CVS bugs (stars, atmosphere,...)

[t00fri]

Now that we are heading towards a pre 1.5.0 release, here are some bugs that are still in todays CVS.

a) - c) below refer to Chris new 'star code' activated with @. I use scaled disk stars throughout.

a) Relative brightness grossly incorrect:

Compare size and brightness of Venus vs the setting sun...



b) Atmosphere bug:

Still the black mismatch regime


c) Black dot bug:

Here is alpha centauri as an example



Next, upon startup using the old stars, the sun still shows the bright "inner sun"



Finally, the depth sorting has improved a number of issues related to visible stars around extended stardiscs
but it's still far from looking realistic.

A few days ago also our "infamous" "hole in the sky" was still around. I havn't checked it again, after the latest fixes.

Bye Fridger

[Fightspit]

There is a problem with our Sun today and it is probably the end of the sun



I am not sure if the sun can be as big as in the scrennshots in 3 or 4 billion year

edit: here the url:
cel://Follow/Earth/2006-09-09T14:51:56. ... 8787&lm=32

[Malenfant]

I've got a version of the executable that Chris kindly made for me (presumably from the CVS code) and noticed a few things too:

- star flares seem to disappear if you get close to the star

- the scaled discs look bloody enormous when you're far away. I've got a trinary M V system I've made and the scaled disc look huge at 1000 AU. When I close to about 1.8 AU they look even bigger, and remain large until you get close enough that all you're seeing is the actual disc of the star. Definitely something wrong there.

- the bright part in the middle of the scaled discs seems to be the actual edge of the star, which means that the size of the star has increased by about 100% just because of the 'scaled disc'

[selden]

Fridger,

This is strange.

I'm seeing neither the "white spot in the middle" nor the "black spots at the edge" using a Windows executable rebuilt just now by invoking makerelease.bat

Can you provide a URL for the black spots?

I deleted my local /src/ directory tree in order to force a complete code update. The Sun looked ok with that version of Celestia.

Then I renamed the celestia subdirectory in order to force a complete code and data update. That looks ok too.

Here's what I see when looking at the Sun:


using this url:
sun

I stepped through all the render paths, with and without the new brightness algorithm. The URL turns on all rendering options just in case it's an interaction problem.

I do agree that the albedo brightness scaling needs to be adjusted.

[edit]Here's my system configuration, in case it makes a difference.
1GB 3.4GHz P4-550, WinXP Pro SP2
128MB GF6600GT, ForceWare v84.21
1600x1200, 60Hz
Celestia from cvs
[/edit]

[Cham]

I think the black dots on the star are related to the FOV. I already have them from time to time on the 1.4.1 version, but they are tiny and always on the star's border, between the surface and the atmosphere.

On version 1.4.2 made by Dirkpitt (Celestia 1.4.2 (20060517)), the glare texture is also abruptly disappearing when I get closer to the star. It's really annoying.

And there's a strange white "dot" on the middle of the star, when I get pretty close. It's always centered on the star, and looks like a small disk.

[Malenfant]

And there's a strange white "dot" on the middle of the star, when I get pretty close. It's always centered on the star, and looks like a small disk.

I think you'll find that's the actual star itself. On that "huge sun" picture (presumably taken with Scaled Discs turned on?), there's a 'white dot' in the middle that is the actual size of the solar disk. Though when I go to that cel link with Scaled disks turned on, I only get a fuzzy disk that's about twice the size of the sun, not one that takes up most of the screen.

Then again, I'm using the normal 1.41 distribution with a recompiled executable and iconv.dll and intl.dll, provided by Chris.

[symaski62]

cel://SyncOrbit/Earth/2006-09-10T05:34: ... 3&lm=-2046

:p sun & venus

[t00fri]

symaski62,

your cel://url does NOT correspond to what I was discussing.

@ and /scaled disk/ stars!

