A Model of Saturn’s Rings

Tips for creating and manipulating planet textures for Celestia.
Avatar
Topic author
FarGetaNik M
Posts: 432
Joined: 05.06.2012
With us: 7 years 1 month
Location: Germany

A Model of Saturn’s Rings

Post #1by FarGetaNik » 24.09.2016, 20:32

After getting my hands on a version of Celestia 1.7 I’m amazed by its rendition of the unlit side of the rings and am looking for a realistic representation of Saturn’s rings, high resolution is desired. Of course you would find the 4k ring texture by Grant Hutchison: http://www.lepp.cornell.edu/~seb/celestia/hutchison/saturn-rings.html and Cassini images like this one: http://photojournal.jpl.nasa.gov/jpeg/PIA11142.jpg

Simply combining the alpha-channel of the texture and the color image gives a very high-resolution ring texture, which looks great at first but then you notice the contrast being too high, with transparent areas being too dark.

Saturn-rings-hires.jpg


To do at least some calibration, I use a different map of Saturn with similar color to many Cassini images and a gamma-corrected brightness corresponding to bond albedo, but found this to be too dark, so I introduced a bit of specularity to simulate a higher geometric albedo. The texture itself was a lot of work to do, and I am still unsure about contrast and color balance. It is the best texture of Saturn I could create, based on Bj?rn J?nssons’s map and filled with details by John Van Vliet and Runar Thorvaldsen.

Celestia’s way of rendering planetary rings is in no way accurate. As a physics student I have some basic understanding of the physics involved, but I am in no way an expert in planetary rings. I’ll try my best to express the way Celestia represents the rings vs. what it should do instead.

  • Celestia takes the color texture and makes it transparent according to its alpha channel. Since the data we have available is not the reflectivity of the ring particles, but instead the apparent brightness of the rings vs. the black background of space, simply combining image data with transparency results in rings being way to dark in transparent areas.
Celestia’s default just uses uniform brightness for each ring, which looks somewhat realistic. The opacity looks similar to real brightness, but the two are still unrelated. For my texture, I divide the brightness by opacity, but processed in a special way I will describe below. The rings should appear like they are photographed when seen in front of the black space.

  • Celestia always uses the same transparency independent of the angle between ring plane and observer. The real rings appear more opaque at low angles, for example when viewed against Saturn. The rings have a thickness, at low angles more ring particles are in the way, blocking more light.

I decided to start from scratch and use data available at Bj?rn J?nsson’s homepage: http://bjj.mmedia.is/data/s_rings/index.html. The transparency texture seems inverted compared to Celestia’s texture, which shows that Celestia is really using an opacity map (opaque rings being white, transparent being black). I used the back-scattered map as a base (Celestia renders the rings brightest at low phase angles when light is back-scattered) but later also processed the forward-scattered.

The textures are only 1500 pixels wide, so is the texture I processed (if you have issues with non-power-of-two textures, you might have to scale it up to 2k, but my graphic doesn’t care). He also posted text files in much higher resolution (5km/pixel). I have no idea how to convert tabulated text to an image file, I really need your help with this one!

I tried many different ways to create a texture that looks realistic. I’ll just describe the way I got the best result: I started with the transparency map, multiplied it with itself and inverted it to simulate a higher opacity due to a low angle with respect to the ring plane. Then I took the forward-scattered map, also multiplied it with itself and divided it by the corrected opacity texture. I gamma-corrected the resulting ring-reflectivity texture. Then I colored it with the relative-color-texture, changed color-balance to fit photographs and united it with the transparency map. Here the ready-to-use texture packed as an addon:

Saturn-Rings.zip
(640.25 KiB) Downloaded 315 times


saturn-rings-calibrated.jpg


But there still remain some rendering errors, probably due to simplification. I can’t work around them, Celestia would need an update to correct that:

