Image Compression as Enumeration

Given the set of all useful images of a reasonable resolution, one could simply enumerate them and use the index as the image encoding.  For example, if there are 16777216^640*480 possible 24-bit 640×480 images, only a small fraction of those would ever conceivably be used as frames in a movie.  In fact, it would take every one on the planet (estimating 6.5 billion people) recording completely different stuff at 24fps, 1.82*10^2219415 years to create that many different frames.

Lets restrict the problem and say that useful images are those that would be used in Hollywood movies.  Assuming this is about 1000 films a year, each of which is around 2 hours at 24fps, this would be about 173 million frames produced per year or 173 billion frames produced if we produced movies for 1000 years.  Over 1000 years, this amounts to about 1.93*10^-2219423 of the possible frames.

To enumerate a single frame out of this set would take about 5 bytes, so a whole movie would take around 864000 bytes (or a little under 1MB).

Of course, it’s not really reasonable to expect one to be able to pick out just the frames that might be used in Hollywood movies.  Still, it might be possible to come up with some other definition of “useful” images, which is still likely to be a very, very small fraction.  Now the only trick is reproducing the image from the enumeration.

It would take a very clever person indeed to actually implement this…

Advertisement