Since there seems to be much comment on my assertation of less than strictly real world results, combined I see with no small measure of imperiousness and invective, I'll point to the document which prompted my comments. You may judge to your own hopefully non-biased liking, as I did.
Go to their homepage and open the "What Is Video Noise?" PDF. Witness the three examples of noise reduction presented in said document. Zoom in closely for best results.
1: Mosquito noise. Take a look at the gray background, behind the cobalt-blue pattern. Specifically, look at the polygonal features. In the bottom half of the image, said features include a border which rises vertically on the right-hand side, then cuts sharply to the left and proceeds to the left edge of the image. It is this horizontal border that drew my attention. It has a few steps: A faint grey line at the top, a more prominent graying further down, a dark brown line beneath that, and finally, another faint grey line well below the rest. Now, this final grey line is the target. Specifically, the glimpse we see of it nearest to the sharp corner on the right, before it passes behind the cobalt-blue pattern. In the heavily artifacted "original image", there is, to my eyes, no information which could indicate that there should have been a faint grey line. The mosquito noise bordering the cobalt-blue pattern has simply erased its existence. And the same could be said about where the grey line proceeds between the vines of the cobalt-blue pattern (although a non-relevant white arrow obscures one area). Only the leftmost spot would seem to bear any indication of this grey line in the artifacted example.
2: Blocking artifacts. Take a look at the bottom, where the building-prominent skyline meets the sky. The blocking artifacting in the "original image" is so heavy that there is essentially no detail. In the "after" image, somehow the blocking has been vanquished to the point that a good measure of mosquito noise, completely missing from the original image, is now resident.
3: Dynamic noise. The restorative results presented here are over the top, at least to someone with a fair measure of Photoshop / Premiere Pro authoring under his belt.
Examples 1 and 2 appear, to me, more the sort of result one gets very specifically with differences in level of compression. The appearance of minute details (including new compression artifacts) in the "after" images, which were completely missing in the "original" images, cannot, to my mind, be otherwise explained. Example 3 seems very much like what one gets when applying a noise filter to an image within an authoring tool, and turning the noise back off. The actual level of noise in the "original image" is so dramatic that it is frankly absurd to suggest that it could be reduced in realtime to the degree showcased in the "after" image while maintaining such crisp detail.
And that is ultimately my point. To me, all three of these examples are meant merely to serve the purpose of illustrating the three kinds of noise being addressed, along with idealized (for the purposes of illustration) examples of reductions of said noise. My hangup is that without anything to indicate otherwise, the results must be assumed to be intended as real. The apparent fact that similar results (minus the magical conjuring of nonextant details) can be had from these devices - in the cases of mosquito noise and macroblocks - is not relevant to these observations.
Did I feel that these observations deserved anything more than a brief mention in support of other reservations? No, not until my hand was forced. Incidentally, I won't be revisiting this thread. I set out to determine one or two things, such as whether or not the Flea HDMI could handle 24 Hz in any capacity (no, apparently) and whether or not the blackening issue could be avoided (no, apparently), and in spite of it all I do still intend to make a few purchases, albiet solely with broadcast sources in mind.