  • Celestia 1.7 does render the unlit side properly in contrast to 1.6.1. Using these corrected texture shows similar structures to reality, only the overall brightness is too low.
  • In reality, the rings appear dimmer when the sun illuminates them from an angle closer to the ring plane, because less ligh can hit the narrower cross-section seen from the sun. Celestia does not take this into account, so most of the time the rings appear too bright.
  • The ring fade to black at very high phase angles. Celestia does not render forward-scattered light (when looking at the illuminated side). In reality the rings become dimmer at high phase angles, but they don’t become completely black (At least I think so, I haven't found photographic evidence)
  • When viewing the unlit side at high phase angles, the visible structures change again due to dust in some parts of the rings, forward-scattering the light.

saturn-rings-unlit backscattered.jpg
saturn-rings-unlit forwardscattered.jpg

Note the difference in the brightness of different ringlets between backscattered (left) and forwardscattered light on the unlit side.

A separate texture would be needed to render forward-scattered light for both the illuminated and unlit side. If this was possible, the rings could be extended to the D- G- and E-rings.

What do you think is the best way to process the data we have to create a realistic ring texture? How to convert high-resolution tabulated data into image files?
Last edited by FarGetaNik on 25.09.2016, 08:01, edited 1 time in total.

Avatar
Goofy
Posts: 238
Joined: 30.08.2011
With us: 7 years 10 months
Location: Italy

Post #2by Goofy » 24.09.2016, 23:25

Hi FarGetaNik.
Have you seen this threads?

http://celestiaproject.net/forum/viewtopic.php?f=5&t=17315&p=133565&hilit=SATURN+RINGS#p133565

http://celestiaproject.net/forum/viewtopic.php?f=5&t=9668&hilit=saturn+rings

I think you'll find them very interesting.
Unfortunately all images are missing.... :cry:
BTW, saying "using Goofy’s version of Celestia 1.7..." is not correct, because I've only shared the work of Chris Laurel and the other members of Celestia Development Team.
Just to address your thanks to the right people :wink:
Bye, good work.
Goofy :smile:
"Something is always better than nothing!"
ASUS G75VX4007H- Intel i7 3630QM, 2.4 GHz-SSD 250GB+SSD 1TB, SATA 3-32GB DDR3 1600 MHz- Nvidia GeForce GTX 780MX 3 GB

Avatar
Topic author
FarGetaNik M
Posts: 432
Joined: 05.06.2012
With us: 7 years 1 month
Location: Germany

Post #3by FarGetaNik » 25.09.2016, 08:00

Thanks, I was searching for those, but found only short, unhelpful threads. Sorry but the search tool on this forum is... special. I was googling :biggrin: So it was intended to include forward-scattered light... If Space Engine could do it, than I see no reason why it would be too difficult for Celestia (I don't have Space Engine installed anymore, I don't know how realistic its rendition is regarding the points I raised).

Goofy wrote:BTW, saying "using Goofy’s version of Celestia 1.7..." is not correct, because I've only shared the work of Chris Laurel and the other members of Celestia Development Team.

Hm if that's the case I'll edit my post to correct that error.
Last edited by FarGetaNik on 25.09.2016, 08:37, edited 1 time in total.

Avatar
Fenerit M
Posts: 1880
Joined: 26.03.2007
Age: 12
With us: 12 years 3 months
Location: Thyrrenian sea

Post #4by Fenerit » 25.09.2016, 08:25

FarGetaNik wrote:So it was intended to include forward-scattered light

Indeed the rings vertex shader doesn't include the ambient occlusion. It "assumes rings made of spherical particles, and no occlusion or shadowing between ring particles." (cit. from "rings.vp" file)
Never at rest.
Massimo

Avatar
Topic author
FarGetaNik M
Posts: 432
Joined: 05.06.2012
With us: 7 years 1 month
Location: Germany

Post #5by FarGetaNik » 25.09.2016, 08:44

Fenerit wrote:FarGetaNik wrote:
So it was intended to include forward-scattered light


Indeed the rings vertex shader doesn't include the ambient occlusion. It "assumes rings made of spherical particles, and no occlusion or shadowing between ring particles." (cit. from "rings.vp" file)

I was reffering to the thread:
chris wrote:It would be fairly simple to create a more realistic photometric model for the rings of Saturn, although tuning it to precisely match how the rings scatter light would be tricky. My plan is at some point to model the rings with two populations of particles: one of larger, primarily backscattering particles and a second one of very fine forward scattering particles. A second ring texture would contain the densities of the two types of particles in the red and green channels. In the region of the E ring, the texture would have red=0 (no large backscattering particles) and green > 0 (some low density of forward scattering particles), so that the rings would only be visible at high phase angles.

