Anyone visiting a multiplex lately will have noticed Kodak's recent campaign – 'See movies the way they're meant to be seen'. A none-too-subtle product puff, yet baffling for an audience not exactly used to bulk-buying film stock. It begs the question - what is Kodak so worried about?

In the April 2001 issue of their promotional magazine, In Camera, Kodak interviewed a DP who had run comparative film/HD24P tests. Guess the conclusion.

Yet despite this, Kodak, owner of Cinesite in the US and Europe, is about to launch an all-digital lab in the UK. So it seems that while Kodak shouts about the integrity of film as the only viable acquisition medium, with a claimed 12 million pixels in every frame, what they're whispering is the level of investment they're making in digital post facilities, though neglecting to mention that with a digital intermediate, you'll only end up with the same 2K resolution as you would get from shooting digitally. The message seems to be don't shoot on tape, stick with film and throw lots of money at the post.

It's a sign of the times when that most esteemed of film journals, American Cinematographer, now carries articles and ads extolling the virtues of High Definition. Any meaningful debate about film vs. HD24P, however, should acknowledge this simple truth: both are equally valid acquisition systems, its what you do with each that counts. HD24P is not a replacement for film, but an alternative.

HD or not HD?

Fundamentally, film is old tech and HD24P is new. But what exactly is HD24P? High definition technology is now over ten years old, in the form of HDTV, a high resolution version of NTSC, but to understand how HD24P for cinema has developed it's useful to know about film. It comes down to frame rate. 24 frames per second is the rate for professional motion picture film, the rate at which individual frames pass through a camera or projector. This standard developed in the 1920s as a consequence of the ‘talkies’ (there was no established speed for silent films). 24fps was deemed to be the lowest frame rate that could successfully reproduce sound on the film's optical audio track. The fact this frame rate made motion reproduction more fluid just happened to help.

Thus, 24fps is perceived of as 'adequate' for mainstream motion pictures and, whatever its flaws, attempts to change the standard have met with resistance; in 1988, a SMPTE report recommending a shift to 30fps (to suit US broadcasters) was overlooked by the industry, not least because of the cost of the additional film stock required. In order to insinuate itself into cinema video has had to accommodate itself to a very fixed idea of the necessary frame rate.

In Europe, where PAL video runs at 25fps, this slight discrepancy in frame rate can easily be overcome. In the US, however, where NTSC video runs at 30fps, attempting to transfer video to film, requiring the loss of 6 frames in 30, is a real headache. Very noticeable motion artefacts can arise. Hence the development of a video camera capable of shooting 24fps in anticipation of film transfers.

More than this, one of the biggest criticisms of tape has been the 'video look’, whereas film offers a more subtle, less 'immediate' image. This too is a consequence of frame rate - in combination with shutter speed. Like the film rate, video frame rates have established themselves in a rather haphazard fashion. They are a consequence of the prevailing electricity supply frequencies; 50Hz in Europe, 60Hz in the US, unchanged for over sixty years. These are the effective shutter speeds of PAL and NTSC video cameras. A single field of interlaced video is captured in either 1/50 or 1/60 of a second. Two fields combine to form one frame, giving a full image rate of either 25fps or 30fps. Effectively though a video frame is exposed for less than half the time of a film frame, giving the flat, sharp, ‘real’ look of video. In film, where the shutter is open for twice as long, very slight motion occurs during exposure leading to the softer, more atmospheric ‘film look’. This is what is known as ‘motion blur’.

Finally, the problems of interlacing had to be overcome for high definition video to become competitive with film. Interlacing, where two fields, each of half the number of picture scan lines, are combined to produce one video frame, developed in the earliest days of television to allow for the very slow phosphors in original TV sets. These could not be made to respond to 1/50 or 1/60 second refresh rates, so alternate ones were excited every 1/25 or 1/30 second. The consequence is that the two fields of a frame are separated in time by the base frequency of the video system. During film transfer this discrepancy introduces motion artefacts and reduces resolution. The solution is ‘progressive scan’, where each frame of video, whether at 24, 25, or 30fps, is captured in a single moment, just like exposing film.

The taste test

Beyond these fundamental matters, much of the rest of the argument is down to matters of taste - and price. It is about how the camera is set, how the shots are lit and framed, and who does your print. The film purists will continue to argue that colours are false, artefacts are still visible, and traces of ‘video look’ are still present following transfer. Those trying to get their movie on screen will stress ease of use, instant results, and cost benefits.

And these can be considerable, even when considering only direct savings from eliminating film. Stock costs are way down, processing, rushes and telecine are eliminated, and tape integrates directly with computer-based grading, FX and CGI.

But there's no pleasing some people. The purists again will argue that shooting on tape erodes the process and traditional etiquette of a production, encouraging a lack of discipline which leads to over-shooting, as if the tools are to blame and not the people using them. Which is no argument. Fact is, the tools are here and the HD market is evolving fast, which presumes a demand for the technology. We will find out soon if HD is a serious contender in cinema acquisition.

