Celestia 1.4.0pre6 for Mac OS X

[dirkpitt]

dirkpitt,

on my system (dual G4 867 MHz, OS X 10.3.3 with ATI radeon 9800 Pro), your previous version (the one you gave me two or three months ago) is FAR superior than this new release. :(

In the "old" version, there are only few minor bugs (I already made some reports to you, two months ago). Now, with the new version, I'm experiencing major screw up, like the models rendering which are all messed up, and those very annoying glitches (lags) while rotating the view.

I hope you'll find the bugs without too much efforts.

Thanks to Chris, the model glitches have a fix, and I think I've discovered what causes the erratic mouse issue
(I thought all along that all of Celestia's brief pauses were due to texture loads, but some don't appear to be).

[dirkpitt]

Dirkpitt, Hank

I use a Radeon 9200 on an iBook G4 933 MHz with 640 MB RAM. OS X 10.3.7, newest driver for Radeon 9200 (from two weeks ago) installed.

I'm encountering the same problems as Cham on his eMac.

Older versions of Celestia worked quite well, but I had to disable the bumpmaps because they crashed the system. There were some minor problems as well (flickering galaxies, ring shadows randomly on/off or with spikes), but they were not bothering me that much.

Ok, just got back from some intensive testing at our local Apple branch (wish I had access to the Apple lab
that's supposedly near Tokyo) and interestingly enough, an iBook G4 800MHz with Radeon 9200
did NOT crash! That is with vertex shaders and everything else all on. The difference? The iBook was
running 10.3.2 (7D24). The ATIRadeon driver is version 1.3.4.7 (1834), while on my PowerBook (10.3.7)
it reads 1.3.36.14 (2598). Somewhere between OS X 10.3.2 and 10.3.6~7, a new bug must have
crept into the ATI driver, i.e., a regression involving vertex shaders.

Now however tempted you might be, I do not recommend downgrading your video driver. Instead I will
file a bug report with Apple. I don't think they want eMacs and Mac Minis (both are ATI 9200) to go
crashing in schools while running Celestia any more than you do.

rthorvald mentioned that Saturn's rings are too dark on the iMac G5. I tried Celestia personally on an
iMac G5 too, and while I thought that the brightness of the rings were fine (non-modified Celestia, medium-res),
I did notice that the "OpenGL vertex program/NVIDIA combiners" render path put a blue tint over all colors.
That, combined with some reports of "pinkness" in some iMac screens and my generally poor experience
with NVIDIA chipsets and... I don't know. I think it's YMMV.

[Marco]

dirkpitt,

Thanks for your investigation! I'll wait with downgrading the driver, I am waiting now more than a year for Celestia to work fine on my iBook, I can still wait some more...

I accidently opened a new topic ("NormalMaps") instead of an answering post to this thread. In short, I'm having some problems with getting NormalMaps to work on a PowerBook. Because there are more VT NormalMaps than VT Bumpmaps available I would really like to play around with them but somehow I can't even get the non-VT NormalMaps to work. Is it possible that it's also a Graphic card/driver/OS/Celestia-problem or is it more likely my dullness?

[selden]

I did notice that the "OpenGL vertex program/NVIDIA combiners" render path put a blue tint over all colors.

What kind of Nvidia card is in the iMac G5?

That render path includes support for "Haze". The exact color will depend on the Haze declarations specified in the SSC file for Saturn, which is pure blue:

Code: Select all

   HazeColor [ 0.0 0.0 1 ]
   HazeDensity 0.25

However, it should only be visible on the planet, not on the rings. Are you actually seeing a blue tint added to the rings?

[dirkpitt]

I did notice that the "OpenGL vertex program/NVIDIA combiners" render path put a blue tint over all colors.

What kind of Nvidia card is in the iMac G5?

That render path includes support for "Haze". The exact color will depend on the Haze declarations specified in the SSC file for Saturn, which is pure blue:

Code: Select all

   HazeColor [ 0.0 0.0 1 ]
   HazeDensity 0.25

However, it should only be visible on the planet, not on the rings. Are you actually seeing a blue tint added to the rings?

The iMac G5 is an fx5200. But now that you mention it, it might have been just the planet. There's
no reason a blue haze should be on the rings, unless it was some overblown radiosity effect
or an optical illusion (probably the latter :)

[dirkpitt]

Ok, good news. I posted on the mac-opengl list and ATI promptly mailed me back about the Radeon 9200 issue:

Hi,

Thanks for posting, the trouble Celestia's getting with the 9200 and
recent OSX versions, and it has filtered down to us here now and is on
our list. ? It's one of our top priorities for the coming days. ? I'd be
glad to let you know directly when we make some progress if you like.

btw, Celestia is a very nice app.

Thanks,
- Steve

--
Stephen Balel
<mail address deleted>
<phone # deleted>
Macintosh 3D Graphics Software Group
ATI Research Inc.