Of course that never happened in any Celestia release. Regarding occlusion (if that's the correct term?) I also think it can't be too hard, you only have to use the right angle-dependant funktion to modify the transparency.

Avatar
Fenerit M
Posts: 1880
Joined: 26.03.2007
Age: 12
With us: 12 years 3 months
Location: Thyrrenian sea

Post #6by Fenerit » 25.09.2016, 14:20

Some ring particles act like the "ambient" which scatter the light that is backscattered by other particles. It is not easy to do because the global illuminance shader must be of a new kind from scratch with respect to the default one, at least for the rings. I don't know if this can help but you can figure out a "multi-level" rings by adding a new tiny spheroids inside Saturn having the MeshCenter a little displaced along the planet's axis, adding them the ring system with different textures/color/transparencies and then to check what happens.
Never at rest.
Massimo

Avatar
Topic author
FarGetaNik M
Posts: 432
Joined: 05.06.2012
With us: 7 years 1 month
Location: Germany

Post #7by FarGetaNik » 26.09.2016, 12:24

I know, the shader needs to be reworked, this would require further developement of Celestia, which seems not to take place anytime soon. The physics can't be to complicated, I think it can be done with simple angular funktions, without considering individual particals.

Multi-level ring systems won't help much besides rendering the unlit side, which Celestia 1.7 can do. Forward-scattered light still won't be displayed, as Celestia doesn't consider the case when it's relevant (high phase angles).

My goal for now is doing the best with what Celestia can do for now, so using the best avaliable data and process is properly.

Avatar
Chuft-Captain
Posts: 1767
Joined: 18.12.2005
With us: 13 years 7 months

Post #8by Chuft-Captain » 01.10.2016, 00:54

Goofy wrote:Unfortunately all images are missing.... :cry:

FWIW,

I re-discovered my original images from that thread : http://celestiaproject.net/forum/viewtopic.php?f=5&t=17315&p=133565&hilit=SATURN+RINGS#p133565, so I have rehabilitated it by restoring the images.
"Is a planetary surface the right place for an expanding technological civilization?"
-- Gerard K. O'Neill (1969)

CATALOG SYNTAX HIGHLIGHTING TOOLS LAGRANGE POINTS

Avatar
Topic author
FarGetaNik M
Posts: 432
Joined: 05.06.2012
With us: 7 years 1 month
Location: Germany

Post #9by FarGetaNik » 02.10.2016, 09:02

So a fellow student was nice enough to convert the tables on Bj?rn J?nsson's website into textures 13177 pixels wide. It turns out only the transparency was in that resolution. Back-scattered light, forward-scattered light and color seem to display the same values for 4-5 pixels, so the resolution is still much higher than the 1500 pixel preview images. I've spend yesterday mostly with calibration, adjusting relative color and brightness of the rings with spacecraft images to fit Saturn. This required changing my atmosphere definition and the color texture again.

Firstly, the full-resolution ring texture:
saturn-rings-blend.png


For the texture I processed it exactly as I described on the first post, except I blended a bit of forward-scattered light into the back-scattered texture, so the rings look more realistic for phase angles other than 0°. Let me kow if you need pure back-scattered (or forward-scattered) light or a power-of-two resolution.

Here my atmosphere definition for Saturn:

Code: Select all

   Atmosphere
   {
      Height      300
      Mie      0.0009
      MieAsymmetry   -0.2
      MieScaleHeight   90
      Rayleigh   [ 0.00005 0.00013 0.0003 ]
   }

Together with my surface texture, here is how Saturn looks in Celestia:
saturn-rings-atmos1.jpg
saturn-rings-atmos2.jpg


Please let me know if color and brightness look realitic, I want to upload it to the Motherlode eventually.

