What determines the image quality of a scene? There’s a vast array of choices of equipment available to videographers. Reviewers are supposed to help narrow the choices.
If a non-technical critic comments unfavorably on the technology used to create a moving image production, that technology must be pretty bad. Mustn’t it?
Here’s a section of a review by movie critic Howard Thompson that appeared in
The New York Times
on May 15, 1965. The Electronovision rush job on Miss Harlow’s life and career is also a dimly lit business technically. Maybe it’s just as well. This much is for sure: Whatever the second ‘Harlow’ picture looks and sounds like, it can’t be much worse than the first.
The reference is to two movies, both titled
, both about the actress Jean Harlow, that were released in 1965. The one that came out first was shot electronically, and it seems from the review that the Electronovision process must not have been very good.
On the other hand, an earlier production using the same electronic cinematography system brought rave reviews. Even today, the DVD of the 1964 Electronovision-shot
receives four stars in customer reviews on Amazon.com.
There is a clue in Thompson’s review that the technology might not have been at fault. The first
was a rush job to beat the second into theaters. In the eternal triangle of videography, if one of the three angles—speed, economy or quality—is emphasized, at least one of the others must be reduced. In this case, it appears that quality was sacrificed for speed.
If the whole truth be told, the technical quality of the 1964
wasn’t sparkling, either. Its higher ratings are based on its performances and on an unusual production.
More recently, critically acclaimed movies have been shot on DV camcorders. In this age of digital cinematography, it might be worth examining what that means.
Worldwide, even where there is no alphabet or the local alphabet doesn’t include letters recognizable to readers of only English, the symbol DVD is known to represent a small disk usually carrying moving image programming. The acronym originally stood for digital video disc, then digital versatile disc, and now nothing more than DVD.
At about the same time that DVD was first considered, there was also DVC. It was to be the term used for digital video cassettes, but the
was soon dropped.
Today, DV has two true meanings, often tied together. A third is erroneous.
One meaning of DV is that of the original DVC. It is a digital video cassette and the recording and playback systems associated with it. The cassette contains quarter-inch-wide tape—enough of it to store an hour of component digital video on something that will fit on a palm.
That form of DV may also be considered the
. HDV recorders use DV cassettes and tape transports.
Squeezing even standard-definition component video information into that small a recording medium required a bit rate reduction system, more commonly referred to as compression. Like JPEG and MPEG, DV compression (the second true meaning of the term) uses a discrete cosine transform (DCT) to move picture information from the spatial domain to the frequency domain, where some of the finest detail can be dropped, if necessary, to reduce the number of bits required. And, like JPEG and MPEG, DV compression has been applied to a number of different media and recording systems.
Sony’s professional DVCAM version of DV increased the track pitch and Panasonic’s DVCPRO increased it even more (before dropping lower in the long-playing LP mode). DVCPRO50 and JVC’s Digital-S (or D9) doubled the bit rate to reduce the compression ratio and increase the color detail. DVCPRO50P allowed progressive-scan recording, and DVCPRO HD doubled the bit rate yet again to allow HDTV to be recorded using DV-based compression.
DV compression was applied to consumer Video8 (8mm) videocassettes in Digital8 (see 10-Q, SMPTE below). It has been applied to memory cards, most famously Panasonic’s P2, and to magnetic and optical disks. One of those optical disk recording systems is Sony’s XDCAM.
XDCAM does not use DV compression exclusively. XDCAM users may choose between DV and different types of MPEG-2 compression. Each has advantages and disadvantages.
Different compression algorithms (and different implementations of each) will react differently to different scenes and compression requirements. All of them should be able to handle a low-detail still image well. At that point, only compression constraints will affect picture quality.
In standard-definition compression, common constraints are 4:1:1, 4:2:0 and 4:2:2. The
represents 720 active (picture carrying) luma (brightness detail) samples per scanning line. The
represents a quarter as many (180) of color;
represents 360. It might seem that 4:2:2 offers better image quality than 4:1:1, and that would be true in a world of unlimited bit rate. When the bit rate is limited, however, 4:1:1 can look better than 4:2:2; the latter could show more objectionable artifacts of insufficient bit rate. 4:2:0 offers 360 active color samples per line but half as many lines of color information as of luma; its information quantity is comparable to that of 4:1:1. Although 4:2:2 can look noticeably better, 4:1:1 and 4:2:0 are quite good.
