Someone is finally writing about what our department has been saying for years. I don't care what the light power or the contrast ratio is. If the projector can't produce a decent picture, it isn't going to get our recommendation. We've relied heavily on the Infocomm Projector Shootout for years to identify projectors that meet three criteria: minimum color cast, maximum brightness, and most accurate video presentation.

I had not connected grayscale response to that criteria, but that is exactly what we were looking for — we just didn't name it as such. Thank you. Now we can show an article to the “if it's in print, it's real” folks.

Steven Douglas
Media engineer
Central Washington University

When I read Barry Braverman's article on DV filters (“The Curse of DV Revisited,” December 2002), I was intrigued. I appreciated his discussion of filters and plan to implement some of his suggestions. I agree that codecs can react differently to the same filter, but his description of the cause of the DV curse is, well, flawed.

The crushed blacks and high contrast produced by many low-end DV cameras are primarily due to the fact that DV is designed to store video at 0 IRE. All digital video, from DV to Digibeta, uses a setup of 0 IRE. The problem stems mostly from the fact that video, when played back through low-end devices, does not have the 7.5 IRE setup added back in. Cheaper cameras and decks do not add it to analog output. The result is a crushed black analog output. A Digibeta deck can add that setup to its analog out; a Canon XL-1 cannot. The fact that he notices it on playback from cheaper cameras falls squarely in line with the fact that cheaper cameras and decks do not apply setup to analog output.

The “DV curse,” as he calls it, has more to do with the design constraints of DV and cheap equipment. When DV source material is acquired, edited, and output properly, it can look great.

There is no doubt that a good filter set is necessary for acquiring an aesthetically pleasing video image, but selecting gear with the necessary controls over input and output and learning how to use it correctly will do a lot more to solve the DV curse than adding a Pro-Mist or a Black Diffusion. A good waveform monitor and vectorscope do not hurt either.

Robert Racine
Vice president of product support
Mission Service Supply

Barry Braverman responds: These are great points, but setup is another matter entirely, albeit a crucial one. Proper setup in digital video (i.e. the setting of the black pedestal to 7.5 IRE) is an ongoing pain for those of us who must use the North American NTSC standard. Lower-end DV models (and some DVCAM cameras and decks) are apparently designed for the rest of the world, where no setup is used, and so far no one has thought to add the 7.5 IRE setup. Thus, monitoring directly from a typical low-end DV camera with 0 IRE setup would present the crushed blacks you describe.

In my evaluations of DV and DVCAM cameras, the source tapes were output through Sony's DSR-1500, which can add 7.5 IRE setup at output. For DV shooters using prosumer gear like the Sony DSR-PD150, which does not add setup prior to output, pedestal and black levels must be scrutinized, preferably with a waveform monitor. Higher-end gear adjusts for the North American setup discrepancy by adding setup to analog outputs and removing it from analog inputs. As you point out, most consumer or prosumer gear doesn't.

My allusion to the DV curse, however, transcends the format's impregnable blacks and lack of shadow detail, although I still maintain that high compression is the principal culprit here. Excessive error correction and clipping are also serious contributors, because highlight and shadow details discarded by a camera's compressor are not recorded to tape and cannot be retrieved in post, irrespective of setup issues.

A diffusion filter may improve a compressor's performance by increasing the redundancy in an image, especially at tough-to-compress boundaries between light and dark pixels. Applying dither in these areas contributes to compressor efficiency and the overall aesthetic, as ugly artifacts are suppressed. I believe the improvement in DV images afforded by a filter is proof in and of itself of the merits of using such a simple strategy.