Note that I use these parameters for the rings:

Code: Select all

   Rings
   {
      Inner    74510
      Outer   140385
      Texture   "saturn-rings-blend.png"
   }

And for Daphnis a semi-major axis of 136505.5 km to fit the texture. A strange phenomenon for Pan and Daphnis i noticed: Their orbital radii are given much smaller values on various sources than HORIZONS tells me. If I use HORIZONS data, they are well outside their respective gap, so I would need to adjust the edge of the ring well beyond what seems acceptable. I am also unsure if my spice orbits for Pan and Daphnis are working, so I don't know what is the case.

Avatar
Chuft-Captain
Posts: 1767
Joined: 18.12.2005
With us: 13 years 7 months

Post #10by Chuft-Captain » 02.10.2016, 21:39

FarGetaNik wrote:Note that I use these parameters for the rings:

Code: Select all

   Rings
   {
      Inner    74510
      Outer   140385
      Texture   "saturn-rings-blend.png"
   }

And for Daphnis a semi-major axis of 136505.5 km to fit the texture. A strange phenomenon for Pan and Daphnis i noticed: Their orbital radii are given much smaller values on various sources than HORIZONS tells me. If I use HORIZONS data, they are well outside their respective gap, so I would need to adjust the edge of the ring well beyond what seems acceptable. I am also unsure if my spice orbits for Pan and Daphnis are working, so I don't know what is the case.

How did you arrive at these particular values?

When I embarked upon a similar exercise a few years ago, I used the following values:

Code: Select all

      Inner   74500 
      Outer     140600   

which allowed both Pan and Daphnis to fit perfectly in their respective gaps - without altering the default SMA of Daphnis (which IIIRC, is 136500.)
Pan_Daphnis.jpg


... but this very much depends on the parameters of the source data used to create your RGB and aplha channels.

CC
"Is a planetary surface the right place for an expanding technological civilization?"
-- Gerard K. O'Neill (1969)

CATALOG SYNTAX HIGHLIGHTING TOOLS LAGRANGE POINTS

Avatar
Topic author
FarGetaNik M
Posts: 432
Joined: 05.06.2012
With us: 7 years 1 month
Location: Germany

Post #11by FarGetaNik » 02.10.2016, 22:23

Chuft-Captain wrote:How did you arrive at these particular values?

On Bj?rn J?hnsson's site, the inner edge for the texture is given 74510 km, the outer 140390 km. On this NASA page: http://ssd.jpl.nasa.gov/?sat_elem#saturn the orbital radius for Daphnis is 136504 km. I ended up with the value on Wikipedia, wich I admit is a random choice. Then I modified the outer edge to fit Daphnis in its gap.
Because of this discrepancy I let HORIZONS plot the orbit: One sample per year (relative to Saturn body center) between 1980-2030. I found the orbital radii only varied by 100 meters (!) around these values:

Code: Select all

Pan      134266.7
Daphnis   137169.5

But this is off by almost 1000 kilometers! Somewhere there has to be a huge error, but I checked the texture and it also fits the data I found with an error of 15 km.

Avatar
Chuft-Captain
Posts: 1767
Joined: 18.12.2005
With us: 13 years 7 months

Post #12by Chuft-Captain » 02.10.2016, 23:36

I'm assuming that you checked that the inner and outer limits of the alpha data matches exactly with the inner and outer limits of the RGB data.
This is important to ensure that transparency and RGB values are combined at matching radius.
"Is a planetary surface the right place for an expanding technological civilization?"
-- Gerard K. O'Neill (1969)

CATALOG SYNTAX HIGHLIGHTING TOOLS LAGRANGE POINTS

Avatar
Topic author
FarGetaNik M
Posts: 432
Joined: 05.06.2012
With us: 7 years 1 month
Location: Germany

Post #13by FarGetaNik » 02.10.2016, 23:49

Chuft-Captain wrote:I'm assuming that you checked that the inner and outer limits of the alpha data matches exactly with the inner and outer limits of the RGB data.
This is important to ensure that transparency and RGB values are combined at matching radius.

