Strange Behavior of Sol when Pluto's Hydra is On-Screen

[jeffmack]

I believe that the following may be a Celestia bug, as I can reliably reproduce it on two completely different platforms:

I "GoTo" Pluto's Hydra, and then "Orbit" about Hydra (Right-Click/Drag) until I bring Sol into view. Then, when I shift the view about, either by changing the camera view (Left-Click/Drag) or by "Orbiting" Hydra (Right-Click/Drag), Sol will exhibit strange behavior. This behavior alternates between up to seven different appearances as the view is shifted. These seven different appearances of Sol consist of: 1) Normal Sol (i.e. Sol itself with its halo); 2) Sol without its halo; 3) No Sol at all; 4) Lower left-hand half of Sol's halo blacked out; 5) Upper right-hand half of Sol's halo blacked out; 6) Upper left-hand half of Sol's halo blacked out; and 7) Lower right-hand half of Sol's halo blacked out.

When Hydra is completely moved off-screen, this strange behavior ceases (Sol consistently appears with its halo).

I witness this behavior on both of the systems on which I have Celestia installed (Celestia 1.6.0 in both cases). The first system is my home laptop - the configuration of which is described in my signature below. The second system is a Dell tower running Microsoft Windows XP Home Edition - Service Pack 3 with an Intel G33/G31 Express graphics chipset. The latest graphics drivers are installed on both systems.

This problem occurs in all available render paths of Celestia.

Attached is a screen capture from my home laptop.

[selden]

Does it happen with any other moons or asteroids? e.g. Nix?
Does it happen with all Star Styles? (type a Ctrl-s several times)
Are the current graphics drivers installed?

I can't reproduce the problem on my computer.
One thing I noticed immediately is that Hydra uses asteroid.cms for its Mesh.
That Mesh is used by many (all?) of the moons defined in numberedmoons.ssc
e.g. Trinculo or Psamathe


System:
8GB 1.6GHz Core i7-Q720; Win7 Pro x64
512MB Nvidia NVS 3100M; Forceware 258.96
Celestia r5135 built with Visual C++ 2008 Express v9.0.30729.1 SP, .NET v3.5 SP1
Celestia v1.6.0 as distributed

[jeffmack]

As noted in my original post, the latest graphics drivers are installed on both systems.

I just experimented with your other questions and found that the behavior occurs with all three Star Styles (the attached screen capture was made with the Star Style of "Points"). It occurs for Nix as well, but it does not occur for Pluto or Charon. I tried the asteroids Juno and Vesta; it does not occur for them either.

It should also noted that with Nix, sometimes the blackout of Sol's halo is pie-shaped, rather than half.

I will try some other objects when I have more time later.

Thanks!

[selden]

Your description suggests that it's somehow related to CMS model files.

p.s. People often believe that they're running the most recent updates but aren't actually. Double check the actual version numbers. Some systems can be updated using current drivers downloaded from the manufacturer of the graphics chipsets, while others can only be updated using older (buggier) drivers provided by the manufacturer of the computer (Dell is always behind). Use the chipset manufacturer's drivers if you can (e.g. intel or AMD). Some Dell laptops can use the chipset manufacturer's drivers and some can't. (Fortunately, mine can.)

Also, sometimes drivers seem to update but don't actually replace the lowest level part of the driver software. To make sure the hardware drivers get updated, you usually have to deinstall the old drivers, reboot, install the new drivers and reboot again.

[jeffmack]

A) I installed the latest drivers on both systems within the last week.

B) I downloaded the drivers from the graphics card manufacturers' Web sites for both systems (ATI/AMD and Intel respectively).

C) I completely uninstalled the old drivers prior to installing the new ones (with intervening reboots).

D) I did double check the drivers' actual version numbers.

They are the most current.

(I am a computer systems engineer - I know how to do this. )

Can anyone else reproduce this?

