Some random Cassini stuff, from the PDS Options

[ugordan]

Here's a very short, grayscale Rhea flyby movie made up from 10 Cassini clear frames. The frames were reduced to 3/4 original size to keep the thing reasonable. One could arguably make a 3d anaglyph because the vantage point changes quite a bit. Note that the sequence is clipped a bit because Rhea was filling the entire field of view. There's a noticeable dimming of surface features as the phase angle increases.

 RHEA.mov ( 972.05K ) Number of downloads: 695

WARNING: Quicktime 7 and H.264 codec required!

[ugordan]

Also, here are a couple of color composites I don't recall were previously released by CICLOPS team:

Prometheus in true color within the rings on the left, Dione RGB composite on the right. Dione's saturation was enhanced, but the thing is still very gray:


An R+B shot of Saturn with the unilluminated side of the rings present:


Another shot of Saturn's cloudtops, R+B, synthetic green. The image on the right was greatly enhanced to bring out the cloud features:

[Bjorn Jonsson]

Great images although I'm not sure the color balance in the Saturn images is correct (however, that's something that can also be said of my own images ). I'm surprised more images processed from the Cassini PDS releases haven't appeared here, they are an absolute goldmine.

As a matter of curiosity, how did you make the synthetic green images ?

[ugordan]

As a matter of curiosity, how did you make the synthetic green images ?

I simply used Photoshop's channel mixer. I used a balance of red and blue that produces this general color. I can't remember OTOH the exact mixes, but I do remember that red + blue add up to over 100% for the green. Generally, it's a pretty even mix with only a couple of percents more red than blue. I was aiming towards something looking like this and to some extent this ground-based image (courtesy of Anne Verbiscer).

You're right, there's a real goldmine in the PDS images, in fact I think the next batch of data should be released very soon, if not already.

BTW, since you're 'here', can you tell me if you're having any problems with flatfielding the WAC images? My code seems to work perfectly on NAC frames, but the WAC frames still end up with excessive static noise. I don't know if my code is faulty (which would be strange because it handles both NAC and WAC images the same way) or the provided wide angle flatfields aren't that good.
That sharpened cloud image above is distinctly noisy as can be seen.

[ugordan]

Several Enceladus shots in stretched color:


Taken on 15 January 2005. Shows dimming with increasing phase angle.



These were taken on 16 February 2005. The brown darkground in the first image is Saturn's disk.
The right image one is a stereo pair intended to be viewed crossed eyes. The gamma of the image was lowered to bring out a little more details on the disc.


These three were taken March 9 2005.

EDIT: I removed two separate images from Feb 16 and replaced them with a stereo pair instead.

[volcanopele]

What do you use for calibration? Some of those enceladus images are stretched very strangely. What filter combos are using?

[ugordan]

I assume you're talking primarily about the uppermost image (3 shots)? That does look oddly colored to me as well. The same procedure I used to get the other images was also used on the first one. I don't know why it turned out with such a distinct hue. Maybe primarily the older, northern terrain is visible. The south polar region does appear bluish, though.
I use my own code which is still far from perfect.
The filter combo is IR1 and UV3. I used the GRN ones where available (which were regularly waaay out of perspective with the other two) to instead select parameters for a synthetic green image. The end result was effectively the same.

[volcanopele]

using a synthetic green may be part of the problem...

[ugordan]

The same thing happens if I use the real green channel, I tried it. Apart from color fringing it made practically no difference.

Here's what I mean:


The left one is using real green, right one is synthetic. I magnified the images 2x to show the color fringing that the green channel introduces (note the small crater).

For what it's worth, metadata for the above image says the sub-spacecraft latitude and longitude are 3.19 and 70.18 degrees, respectively.

[Bjorn Jonsson]

BTW, since you're 'here', can you tell me if you're having any problems with flatfielding the WAC images? My code seems to work perfectly on NAC frames, but the WAC frames still end up with excessive static noise. I don't know if my code is faulty (which would be strange because it handles both NAC and WAC images the same way) or the provided wide angle flatfields aren't that good.
That sharpened cloud image above is distinctly noisy as can be seen.

I don't see any obvious difference between calibrated NA and WA images in my case. The only 'difference' I have noticed is that compared to NA images the WA images are more often compressed aboard the spacecraft using lossy compression and because of this artifacts are sometimes visible.

When I calibrate the images the blue images always get rather dark. To make color composites I have to brighten the blue channel by a factor of ~2 compared to red. I wonder if you get similar results - in fact comparing images might be interesting because without 'groundtruth' images from CISSCAL debugging the code can be difficult when possible bugs are not visually obvious in the resulting images.

[ugordan]

When I calibrate the images the blue images always get rather dark.

