Celestia Forums

Hi all,

I've noticed the discussion on developer mailing list about the "real colors" of saturn. We all know this beautiful Cassini pictures from Saturn which are labled with "slightly color enhanced". Now I ask myself what does it mean and searched for some R,G,B raw images to check this. I've found six raw images "W00000459 - W00000464" and build two RGB pictures.

Cassini optimize the dynamic range of it's 16 bit raw images to 8 bit for transmission to Earth. That means that each received raw image on earth which shall be combined to a RGB picture must be normalized. Now because of the lack of those correction data I used the space background as "black reverence" and the last outer part of the A-ring as "gray reverence". Therefore I can't say that my pictures are really accurate but they should be quite exact.


The bright white area on Saturn is caused by over exposure.



I was a bit shocked how colorless Saturn in reality is. Nasa should more use the expresion "strong color enhanced" for their released Saturn pictures. Further it's interesting how dark the Rings appears from this viewing angle especially compared with Hubble's Saturn pictures.

I'm not sure if it would be an good idea to "drimm" all textures in Celestia to natural (gray) colors. I think a lot of information will be lost. Further I'm absolute sure that a man will never be able to see all this direct with it's eyes. At least he will look through a more or less clear window but more likely he will look on a big screen.

Bye Jens

******

P.S. Since I know that my work is stupid, useless and not worth to discuss on this forum I'm a bit frustrated. At the moment I'm not even motivated to build and release further addons for Celestia.

jim wrote:
...
P.S. Since I know that my work is stupid, useless and not worth to discuss on this forum I'm a bit frustrated. At the moment I'm not even motivated to build and release further addons for Celestia.

Jens,

this certainly does not include me! I have always valued your image work and your original "Ansaetze"/ideas very much.

We just have to be very critical with our own enthusiasm;-) .

As to your new RGB Saturn composite, I think there must still be a normalization problem somewhere. As I suggested in the developer's list, we might determine the normalization from precisely reproducing the yellow equatorial band from the HST Heritage "truecolor" photo. Moreover the darkness of the band should match the photo as well. These two constraints should essentially fix the R,G,B weights. The main issue then is whether the southern regime is pidgeon blue (HST) or yellow-orange (Cassini)...or something else!

Bye Fridger

Fridger wrote:As to your new RGB Saturn composite, I think there must still be a normalization problem somewhere. As I suggested in the developer's list, we might determine the normalization from precisely reproducing the yellow equatorial band from the HST Heritage "truecolor" photo. Moreover the darkness of the band should match the photo as well. These two constraints should essentially fix the R,G,B weights. The main issue then is whether the southern regime is pidgeon blue (HST) or yellow-orange (Cassini)...or something else!

I don't know how exact your suggestion can be realized by useing "legal" manipulation methodes. Therefore here is the raw RGB composite of one of my pictures you can play whit it. "Raw" means that only the raw red, green and blue images were assembled to a RGB picture, the alignment was corrected, the picture croped and nothing else.



Jens

jim wrote:I don't know how exact your suggestion can be realized by useing "legal" manipulation methodes. Therefore here is the raw RGB composite of one of my pictures you can play whit it. "Raw" means that only the raw red, green and blue images were assembled to a RGB picture, the alignment was corrected, the picture croped and nothing else.



Jens

Looks like the "black" background isn't very black there... which is a sign that the images aren't calibrated. So a start would be to find the DN of "space" in each image and subtract that value from each image here. That would get it a bit closer to true colour...

jim wrote:I don't know how exact your suggestion can be realized by useing "legal" manipulation methodes. Therefore here is the raw RGB composite of one of my pictures you can play whit it. "Raw" means that only the raw red, green and blue images were assembled to a RGB picture, the alignment was corrected, the picture croped and nothing else.

...
Jens

,
Jens,

are the above R,G,B grayscale textures really the most detailed/closest R,G,B Saturn views we got in the raw archive right now?