Yes, as far as I could tell transparency and reflectivity matched pretty well, which is no big surprise considering the data is from the same source. Here the textures I used:
backscattered.png

transparency.png


The only discrepancys I can see are due to varying resolution between the files. But they are different data, so structures in transparency don't necessary coincide with structures in reflectivity. But I guess it is the alpha channel that really matters when fitting them to Pan's and Daphnis' orbits.

Added after 11 minutes 23 seconds:
I also was curious if all "official" sources (data we can find anywhere... i. e. from NASA) or HORIZONS was wrong and measured the ring radii myself with Cassini mosaics. Of course this is horribly inaccurate, because mosaics are not guarantied to represent real proportions. But I found the radii to be much closer to "official" values than to what HORIZONS implies (all of them were roughly 300 km less than offcial sources) so it is probably HORIZONS that calculates the orbits wrong. The first time I experienced such a crucial error in that system. Makes me wonder what I should trust, I mean I also used HORIZONS data for many other moons to match them with flyby images...

Avatar
Chuft-Captain
Posts: 1767
Joined: 18.12.2005
With us: 13 years 7 months

Post #14by Chuft-Captain » 03.10.2016, 12:52

I meant to ask yesterday why is your texture 512px high?
You are aware that Ring Textures only require minimal height? - (2 pixels is plenty and reduces the filesize to 10-20K).

Incidentally, I tried your saturn-rings-blend in my setup (no Spice Orbits) with 74510, 140385 and with no adjustments to Daphnis' SMA.
Both Pan and Daphnis seemed to be correctly within their respective gaps. -- Daphnis is always going to be the tricky one as the Keeler Gap is only 42 km wide... equivalent to ~4 pixels in my texture (and presumedly 8 pixels in yours.)

CC
"Is a planetary surface the right place for an expanding technological civilization?"
-- Gerard K. O'Neill (1969)

CATALOG SYNTAX HIGHLIGHTING TOOLS LAGRANGE POINTS

Avatar
Topic author
FarGetaNik M
Posts: 432
Joined: 05.06.2012
With us: 7 years 1 month
Location: Germany

Post #15by FarGetaNik » 03.10.2016, 19:02

Chuft-Captain wrote:I meant to ask yesterday why is your texture 512px high?
You are aware that Ring Textures only require minimal height? - (2 pixels is plenty and reduces the filesize to 10-20K).

Yes I am well aware of that. I blew up the textures to 512 pixel so they are easily visible for processing. I found no need to reducing them back to 1 or 2 pixels. And png compression reduces the file size so there is only a factor of 2, whereas the height is reduced by a factor of 256... I don't think 52 KB is hurting anybody. When uploading it to the Motherlode, I will reduce it to this size, since they require it.

Chuft-Captain wrote:Incidentally, I tried your saturn-rings-blend in my setup (no Spice Orbits) with 74510, 140385 and with no adjustments to Daphnis' SMA.
Both Pan and Daphnis seemed to be correctly within their respective gaps. -- Daphnis is always going to be the tricky one as the Keeler Gap is only 42 km wide... equivalent to ~4 pixels in my texture (and presumedly 8 pixels in yours.)

The outer edge of 140385 is 5km short, and Daphnis is still on the inner edge of its gap. But I guess I want it to be perfect and fit it in the middle. I am also not quite happy that Pan is a few km away from its ringlet. And just because the Celestia data catalogue from 2011 tells us that Daphnis orbits at 136500 km, doesn't mean that it's true. All data I could find give me values from 136504-136505.5 km.

I will try to find precise data for their orbit periods and calculate semi-major axis from there, hopefully the rest of Saturn's moons wont disturb the result too much.

Avatar
Chuft-Captain
Posts: 1767
Joined: 18.12.2005
With us: 13 years 7 months

Post #16by Chuft-Captain » 04.10.2016, 02:27

FarGetaNik wrote:Yes I am well aware of that. I blew up the textures to 512 pixel so they are easily visible for processing.
I thought that was probably the case... just checking. :) -- I also display them larger during processing.

