In the days of "classical" analog video, one could meaningfully count the number of transistors (or vacuum tubes) between the video camera tube and the video display tube, since there would generally be a single path for the signal to take. Even if a video was captured by a CCD or had to undergo something like analog or digital NTSC/PAL conversion, one could still produce a meaningful figure since every portion of the video image would gone through an identifiable path (even though different portions of the video image would go through different paths). A video signal might get converted to video, but all the digital stages it went through were purely combinatorial (sequential circuits would be used to keep track of horizontal and vertical scanning, but the picture data would be handled by combinatorial circuits) and, at any given time, one could point to a circuit and identify that it was handling some particular portion of the original picture.
With compressed digital video, it's impossible to meaningfully count how many transistors the signal goes through, since it goes through a many sequential circuits which munge and reconstitute it. Often times some of these circuits will be running in parallel. If the data from 64 pixels are munged together so as to yield 64 numbers, and then those numbers are processed in parallel, and the results are then reconstituted so as to yield 64 pixels, how many transistors did the pixel data go through? I don't think any particular answer is really "right".