swo17 wrote:How can preserving something inherent in the film go against "what the film really looks like"?
Well, it depends on what the film "really looks like", doesn't it?
I can't comment on
The Third Man as I haven't seen it (and wouldn't be able to play it anyway as it's outside my region), but I've spent a fascinating and hugely educational day in a Soho telecine suite observing the transfer of Jane Arden's
The Other Side of the Underneath from the original 16mm A-B camera neg to HD, during which I had several conversations with people much more knowledgeable than me about the issue of grain, high-def transfers and DNR.
For the record, they'd read this thread (and others in a similar vein), and are fully aware of the current online debates - though equally aware that rather too many of the participants are classic examples of the age-old principle of a little knowledge being a dangerous thing. I freely acknowledge that I'm as guilty of this as anyone - one thing that today really rammed home was that theoretical insights are no match for not just physically handling and processing this material on a regular basis, but having to do it on a tight schedule and budget.
For instance, how do you tell what the film "really looks like"? Surely if you're going from the original camera neg, you can't get any closer to absolute perfection? Well, yes and no. The great advantage is that you're working directly from what actually passed through the camera, and if the telecine operator is doing his job properly and the resolution is set high enough (in this case, standard 1920x1080 HD is sufficient for a 16mm original) it should be possible to register
everything that was ever recorded on the original film.
But, on the other hand, if you're working directly from the original, the picture may well be excessively sharp and grainy. This sounds counter-intuitive, but if you watch a film projected in a cinema, the chances are you're watching something that's three generations removed from the original - a release print struck from an internegative struck from an interpositive struck from the original camera negative. Inevitably, the duplication process results in a certain softening at every stage, with the result that an HD transfer directly from the camera neg may well be significantly sharper and grainier than the director/cinematographer intended.
As a result, while DNR has clearly been grossly misused as a means of eliminating grain (i.e.
Patton, Dark City), it is nonetheless arguably necessary to a certain extent, if only to get the picture to resemble what was originally projected, as opposed to what was originally shot. This is the kind of issue where it's very handy to have someone involved with the original production on board, which happily is the case here (director Jane Arden and co-cinematographer Aubrey Dewar are long dead, but producer, co-cinematographer and long-term creative associate Jack Bond is sitting in on every session) - but obviously that's not always going to be possible.
A further complication is that different film stocks and styles of shooting present different technical and aesthetic challenges, so these issues have to be resolved on a case-by-case basis. As the widely divergent transfers of
Red Desert demonstrate to perfection, there's no "correct" answer in the absence of the director/cinematographer (and even then they can forget or change their minds over time), though the general consensus is that the BFI one currently seems closest to the ideal - and the Blu-ray has a lovely velvety grain that feels much more "filmlike" than the lower-resolution DVD.
I can't comment on "before" and "after" issues yet because I've only seen "before": the digital grading (where the unvarnished HD master is processed shot by shot to ensure visual consistency that may not be present in the original camera neg) isn't till next week. A further challenge comes at the final encoding stage, to make sure the grain isn't swamped by unwanted digital artefacting, though awareness of these issues in advance should ensure they're minimised in practice.
But the crucial points that I came away with today are:
1) DNR is not the tool of Satan, provided it's used intelligently by someone who knows what they're doing and who respects the integrity of the original film. In fact, the chances are that anyone who bridles at the phrase 'DNR' as a point of principle (and I've seen several examples) doesn't really know what he's talking about.
2) The camera neg is not necessarily the best source for a perfect transfer, because although it's closest to "the original", it's also going to be ungraded and the print will be riddled with splices, potentially causing frame judder during the telecine. So digital post-production is always necessary to some extent, requiring informed and potentially contentious but nonetheless unavoidable aesthetic decision-making.
3) However convincing the end result, you can't ultimately get away from the fact that film and digital media are fundamentally different, and even the most conscientious transfer will only be a simulation of a projected image. The trick is to get it as close as possible to an accurate impression - but it may only be possible to achieve this by pulling various digital tricks that some may instinctively object to.
4) The higher the resolution, the less reliable frame grabs will be as a guide to what the final encode actually looks like in motion. How the final digital encode handles grain can often be critical to the success of a Blu-ray project - but you can't really appreciate this without seeing the moving image. (Or, if that's not possible, actually reading the accompanying technical comments instead of just accepting the frame grabs as gospel!)