HDV
From Wikipedia, the free encyclopedia
- HDV can also mean Hepatitis D virus.
High Definition Video (HDV) is a video format designed to record compressed HDTV video on standard DV media (DV or MiniDV cassette tape).
Contents |
[edit] History
The HDV format was developed by Victor Company of Japan, Limited (JVC) and Sony. The format was initially supported by three other companies: Canon Inc., Sharp Corporation, and Sony Corporation. These four companies make up the HDV consortium and are all manufacturers of HDV hardware. They announced their HDV partnership in September, 2003. Sharp has not manufactured an HDV camcorder at this time. They have since been joined by other companies, notably Avid, Canopus, Ulead, Sony Media Software, and Apple, Inc.
JVC was the first to release an HDV camcorder, the GR-HD1. With the GR-HD1, the user could select standard DV (480i), or HDV (720p30, 480p60) mode. Sony was next to launch an HDV camcorder. Introduced in September 2004, the Sony HDR-FX1 recorded in HDV 1080i format. Depending on the TV standard in the released market, the unit would support either 50 Hz or 60 Hz recording (but not both.) The Sony HVR-Z1U, the equivalent U.S. professional model, supports both and adds XLR audio plus 44 additional features, most notably a dual clock that could support either 50 Hz or 60 Hz formats, allowing for both PAL and NTSC on a single camcorder. In 2005, Sony released the consumer-oriented HDV Camcorder, the HDR-HC1. A professional version of the HDR-HC1, the Sony HVR-A1E, was released in September 2005. In 2006, Sony released the HDR-FX7, HDR-FX7E, HVR-V1U, and the HVR-V1E, marking the world's first full 1080p camcorders in a small format, with both 24p and 30p (only 25p on the HVR-V1E) features on the camcorder (except for the HDR-FX7 and HDR-FX7E).
JVC has developed its own extension of the HDV format called ProHD which shoots natively at 720p24.
In September 2005, Canon entered the HDV market with the Canon XL H1, a professional, modular HDV camera system, and in July of 2006, Canon announced their XH A1 and XH G1 models which are similar to the XL H1 but in a smaller form factor and a fixed-mount lens. In 2006, Sony appended the HDR-HC1 with the less-expensive HDR-HC3. The HDR-HC3 features a slightly improved CMOS chip, but omits some features (such as external mic-in) of its predecessor. Canon then introduced the Canon HV10 which is a compact consumer priced 2.76 Megapixel CMOS chip HDV camera system.
Panasonic took a different approach by pushing its DVCPRO HD format for HD acquisition. The Panasonic AG-HVX200 HDTV camcorder records DVCPRO HD (not HDV) on DVCPRO P2 cards. The built-in Mini DV tape recorder is relegated to standard definition DV only. Although the camcorder is a DVCPRO HD camcorder, it is not generally considered to offer the quality that other DVCPRO HD (such as the Varicam) offer; this is likely due to its smaller CCD size.
[edit] Overview
HDV was designed to offer existing video production environments a cost-conscious upgrade path from standard-definition (SD) to high-definition (HD) video. Since HDV operates at the same recorded datarate (25 Mbit/s bitstream rate) as DV, HDV recorders share the same physical (MiniDV ) tape transport as existing DV equipment. For the camera, the main expense is concentrated in the optics and imaging electronics. Compared to HD video equipment built on more professional standards (such as HDCAM and DVCPRO HD), HDV enjoys a tremendous cost advantage. HDV camcorders open high-definition video acquisition to consumers, amateur videographers, and low-budget TV production.
Although 1080i HDV and DV share the same (DV) tape format and the same recorded datarate, they use completely different video compression technology. The DV codec uses strictly an intraframe (spatial) scheme, whereas HDV uses the well-established MPEG 2 video codec. MPEG-2 applies both intraframe (spatial compression) and interframe (temporal compression) to video-compression, allowing HDV to achieve its higher spatial resolution at the target bitrate of 19.7 Mbit/s (720p) and 25 Mbit/s (1080i). Compared to more expensive HDCAM and DVCPRO HD equipment, HDV suffers from significantly more spatial and temporal (motion) artifacts. As a consequence of interframe (temporal) compression, HDV editing is more complex, and introduces greater distortion at the splice point (due to the interdependence of adjacent video frames.) Compared to conventional SD DV, HDV offers a much higher spatial resolution, so most observers are willing to accept the artifacts in exchange for a higher-definition picture. It is worth remembering that standard definition MPEG broadcasts generally use only 2 to 4Mbits/s, and MPEG4/AVC or WMV-HD produce very good results with 6 to 8Mbits/s, compared to which the 25Mbits/s of HDV represents a much lower degree of compression, with much less visible artifacts. HDV audio uses lossy compression (MPEG-1 Layer 2) to reduce the audio bitrate to 384Kbps. DV audio uses uncompressed 16-bit PCM at 1536Kbps. As a result, HDV audio is technically inferior, although MPEG-1 at 384Kbps is regarded as 'perceptually lossless.'