FarGetaNik wrote:The outer edge of 140385 is 5km short, and Daphnis is still on the inner edge of its gap. But I guess I want it to be perfect and fit it in the middle. I am also not quite happy that Pan is a few km away from its ringlet. And just because the Celestia data catalogue from 2011 tells us that Daphnis orbits at 136500 km, doesn't mean that it's true. All data I could find give me values from 136504-136505.5 km.

I will try to find precise data for their orbit periods and calculate semi-major axis from there, hopefully the rest of Saturn's moons wont disturb the result too much.
Ah, a perfectionist. Note however that the implied error of the opacity data (for a 5km dataset) is +/- 2.5km, and possibly the errors are greater than that. From memory, I'm pretty sure that the data I used (which is probably the same that you are using) like all NASA/JPL data includes error and quality statistics, so it might be worth referring back to that to find out just how large the errors might be.
What I think it comes down to is that, although it would nice to have everything lined up perfectly, it's possibly not valid due to the accuracy (or lack thereof) of the ring data. The magnitudes of the corrections you're trying to make to the orbits (~1-5 km) are probably somewhat less than the margins of error in the ring data, so although aesthetically it may be nice to do, ...from a scientific viewpoint, it's already within the margins of error. In fact, you have already spoken of "an error of 15 km" earlier in this thread.
(I'm not saying it's not worth doing from an aesthetic point of view, but that's perhaps just for your personal benefit ... otherwise you're going to require users to change existing long established orbit data for the moons, and I think there needs to be a justifiable basis for that.)

For comparison, I've scaled down and aligned your texture with one of my alpha channels, and noticed that the Keeler Gap is 2 pixels (ie. 20km) further out in your texture (at the top) than in mine.
comparison.png

This is not to say that yours is wrong (as this is only 1 of about 50 that I created and other samples may vary.) It's merely to point out that there are many differences to be found depending on which dataset you choose, so you should ensure that you have absolute confidence in the base data and production of the texture, before adjusting the long established SMA data for the moons. (That's just my opinion FWIW.)

Also, it's probably worth bearing in mind that most users of Celestia have probably never edited the .CFG file, so they will have the default value of 100 for RingSystemSections (in which case your centering efforts will be un-appreciated by them.)

One final general comment. Your rings are looking quite nice asthetically, although (as accuracy seems to be your aim) I do have some reservations about accuracy of the opacity values you have calculated. Overalll, they seem to be just a little too transparent to me, in particular the B-ring which should largely be almost completely opaque.
( cf. Image B from the original thread: http://celestiaproject.net/forum/viewtopic.php?f=5&t=17315&p=133565&hilit=SATURN+RINGS#p133565 )

CC
"Is a planetary surface the right place for an expanding technological civilization?"
-- Gerard K. O'Neill (1969)

CATALOG SYNTAX HIGHLIGHTING TOOLS LAGRANGE POINTS

Avatar
Topic author
FarGetaNik M
Posts: 432
Joined: 05.06.2012
With us: 7 years 1 month
Location: Germany

Post #17by FarGetaNik » 04.10.2016, 15:20

Chuft-Captain wrote:Ah, a perfectionist. Note however that the implied error of the opacity data (for a 5km dataset) is +/- 2.5km, and possibly the errors are greater than that. From memory, I'm pretty sure that the data I used (which is probably the same that you are using) like all NASA/JPL data includes error and quality statistics, so it might be worth referring back to that to find out just how large the errors might be.
What I think it comes down to is that, although it would nice to have everything lined up perfectly, it's possibly not valid due to the accuracy (or lack thereof) of the ring data. The magnitudes of the corrections you're trying to make to the orbits (~1-5 km) are probably somewhat less than the margins of error in the ring data, so although aesthetically it may be nice to do, ...from a scientific viewpoint, it's already within the margins of error. In fact, you have already spoken of "an error of 15 km" earlier in this thread.

