Moon satellite orbit bug

[bdm]

When moons have satellites of their own, the orbits of those satellites is misplaced and drawn with the wrong radius.

Celestia version is 1.4.1.



The smaller round orbit shown above exhibits a lot of jitter. It jumps around the screen a lot and sometimes it is not visible at all. The half-circle orbit is steady and drawn at the right size.

[selden]

Can you provide an SSC?

I can't reproduce the jitter problem under either 1.4.1 or current cvs code when orbiting a moon around Phobos.

p.s. the orbit path code has been rewritten for the next release: the whole orbit is visible in the cvs code although only half the orbit is drawn by 1.4.1.

[Cham]

I'm experiencing his problem very frequently on 1.4.1 and 1.4.2, especially with small orbits objects. I didn't noticed this with the new code, yet, but there are other problems with it (orbital paths gets some holes and dashes ...)

[selden]

In general, if Chris can't reproduce it, he can't fix it.

Please provide some failing SSCs and URLs.

[bdm]

Some notes:
* The fictional planet is a distant world in its solar system
* It won't resemble the screenshot because models and textures are not included
* The moon is distant from the planet (because distant moons are more likely to have satellites of their own)
* When you load the URL below, you may not see the effect immediately. If you hold the right mouse button down and move the mouse around a bit, you should see orbit arcs rendering at random locations as you move around.

URL:
Orbit Weirdness

SSC:

Code: Select all

"Ovatus" "Lalande 21185"
{
    Texture     "neptune.*"
    Color       [0.5 0.5 0.5]
    Albedo      0.26   
   
    Radius      27906.26
    Oblateness  0.145854
    #$Density   1.256

    EllipticalOrbit
    {
        Period           146.9129739063
        SemiMajorAxis    27.4661114354
        Eccentricity     0.02299226
        Inclination      062.317001
        AscendingNode    175.204545
        MeanLongitude    206.910651
        LongOfPericenter 277.953797
    }

    Atmosphere # Copied from Neptune
    {
        Height 130
        Lower [ 0.60 0.65 1.00 ]
        Upper [ 0.50 0.55 0.90 ]
        Sky   [ 0.50 0.70 0.90 ]
    }

    #$AxialTilt          028.903005
    #$AxialNode          063.997314
    Obliquity            077.345645
    EquatorAscendingNode 201.641785

    RotationOffset       000
    RotationPeriod       7.00930204
}


"Ovatus XI" "Lalande 21185/Ovatus"
{
    Radius               21.13
    Texture              "asteroid.*"
    Color                [0.400 0.400 0.400]
    Albedo               0.150

    EllipticalOrbit
    {
        Period           737.1679251216
        SemiMajorAxis    8749267.57
        Eccentricity     0.441316
        Inclination      032.157676
        AscendingNode    352.023682
        LongOfPericenter 313.548387
        MeanLongitude    198.364208
    }

    #$AxialTilt          031.781854
    #$AxialNode          019.193701
    Obliquity            062.941352
    EquatorAscendingNode 003.235570

    RotationOffset       090.000000
    RotationPeriod       14.69696520
}

"Ovatus XIA" "Lalande 21185/Ovatus/Ovatus XI"
{
    Radius               8.14
    Texture              "asteroid.*"
    Color                [0.400 0.400 0.400]
    Albedo               0.150

    EllipticalOrbit
    {
        Period           0.61237355
        SemiMajorAxis    67.661
        Eccentricity     0.013290
        Inclination      000.010496
        AscendingNode    029.493944
        LongOfPericenter 177.002293
        MeanLongitude    093.868304
    }

    #$AxialTilt          000.075468
    #$AxialNode          119.721671
    Obliquity            000.070853
    EquatorAscendingNode 141.824071

    RotationOffset       135.000000
    # RotationPeriod - undefined defaults to synchronous   
}


[selden]

bdm,

I see the problem under v1.4.1 final, but I can't get it to fail using CVS code. So that's one bug that seems to be fixed.

Cham,

Can you provide an SSC and/or URL for what you're seeing?

[chris]

I haven't had a chance to try your SSC, but according to Selden's report, the orbits look fine with the latest CVS code. That's good news, and not surprising--fixing this sort of bug is one of the major reasons that I rewrote the orbit code. We're working on getting the code into a state where we can make an official 1.5.0 prerelease package. In the mean time, you can build Celestia from CVS yourself or try one of the unofficial builds that people have been making available.

--Chris

[Cham]

Chris,

there's a clear major problem with orbits in the new code. I gave a picture of the bug, in another thread. Look there :

http://www.celestiaproject.net/forum/viewtopic ... 8&start=45

and see the picture with that dashed blue curve (orbit with "holes"). This is very annoying, and can be reproduced easily.