Bye Fridger

[cartrite]

Fridger wrote: [quote="t00fri"]


b) Atmosphere bug:

Still the black mismatch regime


I think this may have more to do with the clipping plane then the atmosphere. It doesn't affect cmod meshes as these images suggest. If you notice that they were taken at about 11 meters from the surface and no clipping plane problems. Mabey a clue to tracking down this bug? Cmod meshes and spheres that Celestia uses for planets are rendered by different sections of the code.

http://www.celestiaproject.net/forum/viewtopic ... 6145#76145

I think that the biggest problem with the atmopheres is the dark overhead sky when veiwed from underneath the atmosphere. I think Chris is well aware of this and was working on some fix?

cartrite

[t00fri]

cartrite,

the left hand image was taken at 500m altitude, where everything should certainly be stable and well-behaved.

Bye Fridger

[t00fri]

Fridger,

This is strange.

I'm seeing neither the "white spot in the middle" nor the "black spots at the edge" using a Windows executable rebuilt just now by invoking makerelease.bat

Can you provide a URL for the black spots?

Excellent, Selden.

couldn't do a completely fresh install yet, since Christophe's po file installation gave an error during make install (-> developer list). Yet after going back to the VERY beginning (make distclean, make -f makefile.cvs, configure...make), I also don't get the black dots anymore and also the "bright inner sun" has vanished!

OLD Sun
======


[edited]
NEW Sun
======


Bye Fridger

[Cham]

Fridger, apparently, there is still a "bright inner sun" in the picture above, but larger and fuzzy. Can you confirm ?

[Malenfant]

Yes, I can see a bright "inner sun" in Fridger's last image too.

This is what I see when I go to Symaski's cel URL - on the left is the view with Scaled Discs on, on the right is the exact same view with the stars as Points (it looks similar as Fuzzy Points too).



It should be obvious that the 'white spot' in the middle of each Sun is the disk of the star itself. When "Scaled Discs" is on, there's an extra fuzzy layer added to the outside of the star.

When I zoom into Sol so it's as big as in Fridger's picture, I don't see a white "inner sun" at all, with or without the Scaled Discs turn on.

[ajtribick]

I'm using an LCD screen, and when I look at an angle to the screen at Fridger's image I see a white disc at the centre.

[t00fri]

Fridger, apparently, there is still a "bright inner sun" in the picture above, but larger and fuzzy. Can you confirm ?

++++++++++++++++++++++
Sorry, that's the OLD image, not the new one. Just have no time to make another one without the white spot!
+++++++++++++++++++++++

I thought if I say it in words, it's enough

Bye Fridger

[Cham]

Fridger,

what about the star's glare disappearing when approaching the star ? On 1.4.1, my glare effect abruptly disappears when I go to any star.

[phoenix]

as mentioned before thats what magnetic-earth looks like with current cvs to me:

http://www.purebytes.de/stuff/celestia/cvs_bug1.png

[Cham]

as mentioned before thats what magnetic-earth looks like with current cvs to me:

http://www.purebytes.de/stuff/celestia/cvs_bug1.png

Some of the magnetic field lines should appears in front of the Earth. You should be able to see some lines entering the Earth on its poles.

[kikinho]

The new atmosphere is interesting, take a look at this: http://flavorphoto.com/blog/wp-content/ ... graphy.jpg and this http://home.freeuk.com/m.gavin/dig_stmary.jpg
These are two of many images taken in infrared. Note that Rayleigh scattering is very weak, even when it's day. Infrared images are something different from normal images, and that's why I like IR photos.

Now compare the infrared image with this: http://img147.imageshack.us/img147/1849/earth5nc0.jpg
This is a bug with the new atmosphere code. The sky overhead is nearly black. And if you compare with the other image, you'll see that bug, despite being horrible, is very interesting.

So this means that this bug is the same as seeing the atmosphere in INFRARED WITH COLORS! COOL! HEHE!

That do you think about this?

[t00fri]

Here is how the Sun-venus view from the above cel://url
looks like with the latest CVs compile: