|
| urbanfox.tv > technology articles > format articles > The DV debate |
|
ON THIS PAGE:
|
The DV DebateThe DV family has many offspring (miniDV, DVCAM, DVCPRO, DVCPRO-50, D9 or DVCPRO-HD), but which should you pick and how can it fit in with other formats? by David Fox The DV digital video format is the first to encompass all types of video production, from consumer through to high definition. It has moved beyond tape to DV-native nonlinear editing systems and servers, and to hard disk drive cameras. It is also being used for DVD, with consumer DVD camcorders introduced during 2001, and the likely launch of professional DVD camcorders within the next couple of years. Obviously, then, DV can be the answer to all your video needs? Not exactly. Although they all use the same DV compression system, all DV systems are not the same. Beyond the obvious differences in Megabits per second (25, 50 and 100), there are incompatibilities which can make choosing and using a DV system more challenging than it might first seem. |
|
DV SAMPLE RATES MiniDV and DVCAM sample at 4:2:0 in 625-line versions (4:1:1 in 525-line), while DVCPRO samples at 4:1:1. DVCPRO-50, D9 (DV 50 Mbps) and DVCPRO-HD (100 Mbps) all sample at 4:2:2. Mixing 4:1:1 and 4:2:0, gives 4:1:0. "You get the worst of both worlds," says Olivier Bovis, product manager for corporate, Sony. "It's a lot more complicated than that," counters David Huckfield, general manager, product marketing and sales for Panasonic Broadcast Europe. "4:1:0 hybrids have half the resolution both vertically and horizontally of the same image in 4:2:2. The problem is so bad that the European Broadcasting Union recommended that if 4:2:0 and 4:1:1 sources have to be processed from their decompressed form (say in a linear editing suite without SDTI) then the material should be processed using a 4:2:2 sampling system - which does not hybridise with either 4:1:1 or 4:2:0 sources," says Colin Green, JVC product manager for D9. As most transmission in Europe is going 4:2:0 (MPEG), Bovis claims that using DV or DVCAM makes less difference than DVCPRO. Although, even if a broadcaster doesn't shoot DVCPRO, US material shot on DV will also be 4:1:1, so it is not always possible to avoid it, points out Green. Besides, Bovis admits most viewers would have difficulty spotting the difference, except in an extreme case. "4:1:0 hybridisation does make a difference when it comes to colour space manipulation," says Green. "It's a known fault, but it's one that, if you are aware of the limitations, you can avoid for post purposes." As most DV-based production is for news, he believes that post is not the problem that MPEG hybridisation is for DVCPRO. Huckfield maintains that 4:1:1 "is a better match for the 625-line sample" because while both 525 and 625 have the same 720 active pixels along a standard definition line, 625 has 100 extra vertical lines, "so the ratio of pixels horizontally to vertically changes quite dramatically. For that reason, it is better to sample 4:1:1 rather than 4:2:0, as it gives slightly less emphasis to the horizontal than the vertical for 625." For 525, which has limited vertical resolution, 4:1:1 doesn't matter, but Bovis believes that 4:2:0 is more suited to the greater resolution of 625. Huckfield believes that 4:2:0 was chosen for 625-line DV because it gives a slightly different balance between how the signals are used in horizontal and vertical, similar to Pal and Secam. However, as the EBU found, 4:2:0 in 625 doesn't give a flat frequency response, which makes it necessary to do some pre-filtering. If the signals are being processed, this means going to baseband 4:2:2 then back to 4:2:0, which involves a lot of filtering. With so much processing going on, such a signal will not retain the full 4:2:0 bandwidth of the original. "If we could keep it in the 4:2:0 environment forever, that would be fine," says Huckfield. As that is impossible, he says, the EBU concluded that, for 25 Mbps DV, 4:2:0 "should be confined to special applications."