[selden]

There's no need to get upset. It's just that you have not explicitly stated the versions that you're running, As a result, I have no way to verify whether or not they're the most recent versions. (Many people don't even know how to find out what versions they're using. But presumably that's not the case for you )

As I said, though, so far your symptoms are consistent with there being something wrong with how .CMS models are being drawn. If that's the case, it's not clear if Chris will be fixing the problem. He's been trying to deprecate that format for some time now.

[fungun]

I have seen this also. But with models of starships that I have in addons.
It is really bad when you get up close to said starships and the sun is at a fair distance away.
I just assumed it was the space that the model is occupying crossing the view of the sun.
Never tried it with a moon that is made from a model.

Tim

[fungun]

Sure now I can't find a ship to do it
But Hydra does.

[jeffmack]

I'm not upset Selden. If I were, I wouldn't have included the smiley face.

Thanks fungun. Glad I'm not alone!

Thanks to the both of you!

[Marco Klunder]

Indeed you're not alone...

Sun_black.jpg

Sun_RB_black.jpg

Sun_LU_black.jpg

But I must also say, I only had this reproducable on a DELL laptop, and in Jeff's footnote I also read that he uses a DELL
Reproducing it was also very distance dependent.
When close to Hydra, the problem practically did not occur, or when I saw something, it already was immediately corrected within a fraction of a second.
then I went to a distance of about 3000 km (see pictures), where I could reproduce the problem many times.

On the other hand, back on my normal PC with NVIDEA graphic card (see footnote) I could NOT reproduce it and I really tried hard

Marco

[jeffmack]

I see it on two Dells, correct - but everything else is different about the two systems (processor, mainboard, chipset, graphics card & graphics card manufacturer, version of Windows, etc.).

I believe I also see a relationship with distance (or at least the apparent size of the moon on-screen - the problem seems to be less severe with Nix than with Hydra). I also would sometimes see the "auto" correction with Nix, whereas the behavior would persist with Hydra.

I see that Marco has galaxy rendering enabled, whereas I do not.

Thanks Marco!

[DOJOMO]

I am able to reproduce the same effect. It seems to occur at the 3000 kilometre distance. It also occurs with all star styles. My laptop is a Sony. Celestia svn version 5135.
Regards, David

[jeffmack]

Ahhh, so now we have an example with an Nvidia graphics adapter - as well as a "non" Dell!

Thanks David!

[Marco Klunder]

I see it on two Dells, correct - but everything else is different about the two systems (processor, mainboard, chipset, graphics card & graphics card manufacturer, version of Windows, etc.).

As stated in my previous response, my footnote configuration does NOT represent the DELL laptop configuration and I cannot reproduce the problem on my footnote configuration (untill now, i'm still experimenting). I could identify the same problem you report on my DELL laptop, but not on the configuration stated in my footnote, so this conclusion is WRONG.

Marco

[jeffmack]

My conclusion is based on David ("DOJOMO")'s configuration.

[selden]

I managed to reproduce the clipping of the image of the Sun (all star styles, all render paths, several versions of Celestia) when Hydra is at a distance of 3000 km, so it certainly seems to be a bug in Celestia, perhaps exacerbated by differences in graphics drivers. It just flickers intermittently on my computer, though; the bug isn't visible most of the time.

System: Dell Latitude E6510
CPU: 8GB, 1.60 GHz Core i7-Q720; Win7 Pro x64
Graphics: 512MB DDR3, Nvidia Quadro NVS 3100M; ForceWare v258.96
Disk: 320GB, 7200RPM, Seagate ST9320423AS
Display: 1920x1080, 60Hz, 32bit color, Vizio VX240M
Celestia: v1.6.0, v1.6.1(pre-alpha),
r5135 (compiled w/VS C++ 2008 Express, v9.0.30729.1 SP)

[jeffmack]

Yes, in my case, the blackouts flicker with Nix at the default "GoTo" distance - but persist with Hydra at the default "GoTo" distance.

[jeffmack]

Selden,

Do I need to report this bug elsewhere (i.e. SourceForge)? Or can I consider it already officially reported?

Thanks!

[selden]

I suggest reporting it on SourceForge. Chis hasn't logged in to the Forum for a while, and might not notice the discussion.

[jeffmack]

OK - done.