Printable Version of Topic

Posted by: Exploitcorporations Oct 4 2005, 05:08 AM

Global mosaic:

Posted by: Exploitcorporations Oct 4 2005, 05:10 AM

9-Frame high resolution:

Posted by: Exploitcorporations Oct 4 2005, 05:12 AM

Global with 9-frame inset:

Posted by: dilo Oct 4 2005, 06:02 AM

Impressive work!

Posted by: RPascal Oct 4 2005, 03:55 PM

Very nice mosaics, congratulations!

When looking at the original images we can sometimes see that the dark crater floors are not as dark as the shadow zones in the craters. Would it be possible to preserve this difference in the mosaic by a slightly weaker contrast, so that details of the dark crater floors are still visible? Or do too many of the original images already have a contrast that is too high?

--René

Posted by: Exploitcorporations Oct 4 2005, 07:39 PM

Contrast reduction would probably work. I exclusively used the already dark UV(?) frames for the global view, with all of the high-res pictures adjusted to that, darkening the craters more. As an added handicap, the LCD screen of the scavenged laptop I've been privliged to use was apparently half-frozen in the back of somebody's car in Alaska at some point, leaving the left half of the screen with a pleasant ruby tint and the rest with exaggerated contrast. I've probably overcompensated badly. I'm not even sure what these look like on a normal monitor...by the way, your Huygens work is jaw-dropping, eye-gouging, f@#*ing amazing!!

Posted by: Exploitcorporations Oct 4 2005, 07:41 PM

21-frame (hires inset)

Posted by: tedstryk Oct 4 2005, 07:44 PM

Darn it...this is tantalizing! The thumbnails look great, but I can't get the images open!

Posted by: Exploitcorporations Oct 4 2005, 07:47 PM

Hires in context(detail):

Posted by: Malmer Oct 5 2005, 09:20 AM

started on the same composite but did it in color...

Used IR3 GRN and UV3 as color and CLR as luminance.

http://www.syndicate.se/image/space/HYPERION_IR_GRN_IR_CLR.jpg


It feels like they do some kind of histogram stretching on the cassini "raw" jpgs. (to make the frames easier to wiev for the public I guess)

Makes compositing much more troublesome. (some parts of my composite suffer from clipping in the dark areas due to this stretching)

I guess i should wait until they release the real raws...

/Mattias

Posted by: edstrick Oct 5 2005, 10:10 AM

I think they do a simple automatic contrast stretch. If the orignial data is 8-bit, (some of it is 16), and say 5% of the data is below a data number of 43, and 2.5% is above 229 (arbitrary numbers for example purposes), they do a linear contrast stretch mapping 43 to zero and 229 to 255, saturating the bottom 5% of the grayscale black and top 2.5% of the grayscale white.

I don't know the actual percent values.

That's why pics of small moons against black sky saturate them white.

Limb shots of a moon have black sky and that drives the low end stretch.. if the shot is entirely on sunlight moon, the bottom 5% point might be something like 113 and get mapped to black.

Posted by: Malmer Oct 5 2005, 11:25 AM

To bad... they should do the stretching based on the absolute lowest and highest DN. (or preferably not at all...)

But I bet that I am one of the few with this opinion. Most people just watch the pictures on the net... (and the really serious wait until the pds ones becomes available...)

Why complain. At least they let us see the pictures. (esa has a lot to learn)

M

Posted by: jmknapp Oct 5 2005, 11:44 AM

One thing about the Hyperion craters: they often have "gullies" on only one side, the other side being steep. Moreover, the "gully" sides of adjacent craters are often oriented in more or less the same direction. Any explanation for that?

Posted by: Malmer Oct 5 2005, 12:18 PM

It has to do with the fact that hyperion has an irregular shape. Most crates are situated on "slopes" and therefore the gravity is not normal to the surface. That causes the crater material to flow more from one side of the crater to the other.

this is also a reason for the large landslides on the sides of the very large crater covering almost the entire hemisphere. (they are very steep compared to the gravitational normal.)



/mattias

Posted by: jmknapp Oct 6 2005, 12:23 AM

Interesting... a quick calculation gives a gravitational force on Hyperion less than 1/300th that of Earth, so landslides must be in very slow motion.

