Diving into Titan's Atmosphere with Celestia

The place to discuss creating, porting and modifying Celestia's source code.
Avatar
Topic author
t00fri
Developer
Posts: 8772
Joined: 29.03.2002
Age: 22
With us: 22 years 5 months
Location: Hamburg, Germany

Diving into Titan's Atmosphere with Celestia

Post #1by t00fri » 28.03.2005, 17:06

Hi all,

I went through quite a bit of frustration today, when I tried to model the layers of Titan's atmosphere more realistically.
The present code for the numerical parameters of Titans atmosphere is really full of bugs. Atmospheres were mainly tested so far for cases that do not too much differ from Earth.

Here is a NASA diagram that compares Titans atmosphere layers with Earth's atmosphere.

Image

Strikingly, the height of Titans atmosphere is a factor of 10 compared to that of Earth! Notably, the yellow photochemical haze layer (that hides Titan's surface in visual light!) starts only at 200 km above ground and is ~50 km thick followed by another thin haze layer around 300 km altitude.

After studying Celestias respective rendering code, it becomes clear that all that stuff needs to be thoroughly rewritten in order to even achieve a semi-realistic approach. Chris announced this overdue task a long time ago, but I don't expect that we shall see anything very soon...

In Celestia, the cloud layer is at present (infinitesimally) thin, just a spherical surface that supports a cloud texture and associated transparency map. Hence, around the value of 'CloudHeight', a hole in the (untransparent) clouds opens and exposes the surface in perfect clarity. That's of course very far from what is needed to model Titan's atmosphere.


...and so on...I gave up and will now go have a tasty "Dim Sam" dinner instead ;-)

Bye Fridger

Avatar
Topic author
t00fri
Developer
Posts: 8772
Joined: 29.03.2002
Age: 22
With us: 22 years 5 months
Location: Hamburg, Germany

Post #2by t00fri » 29.03.2005, 12:26

Here is an illustration of a bug that is particularly unpleasant in case of Titan: While the 'cloud/haze surface' around Titan at an altitude of 150 km say, is barely transparent, it does not extend right down to the horizon, if the observer has landed on the surface! Right above the horizon one notices an almost transparent strip of black sky. At daylight times, that strip is colored in the 'Lower' atmosphere color. In any case it is always transparent...

Bye Fridger

Image

maxim
Posts: 1036
Joined: 13.11.2003
With us: 20 years 10 months
Location: N?rnberg, Germany

Post #3by maxim » 30.03.2005, 08:10

Huh. That looks like a severe bug.

maxim

Avatar
Topic author
t00fri
Developer
Posts: 8772
Joined: 29.03.2002
Age: 22
With us: 22 years 5 months
Location: Hamburg, Germany

Post #4by t00fri » 30.03.2005, 15:09

Hi all,

despite the above-mentioned problems & bugs, it is still fun to try and exploit the present possibilities as much as possible, concerning the approach through Titan's clouds towards the surface!

That's what I want to describe briefly next.

First of all, I adapted Titan's atmosphere layout to the actual data seen in the above NASA graph: I placed the yellow haze layer at 225 km above the surface with a total atmosphere height of 550km. Since at present, in Celestia cloud layers have vanishing thickness, I placed the haze layer at 225 km, about midway between the lower (200) and upper (250) boundaries.

Here is my modified titan.ssc file that you may use instead of my original one in 'extras/titan':
    Modify "Titan" "Sol/Saturn"
    {
    Texture "titan.png"
    BumpMap "titan-bump.png"
    BumpHeight 3.5
    SpecularColor [ 0.15 0.15 0.15 ]
    SpecularPower 25.0
    Atmosphere {
    Height 550
    Lower [ 0.675 0.503 0.3 ]
    Upper [ 1.0 0.745 0.444 ]
    CloudHeight 225
    CloudSpeed 100
    CloudMap "titan-clouds.png"
    }
    }


There is one great advantage of the great height (10x Earth!) of Titan's atmosphere in Celestia: there is much more room above ground to explore atmospheric views and still have a reasonably sharp/well defined surface view!

