Satellites stay brightly lit when passing through shadows...

barbarossa2 · Post #1by **barbarossa2** » 11.02.2006, 06:34

I don't really understand why small moons and asteroids (and man made objects) would stay brightly lit and be visible from thousands of miles away when passing through the shadow of a planet. This happens all of the time I think. Am I crazy? Or does anyone else see this as a problem?

Seems the rendering is correct on rings. When they pass through the shadows of planets, they become pitch black (COOL!!)

But not so with the satellites.

:roll:

Graphics Card: Nvidia GForce 6600 GT w 128 MB Ram
Celestia Version: 1.4.0
OS: Windows XP

brokfn · Post #2by **brokfn** » 11.02.2006, 11:08

Yes!
At last, someone who has the same problem as me!

On my machine, it happens all the time (with Hubble for example), except when I turn the render path to basic or multitexture.

Bye,
Brokfn

Post #3by **selden** » 11.02.2006, 12:31

This is an intentional limitation in Celestia. Mesh objects can neither cast shadows nor have shadows cast upon them. The previous code that tried to do so had serious bugs. This may be fixed in a future version of Celestia.

brokfn · Post #4by **brokfn** » 11.02.2006, 14:17

Hi,

Could you please explain why this limitation doesn't appear with Cassini ?

Bye,
Brokfn

Post #5by **selden** » 11.02.2006, 15:28

brokfn,

When do you see Cassini shadowed (eclipsed)?

I see Cassini illuminated when it should be in a planet's shadow: it's not eclipsed. For example, when I set Celestia to the date 18 Aug 1999 03:43 UTC, which is during Cassini's Earth Gravity Assist while Cassini is on the night side of the Earth and in Earth's shadow. The sunward side of Cassini is bright when it should be dark.

The surfaces of a Mesh are shaded depending on the angles of their surface normals with respect to the direction toward a light source. As a result, the side toward the sun is bright, while the side away from the sun is dark. That's not the same as casting or being in a shadow.

brokfn · Post #6by **brokfn** » 11.02.2006, 16:47

Selden,

I'm afraid we're not talking about the same problem.
Here is how Hubble is rendered by my machine with Open GL vertex program / NVIDIA combiners :

[img=http://img206.imageshack.us/img206/9543/hubblefall5wg.jpg]

And now, here is how it looks with basic render path :

url=http://imageshack.us][img=http://img99.imageshack.us/img99/7388/hubblemat1zd.jpg][/url]

(I hope I succeeded in uploading the images - this is my first time)

You notice that in the first image, Hubble is brightly illuminated, so it's hard to see anything on the texture.

Bye,
Brokfn

brokfn · Post #7by **brokfn** » 11.02.2006, 17:00

Aother try :oops:

and

Brokfn

Post #8by **selden** » 11.02.2006, 18:22

brokfn,

The first "overexposed" image is the result of a bug. Somewhere. If you can provide a Cel::/URL for that viewpoint, then maybe it can be tracked down. People have been reporting this problem for a long time, but apparently Chris hasn't been able to locate the cause. It doesn't seem to happen on my current system.

Ctrl-Insert will generate the URL.
Ctrl-V can be used to paste it into your posting.

Also, please specify the type of graphics card is in the system and what version of ForceWare is installed. I have a vague recollection that MX cards tend to do this, but my memory could be wrong.

Vincent · Post #9by **Vincent** » 11.02.2006, 18:28

Selden,

I think Brokfn is talking about this bug : http://celestiaproject.net/forum/viewtopic.php?t=7888&start=58

It has already been pointed out many times... But unfortunately, we don't know whether it will be fixed in a next release or, as Maxim says, we just have to buy a new graphics card... :wink:

Post #10by **selden** » 11.02.2006, 18:40

If Chris can't reproduce it, he can't fix it.

barbarossa2 seems to be reporting this probem for his 6600GT. That's the same model of graphics card that I have, and I don't seem to have the problem.