Bye Fridger

Jens mentions that the Cassini pictures have been converted from 16bits to 8bits for transmission. The algorithms they use have to be optimized for the (nonlinear) characteristics of the detector in order to be able to correctly recover the original values. My suspicion would be that there's no reason to expect that the normalization routines provided in standard image manipulation software packages would perform the correct operations to recover the original luminosities.

selden wrote:Jens mentions that the Cassini pictures have been converted from 16bits to 8bits for transmission. The algorithms they use have to be optimized for the (nonlinear) characteristics of the detector in order to be able to correctly recover the original values. My suspicion would be that there's no reason to expect that the normalization routines provided in standard image manipulation software packages would perform the correct operations to recover the original luminosities.

Quite. When preparing images in ISIS (which is what we usually use to process spacecraft images), we have separate radiometric and photometric calibration stages (see summary here). The radiometric calibration stage basically gets rid of camera-specific artifacts and responses, and the photometric calibration corrects for the photometric aspects of the surface you're looking at (so the brightness is all normalised). Without the files that ISIS would use (which AFAIK haven't been released yet and knowing the USGS won't be for quite a while), it's going to be very difficult to make calibrated Cassini images.

Hi all,

jim wrote:... the dynamic range of it's 16 bit raw images ...

Sorry, a little mistake. The raw data are "only" 12 bit images.

Evil Dr Ganymede wrote:Looks like the "black" background isn't very black there... which is a sign that the images aren't calibrated. So a start would be to find the DN of "space" in each image and subtract that value from each image here. That would get it a bit closer to true colour...

That's absolute correct!
Please read my post comlplete and you will unterstand.

Fridger wrote:Jens, are the above R,G,B grayscale textures really the most detailed/closest R,G,B Saturn views we got in the raw archive right now?

I can't find better textures at the moment. These images were taken with the wide angle camera but we all know that the narrow angel camera has taken much more "R,G,B" images. Now at the moment it seems that these raw images are not released.
Maybe these images helps:
red: http://saturn.jpl.nasa.gov/multimedia/i ... 000375.jpg
green: http://saturn.jpl.nasa.gov/multimedia/i ... 000374.jpg
blue: http://saturn.jpl.nasa.gov/multimedia/i ... 000376.jpg

red: http://saturn.jpl.nasa.gov/multimedia/i ... 000380.jpg
green: http://saturn.jpl.nasa.gov/multimedia/i ... 000379.jpg
blue: http://saturn.jpl.nasa.gov/multimedia/i ... 000381.jpg

Selden wrote:My suspicion would be that there's no reason to expect that the normalization routines provided in standard image manipulation software packages would perform the correct operations to recover the original luminosities.

I agree that it's really a problem not to know which correct operations leads to normalized images.
But now things has changed. It's the first time that I got the raw images for the official released "exquisite natural colour view of Saturn's rings".
http://ciclops.lpl.arizona.edu/view.php?id=227

I hope to find out more about the "secrets" of the normalization routines.

Jens

Hi all,

Good news, I could reproduce this "exquisite natural colour view of Saturn's rings" with very simple methodes. The folowing picture shows both the original NASA release below left and my rebuild with the raw images above right divided by the diagonal.



And for those who unterstand here is what I did in Photoshop:



Beside slight corrections with the gradations tool the color saturation must be inreased by 50%.
Enhanced color saturation - is it that what NASA unterstand with "exquisite natural colour view of Saturn's rings" ? I don't know.

The question is now: Do we get whitout the increased saturation a real natural color picture? Now that's what the next picture shows probably the first real natural color view from Cassini.



There are still a little doubt to me. Does NASA really use "calibration data" for ther current color pictures or do they only adjust these to get resonable results?

Jens

Speaking as someone who's prepared images for publication himself, they most likely do use real calibration data to get their results - at least initially. They may muck around with the histograms and contrast afterwards, but they usually say when they do that.