Again, DV represents either a recording system or a compression system (or both). XDCAM is a recording system. Neither is a camera, though it has become common, unfortunately, to refer to a small camcorder using DV compression and recording as a DV camera and to a camcorder with an XDCAM recorder as an XDCAM camera.
If it is functioning properly, a recording system has no direct effect on image quality (although, arguably, the cost, size, weight and lifetime of storage media may indirectly affect the way the images are shot). If it has enough bit rate and is unconstrained, a compression system has no effect on image quality either (color constraints
have a small effect).
Camera differences can certainly affect image quality, and the way cameras are set up can affect it even more. Both standard- and high-definition (HD) images have been recorded on both DV and XDCAM media, and DV-based compression has also been used for both SD and HD purposes.
Both DV and XDCAM recorders are also available in forms not attached to the backs of cameras. Therefore, almost
camera can be used to make either DV or XDCAM recordings (exceptions are the $30 camcorder available at CVS pharmacies and digital cinematography cameras like the Dalsa Origin).
Moving back from storage mechanism and compression, the next stage of a camcorder is that of image processing. Longtime
columnist Robert Goodman has created guides to such in-camera processing. His photos show how tiny adjustments of such parameters as black stretch or gamma can create major differences in the way pictures look—differences much greater than the subtle ones that might differentiate DV compression from MPEG-2.
Another stage forward is the camcorder’s optical block, containing its prism and imaging chips. Today, camcorders have chips ranging in resolution from 320×240 to 5760×2160 and in size from six millimeters in diagonal to 38. Some cameras use three imaging chips, at least one (the Olympus Octavision) uses four, and others use just one. The single-chip camcorders use mosaic or striped color filters of various types or, in the case of the Foveon imager, a color stack similar to that of film.
Even camcorders with attached DV or XDCAM recording systems vary in this regard. There are SD XDCAM-based camcorders with 2/3-inch imagers and HD versions with 1/2-inch imagers. Similarly, there are DV-based camcorders with 2/3-inch, 1/2-inch and 1/3-inch imagers in SD and HD versions.
Imager size, resolution, color separation technology and maximum optical aperture can affect such things as diffraction-limited modulation transfer (how much contrast the laws of physics allow at different resolutions). That can have a greater effect on image quality than subtle differences between DV and MPEG-2 compression or the essentially non-existent (in terms of image quality) difference between storage media.
The next stage forward is the lens, which, as cinematographers know and videographers sometimes forget, is usually more important than the camera in terms of affecting image quality. Then there are lens adjustments. A briefer shutter time and a wider aperture will provide the same exposure but a different depth of field.
In front of (and sometimes behind) the lens, there can be optical filters: polarizing, star, net, diffusion, etc. They, too, will affect image quality more than differences between DV and MPEG-2.
Going from the storage at the rear of the camcorder to the optical filter at the front exhausts the camcorder-related equipment that can affect image quality, but there is still more. The camcorder’s mount can affect image steadiness, and lighting can have a huge effect on image quality—far more than any subtle differences related to DV or MPEG-2 compression or to DV or XDCAM recording.
It makes no sense, therefore, to say that, as classes, either DV camcorders or XDCAM camcorders offer more or less quality than the other. Lighting, lens and camera image processing adjustments will likely have much greater impact than any recording differences.
Of course, there may be something brewing in a laboratory that could change everything. If the differences between 2/3-inch, 1/2-inch and 1/3-inch imagers can affect image quality, imagine what could happen with an immense imager many times those sizes! There hasn’t been any official announcement of the development yet, and it’s odd that people should refer to an optical block by a storage system’s designation, but it appears that a set of the mammoth imaging chips is imminent. Why else would people talk about a 27-inch DV set?