[selden]

Cham,

It would help a lot if you could provide the SSC and a Cel URL for that viewpoint. It's not in the standard Celestia distribution and Chris really doesn't have the time to type in random values to see if they fail.

[Cham]

Yes, I'll post an SSC code when I'll be back home, later today. Right now, I'm at work.

But this bug pretty general and systematic, as soon as there are many orbits shown at the same time, and/or an object (an asteroid here) seen in front.

[selden]

The only times I see things like that in the current CVS code, they're "real" -- a dark foreground orbit path crossing over in front of a bright background orbit path. So an example of something different would help a lot.

[Cham]

I'm home now.

Selden (and Chris), here's a small addon which is revealing easily the orbit problem on 1.5.0 ("holes" and "dash type" lines), at least on my system :

http://nho.ohn.free.fr/celestia/Cham/pharos.zip (zip file size : 14.4 KB)

Just select any asteroid (Ast 23 was selected below), and show the planets orbits. A seen from that asteroid (a simple sphere with the asteroid.jpg texture), I get holes on the planets orbits. See the picture below :

[selden]

Cham,

I'm sorry to say that the only "breaks" that I see are where dark (brown) asteroid orbit paths are drawn in front of bright planet orbit paths. Actually, the only "breaks" I'm seeing are in the red orbit path of Pharos. When I turn off the asteroid orbits, Pharos' orbit path is continuous as it should be.

I'm not seeing the breaks that are visible in your screengrab. But I'm not seeing asteroid orbits in your screengrab, either.

Can you provide a Cel URL of the viewpoint of your screengrab?

I'm starting to wonder if it's a bug in the MacOS OpenGL driver when it draws OpenGL lines.

Have you looked at your Magnetic Earth since you installed the cvs version of Celestia? Does it show any signs of breaks?

[Cham]

Selden,

I never saw those "breaks" or "holes" on any previous version of Celestia.

Yes, of course, I've tested my Magnetic-Earth, and there's no problem there at all.

The "holes" appears to be related in some way with the asteroids orbits, even when they aren't showing, but I may be wrong here. Until now, I only noticed the "holes" on planetary orbits, and didn't saw any on an asteroid orbit yet. Here's an URL, for a point of view of Ast 23 :

cel://Follow/Ast 23/2004-08-07T00:25:52.13296?x=lCUNMZllFVLzrY0&y=MeKsKnEcsR/R6f/+/////w&z=0f5rd7R32zchHbf//////w&ow=0.022705&ox=-0.633121&oy=0.009793&oz=-0.773658&select=Ast 23&fov=38.300713&ts=0.000000&ltd=0&rf=578999&lm=2082

[Cham]

AAaack !

I just discovered another bug : I can't use my own URL ! It gives me a black screen in Celestia, with everything turned to OFF and a FOV of 0?°. I can't even change the FOV after that and need to relaunch Celestia !?

What the ... ?

[Cham]

Here's the same bug on Earth's orbit :



URL (does the URL above works for you ?) :

cel://Follow/Earth/2006-09-18T21:29:03. ... 78995&lm=7

Please, don't forget to select Earth by clicking on it, if its orbit isn't selected in red.

[chris]

Here's the same bug on Earth's orbit :

URL (does the URL above works for you ?) :

cel://Follow/Earth/2006-09-18T21:29:03. ... 78995&lm=7

Please, don't forget to select Earth by clicking on it, if its orbit isn't selected in red.

It looks like you have two identical orbits overlapping each other, one of them selected. Is that the case here? It almost looks like z-fighting, though I thought I had depth writes disabled for orbit paths.

--Chris

[Cham]

Chris,

yes, in this special case, there's another orbit (Magnetic-Earth) superposed on Earth's one. Not exactly the same, however.

The bug can be reproduced in any system, and not just with superposed paths.

[selden]

Cham,

Those breaks in the Earth's orbit path are "correct." That's the orbit path of your Magnetic Earth being drawn partially on top of the Earth's orbit path.

Remember that orbit paths are actually straight lines drawn between 100 points along the object's orbit. In this case, sometimes a segment of one orbit path is in front, and sometimes a segment of the other. You can change the number of points that are used by editing celestia.cfg. That will change how the "breaks" look. If you don't want the drawing of orbit paths to interfere with one another, you have to place the bodies in significantly different orbits.

I first encountered this problem in my Across to Mars Addon. Its spacecraft are composed of several different models, all following exactly the same orbit at exactly the same location. As a result, when one of the paths might be expected to be red because a body travelling on that orbit is selected, it usually is blue instead because the red path is being overwritten by one or more of the blue "unselected" orbits of several of the other models.

[Cham]

Selden,

your explanation seems to be convincing, but I experienced the same problem with the planet orbit in my "Pharos" addon, while no orbit was intersecting another (the asteroids orbits were OFF). Also, I'm using

OrbitPathSamplePoints 1024

in my config file.