I wouldn't call myself a perfectionist... but you got the point :biggrin: Yes I fear the data is just not accurate enough and I am trying to get an accuracy higher than possible. As a physicist I know well that every measurement contains some error, my problem is I don't know anything about the data I'm using. Even if we could measure the rings up to ~1km accuracy, the rings are still dynamic and can change over time, like the eccentric Titan Ringlet or the ever perturbed F-Ring. But what I learned from the (totally off) HORIZONS data is that the critical points are pretty stable over decades.
I found a table of Saturns rings on PDS: http://pds-rings.seti.org/saturn/saturn_tables.html, I am inclined to just trust this data and made a model of the rings and adjusted opacity to fit it:
opacity-model-comparison.png


Note that brightness in the upper texture does not correspond to opacity, but instead is just for highlighting the rings, ringlets and moons. The biggest adjustment was relocating the F-Ring, which results in the texture being a few pixels smaller (the inner egde is still constant). That was to be expcted of that chaotic ring. If I completely use this data, everything should just line up perfectly.

Chuft-Captain wrote:before adjusting the long established SMA data for the moons. (That's just my opinion FWIW.)

Code: Select all

SemiMajorAxis   136500     # Fit to Celestia's Keeler Gap
#   Eccentricity      0.004827 # Fit to Celestia's Keeler Gap

Sample of Celestia's numberedmoons.ssc . I'm interpreting this so that SMA already was adjusted to fit Celestia's ring texture. So who says it is "long established"? Daphnis was only known for 6 years at the time Celestia 1.6.1 was released. I have no issues in changing Celestia's default at all, I think of it as "imperfect work in progress" (Ok I might be a perfectionist in that point :oops: ) That's my opinion :wink:

Chuft-Captain wrote:Also, it's probably worth bearing in mind that most users of Celestia have probably never edited the .CFG file, so they will have the default value of 100 for RingSystemSections (in which case your centering efforts will be un-appreciated by them.)
Ok you got a point there...

Chuft-Captain wrote:One final general comment. Your rings are looking quite nice asthetically, although (as accuracy seems to be your aim) I do have some reservations about accuracy of the opacity values you have calculated. Overalll, they seem to be just a little too transparent to me, in particular the B-ring which should largely be almost completely opaque.
I discussed that issue in my first post of this thread. The transparency data I got results in the rings to be way too transparent. Since the only thing the rings can cover is Saturn itself, it would make sence to use an opacity consistent with an angle small enough so the rings are covering Saturn. I haven't found a function yet that describes that relation, but then I could plot in a reasonable angle. You can't find a constant, "real" opacity in the way Celestia renders rings.

Avatar
Chuft-Captain
Posts: 1767
Joined: 18.12.2005
With us: 13 years 7 months

Post #18by Chuft-Captain » 04.10.2016, 20:44

FarGetaNik wrote:I wouldn't call myself a perfectionist... but you got the point :biggrin: Yes I fear the data is just not accurate enough and I am trying to get an accuracy higher than possible.
Nothing wrong with being a perfectionist, but pragmatism may be a more useful attribute when dealing with imperfect data. :wink:
FarGetaNik wrote:Sample of Celestia's numberedmoons.ssc . I'm interpreting this so that SMA already was adjusted to fit Celestia's ring texture. So who says it is "long established"? Daphnis was only known for 6 years at the time Celestia 1.6.1 was released. I have no issues in changing Celestia's default at all, I think of it as "imperfect work in progress" (Ok I might be a perfectionist in that point :oops: ) That's my opinion :wink:
You may be right about that, and perhaps the original devs can provide some insight as to how those values were derived. In fact, I checked numberedmoons.ssc and found this comment next to the SMA of Daphnis:

Code: Select all

SemiMajorAxis   136500     # Fit to Celestia's Keeler Gap
... but I don't recall if this is from the original file or if it was modified by me for my rings. A quick look at the sourceforge version should clarify this (or if you see the same comment in your copy).

