Transfering Super 16 to HD
As the broadcasting landscape gears up for the full roll-out of high definition television, the place for Super 16 film has come into question. Discovery and CBS are among the channels that have decided to avoid Super 16 as a production format for HD programs. These broadcasters – along with a number of post-production engineers – argue that when converted to HD, Super 16 doesn’t stand up to the quality expected of HDTV.
Jay Schneider, VP production, operations and engineering services for Discovery Communications, says that Discovery will not target new projects shot on Super 16 because the channel feels viewers who dish out money for HDTV services will not get the quality they pay for when Super 16 is converted to high def. ‘Our programs rely on superior imagery and we feel Super 16 is not the best way to exploit the HD medium.’
Film records images via photo-sensitive grains dispersed over the surface area, which act as the recording element. The problem is, he says, Super 16 has approximately 25% of the surface area of 35mm, and consequently has fewer grains and about 25% of the resolution.
‘When you enlarge Super 16 to the same size as 35mm for transfer to HD, the film grains enlarge proportionately and are dispersed at random,’ explains Schneider. ‘This means the film grains will differ in position from frame to frame. Enlarged size, coupled with random positioning and imperfections in the film, create video noise [elements not in the original signal].’
Vince Forcier, director of engineering and HD services at Virginia-based post facility Roland House, agrees with Schneider that converting from Super 16 to HD is often a gamble when it comes the quality of the final product. ‘I have clients who were converting from Super 16 to HD and were not very happy,’ he says. ‘You can count on some shots looking good but others just don’t. I’ve found clients end up re-editing and choosing shots based on what looks good, not on what editorially makes sense.’
Forcier explains that 35mm has almost three times the resolution of Super 16 – the former has roughly 4,500 lines of resolution, while the latter has around 1,500.
‘You start with a smaller negative in Super 16 and when you convert to HD you transfer to a higher resolution, so anything wrong with the film and any grain structure weaknesses become more apparent . . . whereas standard definition TV has lower resolution than even 16mm, so you could always get a decent picture.’
Schneider says that in the edit suite, the video noise created from the Super 16 to HD conversion may not be very noticeable – it’s in the transmission to the viewers that the loss of image quality becomes evident, and the dispersal of film grains into different positions from frame to frame becomes a problem.
‘The transmission of HDTV to consumers is not at the full 1.5 gigabits per second data rate produced in the production environment,’ Schneider explains. ‘It’s one one-hundredth of that, or 14 to 19 megabits per second. So what we do is put the production video – at 1.5 gigabits per second – through compression to reduce the bandwidth to one one-hundredth of the original bit rate through a process involving interpolation. Interpolation is predicated on predictable images – and noise is the enemy of compression.’
The compression processor diverts power away from reproducing the resolution in an effort to reproduce the noise. As a result, in the post-compression transmission to viewers, the lower resolution becomes apparent, says Schneider. ‘Looking at the video in the edit room you won’t see this. It’s when the video is compressed for transmission that it’s noticed.’
However, advocates of Super 16 say that putting noise reduction equipment between the output of the film telecine and input to the HD recorder reduces the burden on the compression equipment.
Schneider counters, ‘This doesn’t make up for lost resolution, it actually lowers the resolution of the image. Noise reduction equipment is detrimental to resolution, and in an effort to reduce noise it may soften the image.’
Advocates of Super 16 also say new finer grain film creates images sufficient for HD conversion. ‘There are times when that is the case,’ Schneider concedes. ‘It works well in controlled, well-lit environments. The problem is when you shoot in a non-studio environment and there’s less light or imperfect light and the film is pushed and left longer in chemical baths, more film grain noise is created. Once the image is transferred to HD video, and subsequently compressed for transmission to viewers, undesirable artifacts are produced.’
A study undertaken by CBS – and chronicled in the Society of Motion Picture Television Engineers’ SMPTE Journal by Henry Mahler, associate director of technology at CBS – points to similar results. In a carefully controlled environment, tests compared the performance of 16mm, Super 16 and 35mm after being converted to HD video and using top quality transfer equipment.
In the report, Mahler relates that because of the basic size difference between 35mm and the 16mm formats, the additional magnification required for HDTV amplified any imperfections present on the film. At the increased resolution capability and greater viewing angle of HDTV, these imperfections become even more apparent than with the current standard resolution television.
