Tags

, , , , , , , , , , , , , , , , , , , , , , ,

I have pretty much completed my analysis of the available Dimension patents. I do not claim to understand them in complete absolute detail, but with my experiences of the Iterated VDK as a point of comparison and familiarity with programming, graphics, and math, I am confident that I have determined their essential characteristics and value. For a more thorough analysis of PIFS, I recommend this earlier paper at http://www.eng.iastate.edu/ee528/specialtopics/jpeg/fractalreview.pdf

I do not want to unnecesarily downplay the achievements of the VDK’s designers or their skill. For its time, it must have seemed a breakthrough. Despite paring away so much of the PIFS algorithm, they still managed to make a block copying scheme competitive with H.261 or H.263.

But the computer industry moves quickly, and the technology was soon eclipsed by H.264. Lacking any way of making similar improvements, Iterated abandoned the VDK and over the next two decades, no one took the technology any further. Other codecs would still be developed, by Iterated and others, but they would all use some form of DCT or wavelet encoding.

For something so promising as fractals, the reality was — and is — bittersweet. It is not at all hard to understand why some people still believe it can work, and still hold out hope. All the more reason, then, that a careful analysis and understanding of the technology is important. History is replete with ideas either too far ahead of their time, or too impractical, or simply too good to be true.

As an example, PIFS could upscale images far better if one padded them with billions or trillions of pixels containing countless other domain blocks. After decoding, the result would be cropped to the bounds of the original picture. But such a solution would require computer performance far beyond anything available today. It would be agonizingly slow to encode and decode a single picture, let alone a video clip or movie.

The inherent difficulties may explain much of the behavior of Dimension and TMM. Both companies exhibit the classic pattern of dubious technology firms: lots of talk but little or no action, demonstrations which are always private or where the technology is never allowed to be tested fully by the public, no proper independant verification, no formally published scientific literature, grandiose unverified claims, odd accounting practices, stock dilution, etc.

Both companies were also staffed or founded by people whose actions were morally dubious at best, if not fraudulent. Again, the technological difficulties probably played a role. Realizing that what they had was obsolete and could not be improved, the owners found it easy to say “We’re still working on it” and to keep taking people’s money.

Since I forsee no way to improve fractal image technology in any reasonable time frame, I cannot recommend investing in Dimension or TMM. Either firm could of course chart a new direction and tackle something different, but with the increasing ascent of HEVC, their hope of being major players in the video industry are slim to none, and any ROI would be modest.

Advertisements