Canon has started to ship consumer camcorders that are capable of 24pf. That is, 24 progressive 1080 frames are captured per second, and stored as 2 frames in a 1080i transport. This has the advantage, that software will show the 2 interlaced frames as one full resolution 1080p frame (but only at 24, 25 or 30 fps)
Since HDV and DV use the same DV25 tape transport, at the same linear speed, recording times for DV and HDV are identical. That is, a 60 minute MiniDV cassette can store 60 minutes of either DV or HDV footage. As of yet, no HDV cameras can record HDV at LP speed, so the maximum record time on one tape is 80 minutes, as opposed to 120 with an 80 minute tape at LP.
Because HDV uses the same tape form factor as DV, users should be able to use any high quality MiniDV tape in their HDV camcorder. However, because HDV has a lower tolerance for drop-outs because of its long-GOP compression, many HDV users purchase either "master" quality Mini-DV tapes or specially formulated HDV tapes.
[edit] HDV compression
Although HDV and DV share the same tape format and the same recorded datarate, they use completely different video compression technology. The DV codec is strictly an intraframe (spatial) compression. Each DV video frame is recorded as an independent picture, with a fixed bit allocation and uniform placement on the videotape. The HDV codec is based on MPEG-2 video compression, which employs both intraframe and interframe (temporal) techniques. Interframe compressors store only a fraction of the frames in a video as independent pictures -- called I frames -- and encode the remaining frames as changes relative to them. Consequently, HDV frames vary in size depending on their prior and future neighbors. In HDV 1080i, one in every 12 (25 FPS) or 15 (30 FPS) frames is an I frame. In HDV 1080p, one in every 12 (25 FPS) or 15 (24 or 30 FPS) frames is an I frame. In HDV 720p, one in every 6 (24, 25, or 30 FPS) or 12 (50 or 60 FPS) frames is an I frame.
MPEG-2 video enables HDV to achieve a much higher compression ratio than DV, but at the cost of motion-induced artifacts in scenes of complex motion. The artifacts are a limitation of the compression technology and bitrate allocated to the video bitstream. Motion artifacts are imperceptible for static shots and gentle pans, but may become increasingly detracting as motion complexity increases. For example, a moving riverbed may exhibit regions of picture breakup, depending on its portion of the total screen area. It is important to view these limitations in the proper context. Lighting, chroma content, camera motion, etc all play a role in the potential for artifacts. The television series "JAG" shot many scenes using HDV without any incident, shooting over extreme latitudes of sunlight over ocean water, with dark and light content in the subject matter. For the DV codec to approach the spatial quality of HDV, it would require more than four times the storage space. Encoders are constantly improving; The Sony XDCAM HD format is very similar to HDV. MPEG is the standard of the future, and as encoders improve, the potential for artifacts lessens.
Dropouts or errors in the compressed video bitstream affect HDV much more severely than DV. This is an unavoidable characteristic of interframe compression. Since frame data affects multiple frames (and not just the one it originated from), a dropout will impact all dependent neighbors. Frame-accurate editing is also made more difficult by the MPEG-2 codec. Any modifications to the video sequence require the surrounding group of frames to undergo a complete (and lossy) decompression/recompression cycle. However, virtually all professional non-linear editing software is now designed to work flawlessly with HDV. Currently the BBC do not consider HDV to be a high definition video or broadcasting standard due to the compression and errors mentioned above and they instead categorize is as a standard definition video format. BBC will accept only 25% of a program filmed in HDV and only by prior consent. Their preferred choice of format for recording HD is currently HDCAM.
For all its limitations, HDV is quite stunning on HD displays. Although free of motion-induced artifacts, DV tends to look fuzzy when scaled up to HD resolutions. Subjectively, most observers are willing to accept HDV's visual artifacts in exchange for a more detailed picture.