Moreover, I modified my 1k yellow haze layer (titan-clouds) of the Celestia 1.3.2 default distribution to include an alpha-layer, which was carefully tuned: on the one hand, from a distance, Titan's surface cannot been seen through. On the other hand from right under the haze (200 km altitude) both Saturn and the sun can /just/ be made out very dimly! This setup conforms to the latest findings from

http://saturn.jpl.nasa.gov/multimedia/images/image-details.cfm?imageID=1378

wherupon after some image processing of some visible-light images, slight surface details could be made out through the haze. It also conforms to my old personal estimates as to the visible light transmission...

This new 1k 'titan-clouds.png' may be downloaded from here:

http://www.shatters.net/~t00fri/images/titan-clouds.png

Just add it into the extras/titan/textures/medres directory.

Finally, I made a quite sophisticated new titan-bump.png file. In absense of any genuine altitude information, the challenge was to produce a bump-map file from the existing imaging information that was extremely smooth and reproduces the typical 3d impressions we got from photos...

Here is the URL for download:

http://www.shatters.net/~t00fri/images/titan-bump.png

OK and finally, here is a resulting view of the Huygens landing environment from 200 km altitude above Titans surface, i.e. from just under the lower yellow haze boundary. When the whole thing is animated, the slight texturing of the moving haze layer gives a nice "windy" impression with beautiful light effects from the slight inherent transparency of the haze layer...

Image

or nearby from another perspective and exposing a bit the 'bumpy' but /smooth/ terrain

Image

Enjoy,
Bye Fridger

Avatar
Topic author
t00fri
Developer
Posts: 8772
Joined: 29.03.2002
Age: 22
With us: 22 years 5 months
Location: Hamburg, Germany

Post #5by t00fri » 30.03.2005, 15:59

Here at last, a nice shot of the landing Huygens viewed from just under the yellow haze layer. You may also see the landing site pretty clearly...

Image

Bye Fridger

danielj
Posts: 1477
Joined: 15.08.2003
With us: 21 years 1 month

Post #6by danielj » 31.03.2005, 23:32

There are slight bugs in Celestia??s cloudmap near the surface in Celestia 1.4.0 pre 6.Look:
Look in http://br.f1.pg.briefcase.yahoo.com/djansenferreira and look for Titan??s bug.jpg,Titan??s bug 2.jpg and Titan??s bug 3.jpg.
Anyway,it was a tremend improvement towards the previous one.This first bug is only noticed very near the surface(on the surface,actually).The second is actually a swath of unclouded sky that appear and disappear as you move.I don??t know if this structure of atmosphere will make this thing totally normal

jromer
Posts: 33
Joined: 20.08.2004
With us: 20 years 1 month
Location: Columbus, Ohio

Post #7by jromer » 01.04.2005, 02:25

I really like how this looks. I also peered up through the clouds to glimps the dimmest image of Saturn. :)


Image
2.8G P-4 512k NVidia FX-5200
Celestia 1.4.1 running on WinXP (NVidia 61.77)
Celestia 1.4.1 running on Ubuntu6.06 (kind of...)

Avatar
Topic author
t00fri
Developer
Posts: 8772
Joined: 29.03.2002
Age: 22
With us: 22 years 5 months
Location: Hamburg, Germany

Post #8by t00fri » 02.04.2005, 09:29

danielj wrote:There are slight bugs in Celestia??s cloudmap near the surface in Celestia 1.4.0 pre 6.Look:
Look in http://br.f1.pg.briefcase.yahoo.com/djansenferreira and look for Titan??s bug.jpg,Titan??s bug 2.jpg and Titan??s bug 3.jpg.
Anyway,it was a tremend improvement towards the previous one.This first bug is only noticed very near the surface(on the surface,actually).The second is actually a swath of unclouded sky that appear and disappear as you move.I don??t know if this structure of atmosphere will make this thing totally normal


daniel,

the bugs you show are nothing but variations of the bugs I described further up in this thread. Chris will have to do some thorough improvements here. But presumably, it will be quite a while before we can hope for action...

If one views Titan at some higher elevation >few kilometers and below the haze layer (<200 km) things are quite OK, however.

