I originally wrote this in late 2014. This project has stalled due to other, less-involved projects taking precedent, but I hope to get back around to it sometime.
I remembered I had a copy of the 1980 NASA publication Viking Orbiter Views of Mars, a photobook containing a number of mosaics and images of interesting locations and features on Mars. The pictures show their age, though. Despite being processed with state-of-the art techniques available at the time, a lot of them are pale copies of the data that the Viking Orbiters originally returned to Earth. So, I got to thinking, "how would these images look if more modern image processing techniques were applied to them?" Let me start by posting one of the original mosaics that was printed in Views of Mars. *landslide photo* (Source: NASA History Office free online reprint of Viking Orbiter Views of Mars) When this mosaic was originally made, the most sophisticated techniques were mostly still a pair of scissors and an airbrush. The individual subframes could have a little distortion digitally applied to remove some of the geometric distortions created by the operating details and viewing angles of the imaging system. But the task of actually assembling the mosaic fell to a team of people working carefully with scissors and glue to piece the printouts of Viking images into mosaics like the one you see above. Once the mosaic was assembled, it was rephotographed. Those photographs eventually became the mosaics that went on public display at the Jet Propulsion Laboratory and universities involved with the Viking Project. There are a few problems with that approach, chief among them generation loss. Often, the Viking mosaics you see published are 3rd or 4th generation copies. (1st generation - original image; 2nd generation - printout from digital copy; 3rd generation - photograph of the assembled mosaic; 4th generation [if you're really unlucky and get something like the one above] - scan of a print of the 3rd generation copy) Printing generally has a poor dynamic range - you end up losing some of the subtle differences in brightness present in the original data. Bright areas tend to overexpose, dark areas don't expose enough. The end product ends up garish and fuzzy and not at all a decent representation of what Viking actually saw. This loss of dynamic range is part of the reason the edges of the individual frames in the mosaic are brighter than the centers - the subtle differences in brightnesses across the detector are exaggerated by the multiple reprints. Viking Orbiter used vidicon detectors for to image the surface. These were the standard imagers in use at the time, and I've written about some of the drawbacks of them as applied to the Voyager missions in other posts. Those can be found here, here, and here. For funsies, here's a picture of the Viking Orbiter imaging package: *viking camera system* (Source: NASA History Office) See the two telescopes? Each one went to a separate vidicon imager. As Viking Orbiter went around Mars, it would take a picture through one camera, and as it read the image onto the spacecraft's tape drive, it would take another picture through the other. Each camera had a slightly different field of view, with a bit of overlap between the two. In the above mosaic, you can see two tracks of images moving diagonally - each track was taken by a different "eye" of Viking's camera system. Viking didn't have to move to take pictures, it let its orbital motion carry a new portion of the imaging area into its cameras' fields of view. While it saved on the use of attitude control propellant, it meant that it created slight differences in viewing angles between objects in subsequent images. But even that could be turned into a plus. It allowed the Viking team to create stereograms, which allowed project scientists to view small areas in 3D and get an idea of the surface topography. Mosaics could be made larger on by combining images from multiple passes. You'd think this would cause differences in lighting conditions, but Viking was placed in a special Sun-synchronus orbit, meaning that the Sun angle would remain the same over imaging areas on multiple orbits. Mars's orbit around the Sun would carry a new area into view with every orbit Viking completed, with a little bit of overlap with a previously imaged area. Some of the mosaics in Viking Orbiter Views of Mars spanned multiple orbits - one particularly complex mosaic used data from 5 sequential orbits. Anyway, the original data from the Viking Orbiter imagers has a much greater dynamic range than the eventual published mosaics. The original data was archived and eventually stored in NASA's Planetary Data System, and is easily accessible to all interested parties. (Data formatting is an issue, and a topic for another post on another day.) I've downloaded the images used to make many of the mosaics in the book, and I'm starting to get my first results. Here's my digital reprocessing of the mosaic featured at the start of my post: *remastered photo* A couple of hours is all it took to produce this in Photoshop. I've removed all the calibration marks from the images, as well as the dust spots and noise. There's less contrast enhancement from the original, but overall there's a lot more detail since there's almost no generation loss (there is some from jpg compression) between the original data and this image. Despite that, there's a lot more fine detail visible in some of the areas that were blown out in the original image. I'm starting work on another mosaic, this one a lot bigger than the one I've posted here. Already some of my test stitches come in at over 6000 pixels in one dimension, and it's only about 2/3 finished. Can't wait to share it soon!