[edit] Resolution and aspect ratio
In HDV, the video frame is defined to have an aspect ratio of 16:9. Permitted resolutions are 720p and 1080i.
HDV 720p uses a resolution of 1280x720 square pixels. HDV 1080i uses a resolution of 1440×1080 pixels, but is still displayed with an aspect ratio of 16:9 (like SD widescreen formats, it uses a pixel aspect ratio of 1.33 instead of 1.0). This means it has lower horizontal resolution than true 1080 HD formats (1920x1080).
Despite this, the perceived detail of HDV is much higher than that of PAL or NTSC DV formats. 1440 pixels is still twice the horizontal resolution of SD formats. In total, each HDV frame has 1,555,200 pixels, which is 4.5 times the resolution of NTSC DV (345,600 pixels) and 3.75 times that of PAL DV (414,720 pixels).
The numbers above refer to the luminance (brightness) information only; chrominance (color) information is subsampled (4:2:0 for HDV) to reduce the amount of data, as happens with DV and DVD, although NTSC DV uses a different sampling pattern (4:1:1). In other words, in all these formats, the chrominance resolution is one quarter of the luminance resolution. Most professional video formats use a 4:2:2 sampling pattern, and some high-end formats support 4:4:4, which is to say full chroma sampling.
[edit] Notes regarding specific camcorder models
Many HDV cameras support progressive scan (which they denote by "P"). They achieve this by doubling the horizontal scan rate to scan the interlaced sensor twice in one cycle. This lets them use a less expensive sensor at the expense of some interlacing artifacts. There will be fewer artifacts than 50i/60i, but it will not be as smooth as true 25P/30P. Also, various practical tests have shown that resolution in 24P/25P/30P progressive modes is lower than when 1080i interlaced is selected. The exception to this are the Sony HVR-V1 camcorders which scan progressively and store the resultant data using either 2-3 pulldown (in 24p mode) or PsF (in 25p and 30p modes) techniques, so there's no resolution loss.
[edit] Editing HDV
As a consequence of the fact that HDV uses the interframe MPEG-2 GOP (group of pictures) structure instead of a solely intraframe compression system, native editing of HDV footage differs technically from the native editing of DV footage. In DV, as each frame of a video sequence is stored as an independent object, the recorded footage can be spliced at any frame without any loss of quality.
When editing HDV's MPEG-2 data directly, a single frame cannot be changed without re-encoding subsequent frames from the same group. Any editing of the native MPEG-2 video, whether it be a complex transition or a simple scene-change, requires a decompression and recompression of the entire HDV frame group. Especially over many generations, this may result in increased artifacting, for example in the next frame group after a splice. However, because HDV's 1080i bitrate is 25 Mbit/s, these should be not as obvious as those seen when lower bitrates, such as those used for encoding clips for download, are used.
Editing HDV's native MPEG-2 transport stream files also forces the computer system to work much harder to perform even simple tasks of cutting and splicing as frames that don't actually exist as independent cells have to be re-built by the NLE system on-the-fly.
If HDV footage is converted (known as 'Transcoding') to a good intermediate format for editing, these considerations will not necessarily apply, and gradual degradation from generation to generation of edit may be avoided while substantial system performance gains are made. The lossy Apple Intermediate Codec (which runs out of QuickTime) is an efficient, easily usable codec for editing HDV in systems such as Final Cut Pro but lacks the transcoding and generational quality of some third-party HDV Intermediate codecs and offers no realtime performance assistance. CineForm's 'ConnectHD', 'AspectHD' and 'ProspectHD' intermediate codecs and encoding utilities not only maintain higher quality but also function within the rendering engine of some software editing systems (such as Premiere Pro) to boost real-time performance without hardware assistance. CineForm products currently work with Sony Vegas and Premiere Pro editing systems. Lumiere HD offers a similar functionality for Mac based editing systems (namely Final Cut Pro) but without any real-time assistance. BitJazz's SheerVideo offers lossless real-time codecs that speed editing with zero generational loss, although Final Cut Pro does not yet support third-party codecs for real-time effects.