Posted by: Malmer Oct 9 2005, 09:34 PM

I finally finished a color panorama using UV IR GRN for color and CLR for luminance.

I used Philip Stookes hyperion shape model to reproject the images into a common projection.

The color balance is just guesswork. Is hyperion more reddish?

http://www.syndicate.se/image/space/HYPERION_IR3_GRN_UV3_CLR.jpg

Exploitcorporations; You work fast! Excellent work all the way!


Mattias

Posted by: RPascal Oct 10 2005, 04:07 PM

Excellent work!!!
On Wikipedia there is an article about Hyperion with a colorized (?) Voyager image that is said to show the real color - I do not know how correct it is, but then Hyperion should be somewhat more red. Perhaps you can find the origin of this information.
--René

Posted by: tedstryk Oct 10 2005, 04:57 PM

Here is the Wikipedia image


It is the third from the right in my set.


What is significant is that this image, unlike others in the sequence, is only available via a clear filtered view, so it is clearly utilizing color data from other images (and, I suspect, just a general monochromatic overlay)

Using OGV processing combined with super-resolution processing, I constructed this color view of the image that is second from the right. It has very similar color to the Wikipedia image.



I think it is therefore reasonable to conclude the orangish color is due to the difference in filters being used, and is worsened by the fact that the sloppy colorization of the Wikipedia image doesn't allow for any varations across the disk.

Posted by: Malmer Oct 10 2005, 05:20 PM

What software do you use to do superresolution. I can only find something called QE superresolution and it is kindof crappy...

/Mattias

Posted by: volcanopele Oct 10 2005, 05:28 PM

Hyperion has kind of an orangish, golden color.

Posted by: tedstryk Oct 10 2005, 05:32 PM

I find that for it to really work, you have to do it manually in Photoshop...there was a discussion of it in another thread at one time....I will see if I can find it.

http://www.unmannedspaceflight.com/index.php?showtopic=998&st=30

Found it, it was in the Viking thread. My technique has become a bit more standardized. After cleaning the image up, I stretch the contrast if it is on the low side, enlarge it by 500 percent, and then carefully apply unsharp masking (sometimes, I skip this step or use a very high threshold with poor quality images.) I then stack them as layers. Assuming there are, say, five images of equal quality, I assign the first layer 100% opacity, second layer 50%, third layer 33 percent, fourth layer 25 percent, and last layer 20 percent. This yields a 20 percent contribution from each image. As I stack on each layer, I initially enter it as 50 percent, so that I can see it well to precisely line it up (this can take hours, but I find programs like registax just don't do as well with these over-enlarged (and therefore not-so-sharp) images - not well enough for sub-pixel detail. I immediately change it to the opacity it should be. Otherwise, in large stacks, you find that you can progressively slip out of alignment.

In most cases, the images are of varying quality, and I therefore give the better images a higher opacity than they should have, and the lesser image a lower opacity. Once this is optimized, I merge layers. I then tinker with the contrast, if necessary, and apply unsharp masking again. Sometimes, areas of the image get a really blocky appearance, which I try to smooth out at this stage. With the best of datasets, I am done at this point. Usually, I shrink it by half, and sometimes down to only 200 %(or even 150%) of its original size, if the image looks fuzzy and is not fully taking advantage of its large size. This somtimes takes some time experimenting to avoid data-loss. When available, I then create a tri-color image which I scale up to overlay on the new product.

Posted by: Bjorn Jonsson Oct 10 2005, 05:57 PM

I simply use RegiStax. It is mainly used for stacking and processing planetary images obtained using webcams (in that case usually hundreds of images) and I have used it for that but it also works very well for stacking spacecraft images.

Posted by: tedstryk Oct 10 2005, 06:55 PM

Registrax works well with stacking images at original resolution. But, especially with images that are of varying quality and somewhat different illumination (in the case of Viking lander images), I find that better registration can be done manually. I know Phil Stooke and Tim Parker use a similar technique to mine. Tim Parker blows the images up by 1000 percent. For me, that makes them unweildy and does not gain anything better than at 500 percent.

Posted by: Malmer Oct 10 2005, 10:30 PM

Ill give it a shot... It would be nice to get some extra detail into the parts of the hyperion picture with lowresolution material. There are quite a few pictures that can be stacked together... so maybe its worth the extra labour...