The Sony/CineAlta HDW-F900 is the first commercially available high definition camcorder (HDCAM) delivering the Common Image Format (CIF) standard of 1920x1080 active pixels. Principally designed to record at 24 progressive scan frames per second, it will also deliver 25p, 30p and 50i or 60i (interlaced) to integrate with existing post-production and broadcast standards.

The HDCAM format records to 1/2 inch tape in DigiBeta sized cassettes (40 mins. at 30fps, 48mins. at 25p, 50mins. at 24p). Dealing with the greatly increased data rate from the 2.2 million pixels of the 3x2/3 inch CCDs has required the adoption of a relatively high compression rate of 4.4:1, leading to a still considerable 185Mb/s on-tape rate at 30fps. Concerns have been raised about the effect this degree of compression will have upon the image integrity, particularly after several generations - or cycles of compression/decompression. These may be mathematically justified, but what will ultimately prove the worth of the format will be the audience’s perception of the final result, printed to film - an ‘organic’ medium with an inherent capacity to smooth and disperse artefacts. We have certainly not seen anything to worry us.

While all HD cameras offer considerably higher image quality than DigiBeta or its competitors, all HD cameras are not the same. Currently on the market is Panasonic's AJ-HDC20A, described by the manufacturer as appropriate for newsgathering and documentary and, in terms of weight and size, comparable to their DVCPro range. Like the Sony camera, this too is based around 3x2/3 inch, 2.2 million CCDs delivering 1080 horizontal scan lines. It will, however, only record interlaced video at 30fps, onto 1/4 inch tape (DVPro size).

More relevant to cinema acquisition and making impressive claims is Panasonic's AJ-HDC27V ‘HD Cinema’ camera (DVCProHD). On closer examination of Panasonic's literature, however, the story is a lot less clear. The manufacturer claims multiple frame rates, analogous to under- or over-cranking of a film camera, are now possible, offering (up to now elusive) slow-motion effects. But is it really true? Are Panasonic not confusing frame rate with shutter speed?

The effect Panasonic describes as a 'streaking blur’ sounds similar to those obtained on all digital video cameras when shooting at shutter speeds of less than 1/25 second. Holding the shutter open as multiple frames are exposed is not the same as a variable frame rate, where different numbers of frames can be exposed each second. Panasonic also describe ‘pad frames’, suggesting that, whatever the camera is doing, somewhere it is interpolating information, not recording what is really in front of it.

Confused? So are we. Panasonic also assert that their HD VTR system operates at a constant 60fps. Again, if this is so, interpolation will be required, reducing the true information in the signal. Lastly, whilst the camera will record 1080 lines interlaced, only 720 lines progressive seems to be possible. This is no better than standard definition resolution. Hardly a solution for cinema. A possible explanation of these apparent limitations is that these camera are developments within the DVCPro structure, as indicated by the continuing use of 1/4 inch tape. If Sony have been pushed to increase compression to get HD information onto 1/2 inch tape, how much more must be lost in using 1/4 inch?

It’s always the same story - do the tests yourself. For more information go to sources-hardware.

Test department

After hearing third-hand accounts of HD24P warning of colour problems and motion artefacts, Elemental decided to shoot some HD tests to see for ourselves. Having already shot a DV feature and, given our awareness of the creative freedoms and cost benefits offered by shooting on tape, we were keen to extend the experience with a view to shooting our forthcoming feature, Solid Air, on high definition.

In February 2001, with the support of NESTA, we set up a small shoot on location in Glasgow. Sony UK supported these tests, supplying the HDW-F900 camera through Optex, the London hire company, who in addition supplied a full range of lenses and accessories. It's worth knowing that the range of lenses available in the UK is still limited - by 2002 the situation will have improved.

Director of photography, Eigil Bryld, who recently won a BAFTA, was chosen for his considerable film experience and his willingness to experiment. Brian Rose of Optex supervised and tested the equipment before shooting, explaining the camera menus and the lens requirements. As it is a video camera lenses mount and function differently from on a film camera. The focal plane is closer to the lens mount so conventional film lenses cannot be used. The temptation to use video lenses for reasons of cost should be resisted. They are unlikely to have the optical qualities of film lenses, nor do they operate in the same way. Increasing numbers of adapted film lenses or, better, new HD lenses are becoming available. Use them. Also, as the image area (the CCD) is smaller than a 35mm film frame, the effective focal length is increased by approximately 2.3; a 50mm lens functions as a 115mm lens.

The tests were shot over three days, covering interior day/night and exterior day. Rather than shoot as a purely technical exercise, we settled on a loose narrative that offered scope for 'atmosphere' and featuring performance, a relief from the usual material we've seen at demonstration screenings. While we decided to record sync sound on location, the shots were designed ultimately to be set to music.

