DVC— the first consumer-grade digital video format
In 1993, Sony, Panasonic, JVC, Thomson, Philips, Hitachi, Toshiba, Sharp, Mitsubishi and Sanyo joined together to form HD Digital VCR Consortium. One year later, the consortium, which by that time grew to 55 companies, finalized the Digital Video Cassette (DVC) specification.
The original DVC standard stipulated two cassette sizes and a digital interface that allowed copying video from one tape to another or upload it to a computer without loss of quality.
The digital interface was courtesy of Apple Computer. Its engineers started working on it in 1986 and completed the first specification of what Apple has christened FireWire™ in 1987. Later, Sony, Panasonic and Philips contributed to the project.
Sony recognized the capability of Firewire for reliable isochronous real-time transfer of digital video and advocated its inclusion into the DVC specification. Officially designated “DV Interface,” it specified throughput protocols for digital audio, digital video, control and auxiliary data over a single cable.
In 1995, the Institute of Electrical and Electronics Engineers adopted Firewire as IEEE-1394 Serial Bus. Sony branded it iLINK™. DVC launched the same year having Firewire as the standard protocol for transferring DV video.
“DVC” has come to designate consumer-grade VCRs, camcorders and cassettes, while “DV” describes the format in general, including different variations of encoding schemes, professional media and the data interchange protocol.
Unlike analog consumer-grade video formats, DV is a true component digital video recorded at eight bit per color component, with 5:1 compression, having two channels of 16-bit audio that exceed CD-quality, or up to four channels of 12-bit audio. Its color resolution is lower than on professional digital video formats, but four to five times higher than on VHS or Hi8. Integrated time base corrector ensures that the image is rock-solid.
DV encoding is similar to MPEG-1, MPEG-2 and M-JPEG and employs intra-frame compression, that is, each frame is compressed independently. Also, each frame has the same size, which simplifies recording on tape as well as editing.
DV has been designed to be symmetrical, compact, and cost-effective. The same chipset can perform both compression and decompression, a trade-off for its higher bandwidth requirement.
MPEG-2, which was being developed at the same time as DV, employed a significantly more complex encoder compared to the decoder, which made it appropriate for distribution. MPEG-2 found its first applications in satellite broadcasting and in DVD-video.
Like all digital audio and video formats, DV has built-in error correction. If an error cannot be corrected, it is digitally masked. Absent of uncorrectable errors and concealment, each copy is a clone of its parent.
The ability to copy digital video losslessly attracted copyright protection watchdogs, and to keep the format alive, the idea of digital VCR was shelved — as it turned out, indefinitely, only a couple of models were made. Large DV cassette, which was meant to replace VHS cassette and could hold over four hours of video, was repurposed for professional usage, while small or mini cassette was relegated for use in consumer-grade camcorders.
Panasonic PV-DV1000 ($4,200), Sony DCR-VX1000 ($4,199) and Sony DCR-VX700 ($2,999) were the first MiniDV camcorders. At the time of launch, Mini DV cassettes were priced around $20 for 30-min duration and $25 for 60-min duration.
Immediately after the launch, Panasonic, together with Bosch and Philips, introduced DVCPRO — a non-proprietary, as it has been stressed, video format for the production, broadcast, and distribution of digital television video.
Panasonic claimed that a professional version was needed because video tracks on the consumer-grade DV were too narrow for reliable editing accuracy, time code as specified in the DV standard was purportedly not frame-accurate, and tape and cassettes were not robust enough for professional usage.
Panasonic adopted large cassette, augmenting it with medium-sized cassette for portable usage, and stipulated MP tape only. Panasonic also widened the recorded tracks to 18 µm, locked audio to video, and added SMPTE timecode. DVCPRO machines would play large cassettes with standard DV recordings. Small DV cassettes could be played with an adapter similar to a VHS-C adapter.
Along with the fully featured ENG camcorder and studio VTR, Panasonic promised an upcoming compact field recorder, laptop edit system, and high-speed player for “streaming” DVCPRO tapes at a speed four times faster than realtime to a nonlinear editing system. Such faster-than-realtime digital video downloading is standard practice today with modern file-based video formats, but was new thirty years ago. Next year Panasonic integrated tape transport into a standard PC desktop machine, bundled it with editing software and welcomed everyone to the future of video editing.
