Multimedia
Blogs
Forum
Affordable HD
Whitepapers
Newsletters
Events
Advertisers
Blogcast
MillimeterDV Dreams
Apr 1, 2001 12:00 PM, By Bob Turner
It has been nearly 10 years since the DVC tape format was first introduced at gatherings such as the SMPTE Technical Conference and the IEEE/IE Digital Video Workshop. A few of these presentations enthused about how one day this format would be the “universal bit bucket” for our industry.
In short, the presenters explained a plan to create one universal tape format and associated drive mechanism for multi-track audio production, video production, graphics, computer archiving, home recording, etc. The only variation in the format would be a change of digital header information bits (and perhaps a different colored tape shell) to differentiate how the tape was being used. Memory In Cassette (MIC) chips on the plastic cassette shells would provide important metadata, such as new, complex EDLs for edited masters, as well as library and production information. The best part was that this was going to be a universal standard endorsed by all the major players in both the television and computer industries. This broad endorsement would in turn make the format cost less, thanks to economies of scale.
Fast forward to 2001. Few people are using DV tape for anything but audio/video recording; the MIC chip is not the success envisioned; and there are at least half a dozen competing DV tape formats (in several cassette sizes) rather than a single, universal standard.
But even though the DVC format did not evolve as some had hoped, the various DV formats in existence have been successful. Steve Yuhas, president and COO of Panasonic system sales, says 70% of the television stations in North America that have converted from analog to digital use Panasonic's DVCPRO format for acquisition and recording. In fact, more than 750 TV stations in the United States currently use DVCPRO. In the short time since the format was introduced, millions of consumer camcorders and hundreds of thousands of professional DV codec products have been purchased. No previous tape format has had such an immediate and profound impact.
But because digital video formats are still evolving, there are many issues to be concerned with when purchasing a DV-based NLE system. Video Systems recently invited some industry leaders to discuss these considerations.
Participants in this discussion included: Jan Crittenden, group manager, Panasonic Broadcast & Television Systems; Jerry Hergenreder, North American sales director, in-sync; Bernard Lamborelle, technical marketing director, Matrox; Charles McConathy, founder and president of ProMax Systems; Jeff Medeiros, product manager for the DC/DV product line, Pinnacle Systems; Mike Nann, technical marketing, Digital Processing Systems; Mike Skibra, president of Digital Marketing International, the North American representative for Digital Voodoo; Travis White, product marketing manager, Ulead Systems; Tim Wilson, director of marketing and business development, Boris FX; and John Worthington, project leader for CineStream, Media 100.
Video Systems began the discussion by asking, What issues should a potential DV-edit system purchaser consider?
“First and foremost, ask yourself what you'll need a year down the road,” suggested Mike Nann of DPS. “Will DV still be your primary medium? Will you need additional I/O formats, such as component, SDTI, or SDI? Many, if not most, native-DV systems aren't upgradable to support these higher-end I/O formats. Might you need uncompressed later? If so, you should look at a solution that supports more than just DV.”
Jerry Hergenreder of in-sync agreed. “While it is a great acquisition format, by shooting DV you have already thrown away a lot of your color space unless you are using DV [throughout the process].”
Matrox's Bernard Lamborelle said DV has the best price/performance ratio for capturing video, but there are other issues to consider.
“Mastering on DV isn't always the best option because of the fragility of the format and the introduction of artifacts,” he said. “Depending upon your postproduction requirements, it may be advisable to choose a higher-quality format, such as DV50 or uncompressed.”
Tim Wilson of Boris FX said the most important precursor to capturing high-quality video is technique. “Attention to lighting is paramount, but so is solid shooting,” he said.
Ulead's Travis White elaborated: “Image quality starts with the camcorder quality in two areas: glass and CCDs. The good camcorder needs to have a dedicated CCD for each of the three color spaces. Both together will provide a much sharper and truer rendering of color.”
