There seem to be some real imaging wizards around here ( you know who you are ) and since my mosaicing skills can be summarized as: put one image next to the other, I wanted to ask you guys if you know of some good tutorials / tools on the subject.

For years now I have been trying to make mosaics out of Galileo images. There are many, many, many images that haven't been released on the 'general public' sites. Unfortunately the images do not align very well without projecting them properly. (see attached image)

I'd appreciate some pointers.

I know others have their own personal preferences, so i'll just post mine:

The underlying tool set is called panorama tools ( http://webuser.fh-furtwangen.de/%7Edersch/ ) , which has a number of different front end packages to assist in organizing all the steps involved. If you are planning on using a front end, i'd suggest you download the panorama tools version suggested for that particular front end, as each version has its own intricacies.

I've personally found the front end package PTassembler ( http://www.tawbaware.com/ptasmblr.htm ) to be very handy, when grouped with a couple other tools on the side.

PTassembler lets you pick control points where two or more images overlap, and uses the panorama tools package to optimize for lens distortion and lineup between the images. You can take the time to pick control points by hand, but another tool you can download and integrate into PTassembler greatly speeds up this process. Its called autopano ( http://autopano.kolor.com/ ).
Autopano will automatically find control points and import them into PTassembler (and PTGui if you're more partial to that front end) so you can save some time by not picking the control points by hand.
Finally, to stitch together the output images, you can use the PTassembler (which uses panorama tools) to interpolate between the two edges to give a smoother look (problems like vignetting end up distorting the brightness of the edges of an image, sometimes leading to dark/light boundaries). On top of the interpolation built into Ptassembler/panorama tools, there's one last add-on package which can help out with this problem. It is called enblend ( http://enblend.sourceforge.net/ ) and can further blend the edges leading to a much more consistent output image and can be integrated directly into PTassembler.

Thing is - with orbiter imagery - it's not particularly valid to use something like Panotool etc - because it's not a panorama from a fixed camera. The camera's moving. Quite quickly - and infact the camera probably isnt changing it's pointing at any point in that series of images at all.

When that's the case, I'm affraid photoshop is probably your best answer - unless you use one of the panorama programs to reproject the image using the focal length and f.o.v. of the galileo camera which may give you something slightly easier to align!

Doug

I have mainly been processing Voyager and Galileo images. I usually reproject everything to simple cylindrical or polar projection using information I have on the viewing geometry, mosaic the images there and reproject back as a final step.

I usually do this also when making color composites.

Is that a mosaic of the surface of Europa? I miss Galileo lol

oh - and by the way - that lot alligned VERY well as I see it - excellent stuff!

Doug

Not if you look at the right side of each image... I tried to align the images in such a way that when your eyes move from the bottom toward the 'horizon' you don't notice the misalignments as much.

And yes Sunspot, that is Europa on the E12 flyby. As far as I know this was the highest resolution obtained by Galileo of Europa.

Thank you slinted, for the many excellent links! Enough to start reading and experimenting I should say...! But is it true what Doug says that it is not suitable for Orbital images?

Doug, how do you suggest I use photoshop to align the images?

Bjorn. I am a dummie newbie. That's double trouble. How DO you reproject everything to simple cylindrical or polar projection using information on the viewing geometry, mosaic the images there and reproject back as a final step? I have saved the IMG info on every Europa image Galileo made, but I can't use it yet. Too much strain on my little brain.
BTW, when are you going to update that excellent Europa map of yours? ( http://www.mmedia.is/~bjj/data/europa/europa.html ). Been waiting for the eastern hemisphere for ages...

You could simply go along the PTGui/Panotools route - telling it the info about the camera, and then render out a panorama of one image Repeat for all the images and then you're good to start stitching in Photoshop.

you MAY find that Panotools helps - no harm in trying - but I doubt it'll work too well.

Doug

I looked at the site http://webuser.fh-furtwangen.de/%7Edersch/ for the PanoTools but apparently there should be a file called PTStitcher.exe, but you can only download PTViewer.exe from that site... What's up with that?

Sorry about that, the link had the full installation of panorama tools, but apparantly it has been pulled. As I said, if you wanted to try out either PTassembler or PTGui, they provide their own version of panorama tools that is guaranteed to work with their frontends, so that might be the best place to pick them up if you were interested in trying out either of them.