FarGetaNik wrote:The transparency data I got results in the rings to be way too transparent. Since the only thing the rings can cover is Saturn itself, it would make sence to use an opacity consistent with an angle small enough so the rings are covering Saturn. I haven't found a function yet that describes that relation, but then I could plot in a reasonable angle. You can't find a constant, "real" opacity in the way Celestia renders rings.
As long as Pan and Daphnis are in their respectivel gaps (even if not perfectly centered), I would suggest that getting the opacity right is a more important goal. But, this also is no easy task as there are significant differences in the quality and coverage of the existing data, as this example from my processing of some VIMS observations of a number of occultations on different dates graphically demonstrates:
RingGen_VIMS.jpg

One approach might be to average the observed opacities for a selected range of observations after first excluding the poorest quality candidates (From memory, I'm pretty sure that there is a quality flag/value associated with each dataset, and maybe even each measurement.)
EDIT LATER: I reviewed the data files and it looks like there are 2 fields, either of which could be used as a filter of observation quality:

Code: Select all

COLUMN 8: MEAN SIGNAL UNCERTAINTY
"Estimate of the one-sigma
    uncertainty in the mean signal cenetered at the given radius,
    computed based on the number of samples per bin.
    For unconstrained segments of the data, this value is set
    to -9.999."

which is a computed uncertainty derived from:

Code: Select all

COLUMN 13: NUMBER OF SAMPLES PER BIN
"Number of raw measurement
    samples averaged into the given bin. If zero, then the
    signal values are interpolated from adjacent bins."

However, the first and probably most important choice is which instrument to use (RSS, UVIS, or VIMS), and for Celestia the instrument which is closest to visual wavelengths is probably the best choice.

Whatever approach is used however, there will have to be some compromises made (as Gollum discovered. :D )
"Is a planetary surface the right place for an expanding technological civilization?"
-- Gerard K. O'Neill (1969)

CATALOG SYNTAX HIGHLIGHTING TOOLS LAGRANGE POINTS

Avatar
Topic author
FarGetaNik M
Posts: 432
Joined: 05.06.2012
With us: 7 years 1 month
Location: Germany

Post #19by FarGetaNik » 05.10.2016, 13:25

Chuft-Captain wrote:I would suggest that getting the opacity right is a more important goal. But, this also is no easy task as there are significant differences in the quality and coverage of the existing data, as this example from my processing of some VIMS observations of a number of occultations on different dates graphically demonstrates:

I agree, but I don't know how to approach this. First we need to clarify what opacity will be used (hence which obliquity) and than we need to look at the data we have and under what conditions it was measured. Then we know how to process transparency. And I don't think I'm far from reality, here a comparision between a Cassini mosaic and a Celestia render:

transparency-comparison.jpg

My concern was that transparency was to low (Celestia's default seems more opaque than this) but here it seems B is a bit to opaque and A is a bit to transparent.

Your screenshot looks very promising! Is transparency included? Yes averaging out the different measurements is a good idea. But there are some things to consider: there seem to be some data missing (lack of data is not equal to black) and if there are visible derivations in ring radii we would need to average out the radii of the ringlets or the texture will be unnecessarily blurry.

Avatar
Topic author
FarGetaNik M
Posts: 432
Joined: 05.06.2012
With us: 7 years 1 month
Location: Germany

Post #20by FarGetaNik » 06.10.2016, 22:08

I have another update. As described above, I matched the rings to a model made from data on PDS. I processed transparency slightly different than previously, hoping it looks more realistic. In this version, the C-Ring came out more visible. Again, I matched a bit of forward-scattered light in the back-scattered texture for a more realistic look. And as expected, everything lines up just perfectly! I even tried Celestia's default settings for Pan and Daphnis SMA and they fit perfectly! I think I may have reduced the error form ~20 km to ~5 km (texture resolution, at least for relevant details, I'm sure less clear features like density waves in the middle of a main ring are several pixels off, since those structures didn't line up).

Rings-Model.jpg
Rings-Model-hires.jpg


I included Pan and Daphnis in the hires screenshot to hint at their currect location, even if resolution is not high enough to prove.

Saturn-Rings-Model.zip
(1.35 MiB) Downloaded 121 times


Return to “Textures”

Who is online

Users browsing this forum: 3 guests