“Next in importance is quality of the video,” said Mike Skibra of Digital Marketing International. Jeff Medeiros of Pinnacle asked, “Will there be gradient graphics that might create bands of color instead of smooth gradations?” ProMax president Charles McConathy said, “With OS 9.1, the Apple codec is much, much better, thus there's little issue with banding, flashing, or quality shifts that existed before.”
John Worthington of Media 100 added that straight cuts will always give you the best quality because the bits are untouched. “What you get out is exactly what came in from the camera,” Worthington said. “For anything else — transitions, titling, etc. — you are processing the frames, which involves decompressing the DV video, doing some manipulation, and recompressing it. Because DV is lossy, there is some quality loss. In most cases it's not noticeable unless you recompress a large number of times.”
There are differences in codec quality, Medeiros noted. “They are not all created equal. You want to look at the hardware versus software transcoding capabilities,” he said. “What does it do in the raw YUV space when the video is uncompressed? What kind of sampling issues are revealed? Are there quilting issues with multigeneration images?”
Lamborelle added, “Because DV always compresses at a constant bit rate, the image content has a significant impact on the quality of the compression. Simple images can yield amazing results. Complex images contain more details that must be removed. You should always test the quality of a DV system with complex images.
“Although there is only one way to decode a DV stream, many decisions must be taken during the encoding phase. These decisions are taken based on image analysis. Compressor efficiency will therefore vary quite significantly based on the intelligence built into the compressor,” Lamborelle explained.
“Finally, DV usually looks a lot better than JPEG or MPEG-2 I-frame at comparable bit rates because the compression is done on the camera before the noise gets introduced. When capturing an analog source into a DV system, the result can be disappointing.”
Title and graphic edge controls over the top of the DV signal are also very important considerations, agreed Medeiros.
He said a few questions still need to be answered: How clean is the title? How clean are the other production elements? What kind of anti-aliasing do you have? What kind of compositing quality do you get when you perform the compositing or effects? When the editor renders composites, is the software applying anti-aliasing edge controls on the rendering?
“Be careful of claims of ‘realtime’ DV editing,” warned Nann. “Creating a transition between two DV clips requires that for each resulting frame, the corresponding frame of each source clip has to be decoded from DV into an intermediate form. That requires two DV codecs — one for each source clip. Then the two intermediate frames are combined to create the resulting output frame. This can easily be sent to the analog outputs of the DV editing card.
“But if the desired output format is back to DV, a third DV codec is required to re-encode the resulting frame back into DV. Most native-DV editing solutions have only two DV codecs. Thus, they can play transitions between two DV clips out to an analog monitor in realtime, or play a cuts-only DV project out to a DV destination in realtime, but they can't play transitions between two DV clips out to a DV destination in realtime. A separate transcoding step, once the project is complete, is required for output to a DV destination. Those are generally fairly quick, but it's still not realtime for that application.”
Medeiros added, “Realtime is most important during the creative process, where the editor is experimenting and trying various alternatives. Outputting to an analog monitor is all that is required. The short period of rendering is generally required for DV output to tape after all the editing and decision-making is done.”
Another important issue is the timecode accuracy — or, more realistically, inaccuracy — of DV devices, as opposed to the frame-accurate RS-422 control most professionals are used to, Nann noted.
“FireWire can be considered a pretty strict standard, but the reality is that compatibility issues between hardware, drivers, and camcorders can pop up and surprise a new user of DV technology,” said White. “When one is thinking of entering the world of DV editing, they should do their homework on what software they wish to use with what system and which camcorder.”
IEEE-1394 is definitely the connection of choice, but some cameras are not as compatible as they should be, Worthington agreed.
“As time goes on, the incompatibility issues seem to be getting resolved — much like when personal computers first came out and when you bought your printer you hoped it would work,” said Panasonic's Jan Crittenden.
“Over time, special drivers were written for each machine. In today's NLEs, in the driver section you will find you can select the names of specific machines, much like your computer does in working with the multitude of potential printers. This is a feature of the OHCI changes that are taking place.”
