THE CASE OF THE KILLER GRAIN
One last story. There’s an infamous laserdisc out there – you might guess which one it is, but you won’t find out from me – which is known for being a particularly ugly transfer of a major science-fiction blockbuster.
Why do good-looking films wind up looking so bad sometimes? In this case, the director was too busy to supervise the film transfer, so he sent over his official representative, Mr. Y, to do the day-to-day work, then the director would come by once a week to view what had been done and then ask for changes where he felt they were necessary. So far so good.
Unfortunately, the facility chosen for this film transfer wasn’t exactly one of the top transfer houses in LA. (In fact, it’s since gone bankrupt.) Their Rank Flying-Spot Scanners were ten years old at the time of the transfer, and although well-maintained, there’s only so much that old equipment can do with difficult films. This particular Science Fiction epic had numerous opticals, explosions, and effects, which made it particularly difficult to transfer, with widely-varying exposures and density levels.
To make things worse, the telecine operator assigned to this job was a bit inexperienced. That, combined with the poor judgement of Mr. Y, made many of the film’s dark scenes come out somewhat grainy and “pushed”-looking. This is a typical problem when the gamma (mid-range) settings for the Rank are misadjusted, which tends to exaggerate grain while it brings out more detail.
When the director came in to view the finished result, he decided to bring in his own personal TV set to view the transfer. All the engineers and telecine operators at the facility were aghast, and tried mightily to explain to the director that there’s no way a $1,000 TV set can reproduce the subtleties of a $10,000 broadcast monitor, but the director wouldn’t hear of it. “Nonsense,” he retorted. “I’ve watched hundreds of films on this set. This is my own personal standard, and I just want to use it as an additional reference.” Numerous changes were made, just comparing the image on the expensive lab-grade monitor and the cheap consumer set, sometimes averaging a compromised setting between the two. This necessitated even more time and expense, since sometimes, the frustrated telecine colorist could make the image look good on one of the monitors, but not both at the same time.
The director was also unhappy with the grain in the problem scenes. As luck would have it, the transfer was recorded on the component digital D-1 video format*, which is now the standard for the telecine industry. Mr. Y suggested that they remove the grain by dubbing the master tape through a noise-reduction device. These noise-reducers (also known as grain reducers) essentially use a computer to make intelligent decisions on a pixel-by-pixel basis, analyzing which part of the picture is noise and which is actual detail, and then subtracting the noise pixels. Unfortunately, when overused, the noise-reducers tend to add a degree of “lag” and “smearing” to the image, as the overtaxed circuits can’t make their decisions fast enough. This adds artifacts and flaws, and also tends to make the picture soft.
When he looked at the new tape, the director felt that the image was better, but now it lacked the crispness of the original. Now, the second-generation D-1 tape was fed through an image enhancer, which sharpens images by delaying one line of information and adding a subtle black outline to sharp edges. The director viewed this tape, and he pronounced it better still, but now, too much of the grain was back!
Rather than do the transfer over again, Mr. Y made the decision to again feed the enhanced D-1 dub through the noise-reduction box, only this time, it would be done at a much lower setting. At last, the director viewed the fourth-generation D-1 tape, and approved the noise-reduced, enhanced, and noise-reduced image. But that doesn’t mean the picture looked good.
Any audiophile knows that the ideal amplifier is a straight wire with gain. The more processing you add to the circuit, the more distortion and noise gets added to the sound. The same is true of video. All this extra processing to this film image created a very strange, “mushy” kind of image. The grain patterns tended to “float” around the screen in odd, unnatural patterns – one of the artifacts of certain kinds of noise reducers – and the extra enhancement added a harsh edginess that exaggerated details and made them very unpleasant. In short, what you basically had was a terribly over-processed picture. If the original transfer had been made on decent equipment to start with, most likely, none of this would have been necessary.
A few weeks after the laserdisc came out, laserdisc buffs started complaining about how strange the transfer looked. This particular disc began attracting a reputation as one of the ugliest on the market. Someone managed to get ahold of Mr. Y, who concocted a story that the grain was an intentional part of the film, giving it a deliberate “texture” to those shots, and was a result of the high-speed stock used for the production. But even Mr. Y was at a loss to explain why sometimes, back-to-back shots had different degrees of grain and enhancement.
A few months later, after the studio had been deluged with complaints, they decided to try some tests to see if the film could look better with different equipment. A friend of mine was assigned the task of retransferring the first 10 minutes, just to compare it with the original version.
Mr. Y came in and was understandably perplexed and chagrined to find that the new transfer was sharp, crisp, full of detail, and yet had hardly a spec of grain. He was even more embarrassed to discover that the inexperienced operator who had done the first transfer had misframed the Rank and cut off quite a bit of the image on one side of the frame. The new transfer revealed at least 10 percent more picture, showing more detail and more of the sets and characters.
“I don’t care that it’s better,” he snapped. “We’re not going to go back to square one and re-do this picture from scratch. Besides, the director is much too busy to concern himself with this. The old transfer stays as-is.” And with that, he stormed out of the facility.
A few years later, a new transfer was quietly prepared and reissued to much fanfare. It was light-years better than the old one. True, the director again brought in his trusty TV set as his own “personal reference” and – despite the fact that the facility managed to drop it and wound up buying a new one – the new transfer was beautiful. Mr. Y subsequently left the director’s production company and went on to some success in a different part of the industry.
I don’t know if there’s a moral here, except that bad equipment and inexperienced people will invariably result in bad-looking laserdiscs. In fact, I’d argue that a great telecine operator can probably make better pictures on a mediocre Rank than a horrible telecine operator could on a great Rank. But the key is that excessive video processing isn’t the right way to fix a bad transfer.
* The professional D-1 digital component format stores the luminance and color-difference signals separately, providing unusually high detail and very low noise compared to any other format. The D-1 tape signal is virtually identical to that coming out of the Rank itself, and subsequent dubs and tape generations should theoretically be identical to the original.
No comments:
Post a Comment