There are many advantages to editing HDV using a Lossless Intermediate rather than the native Mpeg2 file however the trade off for transcoding to a HDV Lossless Intermediate is that the file size is substantially increased and so large hard drive arrays are required for storage of footage. Avid Xpress Pro can edit using native HDV, and Avid also claims to have the advantage of being able to work with mixed formats in the timeline, without the need to transcode any material, since different formats are coped with seamlessly for viewing and output, with automatic conversion as required. It must be noted though that when DV material, for example, is included in an HDV project, a problem arises because it must be de-interlaced prior to scaling to the HDV format, and then re-interlaced. De-interlacing is generally regarded as a very difficult process to do well. This problem will arise of course in all editing software, regardless of whether a lossless intermediate format is used.
[edit] The Interlace Dilemma and Web Compatibility Problems
Although interlaced video has been heavily criticised by many, including the EBU (who express the hope that in future all original material will be shot in 1080p 50 for easy conversion to either 1080i or 720p), the fact is that it does work well when displayed with current interlaced PDP technology. HDV (50i UK) viewed on some plasma screens with interlaced display technology ALiS (developed by Fujitsu and used on all PDPs with 1024 or 1080 lines) retains both sharpness and smooth motion while having much less motion jitter than a movie (which is 24p or 25p). The motion jitter of movies, though regarded by film enthusiasts as a desirable feature that adds to the feel of film, is actually only as acceptable as it is because film directors know that they must restrict themselves to slow pans. Hand-held material shot in 25p or 24p can look very jittery and confusing.
PCs do not currently support interlaced video very well, and the Windows media player, though it makes some attempt to deinterlace if presented with interlaced files from HDV, fails and produces vertical ripples that appear to depend on the size of the output image and other factors. In general, good deinterlacing yields quite good results if the deinterlacing option is turned on, but deinterlacing of high resolution video in real time with good results can be done only with the help of video hardware acceleration.
Currently the preferred approach seems to be to deinterlace edited material for the web, but this is problematic. Avid Xpress Pro currently appears to fail to do deinterlacing on its export options such as WMV HD, which results in unacceptable 'mice teeth' on moving vertical edges.
Although good quality deinterlacing is possible on HDV files using third-party software, it tends to take a very long time (all night for half an hour or so). HDV has the potential to look extremely good on the Web, using 6 to 8Mbits/s as demonstrated on the Microsoft WMV HD showcase site. Many have tried and failed to get comparable quality on the Web from HDV because of the interlace problem. It is not impossible to achieve good results though, if the correct methods are used to deinterlace the footage. See the examples on this page. The showcase material is of course all from film, and therefore 24p to begin with, avoiding the deinterlace problem.
[edit] Editing software support
- See also Non-linear video editing
For Mac OS X:
- Avid's Xpress Pro HD, Supports a wide range of HDV formats/standards, but no support for true 24p HDV.
- iMovie HD Universal binary, current version does not support 24p HDV
- Final Cut Express and Final Cut Pro 5 Latest version of FCP5.1.2 does support direct HDV 24p, and 720/25p which is used in PAL countries (i.e. needed when working with JVC Pro HD camera in 720/25p mode)
- Lumiere HD Lumière HD (beta) for Final Cut Pro 5. First available software to edit HDV on a Mac with QuickTime based Non Linear Editing systems. This is the only Mac based application (other than FCP) which allows for back to tape in HDV encoding (Including JVC's 24p ProHD). No Universal Binary version and crash-prone when run under Rosetta on Intel-based Macintoshes.
- MPEG Streamclip 1.8 for Mac now a Universal Binary and supports automatic 3:2 pulldown (Converts 24p to 29.97), this program can only perform basic edits such as cutting, copying, pasting and trimming, but is available free of charge, and has excellent tools for exporting, demuxing and converting video, for example from HDV format (usually with the .m2t extension) to MPEG-2 (.mp2) format. MPEG Streamclip can handle most MPEG container formats (including ts, ps, vob, dat, mpg, and mp2) which adds to its usefulness.
- HDVxDV's [1] Soon to be Universal Binary (Beta version of Universal Binary has been in use for months and works great on Intel-based Macintoshes), this program loads in HDV (Including JVC's 24p ProHD) and transcodes to whichever format the user wishes to cut in. Timecode support is inaccurate. Does not allow for back to tape in HDV.
For Microsoft Windows:
- NewTek SpeedEDIT, Supports all SD, HD and HDV resolutions/frame rates up to 2k film. Does not need an intermediate codec for real time editing. First software only editor with real time full resolution/framerate 1394 HDV preview through camera or deck while editing. Can mix formats, resolutions and framerates in a single timeline. Well integrated timeline and storyboard interface.