In 1996, Sony responded with DVCAM, having tracks widened to 15 µm and with audio locked to video. Sony continued using small and large cassettes with ME tape.
JVC took a different road, promoting Digital-S as an answer for field and edit bay use. Digital-S employed DV codec with double the bitrate and double the chroma resolution, recorded on ½-inch tape in a VHS-sized cassette.
In 1997 Panasonic presented its own DV variant with doubled bitrate — DVCPRO 50. Basically, Panasonic modified its machines to record Digital-S on medium and large DV cassettes. The format was supported by Philips and over the years became hugely popular for Electronic Field Production (EFP).
DVCPRO 50 had lower bit depth than Digital Betacam — 8-bit vs 10-bit — and lower data rate — 50 Mbit/s vs 90 Mbit/s — but in practical applications proved to be comparable.
In 1998, Panasonic launched DVCPRO Progressive, prototype of which Panasonic presented a year earlier. Native progressive video better suited computer-based presentations and online video platforms. Several high profile shows have been produced with it, and they still look great.
As of 1998, Sony’s attention started to switch towards MPEG-2, which was more advanced than DV. Normally MPEG-2 was used in interframe or Long-GOP mode, when each second of video would contain several fully encoded frames, and all other frames would store only the differences from one frame to another. It turned out that MPEG-2 was more efficient than DV even when it was used in intra-frame mode. Sony called its variant of intra-frame MPEG-2 as MPEG IMX, and implemented it at constant 30, 40 and 50 Mbit/s rates on Betacam hardware.
With Sony pivoting to MPEG-2 and JVC trying to convert VHS-based hardware to digital in the form of Digital-S and D-VHS, Panasonic became the leading force in further development of DV.
In 1999, Panasonic launched the much-anticipated DVCPRO HD equipment also known as DVCPRO 100. The 100 Mbit/s recorders supported both 720p and 1080i resolutions, could play back DVCPRO, DVCPRO 50, DVCPRO P, as well as consumer DV/MiniDV tapes. Standard definition could be internally upconverted to high definition.
The second half of the 1990s saw a rapid transition of video production to a networked computer-based operation. In 1998, RAM price dropped to just 1/100 of 1992 price, while the capacity of HDDs continued to grow steadily.
The Giant Magnetoresistive Effect (GMR), discovered in 1988, culminated in developing of a high-capacity HDD — in December 1997 IBM announced an HDD with maximum capacity of 16.8 GB. For the first time, a single HDD could store full content of a 60-minute MiniDV cassette.
The rest of the industry began commercialization of GMR heads during 1998 and by 2000, 100% of HDD production worldwide included GMR sensors.
This does not mean, of course, that computer-based editing had not been tried before, after all, Avid/1 Media Composer was introduced in 1989. The system looked very much like a modern NLE, but hardware limitations did not allow to edit and render movies at full resolution. The Media Composer used low-resolution M-JPEG clips with embedded timecode, so when you wanted to render the final video, you would produce an Edit Decision List (EDL), which is basically a list of in/out points with timecodes. Then you would run this EDL on a system that directly controlled source and target VTR. If your sources came on several tapes, the system would stop and ask you to replace the tape, then it would wind the tape to a correct location and continue the edit.
DV, with its relatively low data rate and with Firewire digital link being an integral part of the standard, allowed to pass video between a camcorder, a VTR and a computer without recompression, which meant no generation loss. This also reduced the CPU load, and significantly lowered the storage requirements. Systems that started to ship in the mid-1990s, allowed to edit DV natively at full resolution, but needed a whole farm of hard drives to store content just of a single tape. New generations of HDDs developed by the end of the 1990s allowed to whittle down the number of drives, and computer-based editing became affordable.
In 1999 Sony ported DV to 8-mm hardware, releasing Digital8. It is interesting to compare Digital 8 with DVCPRO.
Could the DVC Consortium reuse 8-mm hardware for DV instead of creating a whole new family of cassettes and new tape transport?
