Which Photoshop filter does this? I've looked around and I can't find an obvious one.

My trick for this has been to take a mask that I've made with alternating black and white lines, paste it onto the image, erase away the part of the mask that I don't need (covering the part of the image that is not affected by the truncated lines, then use the wand set to no antialiasing and non-contiguous pixels to select the lines from the mask that cover up the black pixels in the underlying image, then go to the underlying image, shift the selection by one pixel up or down to get to the good pixels, copy and paste.

Using a filter that takes the maximum pixel value would be much much easier than this!!

--Emily

What you need is to deinterlace.

For this image:
http://saturn.jpl.nasa.gov/multimedia/imag...3/N00090982.jpg

Filter > Video > De-Interlace > Even Fields and Interpolation.

Not perfect, but it does the job.

Doug

Ideally, if the missing lines were all totally black, you could expand the canvas size to the right by say 1 pixel using black color and magic-wand the entire right side and select all the missing lines in one go. Since jpeg messes this up, you can increase the tolerance somewhat to compensate.
I don't know if it's me, but sometimes the deinterlace filter needs to be set to odd lines instead, could depend on the selection or something.

It's not a filter per se -- I just create a second layer and paste it on top of the original layer. Then under the layers menu, you have many options for which logical rules apply pixelwise to the layers, and one of them is Maximum/Brightest.

That's a whole wonderful side of Photoshop, playing with layers, logical rules, and transparency. The other thing it's useful for is colorizing high-res BW images with a low-res color layer.

I'm away from the computer on which I have Photoshop installed, so if this description seems unhelpful, let me know and I'll repost while running Photoshop.

OK, I've now done a side-by-side comparison of three methods of removing the every-other-line truncation: de-interlacing, taking the maximum of two layers, and my laborious method described above. De-interlacing seems to do the best job of the three because of the interpolation step. And you can even get away with applying it to the whole image (which effectively throws out half of the image's original pixels, replacing them with pixels interpolated between the remaining pixels) without very much loss of detail, even for these Rhea-crescent pictures, which are full of detail. It's better to select the area with the truncated lines and just apply the filter to that, but if you want to do a quick-and-dirty job the quality doesn't suffer much if you batch process a whole folder's worth of pictures.

The three strips below were produced from N00090982.jpg, with the Maximum, De-Interlace, and my methods applied to the corner that had the truncated lines.
Click to view attachment

--Emily

Hi all,

I've been working in IDL so I wrote a procedure to find how much of the line is black (or near black beyond a threshold), then take the average of the two surrounding lines. This is similar to one of Emily's methods I think.
I'll post it at this URL if you'd like to sort through the logic...

http://laps.noaa.gov/albers/idl/clean_ortho.txt

A good input threshold to use I find is around 27.

Very sneaky, Steve! Wish I had IDL. I know it's possible to export freestanding IDL applets (.sav files) that can be run with IDL Virtual Machine...don't suppose you'd want to give that a try...

--Emily

Interesting as I'm a novice with the IDL virtual machine. I tried making a .sav file if anyone would like to test it, assuming it can be downloaded for that purpose from my site:

http://laps.noaa.gov/albers/idl/clean_ortho.sav

Thanks for the attempt! It didn't work; I got an error:Oh well! It was worth a try, but I suppose there's no need to get fancy; De-Interlace works just fine for my purposes.

--Emily

Great explanation of how a mosaic is made and how Cassini images are compressed on your blog. One little suggestion. I select only the area that contains truncated lines to deinterlace. That way, the resolution isn't needlessly impacted for the rest of the image.

My trick - select the area containing truncated lines. Do a select color range to get the almost or all black pixels. Move the selection area (not the pixels, just the selection) up one line so it's over the good lines. Copy and paste, and move down over the bad lines. Then repeat, moving the selection down a line, copying and pasting and moving it up a line.

Now I have three layers. Original with black lines, and over the black lines, one copy of the line above, one copy of the line below. The top one I make 50% transparent so the "line fill" is an average of the line above and the line below. Flatten and serve with a glass of red wine.

Phil

Another way is to use a boxfilter to null the pixels affected by the truncation. Then use low pass filter with a box size of samp=1, line=3 to have each nulled pixel be an average of the good pixels above and below it...

Wow, great Rhea mosaic there, Emily!

Crater Chains:
Ive noticed there are a lot of these long linear features as well on Tethys, Rhea, and Iapetus, and yes even Dione. Ive started doing some fits to them and most do line up as radials to big basins, tho not always the biggest one such as Odysseus. Smaller basins and craters produce such secondary impact chains. We see something similar associated 100-km size craters on Callisto. It is when they fail to line up that they get interesting. that is why the big chains on Callisto must be tidally split comets, since there are no basins they can be linked to. ill have more to say on all this later but am currently working on the paper so have to focus on that. i note that the theory of disruption predicts that there will be no split comets at Saturn, due to its very low density.....
cheers,
paul

I'd like to know more about this. Does Jupiter break up comets in the vicinity of its "surface" or much farther out? If the former's the case I can understand how Saturn's density plays in, but if the latter - does a less dense object behave differently than a dense one (given same mass) at great distance, i.e. do they both not act as point gravity sources?