[dirkpitt]

I accidently opened a new topic ("NormalMaps") instead of an answering post to this thread. In short, I'm having some problems with getting NormalMaps to work on a PowerBook. Because there are more VT NormalMaps than VT Bumpmaps available I would really like to play around with them but somehow I can't even get the non-VT NormalMaps to work. Is it possible that it's also a Graphic card/driver/OS/Celestia-problem or is it more likely my dullness?

Oops, just spotted this post today. If you have an aluminum powerbook (fx5200 or Radeon 9600/9700),
it ought to do both kinds of normal map just fine. Just make sure that, for VT, you name your .ctx file to
match the normal map def in your solarsys.ssc (or you can edit your solarsys.ssc). For example,
in my solarsys.ssc, I have a line that reads: NormalMap "earth-normal.*" and in my hires textures folder
I have an earth-normal.ctx along with the folder containing the level0, level1, etc images. For non-VT,
you would need to name your normal map image file something like "earth-normal.jpg".

[Matt McIrvin]

Great job, dirkpitt. It's nice finally having the object/location browser!

Unfortunately, now that bump maps and normal maps actually work on my Mac (they are beautiful), my need for a more powerful video card becomes all too apparent... the gap between what it's capable of doing at all and what it's capable of doing quickly has increased. Maybe it's time to treat myself to a new card.

Ring shadows inch closer and closer to working correctly on my old Radeon 7500 with each update to Celestia and the ATI drivers. A point version or two ago, the shadow of Saturn's rings on Saturn started appearing correctly. The shadow of Saturn on the rings is still wonky: frequently it doesn't appear at all, sometimes it is blocky and distorted, sometimes the same but with half of it missing, and-- shockingly-- in the latest version sometimes it looks absolutely correct! This is the first time this has ever happened. There doesn't seem to be any rhyme or reason to it. It might depend on memory usage or some such thing.

[Matt McIrvin]

Whatever happens on the screen is reflected on the surface of Earth! I didn't check, but I assume the cube had an ATi Rage 128, the facotry default installed. It was running 10.3.7

This is a longstanding problem with Celestia in recent Mac OS X versions; it is probably a bug in Apple's "Quartz Extreme" UI acceleration, since it first appeared when Quartz Extreme did. It seems to appear whenever the video card is being pushed close to its memory limit.

The workaround is to not use very large JPG or PNG textures (which have to be loaded all at once); replace high-resolution textures with VT or DDS versions. There are lots of these at the Motherlode.

[Marco]

dirkpitt,

just wanted to tell you that everything is working now.

On the PowerBook it was my mistake that the NormalMaps didn't work, thanks for your help.

On the iBook however, I don't really have a clue to why it works now. Bumpmaps, Normalmaps work great and all the new features work as well along with the old features.

Is it possible that the driver was upgradet without me being notified by the system? I guess not...
Or did you change something in Celestia 1.4 pre 6 in the last week?

If not, then this is really magical... That's what I did:
I downloaded tons of Addons and installed them properly on my iBook. Then I told Celestia to IgnoreGLExtensions [ "GL_ARB_vertex_program" ] by deleting the #. After having started Celestia, I switched the rendering path to OpenGL by typing CTRL+V and - it worked! There was not one feature missing. Then I added the # again and Celestia now still works.

There is absolutely no difference between what I see on the PowerBook and on the iBook.

Do you have any idea of what has happened? If you need more info or if you want me to test something, just let me know.

bye
Marco

[dirkpitt]

On the iBook however, I don't really have a clue to why it works now. Bumpmaps, Normalmaps work great and all the new features work as well along with the old features.

Is it possible that the driver was upgradet without me being notified by the system? I guess not...
Or did you change something in Celestia 1.4 pre 6 in the last week?

That's probably just a fluke. ATI told me they're working on the 9200 bug this week and they haven't released any public updates yet. And I sure haven't changed anything in Celestia that would work around the issue.

[dirkpitt]

I've uploaded a new build of Celestia 1.4.0pre6 for OS X. You can get it here:

http://www.celestiamotherlode.net/creators/dirkpitt/celestia-osx-1.4.0pre6-20050201.zip

Changes:
- Erratic mouse issue fixed
- Model aliasing fixed
- Is able to find CelestiaResources folder on localized systems

(The CelestiaResources data files are not included with this build)

Any G5 owners are encouraged to test this build out to see if there is a speed difference from the
older 1.4.0pre6 build (press the backtick ` key, below the ~, to show frames per second).

As previously noted in this forum thread, remaining issues (with workarounds) are:
- Screenshot appearing on planets on Rage 128 (workaround: switch monitor to Thousands of Colors)
- Hard crash on Radeon 9200-based iBook, eMac, Mac Mini (workaround: turn off vertex shaders as
explained in the Readme)

[Cham]

Dirkpitt,

thanks a lot for this version. The two major problems I experienced in the previous version appears to be solved now, on my system. But there's a new annoying bug ! The browser has a drawing problem. See there :



If I try resizing the window, it may get worst. The content may completely disappear !