Bye Fridger

rthorvald
Posts: 1223
Joined: 20.10.2003
With us: 20 years 11 months
Location: Norway

Post #9by rthorvald » 03.04.2005, 18:43

t00fri wrote:Hi all,

despite the above-mentioned problems & bugs, it is still fun to try and
exploit the present possibilities as much as possible, concerning the
approach through Titan's clouds towards the surface!


Very nice work, Fridger!
I have used your new SSC values, exept that i feel the extreme height of
the cloudmap makes Titan look a little artificial - it kind of gives me an
impression of it being enclosed by a dyson sphere... ( i don??t use your
cloudmap, so that might account for it...). However, i am
experimenting with multiple, stacked cloudlayers, at least for parts of the
atmosphere. That might possibly bring about something interesting...

Just to complement your great screenshots, here??s one of my own...
Getting all the way down to the surface, and looking at Saturn through
the clouds (barely visible at dawn and nightfall, in the daytime it??s almost
ompletely hidden). This scene is far from finished, but it gives an idea of
where i am going. Very un-scientific, though :-)

Image

-rthorvald

Avatar
Cham M
Posts: 4324
Joined: 14.01.2004
Age: 60
With us: 20 years 8 months
Location: Montreal

Post #10by Cham » 03.04.2005, 18:53

Runar,

this scene looks SUPERB ! :-)
"Well! I've often seen a cat without a grin", thought Alice; "but a grin without a cat! It's the most curious thing I ever saw in all my life!"

ElChristou
Developer
Posts: 3776
Joined: 04.02.2005
With us: 19 years 7 months

Post #11by ElChristou » 03.04.2005, 19:09

rthorvald wrote:...multiple, stacked cloudlayers, ...


Hello, rthorvald

Can you explain a bit more about your multiple cloudlayers? I think it's the first time I read something like this... and it sounds very interesting...

Tx.
Image

Avatar
Cham M
Posts: 4324
Joined: 14.01.2004
Age: 60
With us: 20 years 8 months
Location: Montreal

Post #12by Cham » 03.04.2005, 19:14

I'm pretty sure those clouds (the cloud layers) are parts of the mountains model. Transparent planes floating on top of the mountains, isn't ? Or maybe just moving planes around the planet, like the solar flare trick. I'm wondering on the depth sorting problem, however.
"Well! I've often seen a cat without a grin", thought Alice; "but a grin without a cat! It's the most curious thing I ever saw in all my life!"

Avatar
Topic author
t00fri
Developer
Posts: 8772
Joined: 29.03.2002
Age: 22
With us: 22 years 5 months
Location: Hamburg, Germany

Post #13by t00fri » 03.04.2005, 19:25

Excellent, Runar!

this thread looks like becoming pretty creative...

*************************
Perhaps this helps motivating Chris a bit to get the atmosphere
stuff right...at last (provided he still cares to read this :roll: )
*************************


Yes how about those extra low clouds?

Bye Fridger

Avatar
Topic author
t00fri
Developer
Posts: 8772
Joined: 29.03.2002
Age: 22
With us: 22 years 5 months
Location: Hamburg, Germany

Post #14by t00fri » 03.04.2005, 19:35

rthorvald wrote:...
Very nice work, Fridger!
I have used your new SSC values, exept that i feel the
extreme height of the cloudmap makes Titan look a little
artificial - it kind of gives me an impression of it being
enclosed by a dyson sphere... ( i don??t use your
cloudmap, so that might account for it...). However, i am
experimenting with multiple, stacked cloudlayers, at least for
parts of the atmosphere. That might possibly bring about
something interesting...

...


Runar,

indeed, I had to put the atmosphere height to 550 km in
order to have its tail washing out a bit the unnaturally sharp
boundary from the (infinitesimally thin) yellow haze layer.

All these "tricks" are workarounds, given the unsatisfactory
status of Titan's atmosphere rendering...

Bye Fridger

rthorvald
Posts: 1223
Joined: 20.10.2003
With us: 20 years 11 months
Location: Norway

Post #15by rthorvald » 03.04.2005, 20:28

Can you explain a bit more about your multiple cloudlayers? I think it's
the first time I read something like this[/quote]

Cham wrote:I'm pretty sure those clouds (the cloud layers) are parts
of the mountains model. Transparent planes floating on top of the
mountains, isn't ?