Using 8-mm cassette and having the format backwardly-compatible with all those Sony 8-mm camcorders would give Sony an unfair advantage. The consortium instead developed the smaller 6-mm cassette for DVC. — Steve McDonald
Let us step back and try separating reliable information from hearsay. Here is what has been reported in press:
- In March 1993, The New York Times reported that during the next few years, the consumer VCR would begin its conversion from analog to digital technology. Also coming were “video recorders capable of capturing and playing high-definition television images”. Leading electronics companies started talking about creating a common format for digital VCRs to avoid another format war like that between Beta and VHS. The systems would store at least four hours of conventional programming or two hours of high-definition programming on a videotape a little larger than a standard audio cassette. Digital VCRs could also become storage devices for computers.
- In 1994, a conference regarding “Specifications of digital VCR for consumer use” was held in Tokyo.
- In April 1994, The New York Times reported the outcome of the conference — the committee, representing 50 electronics companies from around the world, had worked out the technical details and had agreed on a proposed standard for VCRs that would use digital technology.
- The new machines would use quarter-inch wide tape. The cassette would be a little bigger than a standard audio cassette or an 8-millimeter videocassette. Yet it would be able to record four and a half hours of conventional television. A smaller cassette with only one hour of recording time would be offered for use in camcorders.
It is evident from the reports that within one year — from 1993 to 1994 — the new domestic VCR format was scaled down from high definition to standard definition. Two sizes of cassettes — for VCR and for camcorder — were agreed upon, along with narrower tape compared to readily available 8-mm cassettes.
Next we are entering the territory of hearsay and rumors.
- Supposedly, Sony wanted to use 8-mm hardware for the new digital VCR, but other companies rejected this idea on the grounds that using 8-mm cassette and having the format backwardly-compatible with Sony 8-mm camcorders would give Sony an unfair advantage. Comparing the DVCPRO and 8-mm cassettes it is obvious their specs are very similar, so Sony’s suggestion made sense. On another hand, could 8-mm cassette be scaled down to MiniDV size? If it could, what would the recording time be? 20 minutes? 30 minutes? It would be comparable to VHS-C, but nowhere close to 120 minutes that have become standard for 8-mm video.
- Another rumor is that Sony was allowed to port DV format to its 8-mm hardware five years after the DVC specification was inked. This explains why Sony released Digital 8 in 1999 instead of doing it earlier. By that time Sony had become one of the leading manufacturers of DV equipment and introduced its own professional version of DV — DVCAM — so 8-mm hardware fell by the wayside.
In 1999, not being very successful with its digital ½-inch equipment, JVC presented a professional shoulder-mount DV camcorder, JVC GY-DV500 branded as “Professional DV”. The camcorder was based strictly on the consumer DV format. JVC justified the “Professional DV” moniker by using a professional shoulder-mount camera body, controls and a removable lens in a combination with a consumer-grade tape transport.
In 2000, Roo Productions, an independent production company in Arizona, United States, made a commercial film shooting it on the GY-DV500 and using an editing system based on the Matrox RT2000 card. Reportedly, the movie was filmed out and exhibited in theaters, but I could not find a single still shot much less a clip or a trailer for it.
The RT2000 came with the Adobe Premiere editing program, which we had already used. By creating this production entirely in DV, we created a Hollywood-style film for a fraction of the cost. — Benjamin Freedman, owner of Roo Productions
Sadly, JVC did not offer film-like frame rates on the GY-DV500, despite that some of its consumer-grade models had 25/30-fps progressive mode depending on region.
Sony did not bother much with film-like frame rate either, but for California porn producers and skateboard filmers, who were the major purchasers of the famed VX1000, this was not a deciding factor. In fact, the more real footage looked was the better.
This is why it saddens me when I see some recent remasters of old videos that have completely missed the boat. In particular, a 4K remaster of an iconic skating movie “Yeah Right” drops every second frame of the original video, reducing image rate from 60 to 30 per second.
The VX can only shoot at 30 fps (25 fps for PAL region), interlaced, this means 60 distinct images (50 for PAL region), called fields, per second. Interlaced video looks fine on a CRT TV, but it must be converted for presentation on modern progressive-scan TVs, for watching on a computer or on Youtube. The technical word is “deinterlacing”, and there are several ways of doing that. The most truthful method that preserves as much spatial and temporal resolution as possible is converting each field into a complete frame, so you get 60 full frames from 30 interlaced frames.