System:
1 GB, 3.4GHz P4-550, WinXP Pro SP2
128MB GF 6600GT, ForceWare v77.72
Celesita v1.4.1

brokfn · Post #11by **brokfn** » 11.02.2006, 18:51

Apparently it's the same bug.
Here is the Cel URL :
cel://Follow/Sol:Earth:Hubble/2006-02-1 ... 17335&lm=0
and here is some data about my graphic card :
Renderer: GeForce4 420 Go 32M/AGP/SSE2/3DNOW!
Version: 1.5.3
BIOS video : 4.17.20.49.25.
Unfortunately I won't be able to change it (it's a notebook), and as the drivers are procided by HP, it will not be possible to update them as well.
Snif. :cry:

Bye,
Brokfn

Post #12by **selden** » 11.02.2006, 19:33

Unfortunately, that Cel:// URL is more than 9,000km away from Hubble in the default Celestia v1.4.1 configuration on my system.

Did you modify Hubble's orbital parameters?

Can you generate a Cel:// URL with just a default Celestia installation?

p.s.
It would still help if you could provide the version number of the drivers that are installed on your system, both as shown by the Display Properties popup menu and by Celestia's Help menu.

brokfn · Post #13by **brokfn** » 12.02.2006, 10:05

Sorry, it was Terrier's model.
Anyway the problem is just the same with the default model.
Here is the url :
cel://Follow/Sol:Earth:Hubble/2006-02-1 ... 17335&lm=0

and more information about my graphic card:
Vendor: NVIDIA Corporation
Renderer: GeForce4 420 Go 32M/AGP/SSE2/3DNOW!
Version: 1.5.3
Max simultaneous textures: 2
Max texture size: 2048
Point size range: 1.000000 - 63.375000

driver 7.2.3.0 (provided by HP)

Thanks for your help.

Bye,
Brokfn

Vincent · Post #14by **Vincent** » 12.02.2006, 10:45

On many desktop systems, this bug appeared with the latest versions of NVidia drivers. So even if your laptop's graphics drivers are not up-to-date, I'm not sure updating them will solve the problem.

Maybe Chris can help here since the problem seems to be related to NVidia cards rather than to Celestia.

Post #15by **selden** » 12.02.2006, 14:21

Of course, Hubble looks fine to me with that URL.

Have you tried adjusting the AmbientLightLevel?
A value of 100 makes things very bright, and the menu doesn't always fix it.

When it's gotten botched on my system I've had to use a script to adjust it.

AmbientOff.cel

Code: Select all

{ set { name "AmbientLightLevel" value 0 }}

brokfn · Post #16by **brokfn** » 12.02.2006, 14:49

Have you tried adjusting the AmbientLightLevel?
A value of 100 makes things very bright, and the menu doesn't always fix it.

I tried: no effect.

Bye,
Brokfn

Adirondack · Post #17by **Adirondack** » 17.02.2006, 12:09

brokfn,

I don't know if it helps in your case, but does it help when you switch off the clouds rendering?

Unfortunately I'm not at home right now, so I can't reproduce this bug and try to check if 'clouds off' will take any effect.

Adirondack

Vincent · Post #18by **Vincent** » 17.02.2006, 12:34

Adirondack,

Switching off the clouds rendering indeed solved the problem on my system. That means that this bug can possibly be solved from Celestia's code. I thought it was definitively unsolvable. Thanks a lot !

Adirondack · Post #19by **Adirondack** » 18.02.2006, 01:24

Vincent,

this is an old (chipset-)bug I know since 1.3.1.
It occurs with the nVidia 4 (MX) chipset.
I can't confirm the bug with my GF 6600.

Adirondack

brokfn · Post #20by **brokfn** » 18.02.2006, 09:22

Hi,

Thank you Adironback.
Switching off the clouds makes the satellites appear normal on my system too.
I hope it'll help to fix the problem.

Bye,
Brokfn