Progress on the Phoenix Mission Success Panorama

I mentioned on Wednesday that Mark Lemmon said he had pronounced the data acquisition for Phoenix' Mission Success Panorama to be complete -- and I predicted we'd see a version from one of the amateurs out there on the Internet before it came out from the mission team. And here it is, by James Canvin. He's not done fiddling with this yet, but it looks pretty awesome as it is; visit his website for a higher-resolution version.

I'm probably speaking for most of you when I say that space enthusiasts just love looking at the imagery that comes from space missions. We'd prefer to see pictures that are seamless and smooth, with no annoying gaps in data. When people like James assemble images into panoramas, they use software that warps images to align features that are common to adjacent images, then blend the two images together to make a seamless product. They also fiddle with the brightness and the color in images to make adjacent tiles match perfectly, and sometimes to make sure they match our preconceptions for what the colors of Mars (or Saturn, or whatever) should look like. And if something causes a gap in the data, it's not hard for an expert image processor to do stuff to fill in the gap -- copy some rocks or soil from one place to fill in another place, or copy some parts of images taken many days previously to fill in spaces that weren't covered in one observation. All of this can be quite a lot of work, but once you get good at it there is a process you go through that can allow you to take a data set like the 150 different sets of images that James used to create the panorama above in your spare time in just a couple of days.

Scientists have totally different goals for their data. As much as we love the images from spacecraft, the cameras are not there to send us pretty pictures. They are there to send us data. Each pixel in an image is a number, a measurement of how many photons of a certain wavelength range of light reached the camera detector. When scientists process images, they are very, very careful about what they do with those numbers. What the amateurs do to blend images together -- to a scientist, that's fudging the data. Not that the scientists dislike what the amateurs do -- in fact I have seen in many operations facilities where science teams have printed out the image products produced by amateurs and posted them on their walls!

But the science team can't use for their research the images that the amateurs are making. Scientists have to mosaic images by a very different process, employing more mathematical transformations and less aesthetic judgment. Sometimes the process fails on the first try, which usually means that the internal geometry of the instrument or the observation wasn't perfectly understood; the scientists have to refine their parameters and try again, and sometimes again and again. And they don't like to release preliminary or intermediate results. They prefer to wait until they are certain they have produced the best result possible from their data. So the science team's releases tend to come at a slower pace than amateurs'.

A couple of other random Phoenix items to mention. There's a neat article at O'Reilly.com about "The Software Behind the Mars Phoenix Lander," an interview with project software systems engineer Peter Gluck. And I updated my robotic arm camera raw images page last night and noticed that they took more photos of Holy Cow and Snow Queen. I didn't see any changes to speak of since sol 31, which was the last time they re-imaged Holy Cow, so I decided it wasn't worth the time to update my animations at this point.