We opted to shoot 25fps (for compatibility with a standard definition off-line) in progressive scan ‘shutter on’ mode with a true 16:9 aspect ratio. ‘Shutter on’ is the closest analogue of a film camera, introducing a moderate amount of motion blur. Although Eigil became accustomed to the monochrome viewfinder, it is very limited compared to a film camera viewfinder and an additional camera mounted 6 inch LCD monitor was invaluable. We were supplied with a 28 inch widescreen HD monitor as a video assist but it proved unwieldy. It was necessary for line-up purposes, though, to ensure the camera was correctly set. It also revealed a frighteningly high degree of detail. A 14 inch widescreen monitor is probably sufficient for assist.

Of course, lighting and shooting is a subjective matter agreed between the director and DoP. There are those who would recommend shooting HD using onboard settings or external filters to manipulate the look. We chose not to. We wanted to keep the footage as ‘clean’ as possible to allow for the greatest choice in post. Though we did light in a conventional film style, we kept it simple, exploiting daylight whenever possible and combining it with small fills. Kinoflos worked well with the camera. The significant difference to working on film is the opportunity to use lower output lamps, if not fewer of them. Similarly, we were able to use pracs more often for the night scenes than we would have for film. Over two days we shot over 3 hours of material.

Having shot HD we recommend:

• As a general rule, expose for highlights. Like conventional video, HD behaves inversely to film, clipping (overexposing) very quickly but holding more detail in the shadows. If the image happens to be slightly underexposed it can be recovered.
• With lighting, less is more. A great advantage of the format is that you can shoot in low-light situations. That's not to say you don't need to light as carefully as for film. You just don't need to stand that 12K all the time. Don’t push it too far, though; don’t turn up the gain.
• Unless you're sold on a particular look, avoid using filters or the menus. Be aware of what can be achieved in post, and grade it then. Filtering can't be undone at a later stage, so it must be used judiciously. Be very cautious with diffusing filters. They may be needed, but use smaller values than for film. High values can corrupt the image.
• Just because it's cheap, tape's not an excuse to overshoot. Your editor won't thank you for it. Shoot exactly as you would with film, but be aware that longer takes are possible, which makes it useful for certain styles of documentary work or long takes in narrative work.

Four DAT quality audio tracks can be recorded to HDCAM tape. Two inputs are available directly on the camera. Four are available with an adapter. In most respects it is the same as for any other pro video camera. Where we had problems was in attempting to use record-run timecode. We had to settle for free-run, which is problematic in post. We are not alone in this. We have yet to pin down the cause - it may just have been inexperience with the camera. It was a surprise, though, as we had chosen to shoot 25fps - we had expected 24fps to be problematic. Again, test it yourself.

The material shot has since been transferred to DVCam for offline digitising. The next stage will be the online and grading at HOME, a dedicated HD facility in London, then film transfer at Hokus Bogus, Copenhagen. So far we have seen nothing to suggest the result won’t be indistinguishable from material originated on 35mm negative.

In March, following our test shoot, we were invited to Panavision's UK headquarters to see their developments in HD Digital Cinematography. Panavision, a company renowned for their precision motion picture equipment, have rebuilt Sony's camera body to their own design and developed their own lenses, the Primo Digital range. At a test screening, we were impressed with the results, particularly with footage printed to an approximate 'Cinemascope' ratio of 2.2:1 using a process known as DigitalScope. This is done by cropping the 1080 horizontal lines to around 900 and electronically introducing an anamorphic squash to suit the 1.85:1 35mm film frame ratio.

Famous last words

As one of the few UK production companies investigating new technologies, we're aware of the perception of digital/HD movies as low budget and therefore inferior to film. We believe this is false; the success of any film will always come down to the individual talent and skills of those involved. Of course, the means of acquisition remains a matter of choice for any producer, but many believe that shooting on video is no more than a way of saving on stock and process costs. Naturally these represent only a small part of the overall spend on a feature-length project.

For tape shooting to be truly effective, the entire process has to be reconsidered. For example, in our test shoot, we had no operator, focus puller or clapper loader. In a 'real' shoot using HD, we would expect to use one or two camera assistants, as opposed to the three or four normally employed on a 35mm film shoot. There are other areas in which intelligent restructuring can lead to truly significant savings. While we acknowledge that every project is different, there are consistent potential benefits in HD. It is not a case of undermanning, it is practical and creative decisions arrived at through successful integration of digital technologies. Beyond this, the flexibility and convenience of video are of enormous benefit in sustaining an intimacy supportive of the artists’ performances.

With the average UK feature budget pegged at around 3-4m, any efficiency is worthwhile. As posited in the film counsel section, British films are almost entirely niche product in the international marketplace; only the very few exceptions ever recoup beyond expectations. Which brings us back to an unpalatable truth, but one which bears repetition: that in order to create a viable and self-sustaining UK film industry we must make movies cheaper. HD24P, combined with intelligent and cost-effective production and post production methods offers that opportunity. Film-makers ought to welcome it.