If you were just curious about the panorama tools software itself, here's another link : http://panotools.sourceforge.net/

As to its usefulness in working with orbital images, I definetly agree that its not a perfect fit since the orbital images are taken from different locations with different surface resolutions, (which is not what panorama tools was designed to do) but I do find it useful to use it to get a low res layout image automatically using the auto-control point features of PTassembler. If you are looking for something that will produce true map projections (taking into account phase angle / distance to target etc.) this won't do the trick, but it might be a good place to start if you're just hoping to quickyl and automatically get a feel for what the area looks like, or how the images will be stitched in the end product.

As Doug mentioned, it might also be useful for you to just pull out specific pieces of this to help you along the way. For example, if you were using photoshop or something similar to align every image as its own layer, you could output each of those layers to the enblend tool I mentioned before, and have it to the blending at the edges. You'd want for those edges to be as exact as you can, but in the case that there are still distortions / misalignments, you would at least be able to blend them into a visually consistant scene.

I don't mean to sound like an expert at this, I'm far from it, but I've found these tools useful in the past. I hope they can help you as well, what you've got already looks great, and I can't wait to see what else you're able to do.

Unfortunately PTGui and PTAssembler are not free (anymore). But I found a free package called PTOpenGui ( http://www.nic.fi/~juhe/ptbcbgui/ ). Still needs the complete Panorama Tools though, including PTStitcher.
I really want to experiment with some MER-panoramas as well, so... ehm... is there an 'alternative' way of obtaining this PTStitcher and co.? ...Say no more, say no more, know whatahmean, nudge nudge?

I bought PTGui a year or so ago as a matter of course. I find it SO good that I considered the price to be a bargin

Doug

Just a thought - you could use PTGui with like, a 1 degree f.o.v. setting (or lower ) and then let autopano do the matching for you How about posting a few of the frames in that stitch here, and we can all have a go and compare notes on results?

Doug

Tonight I'm going to see if I CAN work with PTOpenGui / PanoTools and or Photoshop, so no results from me yet...

Ok, I will attach the E12MOTTLE02 observation here. Don't know if this will eat into the storagespace on this forum ...?? You'll still have to cut and paste the images though, because they are all quarter sized.

In the meantime, here is a mosaic I did some time ago based on the Galileo E4ESMACSTR02 observation. Worked well because there are so many datagaps you don't notice the misalignments! Nasa released part of it in this release: http://photojournal.jpl.nasa.gov/catalog/PIA00543

This one is Melkart crater on Ganymede, from the G8GSMELKRT01 observation.

Again on Ganymede, Nun Sulcus. Observation G7GSNUNSUL01:

The southern latitudes of Europa (17ESREGMAP03):

This mosaic of Buto Facula on Ganymede was released by nasa: http://photojournal.jpl.nasa.gov/catalog/PIA01659 . It is not full-resolution though. But this one is :

a pit crater on Ganymede:

Pwyll crater on Europa. Same as http://photojournal.jpl.nasa.gov/catalog/PIA00586, but it covers a slightly larger area:

This one is the high-resolution companion to the former image. Look to the right of the Pwyll ejecta-blanket and you'll recognize the same features as in this image:

Erech Sulcus on Ganymede. Same as http://photojournal.jpl.nasa.gov/catalog/PIA01615, but a wider area:

Palimpsest named 'Zakar' on Ganymede:

Hi, thanks for the links to those great apps.
I downloaded them and started tinkering but I have had one problem that I'm not sure how to handle.
I couldn't find reference to it anywhere in the help page.

After assigning Control Points and all that and I go on to create a preview and the final pan but I get the following error:
"Image Must be RGB"

I'm using the raw images from the MER website which I I figure are single channel,
Do I have to try and combine three images to use with the Pano tools?

I'm a total newbie to this so sorry.

Thanks again
jG

I have attached my latest incarnation of Galileo's E26 image of Amalthea. It was taken from a range of 238,000, similar to many of the high quality images of moons that Cassini has taken, but Amalthea is so little it still doesn't extend very far.

That's a pretty cool image, Ted.

Thanks. I have really tried to make this image look good at a size that visually is presentable, as it is the only view of Amalthea with which one can do so without it looking blurry or artifically smooth (it shows the disk about 60 km across, while the next best views show it at about 34 pixels across in the same dimension). However, the original overexposed the bright regions, making it look crappy, and suffered from compression effects. I used a mix of two versions Galileo transmitted to mitigate the lack of clarity caused by enlarging the original image from 800x800 pixels to 1024x1024 pixels. This merging did a lot do de-emphasize artifacts. I also removed noise. There is an image of Thebe with nominally higher resolution (2.0 km/pixel compared to 2.4 km/pixel), but it is much noisier, losing its advantage. Eventually I hope to colorize this view using color data from Voyager and other Galileo orbits.