Whoever did the remaster either did not know how to deinterlace correctly, or did not care, or intentionally reduced image rate to achieve “film look” from what originally had “live look”. The “film look” craze reached fever pitch twenty years ago, everyone wanted their video to look like film, and using film-like frame rate is one checkbox in the long list of things needed to achieve film look.
Movies are usually shot at 24 fps, 25 fps is common in 50 Hz countries, and 30 fps can look filmic enough. So, the frame rate reduction could have been done intentionally, but my point here is that the look of the video has changed. Some people do not notice it, but many do.
Throwing away every second frame produces stuttery motion, especially when shooting with high shutter speed and large depth of field, which is how camcorders like the VX operate unless the filmer puts on a neutral density filter, slows shutter speed down and blurs the background, which is hard to do on a small-sensor camera. So, the 4K remaster, while improving sharpness and detail, destroyed the motion portrayal of the video, and as such, its character.
To add insult to injury, the deinterlacer used for the remaster occasionally farted, producing ugly combing.
Big-name Hollywood directors can make similar mistakes as well. David Lynch used a Sony PD150 DV camcorder to shoot “Inland Empire”. The DVD booklet describes the process:
Inland Empire was shot in SD 4×3 48029.97 on DV tape. After being captured, the footage was edited in the same format in Avid. The footage was then converted to HD 16×9 1080p 23.98 and input into Clipster for remote DaVinci 2K grading. During this process, a Teranex converter was used to treat video noise on selected shots. With the digital-intermediate (DI) master complete, it was output to 35 mm film and HDCAM-SR tape.
In other words, Lynch shot his movie:
- in 4×3 frame aspect ratio (the PD150 has widescreen mode, but the sensor is native 4×3 anyway);
- at 29.97 fps (the PD150 has no film-like 24 fps mode);
- interlaced (the PD150 can shoot in progressive mode… at 15 fps, which is a joke, so Lynch had no other choice than to shoot interlaced).
Converting 30 interlaced frames into 24 full frames never works well, you get a combination of stutter, combing, ghosting and blurring. Cropping 4×3 into widescreen means losing quarter of the pixels. But it seems that Lynch did not care much about technicalities, he just loved digital workflow and he liked his camcorder.
Could Lynch do better than the PD150? He sure could. He could buy a Canon GL1 and shoot at 30p. He could buy its European counterpart, the XM1, and shoot at 25p, slow it down to 24p in Avid and film out without motion artefacts. Or, he could buy the DVX.
David Lynch made “Inland Empire” in 2005, three years later after the Panasonic AG-DVX100 had been announced and two years after it had been available on the market.
In 2002 Panasonic announced the AG-DVX100 handheld camcorder, which could record at the film-like 24-fps rate. This was huge!
The DVX injected fresh blood into DV format, converting it from something that amateurs used to shoot family videos, and journalists used to shoot news pieces, into the first affordable digital filmmaking machine.
Everyone started talking about DV filmmaking and shooting movies with the DVX. One of them is “Iraq in Fragments”, premiered at the 2006 Sundance Film Festival, winning three awards: “Directing Award Documentary”, “Editing Award Documentary” and “Excellence in Cinematography Award Documentary”.
The documentary has been shot in Iraq over a two-year period in widescreen at 24 fps. The DVX’s sensor is not native widescreen, so some vertical resolution has been lost. Even with custom gamma settings and other tricks, 8-bit DV could not capture the full dynamic range of the blistering Middle East setting, and fast shutter speed produced stuttery motion at times, still it looked as close to 16-mm film as digital could in the early 2000s.
If I had to continue shooting movies in standard definition, as long as I was able to do it at cinema frame rate with progressive scan images I would be able to accept it, because if you can do that you’re basically working within the same kind of visual parameters as your basic 16-mm film stock. It gives you the ability shoot in a cinematic way. — James Longley
James Longley, the author of the film, has uploaded it on Youtube, go check it out both for the technical qualities, but primarily for the content. The message drives the film forward, and there are no video artifacts to distract you.
One may think that by mid-2000s only cash-strapped filmmakers used DV to shoot movies. Enter the BBC.
