New Improved Rings for Saturn ... anyone?

[Chuft-Captain]

Is anyone interested in improved rings for Saturn?

Here's what the standard ones look like for me... (pretty bad in my opinion):

Saturn Rings STD (lores).jpg

And here's a couple of options for new improved rings which I'm working on:

Saturn Rings Darkside (medres).jpg

Saturn Rings Sunside (hires).jpg

Based on Cassini Imagery Courtesy NASA/JPL-Caltech: http://saturn.jpl.nasa.gov/photos/imagedetails/?imageid=3381

Comments?
Preferences?

CC

[chris]

Something seems to be wrong with the opacity in your alternate ring textures. The ring texture in the standard Celestia distribution uses transparency information derived from measuring stellar occultations. So, you should not try and improvise an alpha channel. I'm less certain about how the color information was obtained--Grant or Fridger may know more. Nevertheless, the color of the rings in your test images seems much too saturated. I know that it's based on supposed natural color images from Cassini, yet I'm suspicious... The rings exhibit much subtler colors in other natural color images of Saturn such as this one:

http://photojournal.jpl.nasa.gov/catalog/PIA05389

One thing that really does need improvement is the way that Celestia renders the unilluminated side of the rings. In Celestia, Saturn's rings look roughly the same whether the viewer is on the lit or unlit side. In reality, the rings should look dramatically different. This is not something that can be fixed by changing the ring texture--the shaders need to be modified to use a more sophisticated illumination model.

--Chris

[Chuft-Captain]

Something seems to be wrong with the opacity in your alternate ring textures. The ring texture in the standard Celestia distribution uses transparency information derived from measuring stellar occultations. So, you should not try and improvise an alpha channel. I'm less certain about how the color information was obtained--Grant or Fridger may know more. Nevertheless, the color of the rings in your test images seems much too saturated. I know that it's based on supposed natural color images from Cassini, yet I'm suspicious... The rings exhibit much subtler colors in other natural color images of Saturn such as this one:

http://photojournal.jpl.nasa.gov/catalog/PIA05389

Bear in mind that it's a WIP (just started last night), and yes I have improvised the alpha channels for now because I don't know how it was set in the original. (I just wanted to get something up to get some initial feedback)

If anyone can give advice on how to get the alpha channel right, I'd appreciate it. I'm using The GIMP, but I'm a bit of a "bunny" in GIMP.

The second image is based on the un-illuminated image here: http://photojournal.jpl.nasa.gov/catalog/PIA08389

and the third is derived from the sunlit image here: http://photojournal.jpl.nasa.gov/catalog/PIA11142

I've not altered the colours other than to adjust the opacity down to a value in the range 50 - 70%. (just to approximate the transparency of the original)

One thing that really does need improvement is the way that Celestia renders the unilluminated side of the rings. In Celestia, Saturn's rings look roughly the same whether the viewer is on the lit or unlit side. In reality, the rings should look dramatically different. This is not something that can be fixed by changing the ring texture--the shaders need to be modified to use a more sophisticated illumination model.

I agree.

[ElChristou]

As Chris pointed, you should use the alpha channel of the default texture.
Here some basic steps under PS (you should be able to do the same with Gimp):

- Open default ring png, you'll find a semi transparent layer, let's say it's "layerA"
- Create "layerB"
- Do a selection of what is inside layerA
- Keep the selection active, switch to layerB, fill the selection with pure black
- Create layerC, fill it with pure white and stack layerB on top of layerC
- Delete layerA
- Flatten image
- Do a negative, and tada! here is the Alpha you need.

Open a new doc (same dimension - 1024x2), tweak the nasa pict to align perfectly the features with default textures, use the previously made alpha to get a selection, copy/paste in a new doc, here you are. Eventually at this stage you can desaturate the result as you like.

You should be able to do that in less than ~15 minutes...

[t00fri]

Is anyone interested in improved rings for Saturn?

And here's a couple of options for new improved rings which I'm working on:

Saturn Rings Darkside (medres).jpg

Saturn Rings Sunside (hires).jpg

Based on Cassini Imagery Courtesy NASA/JPL-Caltech: http://saturn.jpl.nasa.gov/photos/imagedetails/?imageid=3381

Comments?
Preferences?

CC

CC,

I have done precisely this exercise about 1.5 years ago and displayed the results in this forum. Here is the reference:

Saturn ring thread:
http://celestiaproject.net/forum/viewtopic.php ... gs&start=0

My improved rings (Ciclops ring data):
-------------------------------------------------
http://celestiaproject.net/forum/viewtopic.php ... s&start=19

Did you also cross-check the scaling of your rings by making sure that the tiny moon Pan always moves correctly within the empty Encke division!?