Expert observers looking at the HDTV tape transfers found that the Super 16 and 16mm pictures were noticeably softer than the 35mm and HDTV sequences, which appeared very much similar. Observers also found that the weave, judder and grain noise were more evident in the 16mm formats, and they concluded that the film motion artifacts may be the result of fewer sprocket holes in 16mm film (there is a 2:1 ratio between the number of sprocket holes per picture in 35mm and 16mm).
In terms of costs, Forcier says the price of converting from Super 16 or 35mm to HD is essentially the same. However, the cost of a 35mm shoot is usually double the price of a 16mm film, and the large crew and cumbersome equipment make it far from ideal for doc projects. On the other hand, HD cameras are small enough to shoot anywhere, they hold up under severe weather conditions, and there are no transfer issues or costs, says Forcier.
Shooting in HD can be less expensive than Super 16, he adds, particularly when it comes to docs where a lot of film is used. ‘To go from shooting to loading the AVID for a 20-minute show costs about US$300 for an HD project, whereas in Super 16 this would cost about $1,500,’ says Forcier. ‘And the latest in HD cameras, new post-production tools, and proper lighting can bring an HD project a lot closer to the filmic look, which many filmmakers feel they lose when turning their back on film.’
Pal versus NTSC: Which format should rule?
Producers have long accepted the fact that selling their wares between Europe and North America has meant factoring in time and money for transmission conversions between PAL and NTSC.
The NTSC (National Television Systems Committee) TV transmission standard is used in the U.S., Canada, Japan and Mexico, and is characterized by 525 lines in each picture, and 30 frames and 60 fields per second. PAL (phase alternate line) is utilized in Europe and Australia, and has an extra 100 lines of resolution – 625 lines, five fewer frames (25 frames) and 50 fields per second.
NTSC was invented first, in North America, and came into widespread use prior to the European development of PAL. As a result, NTSC is compatible with the North American 60 cycle volts, while PAL runs smoothly in a 50 watt electrical system.
The consensus among European and North American producers is that PAL – with more lines of resolution – is the better of the two formats. As a general rule, producers find that conversions from PAL to NTSC result in higher quality images than from NTSC to PAL, in which some image degradation often occurs.
When converting from NTSC to PAL, you are converting from a lower to a higher resolution which requires the image to be stretched. ‘This causes the image to become softer and not as clear. It loses resolution,’ explains Lynn van Rooyen, post-production supervisor at CineNova Films in Toronto. When converting programs to PAL for European distribution, she has found that fine, detailed images, such as maps, lose some of their quality and have artifacts.
‘PAL is absolutely, definitely better,’ says George Panayiotoo of Films at 59 in Bristol, England. For European producers, he says converting to NTSC causes little quality loss because the conversion process begins with a higher resolution image.
However, Jeff Dewolde at International Image in Santa Monica argues conversion technology has improved tremendously over the years so conversions from NTSC to PAL can be quite good.
Says Dewolde, ‘Standard conversions today do not have to add artifacts if done properly and the facility is using top technology. The most important factor is the quality of the master. It’s all how you shoot, and the quality of the post-production.’
For North American producers, it is more costly to post in PAL, and the equipment is difficult to find. For example CineNova is currently producing several programs in PAL and has had to convert an AVID and a VTR machine to the PAL format. As well, van Rooyen says you can’t do a linear on-line in PAL in North America, only a non-linear on-line. Final sound post-production must also be formatted for PAL. In general a US$17,000 post budget can increase by $10,000 per hour when working in PAL, she says.
CineNova has to post in PAL for Lost Cities (w/t), for example, because the doc is based around underwater footage shot by a French crew in Egypt who only had access to a beta PAL camera. Converting the images to NTSC – to allow for the inclusion of footage shot in North America – would not have caused any quality loss, but when re-converting to PAL for European distribution, the double converted PAL images would be degraded, she says.
There is consensus on both sides of the ocean that having either North America or Europe switch transmission standards would be daunting both technically and financially – especially at the current time when the challenges of rolling out high definition television are yet to be met. So, a place for both NTSC and PAL is assured in their respective territories – at least for now.