Celestia Forums

Hi folks,

have been lost in the "outer space" of business for a long time... just a few days before I finally found some time to install the long-ago downloaded 1.4.1.

A LOT OF THANKS TO ALL!!!

@Fridger & Toti: Galaxies look very amazing now. I guess I will set up at least part of the Virgo Supercluster now, sooner or later...

@Chris: the multiple star .stc is what I dreamed of. Really fantastic what a few lines of .stc code now put on the screen. One of the first things I tried was something I rambled about long time ago:

...and it simply looks *RRRRRRRRRR*
(maybe somebody else already has done this ...meanwhile).
BTW, is it something known the texture "jumps" from one star to the other when the system rotates? It can be best observed if looking along the rotation axis.
(I've left all config files yet untouched since compilation of 1.4.1, just dropped in some other addon files and hires textures).

~Diane.

Wow, good to see you back. If you think 1.4.1 is good wait till you see the 1.5 pre. New galactic structures and atmospheric rendering are excellent.

Oh yeah and just an FYI to all interested, I am still working on my Beta Lyrae addon. I just got caught up in my work with other cataclysmic binaries

See http://celestiaproject.net/forum/viewtopic.php?t=8797

here is a look at DQ HER and intermediate polar...

EDIT (I don't know why I have that funky square artifact there...)

Diane,

Unfortunately, I don't see anything that could be described as "texture jumping". By default, all stars now use the same texture and differ only by the colors generated from their classifications, for what that's worth.

There is some noticable jitter if I turn up the simulation time rate, but that's because I'm only getting about 18 fps.

There is a distance in "Star Style: scaled disc" where the two of them start to blur together rather poorly. Maybe that's what you mean?

The scaled disc algorithm isn't all that great and will be replaced in v1.5 with a better one based on HDRI (high dynamic range imaging).

Hehehee !

Medusa, you really are happy with Celestia 1.4.1 ?? Really ? You're late ! Wait to see 1.5.0 which should be out soon. Your jaw will drop ! Hehehehe !

Selden,

what I see is that the texture disappears for the star which is more far away from the observer. This happens if the system rotates (timerate 1000000x or so). There also seems to be something like a radius change (very little but noticable if close enough).

System Config:
I 've Debian Linux 3.1, installed everything necessary to get Celestia compiled.
Nvidia driver 8762, standard config done by dexconf & nvidia-xconfig, then entered a few lines in the device section of /etc/X11/XF86Config-4 to get two monitors running:



With that, I usually start Celestia in a window approx 2000x768 pix wide.
All Celestia config files I left untouched so far, Celestia itself compiled with GTK interface (I use the tiny twm window manager to save as much as possible memory for Celestia).

Maybe it's a problem of the big window? I saw disappearing (hires) textures (black patches) before -- since I use two monitors but usually just a few moments before Celestia crashed due to excessive memory usage.
(This happens if the virtual memory usage of Celestia has grown in size to approx 2.7 GB - I have 2 GB of physical RAM and 1 GB of disk swap space).

~Diane.

Diane,

Traditionally one would allocate swap space 3-4x the size of main memory. That might help your crashing problem, although the performance at that size wouldn't be very good.

Sorry, I don't see the texture problems you report on the Windows systems that I use. They have more recent drivers than you mention (v91.47 ) but I don't think that's the problem.

The only thing similar that I see is what I described for when the star images are extremely small on the screen. When Celestia switches to using its algorithm for unresolvable objects, the texture is no longer drawn and the size of the circle that it draws is not perfectly matched to the size when the object is resolved. This is consistant with it being the more distant object that has the problem.

It could be that Celestia is confused by the width of your display, I suppose. It uses the display height to calculate the size of the display and might be switching algorithms sooner than it should.

Selden,

I just performed some tests. First I set back my system to one monitor to exclude it was a twinview problem. One star texture still disappears, and it happens even if the stars fill the entire window and in only 1000x time.
For testing, I started nothing but X and the twm window manager and a xterm. Single screen, 1024x768 which should be quite common today.
Then I started Celestia, selected "Sheliak" and went to it by the "G" key.
It's always the texture of the star more away which misses. I took two screen shots (.jpg) just before and after such a "switch":





A strange thing is I could not take the screenshots in .png format, this immediately crashed Celestia with a "Segmentation fault". Maybe a hint?

BTW, I know swap space could be a bit larger but when I set up that machine it had only 512MB so 1GB swap seemed pretty much. Upgrade was later step by step.
For the present test it cannot be a matter for I didn't switch to hires so mem usage is quite low. The output below by the "ps" command with some options was taken at the same time as the screen shots:



I don't know if that helps in any way...

~Diane.

Diane,

I think you need the help of a Linux expert. I'm not one.

However... what version of gcc did you use to compile Celestia? There have been problems compiling with the v4 compiler, although I thought those were fixed.

It also sounds to me like you might not have built Celestia with the appropriate jpeg and png libraries, which suggests that other libraries might be incompatible, too. V1.4.1 uses astar.jpg for the surface texture of all stars, and it certainly shouldn't crash when writing a png screengrab.

Unfortunately, I don't know which version are the best to use. Maybe someone else can help.

Selden,

I made some furher test.

First, the effect also is visible when turning Beta Lyrae around by mouse movements. It then looks if not the texture of the farther star disappears but something else appears in front of it.

When I set "Sheliak B" to a wider orbit

for example, then both stars have their texture but the one of the farther star is somewhat hazy.
When "Sheliak B" has still larger orbit

then both stars always have a clear texture.

So I guess it's a matter of stellar atmosphere or star flare rendering. Maybe a macroscopic variant of the behaviour you mentioned in your first reply - let's hope it will be gone with Celestia 1.5.

~Diane.

P.S. I have gcc 3.3.5 and libpng 1.2.8 - ./configure didn't drop out any error or warning about.[/b][/code]

Diane,

Please accept my apologies. I just realized that I'd been trying to see your problem while running Celestia built from current CVS code. Under v1.4.1 I do see something similar to what you do.

The v1.5 code seems not to be using the flare code any more. Here's what I see:



So, it does seem to be fixed.

Please accept my apologies. I just realized......

.


... never mind.

~Diane.