See also this!
http://celestiaproject.net/forum/viewtopic.php ... gs&start=2

It is crucial to get the ring transparency right. This is done by entering into the alpha channel the earthbound hires measurements of the ring /transparency/ as a grayscale image. The transparency was obtained by measuring the amount of transmitted light during the passage of the rings in front of a star (of known brightness)!
It is a published scientific paper with tabulated data. Easy to find...
Since we used it for the default texture, you should have just looked up the README in the Celestia distribution: http://pds-rings.arc.nasa.gov/ringocc/ringocc.html

++++++++++++++++++++++++++++++++++
If you don't include the transparency/opacity data you will get too transparent rings as it seems to be the case on your above image...
++++++++++++++++++++++++++++++++++

A lot more work than one night went into the default rings actually

Fridger

[ElChristou]

BTW, where can I find the original 4k texture?( I want to give a try but it would be cool to use the 4k alpha...)

[t00fri]

BTW, where can I find the original 4k texture?( I want to give a try but it would be cool to use the 4k alpha...)

Christophe,

see my post above yours. I have done it all already 1.5. years ago...

Fridger

[t00fri]

As Chris pointed, you should use the alpha channel of the default texture.
Here some basic steps under PS (you should be able to do the same with Gimp):

- Open default ring png, you'll find a semi transparent layer, let's say it's "layerA"
- Create "layerB"
- Do a selection of what is inside layerA
- Keep the selection active, switch to layerB, fill the selection with pure black
- Create layerC, fill it with pure white and stack layerB on top of layerC
- Delete layerA
- Flatten image
- Do a negative, and tada! here is the Alpha you need.

Open a new doc (same dimension - 1024x2), tweak the nasa pict to align perfectly the features with default textures, use the previously made alpha to get a selection, copy/paste in a new doc, here you are. Eventually at this stage you can desaturate the result as you like.

You should be able to do that in less than ~15 minutes...

Christophe,

I don't understand why you do this complicated procedure. Both in PS and GIMP you just do a RGBA decomposition with ONE click. That's all. In GIMP there is a menu entry for this. Then you have the R,G,B and the alpha channel individually at your disposal.


Fridger

[ElChristou]

I don't understand why you do this complicated procedure. Both in PS and GIMP you just do a RGBA decomposition with ONE click. That's all. In GIMP there is a menu entry for this. Then you have the R,G,B and the alpha channel individually at your disposal.

Because in one click you don't understand what's going on! Here at least Chuft as learn something... (I think... )

[ElChristou]

BTW, where can I find the original 4k texture?( I want to give a try but it would be cool to use the 4k alpha...)

Christophe,

see my post above yours. I have done it all already 1.5. years ago...

... So I cannot give a try? even for the pleasure?
... too bad...


(Still if you or someone knows where to find the 4k, I take!)

[t00fri]

BTW, where can I find the original 4k texture?( I want to give a try but it would be cool to use the 4k alpha...)

Christophe,

see my post above yours. I have done it all already 1.5. years ago...

... So I cannot give a try? even for the pleasure?
... too bad...


(Still if you or someone knows where to find the 4k, I take!)

Of course you can try. That's all up to you. The textures are readily available.

Here is the rings from Ciclops (11k)
http://ciclops.org/view/3858/Expanse_of_Ice

I have quoted the transparency paper above. The difficulty is NOT in assembling these data into the alpha channel of the Ciclops RGB data. That's really trivial. It's the correct relative normalization of transparency versus albedo! So it's easy to make a nice-looking image, but it's hard work do do a CORRECT image Also the correct rescalings are not quite so easy...

++++++++++++++++++++++++++++++
Then there is the shader deficiency Chris mentioned. That's why I have not yet published my rings up to now.
++++++++++++++++++++++++++++++

Note that the original 4k transparency will not do, since there is now a new faint outer ring
Here is my personal alpha texture only for you (4648x500, unpublished):

http://www.celestiaproject.net/~t00fri/images/ ... -alpha.png


Fridger

[t00fri]

I don't understand why you do this complicated procedure. Both in PS and GIMP you just do a RGBA decomposition with ONE click. That's all. In GIMP there is a menu entry for this. Then you have the R,G,B and the alpha channel individually at your disposal.

Because in one click you don't understand what's going on! Here at least Chuft as learn something... (I think... )

Sorry, but the menu command does something VERY simple! I doubt it becomes more intuitive by exercising half a page of operations...

Since the colors (including alpha) are arranged in interlaced mode in the RGBA texture, i.e.

R G B A R G B A R G B A ...

the menu command (click ) just pics up every 4th entry and stores it sequentially in a separate file. This makes up for the red, the green the blue and the alpha texture, respectively. What on Earth is complicated or more instructive beyond this simple fact?

Fridger

[ElChristou]

...What on Earth is complicated or more instructive beyond this simple fact?

The problem is that you take a very specific feature from a specific soft and generalize it. For example I have started using PS many, many years ago and at the time they were no fancy one click options. You had to do yourself drop shadows, bezels etc... Today indeed many tools propose fast solution but what if someday you don't have the precious one button tool at the glance?
Today I don't work anymore in design so I'm not up to date with the last tools; I guess CS4 have such one click extraction feature, but if I remember well CS2 no.
So I think that these few steps can be kind of instructive for those who won't have the last fancy tool or won't be able to write down a script to do the job... Now of course it's just my opinion...

(Tx for the 4k alpha!)

[t00fri]

...What on Earth is complicated or more instructive beyond this simple fact?

The problem is that you take a very specific feature from a specific soft and generalize it. For example I have started using PS many, many years ago and at the time they were no fancy one click options. You had to do yourself drop shadows, bezels etc... Today indeed many tools propose fast solution but what if someday you don't have the precious one button tool at the glance?
Today I don't work anymore in design so I'm not up to date with the last tools; I guess CS4 have such one click extraction feature, but if I remember well CS2 no.
So I think that these few steps can be kind of instructive for those who won't have the last fancy tool or won't be able to write down a script to do the job... Now of course it's just my opinion...

(Tx for the 4k alpha!)

It was always possible to do a 1-click RGBA decomposition as I described with the GIMP.
CC obviously uses the GIMP not PS. Only the position of the command in the pop-up menu has changed a bit recently in version 2.6.x.

Fridger

PS: NOTE the 4k-alpha is NOT 4k

[Chuft-Captain]

EDIT:

I'm less certain about how the color information was obtained--Grant or Fridger may know more. Nevertheless, the color of the rings in your test images seems much too saturated. I know that it's based on supposed natural color images from Cassini, yet I'm suspicious... The rings exhibit much subtler colors in other natural color images of Saturn such as this one:

http://photojournal.jpl.nasa.gov/catalog/PIA05389

Chris,
I wonder if the difference in saturation is a consequence of the different positions and phase angles under which the images were acquired.
For example, the image you refer to was taken when the spacecraft was "47.7 million kilometers (29.7 million miles) from the planet. The image scale is 286 kilometers (178 miles) per pixel.", whereas the images I used were taken "at a distance of approximately 1.1 million kilometers (700,000 miles) from Saturn and at a sun-Saturn-spacecraft, or phase, angle of 28 degrees. Image scale in the radial (horizontal) direction is about 7 kilometers (4.3 miles) per pixel."

(It's also possible that I may have introduced some colour artifacts if I didn't save in the correct way, however that wouldn't explain the pronounced difference in saturation)

I don't understand why you do this complicated procedure. Both in PS and GIMP you just do a RGBA decomposition with ONE click. That's all. In GIMP there is a menu entry for this. Then you have the R,G,B and the alpha channel individually at your disposal.

Because in one click you don't understand what's going on! Here at least Chuft as learn something... (I think... )

Sorry, but the menu command does something VERY simple! I doubt it becomes more intuitive by exercising half a page of operations...

Since the colors (including alpha) are arranged in interlaced mode in the RGBA texture, i.e.

R G B A R G B A R G B A ...

the menu command (click ) just pics up every 4th entry and stores it sequentially in a separate file. This makes up for the red, the green the blue and the alpha texture, respectively. What on Earth is complicated or more instructive beyond this simple fact?

Fridger

Guys,
I'm quite happy to know about both techniques.
Thanks for the advice, both of you.

EDIT:
FWIW, I can't find an RGBA decomposition option in the verion (2.2) of GIMP which I have, so it looks like I will need to follow ElChristou's method. (unless someone can direct me to the appropriate menu location.

Did you also cross-check the scaling of your rings by making sure that the tiny moon Pan always moves correctly within the empty Encke division!?

I adjusted scaling for the position of Pan and also the even tinier moon Daphnis...I'm not sure if Daphnis is perfectly positioned, but I think the Keeler Gap is the appropriate location. If you see any issues with this let me know. (Getting the Enke gap in the right place is fairly easy, but getting both moons right is more difficult!)

scaling.jpg

A final version would require this scaling work to be re-done anyway.

Then there is the shader deficiency Chris mentioned. That's why I have not yet published my rings up to now.

Fridger, I don't understand why you don't publish your 1.5 year old rings. Even with the shader deficiency, they are surely an improvement on the default.

It also seems to me that depending on what Chris has in mind, any real solution to the shader problem would probably change the ballgame wrt. transparency/reflectance of the rings thus making all current rings invalid/obsolete (including your new ones).

So it seems to me that the shader deficiency is not a reason to not publish.

CC

[t00fri]

CC,

while I don't understand why you use such an old version of GIMP (current = 2.6.6), the decompose feature is easy to find also in version 2.2:

Open your RGBA texture and then click on that image with the right mouse button. Follow the popup menu sequence

Filters -> Colors -> Decompose...

Click on Decompose..., a dialog opens and you activate RGBA (instead of RGB). Hit OK.

But don't you have the Help file installed?? The decompose feature is easy to find in the manual and then you don't need anyone to tell you once more...

Fridger

PS:

Fridger, I don't understand why you don't publish your 1.5 year old rings. Even with the shader deficiency, they are surely an improvement on the default.

I have my high standards for publishing. Preliminary download links only confuse people. Also, I am not involved with making add-ons and the ML is NOT for me. However, my forthcoming Celestia.Sci package will of course have improved Saturn rings and much more...