|
|
|
GETTING THE SAMPLING RIGHT As switchers and special effects devices work uncompressed, all 4:1:1 and 4:2:0 signals have to be translated to CCIR 601format for processing, and both will have to be interpolated (horizontally or vertically) during post. Panasonic chose to go for 4:1:1 because it believes that 4:2:0 suffers more degradation during filtering or subsampling than 4:1:1. But, to get the benefit of this, Green insists the 4:2:2 sampling system must be the right type. "If you use a low compression, intraframe based system such as AVR77 MJPEG or DV50 (D9 or DVCPRO-50) or Digital Betacam (about 2:1 compressed, Intraframe 4:2:2) then no problem. In fact the end result after 'x' number of generational dubs can look cleaner than even staying with the original acquisition format for the same amount of processing," except where loss-less transfer of the compressed video takes place during dubs using DV-native Firewire systems or SDTI. "Unfortunately, most editing involves some form of picture manipulation such as a dissolve or wipe which requires decompressed video, so causing picture degradation courtesy of the decode/encode process, particularly where the compression system uses high compression rates, 4:2:0 or 4:1:1 sampling, and, worst of all, analogue signals for re-encoding," so the FireWire/SDTI exception only applies where cuts-only editing and loss-less transfer is used between editing devices. Green also warns that using Betacam SX (which is 4:2:2) in the chain can compound the problem by adding temporal compression artefacts, because SX is interframe IB sequenced MPEG422P, which he claims wasn't understood fully at the time of the EBU's recommendation. He believes that 4:1:0 hybridisation will have considerable implications for archive material, and says that future formats, like HD and beyond, will probably have difficulty with 4:1:0 material. "In an ideal world, everyone would use D-5, but it costs £50,000 upwards and not everyone needs 200 generations." He, naturally, believes D9, which uses 3.3:1 compression, or Digital Betacam (with which D9 has been favourably compared by the EBU), are the best choices. If you need to shoot on a smaller format, D9 can interchange seamlessly with DVCPRO-50 via SDTI. |
|
|
TAPE: TRACK WIDTH The other most significant difference between the formats is that DV and DVCAM both use metal evaporated tape (with track widths of 10µm and 15µm respectively), while DVCPRO uses metal particle tape and needs a wider track width (18µm) to compensate for its lower coercivity (signal to noise). With miniDV, the more expensive ME tape had to be used to allow narrow track width and longer recording times (which would be only 20 minutes with MP compared to 30 with ME). ME gives about 20% increase in signal off tape than MP of the same track width, although this is not necessarily 20% better signal to noise, says Green. Huckfield maintains that choosing 18µm (the same track pitch as D3 and D5), adds reliability, as does the use of MP tape. "We have customers getting 7,000 and 8,000 head hours and the machines are still going strong [with no head changes], so who knows what the limit is," he says. When DV first came out, MP tape was cheaper than ME, which is more complicated to manufacture. However, with the huge volumes of DV tapes produced now, prices have dropped to about the same level as MP. Green points out that ME tape wasn't designed for multi-pass editing, and that MP is the better editing platform. It survives continued shuttling back and forwards better, adds Huckfield. However, this merely means ME tape has to be edited onto disk or a 4:2:2 tape and as Green says no tape at 25Mbps will give good multi-generational capability, this is how almost all DV-based editing is done. Huckfield points out that today's tape is a lot more likely to be re-used, to save money. "The number of tapes sold per VTR is reducing. Because of error correction, you can re-use a tape a lot more. That's not good for the tape manufacturers, but that's good for the user."
|
|
|
SIGNAL TO NOISE Sony chose 15µm for DVCAM because at 10µm, DV is too sensitive to tracking differences between individual machines, which is why Bovis says the track width must be wider for broadcast. He argues that despite any signal losses (about - 0.8dB) due to its narrower track width (than DVCPRO), DVCAM's ME tape gives 4.2dB better signal to noise overall than MP. It can also record 50% longer time than DVCPRO (which has to use a thinner tape which won't play in all machines if it is to go for the 184 minute maximum). This is partly because the ME tape is slightly thinner as the magnetic coating is directly fixed to the base layer, and partly due to its slower running speed. Bovis claims the ME tape will also last for at least 60 years. Besides better signal to noise, Bovis claims ME tape also has a better protective layer, so it is less harsh on the drum, increasing head life. "We have some customers still using heads at 6,000 hours," he says. The track pitch also means that DVCPRO can't do pre-read, which Bovis says was needed for the news market. "DVCAM is the limit we can do technically. DV would be easier," he says. JVC has pre-read on D9, but Green believes it is something very few people use except for compositing. Panasonic doesn't have pre-read on any of its DVCPRO systems, because there is no demand for it. If there was, there would be a technical work-around to make it happen. |
|
|
PRODUCTION FORMATS Huckfield believes one of DVCPRO's advantages is being standardised on a single cassette size, which means any of its VTRs can replay a cassette created on a lower format, and the SD and HD worlds can co-exist in the same product line. "We've got the widest range of professional DV formats of any manufacturer, from Professional DV to DVCPRO-HD," says Huckfield. "We believe it is the best format for production." Sony originally didn't want to market DVCAM to broadcasters, who it wanted to buy SX, but it gave in and launched the Master series to combat DVCPRO when it saw broadcasters wanted something less expensive than SX or today's IMX. The first generation DVCAM models didn't have very good multi-generational capabilities, "but today there is no difference between 4:2:0 and 4:1:1 machines," says Bovis. |
|
|
COMING SOON At NAB 2001, JVC brought out CineLine versions, developed with OpTex, of both its widescreen Professional DV and D9 camcorders. These have the colour matrices and gamma of film built in, together with long viewfinders and all the other add-ons associated with film production. As low-budget movies are increasingly being shot on DV, it sees this as a growing market. It also showed a hard disk drive add-on at NAB which allows recording to disk and tape at the same time. This may also be used by Panasonic, although Huckfield wonders if the price of disk packs will restrict it to niche applications. Green believes that HDD will be a better way of recording for 50 or 100 Mbps (once disk capacity increases), because it can overcome the limitations of tape.
|
|
|
|
|
|
|
|