Quote:
Originally Posted by man on the moon
Welcome to the board, btw, if you're planning to stick around.
|
man on the moon, thank you for welcoming me.
One of the reasons why I chose this forum (aside from the obvious connection between badly processed images, and Bad Astronomy) was a desire to face well-reasoned objections, if any could be found. Aside from all the time wasted by basic confusion (which probably would have happened anywhere on the Internet) this has been mostly a good experience so far.
My only real frustration in this discussion is the fact that so many have continued to refer to the selective suppression of shading as a "claim," which seems to mean that they haven't been able to see the effect with their own eyes, even after I attached before-after-slope.gif. I don't understand how anyone could look at that animation and still be unable to see that the 3D shading of the slope has been suppressed, while the 3D shading of the small lump or boulder has been enhanced.
I'm attaching a new image which may be able to help those who have been unable to see the effect, even when they look at before-after-slope.gif. The two frames in this new image are the power spectra (*) of the corresponding frames in before-after.gif. The left frame is from the image that was returned by the spacecraft; the right frame is from the image that was used in the press release.
Can everyone see that the pixels of "snow" have been substantially darkened, in a well-defined band, across the middle of the power spectrum? The fact that pixels in the power spectrum have been darkened means that brightness variations have been _suppressed_. The fact that the darkening in the power spectrum has been mostly confined to a well-defined region means that the suppression was _selective_. The fact that the darkened region is in the middle of the power spectrum means that the _large-scale_ brightness variations are the ones that were selectively suppressed.
Can everyone see it now?
I also want to correct another misconception which seems to be very common here: the idea that the raw image needed some kind of difficult, time-consuming labor-- that it needed to be "carefully calibrated" or "highly processed"-- before it could be understood by normal human viewers. This is completely untrue. Only extremely minimal processing is needed: left/right flip, streak removal, and contrast normalization. My first script only took about ten minutes to write, using 1970s and 1980s technology (Unix shell and NetPBM). I didn't save it, but it was something like this:
Code:
tifftopnm cydonia1.tiff | ppmtopgm | pnmdepth 65535 \
| pnmscale -reduce 2 | pnmflip -lr > flip50.pgm
pnmscale -height 1 -width 512 flip50.pgm | pnminvert \
| pnmscale -height 4800 -width 512 > streaksneg.pgm
ppmmix 0.5 flip50.pgm streaksneg.pgm | ppmtopgm | pgmnorm > test.pgm
Try it yourself, if you have NetPBM; the result is quite decent.
I just ran the above script on a low-end PC from 1997. Total run time was 2 minutes, 11 seconds-- and that's for processing the whole swath, not just the part with the "face." If I can do it in minutes with freeware and a low-end 1997 PC, then surely a professional with high-end 1998 software and hardware also could have done it in minutes, or possibly in seconds.
(*) power spectrum: each pixel depicts the logarithm of the modulus of the corresponding coefficient of the discrete Fourier transform of the image, shifted to place the wavelength=infinity coefficient as close as possible to the center of the power spectrum.