In 2005, one year before BBC HD inauguration, the BBC filmed several TV series, including “Rome” and “Space Race” with Panasonic DVCPRO 50 camcorders. The BBC felt that SD camcorders were more light-sensitive and less critical to focus errors, and delivered fitting “gritty” look.
During the 2000s, HD became the norm, but DV remained a viable option for amateur filmmaking, ENG and long programs. Panasonic continued developing the format, launching P2 solid state media and announcing that plans for DVCPRO P2 would follow the historical pattern of the DVCPRO25/DVCPRO50/DVCPROHD tape-based format.
The P2 card (Professional Plug-in card) incorporated four SD Memory cards and originally came in 2 GB and 4 GB capacities. The 4 GB card could record 16 minutes of DVCPRO25 or 8minutes of DVCPRO50 and was sold for $600.
The prices went down over the years, as flash memory capacity grew and its cost fell, but still remained high compared to tape. In 2007, a 32GB P2 card had a list price of $1650. In 2008 the price for a 32 GB card went down to $1500. Then to $625 in 2009. In 2011, the 32GB dropped to $480.
The professional video industry has wholeheartedly embraced solid-state recording, and we like to think we initiated that movement (with P2 in 2004) — Joe Facchini, vice president of sales and product management for Panasonic (2011).
Just think about it, a 32 GB card costs less than $10 today. Sure, an SD card is not as rugged as a metal P2 card. Still, I think that Panasonic may have started the switch to solid-state media a couple of years too early.
As if recognizing that not everyone wanted to switch to tapeless worfklow, especially at such high prices, Panasonic launched DVCPRO XL — Extra Large — cassette in the same year it unveiled P2 card. The cassette almost as large as a VHS cassette allowed to record two hours of DVCPRO 50 and four hours of standard DVCPRO.
In 2006, Panasonic launched DVCPRO HD EX, also known as DVCPRO HD LP — a long-play version of DVCPRO HD. Track pitch has been reduced from 18 μm to 9μm, becoming 1μm narrower than with consumer-grade DV. This allowed to double HD recording time, reduce tape running cost and make it viable to record HD onto medium cassette.
Sony also did not want to abandon tape, yet. In 2003, Sony together with Canon, Sharp and JVC announced HDV — a consumer-grade HD format ported on DV hardware. Sony used both large and small cassettes, while Canon and JVC used small cassettes only. Sharp did not produce any harware. Firewire was used as a digital interface.
It is worth noting that Panasonic engineers ensured that MPEG-2 packets could be sent over Firewire way back in 1996. With the same track pitch and the same tape speed, you could have one hour of standard definition DV video or one hour of high definition HDV video on the same cassette, it was almost a miracle.
Throughout the 2010s, DV continued to be supported by Panasonic, Sony and JVC on their tapeless equipment: P2, PD, SxS, MemoryStick and SD cards. The ENG camcorder, launched by Panasonic in 2018 and still available now, records from 200 Mbit/s AVC Intra all the way down to 25 Mbit/s DV.
DV is still supported by major computer platforms and operating systems, unlike some other codecs that suffered during the transition from 32-bit to 64-bit technology in macOS, including popular intermediate codecs from Avid (DNxHD / DNxHR) and GoPro (CineForm). Professional editors were livid when they found out they could no longer edit their shows using NLEs like Premier Pro. The only option that worked was running Premier under Rosetta translation layer. As of 2023, the issue has not been solved.
You can shoot DV in the 2020s, the tape is no longer produced, but is available at places like eBay, and even at $5 per one hour cassette it is cheaper that it was at the time of the launch, and tapes can be re-used. DV is digital video through and through, with sufficient resolution, no jitter, it supports 4x3 and 16x9, supports film-like progressive scan rates. More advanced camcorders have gamma curve control and HDR-like dynamic range compressors. All DV camcorder have CCD sensors that use global shutter, which means no skew and no jello.
And the cassettes are so cute. They come in different colors and textures, and are fun to fiddle with. The cassettes are small, but not too small for you to lose them like a MicroSD card.
So, take your old MiniDV camcorder out of a dusty box and go shoot some digital video on tape. I know I am going to! ⬛