WA only or both NA and WA images?
I noticed wide angles are pretty tricky because they use on the order of 10x shorter exposures than NA (speaking OTOH). This makes that shutter offset fudge factor in cisscal quite important. I've experimented with Jupiter and Dione WA images, comparing them to similar NA images to come up with a different offset than what is used in cisscal (to match the resulting colors). Using the default offset definitely makes the WAC images look weird. I'll have to check the code when I get back home and maybe put up a few examples of what a 2 ms shutter offset difference does to a WA image.
As for NAC channel being too dark, does it happen to you all the time? When looking at a grey target, such as Dione, I get pretty normal results, not brownish or yellowish.

I do often wonder just how much CICLOPS team manipulates the color images to get a greater "awe factor".

[Bjorn Jonsson]

I always get similar results for NA and WA, i.e. the blue images are dark. To get nice color composites I usually need to multiply G by a factor of ~1.3 and B by a factor of ~2 (this is from memory - the exact numbers are a bit different).

I suspect CICLOPS may manipulate the color to get more realistic results but having seen Jupiter and Saturn through a telescope I suspect the color may be too saturated even though something like Saturn would look different at close range - the effects of contrast (black sky) are remarkable. I remember how strange Jupiter and Saturn looked near the Moon's limb several years ago - they looked dark, very dark.

Processing color images of Saturn to roughly match what I see through a telescope and then applying identical processing to satellite images is one of the many things on my 'PDS/Cassini TODO list'.

IIRC I corrected a bug involving shutter offsets in my code several months ago. I'll have to check that part of my code tonight to see exactly what I did.

[ugordan]

Here's an example of the effect of small changes in shutter offset. The top row is a Dione image using W1481738584_2.IMG, W1481738622_2.IMG and W1481738603_2.IMG. Three cases are shown and the resulting brightness ratios are also preserved. Note the brighter results when subtracting the value from the offset slope.
The second row is a Jupiter RGV image using W1352839344_2.IMG, W1352839385_1.IMG and W1352839450_1.IMG.
The brightness trend is even more obvious here. I also included a NAC frame taken at around the same time, but using a slightly different filter combo, CB1/GRN/BL1, primarily to compare the brightness and to a lesser extent color.

Based on the assumption the WA and NA images should be equally bright once the optics areas are taken into account as well as the ugly yellow color of both Dione and Jupiter, I'm inclined to say a -1.0 ms offset (which cisscal uses) is definitely wrong.
As I said before, WAC is very sensitive to this offset because the red and green exposures are typically 10 and 15 ms (compared to around 70 for blue) that a small shift makes a large change in output. NAC is much less sensitive to this as it uses much longer exposures.



Finally, I can't seem to reconcile the Jupiter and Dione appearances - gray Dione appears to suggest +1.0 ms is correct, while very bluish Jupiter at that setting suggests something slightly higher than 0.0 ms.

I'd definitely be interested if you come up with similar results or if my code is seriously messed up somehwere.

In short, I wouldn't trust the color of the wide angle shots without much more precise knowledge of the shutter response. The images officially released probably contain a fair bit of artistic freedom to them.

[Bjorn Jonsson]

Regarding shutter offsets, I had a look at some code I modified several months ago and found this:

constoffset=1.0; // See comment in line 65 in cassimg__dividebyexpot.pro
// There is a bug in CISSCAL, the following line is missing. See section 4.3 in the Ground Calibration report.
if (cam=="ISSWA") constoffset=0.0;
for (i=0; i<numsamps; i++) buf[i]=expo-buf[i]-constoffset; // The true exposure at sample j

So apparently the shutter offset is 0 for the WAC.
IIRC the Ground Calibration report is on one of the calibration volumes (probably vol. 11).

Based on the assumption the WA and NA images should be equally bright once the optics areas are taken into account as well as the ugly yellow color of both Dione and Jupiter, I'm inclined to say a -1.0 ms offset (which cisscal uses) is definitely wrong.

Apparently this is correct and the offset should be 0.0 - see the code above.

[ugordan]

So apparently the shutter offset is 0 for the WAC.

Ah yes... that would explain why I also at one point included the check if the camera was NAC to subtract the constant offset. I failed to make a comment in the code why that was the case and later removed the check...
Thanks for finally clarifying this mess.

About the point of multiplying G and B by constant factors, I guess it all depends on what you consider realistic. What color do you get for example for low phase Enceladus shots without the factors? If it still turns out white, I'd say that's realistic. Of course, Saturn may wind up brown-yellowish then, but again -- how do we really know it doesn't actually look that way from those angles? I try to limit my processing in terms of color to changes in gamma/saturation, without multiplying individual channels by these magic numbers. Additional color balancing might be performed with weighed mixing of the 3 filters, but you need to have a correct white point for that as well.
The calibration procedure already has correction fudge factors built in so I take it they're the correct ones. The end result might look more dull and less colorful (barring any errors in my calibration code), but I prefer realistic (whatever that actually is) to plain pretty.