Friday, August 26, 2011

Some projects don't work out...

When I do work with limited and/or poor quality datasets, the failure rate is high.   Such was a recent attempt to make a perspective view from images of the surface of Venus taken in 1975 by Venera-9, the first spacecraft to take pictures from another planet.   Below is my enhanced version of the panorama it took.
Given that the spacecraft was only expected to last around an hour on the surface and its  slow transmission rate (by modern earthly standards), the view was only 128 pixels high and about 512 pixels wide.  In order to sample both the horizon and nearby areas with good enough resolution to distinguish what was being seen, the camera scanned a 180 degree panorama that started out looking at the horizon, dipped down in the center to the foot of the lander, and then went back up to the horizon on the other side.   My goal, following work that Don Mitchell has done with the much better datasets from Veneras 13 and 14, was to make a picture that would show Venera 9's surroundings from a more normal perspective.  I had reprojected the corner images before, as seen below.
These views, while from a somewhat more human perspective, are still awkward to look at.  My goal was to sample features at different distances in order to create something resembling an ordinary photograph.  The problem was that unlike Veneras 13 and 14, which took similar scans on both sides of the lander, Venera 9 only took one, and at significantly lower resolution.  In fact, the compilation is so obvious that some blurring was necessary to disguise the fact that the scan lines and pixels were going in different directions in different places.   Also, while I intended the image to be 480x640 pixels, the largest presentation it could sustain and still look half way presentable was 275x360.  Still, at about 99,000 pixels, it is still oversampled (the original data comes to about 65,500 pixels).  In other words, the best that could be managed was an image slightly less than 0.1 megapixels.   It also took some inventiveness.  One side of the panorama shows a ridge in the distance,  the other side does not.  The transition had to be guessed at.  Also, filtering was done to reduce the effects of differing illumination angles, improving the believability factor but at the expense of yet more resolution.

This is the finished product.  It turns out that there was not enough data in the Venera-9 set to make this work.
Finally, for fun I made a colorized version based on the Venera 13 and 14 color images.
Data courtesy the Russian Academy of Sciences.   Processed images Copyright Ted Stryk.

1 comment:

Bill said...

'Failure' or not, those are fascinating images. Thanks for your work on them.