Page 1 of 1

Comet's coma

Posted: 22.09.2008, 19:17
by ElChristou
I have a problem with comets since yesterday build. Coronas are not displayed anymore. Someone can confirm?

Re: Comet's corona

Posted: 22.09.2008, 19:25
by chris
ElChristou wrote:I have a problem with comets since yesterday build. Coronas are not displayed anymore. Someone can confirm?

I'm still seeing comet tails.

--Chris

Re: Comet's corona

Posted: 22.09.2008, 19:43
by ajtribick
chris wrote:I'm still seeing comet tails.
Me too.

Re: Comet's corona

Posted: 22.09.2008, 19:45
by selden
I just now downloaded 4455 and rebuilt it from scratch.
Comets still look the same as before.
When I set the date to 9 Feb 1986, Halley has both a coma and a tail, and Hale-Bopp has a coma (although it really shouldn't at 20 AU from the Sun).

Make sure you have comets enabled.

Re: Comet's corona

Posted: 22.09.2008, 19:53
by ElChristou
I see the tails but not the coronas... I'll try a new build to see...

Re: Comet's corona

Posted: 22.09.2008, 20:02
by selden
I think you mean "coma" not "corona".
The coma of a comet is the brightest part closest to the comet.
The corona is the outer layer of the Sun, visible during an eclipse.

Re: Comet's corona

Posted: 22.09.2008, 20:08
by ElChristou
Yep, sorry Selden. Here is what I have with a 2 minutes ago build:

bor.jpg


No coma at all... :?
How can I debug this?

Re: Comet's corona

Posted: 22.09.2008, 20:19
by chris
ElChristou wrote:Yep, sorry Selden. Here is what I have with a 2 minutes ago build:

No coma at all... :?
How can I debug this?

Can you provide a cel URL?

--Chris

Re: Comet's corona

Posted: 22.09.2008, 20:47
by ElChristou
chris wrote:Can you provide a cel URL?

I have no comas at all whatever comet I visit. Anyway, here is one:

cel://Follow/Sol:Borrelly/2008-09-22T20 ... 2306&ver=2

Re: Comet's corona

Posted: 22.09.2008, 21:07
by selden
That url looks (almost) OK under Windows XP when running Celestia r4460.
It does have a dark region between the coma and the tail, but the coma is being drawn.

You might try deleting all of the .h and .cpp files and doing an svn update to make sure you have all of the correct files, then rebuild. Otherwise, it might be a Mac graphics driver problem :(

Re: Comet's corona

Posted: 22.09.2008, 21:25
by t00fri
ElChristou wrote:Yep, sorry Selden. Here is what I have with a 2 minutes ago build:

bor.jpg


No coma at all... :?
How can I debug this?

Nope. everything fine under Celestia-KDE (build 4450)

Fridger

Re: Comet's corona

Posted: 22.09.2008, 21:29
by ElChristou
selden wrote:You might try deleting all of the .h and .cpp files and doing an svn update to make sure you have all of the correct files, then rebuild. Otherwise, it might be a Mac graphics driver problem :(

I'll try that...

Re: Comet's corona

Posted: 22.09.2008, 21:41
by ElChristou
Nope, I delete all src and update my source, same thing.
Must be something with the Xcode project, but I think I will wait the next changes by DW to rebuild with a fresh Xcode file...

Re: Comet's corona

Posted: 22.09.2008, 22:26
by t00fri
Christophe,

how about correcting the misleading title of your thread: corona -> coma?

Fridger

Re: Comet's corona

Posted: 23.09.2008, 00:19
by Cham
I don't have any problems with comets.

Re: Comet's coma

Posted: 26.09.2008, 03:11
by chris
Reposting from the dev list:

Here's a patch (and replacement render.cpp) that should fix the comet bug reported by Christophe:

http://www.celestiaproject.net/~claurel/celest ... /cometbug/

The calculation of comet tail normals was incorrect--for the front of the comet tail, it was taking the square root of a negative value. Eventually, a negative infinity was passed to the graphics hardware. The floating point units on all but the most recent graphics hardware is not IEEE754 compliant, so the handling of infinities and NaNs is essentially unpredictable. That's why the bug appeared on Christophe's system, but not on anyone elses.

The patched code also fixes the problem where the comet's coma was being rendered regardless of the distance from the Sun. The rather odd looking bright 'cap' is also gone.

--Chris

Re: Comet's coma

Posted: 26.09.2008, 06:17
by ElChristou
chris wrote:Reposting from the dev list:

Here's a patch (and replacement render.cpp) that should fix the comet bug reported by Christophe:

http://www.celestiaproject.net/~claurel/celest ... /cometbug/

The calculation of comet tail normals was incorrect--for the front of the comet tail, it was taking the square root of a negative value. Eventually, a negative infinity was passed to the graphics hardware. The floating point units on all but the most recent graphics hardware is not IEEE754 compliant, so the handling of infinities and NaNs is essentially unpredictable. That's why the bug appeared on Christophe's system, but not on anyone elses.

The patched code also fixes the problem where the comet's coma was being rendered regardless of the distance from the Sun. The rather odd looking bright 'cap' is also gone.

Do you mean Selden's previous image with the bright cap wasn't a wanted effect?
I've build upon 4466 and do have the coma, but it react now as the rest of the tail in transparency or brightness. The coma itself alone is no more visible... (is that what you wanted?)

Re: Comet's coma

Posted: 11.03.2009, 12:42
by Dentarthurdent
Where do the patch and replacement render.cpp files go? I've had this same problem with comets and their comas. Much appreciated.

Re: Comet's coma

Posted: 11.03.2009, 13:46
by selden
Dentarthurdent,

You need to build Celestia from scratch. It's not a patch that can be applied to v1.5.1. This bug is fixed in the current SVN revision of Celestia. I don't know if anyone is still providing a pre-compiled copy of that revision. For what it's worth, I'm feeling optimistic about the availability of v1.6. If things go well, a "release candidate" might be available within the next month or so.

There are notes elsewhere on this Forum and in the Celestia Wikibook describing how to get started compiling Celestia. http://en.wikibooks.org/wiki/Celestia#Development