More or less... It is a 2k regular cloudmap for the model, and is only used
to hold the clouds in the foreground. It is just a test; later on i??ll separate
it completely from the model, to facilitate movement - which doesn??t
work very well when it is defined like this. Depht sorting isn??t a problem,
as it is far below Titan??regular cloudmap.

But, i am also working on larger maps at different heights, attached to
invisible objects that orbits inside Titan. That is more difficult, as it seems
the depht sorting gets messed up when the objects are closer in size -
and close to - each other... There are ways, though; the Sunflare
technique works - that is, using many small ones instead of one big one.
But this method will limit movement.

t00fri wrote:Perhaps this helps motivating Chris a bit to get the
atmosphere stuff right...at last (provided he still cares to read this


That would be nice...
What i hope to do is to simulate three-dimensional clouds, at least in a
few spots. I even tried building real 3D models, but they do not look as
good, and adds too many megabytes to the soup to be practical...

-rthorvald

Seb
Posts: 44
Joined: 12.09.2004
With us: 20 years
Location: Wiltshire, UK

Post #16by Seb » 14.04.2005, 12:59

I'm very impressed. Could we make sure your work is in the next version of Celestia?

Avatar
Topic author
t00fri
Developer
Posts: 8772
Joined: 29.03.2002
Age: 22
With us: 22 years 5 months
Location: Hamburg, Germany

Post #17by t00fri » 14.04.2005, 14:47

Seb wrote:I'm very impressed. Could we make sure your work is in the next version of Celestia?


As a developer and author of Celestia, I usually include my contributions in the official version. However, there, we do have serious constraints as to the size of the archive. Hence I shall presumably rescale my default Titan texture to 1k.

Bye Fridger

Rassilon
Posts: 1887
Joined: 29.01.2002
With us: 22 years 7 months
Location: Altair

Post #18by Rassilon » 14.04.2005, 23:26

t00fri wrote:Here is an illustration of a bug that is particularly unpleasant in case of Titan: While the 'cloud/haze surface' around Titan at an altitude of 150 km say, is barely transparent, it does not extend right down to the horizon, if the observer has landed on the surface! Right above the horizon one notices an almost transparent strip of black sky. At daylight times, that strip is colored in the 'Lower' atmosphere color. In any case it is always transparent...

Bye Fridger

Image


A quite bit late in response but I think I might be able to shed some light on this....

If I am referring to the same issue here...I believe Celestia has always had an issue with cloud layers at very close distances to the surface....I do not think its a bug but an artifact of precision....or possibly the near clipping plane....When you get too close to an object in opengl there is a tendency for parts of the object to dissapear...
I'm trying to teach the cavemen how to play scrabble, its uphill work. The only word they know is Uhh and they dont know how to spell it!

Avatar
Topic author
t00fri
Developer
Posts: 8772
Joined: 29.03.2002
Age: 22
With us: 22 years 5 months
Location: Hamburg, Germany

Post #19by t00fri » 15.04.2005, 18:07

Rassilon wrote:
...
A quite bit late in response but I think I might be able to shed some light on this....

If I am referring to the same issue here...I believe Celestia has always had an issue with cloud layers at very close distances to the surface....I do not think its a bug but an artifact of precision....or possibly the near clipping plane....When you get too close to an object in opengl there is a tendency for parts of the object to dissapear...


Ras'

Yes you are right. I had also looked into the code. But for me this is nevertheless a BUG. I have often been fighting in my scientific programming 'life' against such 'BUGS' that often are worse to cure than simple code bugs ...

Bye Fridger

Rassilon
Posts: 1887
Joined: 29.01.2002
With us: 22 years 7 months
Location: Altair

Post #20by Rassilon » 15.04.2005, 22:06

I suppose chris could fix this 'bug' by recoding the cloud rendering...Im not sure how it is done in Celestia but I think what has been done is hidden surface removal instead of drawing a complete sphere for the cloud layer. This becomes evident when close to the object. Or if the problem lies in the clipping plane then all need be done is assign a seperate clipping plane for the cloud sphere and the main sphere that is the surface....

Really is a guess since I have no clue how chris coded this routine...
I'm trying to teach the cavemen how to play scrabble, its uphill work. The only word they know is Uhh and they dont know how to spell it!


Return to “Development”