I wasn't aiming to register all of all three images, just the locale around the big crater and the putative shadow, so the trapezoidal distortions weren't important to me. Sounds like we both used exactly the same reference points to rotate and resize the images. (I remember thinking while I was working that Mars orbital images are just chock full of useful reference points--craters.) I did the gross rotation and resizing, and then fine-tuned the registration by making the upper image temporarily semi-transparent so that I could align it by eye by registering just the big craters. Close enough for government work...

Oh, and you said "...but it was shifted enough from the other two that the Sun was probably at a much different angle." - Ithink you'll find that the Sun was at the same angle to Mars, just that the spacecraft was at a different position/angle. The Sun don't move much in the time it takes a parachute to drop 500m in a 6 mbar atmosphere! :-)

Doh! Yeah, yeah...really poor wording. I was trying to express that Op was moving horizontally--apparently a lot between the first and second frames, not much between second and third; and so (with the Sun angle constant!) its shadow in the first frame was probably just off the right edge of that frame (and assuming that the descent imager was pointing straight down). Which doesn't sound much like "Sun was probably at a much different angle", I know, but trust me, that's what I was thinking!