Win32 Executable from SVN

[Dirl]

Celestia SVN executables from Mr. Guillermo Abramson:
http://cabfst28.cnea.gov.ar/~abramson/celestia/svnexec/

[abramson]

Celestia SVN executables from Mr. Guillermo Abramson:
http://cabfst28.cnea.gov.ar/~abramson/celestia/svnexec/

Due to a change of servers, my webpage has changed address. Find the binaries here: http://fisica.cab.cnea.gov.ar/estadistica/abramson/celestia/svnexec (updates every night).

Guillermo

[BobHegwood]

Again, thanks very much for doing this my friend.
Much appreciated by the Brain-Dead

[Derek]

Guillermo,

Appreciate your SVN I seem to have a problem when viewing the orbits of the planets and asteroids.
1. Pluto appears as a planet not the dwarf as in 1.6.0
2. Pluto's orbit in svn 4880 is broken and in 4903 is a strange shape, Ive eliminated any extras causing problems.

[abramson]

2. Pluto's orbit in svn 4880 is broken and in 4903 is a strange shape, Ive eliminated any extras causing problems.

Hi. I just tried my build 4903 in place of a fresh installation of 1.6.0, and observe that Pluto-Charon orbit is indeed not closed. There is a gap ~120 degrees ahead of Pluto-Charon (see figure).

Custom built svn-4903. See the gap in Pluto-Charon orbit (red), on the right of the figure.

The same happens in 4898, 4890, 4881, 4870, 4863, 4855. And NOT in 4787 or earlier. I don't know why.

If Derek meant a gap when he says "broken", that's what I see. I don't see any other "strange shape". All svn versions with the problem behave exactly the same. I don't remember noticing this before.

Does anybody else see this in their own compiled versions? Is it a bug, or a glitch in my own compilation?

Svn 4880 (mentioned by Derek) is not among my builds.

Guillermo

[Derek]

Hi Guillermo,
Sorry 4880 was a mistake should have read 4890, further investigation I've found it's a combination of addons together with your svn 4898 and 4903 which are causing the other problems.

[selden]

The orbit path code has been completely rewritten.
The new code was added to Celestia as of r4825.

Hopefully its bugs will be fixed in the not too distant future.

r4825 | cjlaurel | 2009-08-06 22:15:26 -0400 (Thu, 06 Aug 2009) | 2 lines

Rewrote orbit drawing code to fix multiple shortcomings. New orbit code uses adaptively subdivided cubic curves to draw orbits more accurately and without numerical precision artifacts. The orbit rewrite uses Eigen throughout (instead of the Celestia vector classes.)

[abramson]

OK. Selden, should I report it as a bug, or does everybody already know about this?
G

[chris]

OK. Selden, should I report it as a bug, or does everybody already know about this?
G

The gap is intentional, not a bug. Because of perturbations from other bodies, Pluto's orbit is not a perfect ellipse. Celestia now draw planet orbits over a duration of one mean orbital period, starting at the current time minus one half the orbital period. Thus the gap appears approximately 180 degrees away from the body, though this will vary for significantly eccentric orbits like Pluto's.

This may not be the best way to draw the orbit. Indeed, the fact that a number of people are reporting this as a bug suggests that Celestia should have some other strategy for rendering orbits that are approximately periodic. One possibility is to draw the orbit starting from the current time back to one orbital period earlier, fading out the tail end of the orbit so that the gap looks intentional.

The gap should not be confused with a very real bug when drawing highly eccentric elliptical orbits. I have a partial fix for this already, and hope to have it completed today.

--Chris

[abramson]

Wow! Such a detail! I wasn't aware of this delicacy. Thanks, Chris.

The fading of orbits boud be a fine thing. Even to draw them shorter than a period: for example, to show the apparent path of planets from a vantage point.

Guillermo

[ajtribick]

The gap is intentional, not a bug. Because of perturbations from other bodies, Pluto's orbit is not a perfect ellipse. Celestia now draw planet orbits over a duration of one mean orbital period, starting at the current time minus one half the orbital period. Thus the gap appears approximately 180 degrees away from the body, though this will vary for significantly eccentric orbits like Pluto's.

This may not be the best way to draw the orbit. Indeed, the fact that a number of people are reporting this as a bug suggests that Celestia should have some other strategy for rendering orbits that are approximately periodic. One possibility is to draw the orbit starting from the current time back to one orbital period earlier, fading out the tail end of the orbit so that the gap looks intentional.

Another approach that may be interesting would be to draw the osculating orbit. Not sure how feasible that would be though.

[chris]

The gap is intentional, not a bug. Because of perturbations from other bodies, Pluto's orbit is not a perfect ellipse. Celestia now draw planet orbits over a duration of one mean orbital period, starting at the current time minus one half the orbital period. Thus the gap appears approximately 180 degrees away from the body, though this will vary for significantly eccentric orbits like Pluto's.

This may not be the best way to draw the orbit. Indeed, the fact that a number of people are reporting this as a bug suggests that Celestia should have some other strategy for rendering orbits that are approximately periodic. One possibility is to draw the orbit starting from the current time back to one orbital period earlier, fading out the tail end of the orbit so that the gap looks intentional.

Another approach that may be interesting would be to draw the osculating orbit. Not sure how feasible that would be though.

I've considered this before, and it's completely feasible. The new code draws orbits a series of cubic curves, and there are well-known algorithms for approximating an ellipse as cubics. Only a small number of cubics are required to get quite good accuracy. See for example http://www.tinaja.com/glib/ellipse4.pdf

I see two drawbacks with using osculating orbits:
- It's a different approach than used for non-periodic trajectories, where the actual path through space is shown
- The shifting, precessing ellipses might be confusing to users without some celestial mechanics background

Regardless, the osculating orbits would be very interesting to see, as they should show the orbital evolution of bodies more clearly than other schemes for drawing orbits. If you have some time to write code to calculate a piecewise cubic approximation to an ellipse, I'd add an osculating orbit view to Celestia, at least as an option.

--Chris

[Teto]

Celestia SVN executables from Mr. Guillermo Abramson:
http://cabfst28.cnea.gov.ar/~abramson/celestia/svnexec/

Due to a change of servers, my webpage has changed address. Find the binaries here: http://fisica.cab.cnea.gov.ar/estadistica/abramson/celestia/svnexec (updates every night).

Guillermo

This built is very appreciated.
I've found one problem with built svn4903: When you're approaching a star, its texture is half cutted. Very funny. Is it a bug of the built or of the program itself? I don't know, I'm just a final user.

Teto.

[abramson]

I've found one problem with built svn4903: When you're approaching a star, its texture is half cutted. Very funny. Is it a bug of the built or of the program itself? I don't know, I'm just a final user.

I also see it, Teto. That's definitely a bug, and it's not obvious what's happening.

What you are seeing is the Sun from the inside, which is transparent in Celestia (textures are only visible from the outside). Solar radius is 696000 km, one shouldn't see that until the distance is that value, not more than 1 million as it happens. The size of the Sun is right, though (judging from it's apparent size). Something is been wrongly calculated here.

This is definitely a bug. Strangely, it seems to happen only after visiting Earth! If you interrupt the startup script before going to Earth (with ESC) you can go right to the surface of the Sun without seing this artifact. You can even go to other planets and return to the Sun. But after going to Earth (or the Moon, but the Earth always gets in sight) the next time you go to the Sun you see it. Isn't it weird? Can anybody confirm this? I think you should report it as a bug, and I hope it doesn't give Chris a headache.

Guillermo

[Teto]

Done!

Teto.