/Mattias

Posted by: tedstryk Oct 11 2005, 12:38 AM

One word of caution. I have had little success using this technique with the Cassini jpegs. It has worked well with pds-released Cassini images, but the compression in the jpegs negates the improvement in resolution. I end up just enhancing the jpeg artifacts.

Posted by: AndyG Oct 11 2005, 09:27 AM

Yes...they must be.

I threw together a Flash app which returns handy gravity info on small-to-large terrestrial bodies. It's [a href="http://personal.strath.ac.uk/cjis28/gravity.swf"]here[/a], and requires the latest Flash player.

Setting Hyperion as density 1440kg/m3, radius about 130km, I see I could jump across any 600m craters present... ;-)

Andy G

Posted by: Malmer Oct 11 2005, 10:44 AM

I use a homebrewn jpg artifact removal filter that might work... (its basically a median filter that operates on the edges of each 8*8 pixel jpg block)

/m

Posted by: jmknapp Oct 11 2005, 12:49 PM

Hi Andy,

I wasn't able to get that app (http://personal.strath.ac.uk/cjis28/gravity.swf) to work on my machine. Sounds good though. I tried updating my Flash player but that didn't help.

But I did see your other page of Javascript calculators--particularly one on crater sizes. Very handy considering I was just wondering what size cometary objects might have created the string of craters on Tethys:

http://saturn.jpl.nasa.gov/multimedia/images/raw/raw-images-details.cfm?feiImageID=50034



I figure the biggest ones in the string are about 2km across, & your calculator indicates craters that size could be made by a 60m chunk of ice traveling at 10 km/s, or a 20m chunk traveling at 50 km/s, assuming Tethys was mostly ice.

Posted by: Phil Stooke Oct 11 2005, 02:05 PM

Love that big colour mosaic. One day I may get around to reprojecting the Cassini Hyperion data using that shape model, to make a global mosaic. But I have really got to finish my book first. Aaargh! There's too much going on these days. It almost makes me nostalgic for the 80s when you could go a year without any new planetary data at all!

Phil

Posted by: elakdawalla Oct 11 2005, 02:42 PM

Yikes! Don't say that! Don't even think that! (I know you don't really think that.)

Seriously, there no mission yet planned that is going to provide data like Cassini's at Jupiter or Saturn or Uranus or Neptune once Cassini is dead. It's a flood right now, for sure, but in a few short years we'll all be herding at the water hole...

--Emily

Posted by: AndyG Oct 11 2005, 03:30 PM

I had to give my Firefox a kick to run it: it was written in the latest Flash-8. Maybe try a different browser?



Ah yes...that's an old one, drawing on Shoemaker's stuff and various H-bomb studies: though there's not much chance of "igniting cotton" on Tethys!

But it is interesting that the strings/chains of impacts really don't need much in the way of material - just a disruptive gravitational influence on a loosely-bound "hundreds of metres" body sometime before the impact. I suppose, from the direction and layout of the strings, it might even be possible to suggest likely culprits for these disruptions. Interesting idea...

Regards,

Andy G

Posted by: JRehling Oct 11 2005, 03:37 PM

Cassini might be continuing action for a very long time to come. I saw one estimate (before the Huygens relay snafu correction) that Cassini might survive for (this was the figure) 200 years! That correction ate up a lot of fuel, and I get the feeling they are being a bit profligate in making added corrections now, but all told, and given that Voyager launched in 1977, we might really expect to see Cassini still operating 25 or even 50 years from now. Well, "see" it for those of us who live that long ourselves!

Posted by: ugordan Oct 11 2005, 03:47 PM

I really, really find it hard to believe Cassini could last that long. The first problem is it's likely to run out of propellant long before that. Without it, there will be no way to do new close encounters with any of the moons and the spacecraft would essentially be in a uncontrolled orbit, freely perturbed by the moons' gravities.
The second problem are the 3 RTGs -- they slowly, but continually produce less and less electricity. In a few years time, there will probably be a serious energy budget where not all intruments will be allowed to operate at the same time.

IMHO, it's reasonable to expect at least 6 years of good science past the nominal mission, but anything over that will depend largely on the number and magnitude of orbital trim maneuvers conducted.