am using the KCelestia version of 1.3.1. As you can see, it results in very little of the ugly hatch marks on the rings. The second shot is 1.3.2.

Ah . . . In KCelestia, Kendrix increased the resolution of the shadow texture. In the next prerelease, you'll see a ShadowTextureSize option in celestia.cfg. Adjusting this value to 1024 or 2048 will improve the quality of the shadow textures at the cost of using up some additional texture memory. I feel that 2048 is overkill; 1024 is probably the right setting for a 64 or 128MB card. Of course, if you have a GeForce FX you should leave this setting alone and use the pixel shader shadows instead.

--Chris

Avatar
selden
Developer
Posts: 10192
Joined: 04.09.2002
With us: 22 years 2 months
Location: NY, USA

Post #12by selden » 16.03.2004, 12:28

Chris,

I noticed that the FX rendering path is not the default when one has an FX card. Is this an oversight or was there a reason?

Pre6 always uses the "Opengl+combiners" path when it starts and one must type a Ctrl-V to enable the FX shadow features.
Selden

Toti
Developer
Posts: 338
Joined: 10.02.2004
With us: 20 years 9 months

Post #13by Toti » 16.03.2004, 14:27

Selden:
That URL hangs up Celestia 1.3.2 pre5 on my system, too. (Detonator 44.xx drivers) (Message = Celestia caused an error in NVOPENGL.DLL...)
I upgraded to 53.04 drivers and it still crashes (Message = Celestia caused an error in <unknown>...)

Intel P3 1Ghz 256Mb RAM NVidia RIVA TNT2 M64 32Mb VRAM Windows Me

Avatar
selden
Developer
Posts: 10192
Joined: 04.09.2002
With us: 22 years 2 months
Location: NY, USA

Post #14by selden » 16.03.2004, 14:36

Toti,

That's to be expected. According to the posting by Chris above, which he wrote after pre6 was made available, it'll be fixed in pre7, which may be downloadable in the not too distant future.
Selden

chris
Site Admin
Posts: 4211
Joined: 28.01.2002
With us: 22 years 9 months
Location: Seattle, Washington, USA

Post #15by chris » 16.03.2004, 17:15

The FX rendering path isn't the default yet because it's still somewhat experimental. I hope to make it the default before the final release of 1.3.2 however.

1.3.2pre7 is now available with (hopefully) a fix for the crashes at Saturn's rings and other locations. It also includes celestia.cfg fields to turn on smoother shadows (for those without GeForce FX cards), orbits, and rings.

Here they are:

http://www.spacesim.net/prerelease/cele ... .2pre7.exe

And the exe only:

http://www.spacesim.net/prerelease/cele ... e7-exe.zip

--Chris


Return to “Bugs”