This is occuring on a G4 dual 867 MHz with 1.5 GB ram, OS 10.3.3, ATI Radeon 9800 pro (128 MB vram).

[Cham]

For the moment, I'm unable to reproduce the drawing problem in the browser.

There's a thing which annoys me in the browser, however : when I click in some item, the columns are moving in a laggy, or jerky way. They appears to move to the left, then to the right, and again to the left, very fast. This is confusing. It's not exactly like in any other OS X window with columns. I think there is some polishing to do here.

Also, I'm not sure to understand the browser : it appears to be lacking a LOT of exoplanets systems. In my exosys.ssc file, I have about one hundred of systems. The browser is showing only half of that (approximately).

[hank]

Cham,

The browser should be considered experimental at this time. It obviously needs improvements. Currently only the first fifty objects in each category are included, which is why you're not seeing all your exoplanetary systems. The column repositioning was done in an attempt to avoid wasting space with useless empty columns. It was not so distracting under 10.2 because the intermediate states were not visible. If this can't be corrected, it's easy to remove.

- Hank

[Cham]

Thanks for the explanations, Hank.

I really appreciate your efforts, Hank and Dirkpitt, to make this free software available to all. Celestia is a so nice product, we can feel it's close to perfection. I want to help to improve it. So here's my comments on the latest build (these aren't complains, but just some observations)

1- the "okay" button at the starting sequence is useless. IMHO, it should be removed.

2- The quiting dialog box, when you quit Celestia, is annoying. I prefer the way it was before. Celestia isn't a game, so I prefer that it quit without asking, like most applications. But this is really a minor complain.

3- Yes, the browser needs more work. It's movements are very annoying and confusing. It should be like the standard OS X way (maybe like the open/save dialog box). Of course, it should be able to show all solar systems, spacecrafts, etc.

4- The nebulae and the galaxies needs a separate option to show them separately. This may be more of a concern to Chris instead.

Well, that's all for the moment. All other comments I may have are about new features.

[dirkpitt]

1- the "okay" button at the starting sequence is useless. IMHO, it should be removed.

2- The quiting dialog box, when you quit Celestia, is annoying. I prefer the way it was before. Celestia isn't a game, so I prefer that it quit without asking, like most applications. But this is really a minor complain.

3- Yes, the browser needs more work. It's movements are very annoying and confusing. It should be like the standard OS X way (maybe like the open/save dialog box). Of course, it should be able to show all solar systems, spacecrafts, etc.

It's good to get feedback on builds. Q1: I think Hank had already answered this (OK button probably
should be a "Stop" or "Cancel" instead, but it might not be so easy to implement).

Q2: I thought that a Quit confirmation might not be so useful either, but there were times using the old
1.3.2 version when I was meaning to bookmark a great view but I quit Celestia and lost it.
Maybe if the history got saved to the disk automatically (like Safari) the quit confirmation might not be
necessary. Just an idea though at this point.

Q3: Hank has already answered this too. The browser is a temporary solution that will be snazzed up.

[Cham]

Hank and Dirkpitt,

there's another small but irritating bug in this version, that was also in 1.3.1 and 1.3.2, but NOT in the version Dirkpitt send be privately three months ago. Here's a description of that weird bug :

The bug is only appearing at startup, after about 10 seconds, at the startup box level. Celestia then unexpectedly quit and I get that message from the system asking me if I want to send a report to Apple. It's occuring only one time in about 10 or 15 starts. But systematically, when I launch Celestia for, say, the 10th time, it then crash. If I restart Celestia after that, it's loading correctly without crashing. It appears to be a reproductible bug, but I need to quit and restart Celestia about a dozen of times before reproducing the bug.

Like I said, I was experiencing this problem with the 1.3.1 and 1.3.2 versions, but Celestia NEVER crashed with the Dirkpitt version made three months ago. So there's definitely something that is different in his version which makes Celestia more stable. Can you investigate on this ?

This was occuring on a dual G4 with 1.5 GB ram, OS X 1.3.x (any version), and with BOTH my previous NVidia GeForce 4 MX and my recent ATI 9800 pro video cards, so I don't think it's related to the video card.

[hank]

Cham,

What messages do you get in '~/Library/Logs/CrashReporter/Celestia.crash.log' when this startup crash occurs?

- Hank

[Cham]

I get a long text file, with all my Celestia crashes since my first version. For the last entry about Celestia 1.4, I get something which start like this :

Command: Celestia
Path: /Applications/Astronomie/C?©lestia/Celestia 1.4.app/Contents/MacOS/Celestia
Version: Celestia version 1.4 (6-20050201)
PID: 10716
Thread: 2

Exception: EXC_BAD_ACCESS (0x0001)
Codes: KERN_INVALID_ADDRESS (0x0001) at 0x15fecb68


Is this enough, or do you need all the text which is following this ?