- Avid's Xpress Pro HD,
- Avid's Liquid
- Adobe's Premiere Pro
- Adobe's Premiere Elements (from version 3.0 onward)
- Canopus' Edius
- VirtualDub called VirtualDub-MPEG2
- Sony Media Software's Sony Vegas.
- Sony Media Software's Sony Vegas Movie Studio Platinum Edition.
- CineForm Aspect HD + Connect HD
- Avid Adrenaline HD + Xpress HD + Avid DS/Nitris
- Ulead MediaStudio pro 7, 8 and VideoStudio 9 and 10 Plus
- CyberLink PowerDirector
- MPEG Streamclip 1.0 for Windows, which has the same features as its Mac counterpart.
- MPEG Video Wizard DVD.
- Serif MoviePlus 5
Under Linux:
- Cinelerra - free software, allows HDV editing and composition
[edit] Specifications
Media | DV or MiniDV Tape |
Video signal | 720/60p, 720/30p, 720/50p, 720/25p, 720/24p, 1080/60i, 1080/50i, 1080/25p, 1080/30p, 1080/24p |
Video sampling frequency for luminance | 74.25 MHz (720p), 55.6875 MHz (1080i) |
Video Chroma subsampling | 4:2:0 |
Video sample quantization | 8 bits (both luma and chrominance) |
Video Compression | MPEG2 Video (profile & level: MP@H-14) |
Video compressed bitstream rate | Approximately 25 Mbit/s for 1080i / 19.7 Mbit/s for 720p |
Audio sampling frequency | 48 kHz |
Audio quantization | 16 bits |
Audio compression | MPEG-1 Audio Layer II |
Audio compressed bitstream rate | 384 kbit/s (192 kbit/s per channel) |
Audio mode | Stereo (2 channels); optional 4-channel MPEG-2 Audio Layer II at 96 kbit/s per channel mode |
MPEG Stream type | Transport Stream Packetized Elementary Stream |
Stream interface | IEEE 1394a Apple FireWire 400 or Sony i.LINK (MPEG-2 TS) |
File extension | Usually saved as .m2t |
[edit] See also
- ProHD - a variant on HDV
- Digital cinematography
- Filmizing
[edit] External links
- Latest HDV camcorder reviews, news, and product information at DigitalCamcorderNews.com
- HDV Format Legal Information site
- HDV Video production at *FilmDailies.com - a filmmaker's blog
- Sony HDV Info
- HDV Cafe Hi Def Video Resources
- German Forum with HDV focus
- / HDV information, HDV FAQ
|
|
---|---|
Magnetic tape |
VERA (1952) - 2 inch Quadruplex videotape (1956) - 1 inch type A videotape (1965) - 1/4 inch Akai (1967) - U-matic (1969) - Cartrivision (1972) - Video Cassette Recording (aka VCR) (1972) - V-Cord (1974) - VX (aka "The Great Time Machine") (1974) - Betamax (1975) - 1 inch type B videotape (1976) - 1 inch type C videotape (1976) - VHS (1976) - VK (1977) - SVR (1979) - Video 2000 (1980) - CVC (1980) - VHS-C (1982) - M (1982) - Betacam (1982) - Video8 (1985) - MII (1986) - D1 (1986) - S-VHS (1987) - D2 (1988) - Hi8 (1989) - D3 (1991) - D5 (1994) - Digital-S (D9) (199?) - S-VHS-C (1987) - W-VHS (1992) - DV (1995) - Betacam HDCAM (1997) - D-VHS (1998) - Digital8 (1999) - HDV (2003) |
Optical discs |
LaserDisc (1978) - Laserfilm (1984) - CD Video - VCD (1993) - DVD-Video (1996) - MiniDVD - CVD (1998) - SVCD (1998) - FMD (2000) - EVD (2003) - FVD (2005) - UMD (2005) - VMD (2006) - HD DVD (2006) - Blu-ray Disc (BD) (2006) - DMD (2006?) - AVCHD (2006) - Tapestry Media (2007) - Total Hi Def (2007) - HVD (TBA) - PH-DVD (TBA) - SVOD (TBA) - Protein-coated disc (TBA) - Two-Photon 3-D (TBA) |
Grooved Videodiscs |
Baird Television Record aka Phonovision (1927) - TeD (1974) - Capacitance Electronic Disc aka CED (1981) - VHD (1983) |