Wilson said that there are feature sets on the VCR, and some say that camcorder transports were not designed primarily to search or shuttle. “Cameras tend to be much slower, and somewhat less accurate, than decks. I'm hearing that head wear is less of an issue than I thought it might be, but I'm a strong believer in decks, regardless.” Medieros added that Pinnacle users almost always connect camcorders.
“Camcorders are built to be camcorders and do not have the ruggedness or the ballistics built in to be both,” Crittenden said. “It will work, but you prematurely wear out the camera deck and the batch digitizing will not be very accurate. VCRs will generally handle tape in a more gentle manner, and search to commands are likely to be carried out with the tape characteristics in mind.”
Furthermore, many producers taking the plunge into DV don't realize that the FireWire connection carries with it all the information, and it's all locked, said Wilson. “Neither video nor audio can be altered on the way in,” he added. “This can be immensely frustrating for producers. For example, audio can't go through a mixer on the way into the workstation.”
“There are a number of audio issues, including two-channel versus four-channel audio, plus several different sampling rates,” Medeiros noted. “Can you support all the channels and all the sampling rates natively in the editor, including locked versus unlocked audio?”
Skibra added, “Some products are limited to 16-bit audio. QuickTime currently has support for only two channels.”
This is the single biggest hazard, according to Wilson. “Some default to 32K, others to 48K,” he said. “Producers must be aware of which is which, since almost no editing systems can handle mixed sampling rates in the same timeline. It becomes an even larger issue with mixed digital-analog environments, where legacy audio and certainly CD audio has almost certainly been digitized at 44.1K. Something has to be converted, and the time to figure that out is not on a deadline, with a client looking over your shoulder.”
“Storage is a huge issue,” said Medeiros. “It's all about speed and data rates. We are certainly in a market today where ATA 100 drives are used in the field broadly to move DV. We still sit on standards such as SCSI data rates, which support dual-stream DV editing. If you use SCSI, the cost goes up. If you use ATA 100, the price goes down but there are concerns.”
Nearly any modern internal drive has that level of performance, as long as it doesn't also have OS software on it, Wilson said. “That is, it should be a drive dedicated to DV capture and editing. But you can get 60GB or 80GB of video-quality drive space for a few hundred dollars.”
A fast and efficient backup solution should be considered, said Lamborelle.
Crittenden added, “I think there is one more consideration here, and that is archive issues. How long does your tape need to be playable? Metal Evaporated tape doesn't have the same shelf life as Metal Particle. Ask any broadcaster.”
And don't forget the support issues when looking at various editing systems, warned Crittenden.
Nann agreed: “Technical support is important, but exactly how important depends on the user. But tech support will always be an important criteria.”
“Training is becoming more and more important, as well,” added Lamborelle. “Solutions are now becoming a lot more complex and sophisticated. As they evolve so rapidly, people don't have time to keep up on their own. They need ways to learn how to work efficiently, not just how to operate the various functions.”
Worthington agreed: “Because most DV users don't have a lot of video experience, I'd add user groups to the list as well. There are a variety of good email lists for DV editing.”
“In the immediate future, realtime on the Mac is going to be big, but I am not sure about where DV is going,” said McConathy.
Lamborelle concurred: “I think DV is a great format for today's needs. DV25 will not necessarily evolve. DV50 and DV100 are already available to provide higher quality,” he said. “On the other hand, with the continuous increased chip and system-level performance it may be likely that newer formats, such as MPEG-4, may someday be used from acquisition through editing and distribution.”
Crittenden added, “The best news is that any of the ¼-inch DV formats can be played in the DVCPROHD machine and the signal upconverted to an HD signal. The DVCPRO formats have guaranteed that DV will be around for the next 10 to 15 years, due to the adoption of the product by the broadcasters and the development of an SD-to-HD solution.”
Contributing editor Bob Turner is a 25-year veteran film and video editor and nonlinear editing system consultant. Reach him at bobturner@mediaone.net.
Continue the discussion on “Crosstalk” the Millimeter Forum.