How about a plunge through the rings, especially at a low angle? Could that not disrupt a comet?

Ugordan you're right about planets acting more or less as point masses, so I guess the theory Dr Shank mentions envisages Jupiter's disruptions happening pretty close in. The most uncertain parameter in any such theory must be the cohesive strengths of the comets. I wouldn't be surprised if there is a huge spread, and you might only need a few atypical ones at long intervals to produce the observed alignments on the moons. Therefore I'm a bit wary of the prediction quoted, even without the rings issue.

Interesting. I look forward to reading that paper - maybe here???

i have to admit im short in the theory department. im only interested if they land someplace!
this all goes back to shoemaker-levy 9 days, now almost 15 years ago. breakup requires passage within about 1.5 jupiter radii of the surface of the planet, and something similar for saturn, except the density plays in somehow. the articles by asphaug, benz and others back in the mid-1990s tell all, otherwise start with the Schenk paper in Icarus in 1996 which includes such references. i dont have a copy with me at the moment, alas!
cheers,
paul

If anyone is still interesting in looking at Rhea the closeup images that were mangled have been replaced by much improved versions.

Click to view attachment

Did someone try to make the Saturn-shine mosaic of Rhea ?
Marc.

Well, I'm sure this is obvious to you but for others who may be interested I'll just expand on the point I made in my last post.

Any comet which is going to collide with one of Saturn's inner moons after being disrupted at Saturn has to have been orbiting pretty close to the ring plane in the first place. That particular subset of comets is precisely the one with the biggest chance of a lengthy and damaging encounter with the rings. Saturn's low density means that its surface is further out than the radius at which the gradient of it's gravitational field is deemed sufficient to disrupt a comet (based on certain assumptions about the material strength of comets). I suggest that an encounter with the rings presents a viable alternative mechanism whereby a comet in the right kind of orbit could be disrupted and then form a crater chain, streak or whatever on an icy moon.

In fact it could well be that for the special case of comets orbiting in their respective planet's equatorial plane Saturn (plus rings) is better at disrupting them than Jupiter.

Unless you consider those phase angles that make the concave look convex to be "mangled." This may motivate me to actually download some photos to invert them.

Alan - glad to see the raw image improvements. I had sent in an email to the link on raw images page on Sept 2 about the raw images stretching issue. I received a nice reply on Sept 10 saying they were about to fix the stretching and would retroactively reprocess the S33 images. This was caused in June as an inadvertant side effect of stretching improvements they were implementing.

They thanked me for "catching" this. I wonder how many others had reported it to them (either internally or externally).

You also have to wonder whether or not the whole fiasco was indeed a simple error. After all, we know that many people on the Cassini imaging team have been less than happy about feeling "forced" to release real-time images.

I'm not a big fan of paranoid speculation, truly... but I can just see a few specific people snickering to themselves and gloating, "Just let those &^%#@! UMSFer's try and scoop us to our pretty processed images with THIS stuff!"



-the other Doug

I don't believe the conspiratorial stuff for a second I'm afraid. I know the head of Cassini outreach personally, and she wouldn't take that sort of nonsense - believe me.

The UMSF admin team (Well, Bjorn and Emily) asked about it and it got fixed shortly thereafter.

Doug

Oh, trust me, I don't seriously believe there was anything like that going on. Just that some peoples' public statements on the general subject could lead the conspiracy-minded to that kind of conclusion.

The lesson, I guess, is that when you go on record as having a given minority view, and you could possibly have some influence on future events surrounding related issues, you have to be more careful than normal to avoid even the appearance of wrongdoing...

But, as I said, I'm not seriously suggesting anyone actually messed up the raws in order to make then unusable by amateurs. Just that the appearance of wrongdoing is invited, and could be pursued by the conspiracy-minded, because some people have aired such views. Actually, on the larger conspiracy front, I'm a little surprised that the Hoaxland crowd never jumped on the image quality issues by screaming "What are they trying to hide?????!!!!!"

-the other Doug

You're giving them too much credit. You think they're capable of noticing such a mundane thing over all the alien artifacts and technology and satellites visible as tiny specks in each frame?

Hoglanights don't need a image problems/sever issues to find artifacts/alien buildings.


Entertaining at the most but that's where the line is drawn.

I can get a good laugh with his "Theory's"

Updating this post about de-interlacing Cassini images, I can suggest trying to use the freeware software GDL to see if it is a viable alternative to the pricier IDL license.

http://gnudatalanguage.sourceforge.net/

Steve