Reference: StudentFilmmakers Magazine, April 2007. HD and HDV – An Addendum: A Clarification of Terms and Formats by David J. Weinberg. Pages 48- 50.
Sam Kauffmann’s “HD and HDV” (StudentFilmmakers, January 2007) provides a useful introduction to the subject, but some clarification will be helpful.
Given the widespread confusion about D-Cinema and DTV, and knowing that Kauffmann’s article was aimed at those creating material for theater presentation and DTV broadcast, a reference is needed against which you can compare the digital video camera formats’ specs in Kauffmann’s article, thus enabling some measure of the compromises these formats inflict on the source before it gets to the viewer.
There are differences between generally accepted production standards and the FCC’s officially mandated DTV broadcast transmission standard. Since material is produced to be broadcast, I take the broadcast standard as my reference.
The FCC’s DTV standard table 3 delineates a set of recommended formats that have been universally adopted for transmission and consumer display. The formats are recommended because they are in the appendix, which is supplemental information, not mandated. However, manufacturers treat most of them as required.
From the FCC’s DTV standard’s table 3 (in this context, specifying “p” or “i” is irrelevant) and addressing the active pixels (those showing image content):
* 480 rows of pixels: The 4:3 aspect ratio calls for 640×480 square pixels (=307,200 pixels per frame); for either a 4:3 or a 16:9 aspect ratio, the table lists 704×480 horizontally squeezed/stretched rectangular pixels (=337,920 pixels per frame).
* 720 rows of pixels: the specified 16:9 aspect ratio calls for 1280×720 square pixels (=921,600 pixels per frame).
* 1080 rows of pixels: the specified 16:9 aspect ratio calls for 1920×1080 square pixels (=2,073,600 pixels per frame).
The formats mentioned in Kauffmann’s article are not an exhaustive list from the DTV standard’s table 3, which lists 18 formats — five sets of horizontal/vertical pixel counts, two aspect ratios, driven at three or four frame rates, some interlaced, some progressively scanned. This does not consider the frame-rate reduction of 0.1% to accommodate NTSC color, which adds another 18 formats.
Referring to the formats listed in Kauffmann’s article:
* DV: Kauffmann’s “pixel aspect ratio” (“pixel count” would be a better descriptor) of 720×480 yields 345,600 pixels per frame, yet he lists 210,000 pixels per frame, which computes back to about 530×396 square pixels. Plus, 720×480 square pixels yields a 1.5:1 aspect ratio (since no pixel shape is specified, square is a rational assumption). None of this matches the FCC standard’s recommendations, although 720×483, 720×486 and 720×586 (all with non-square pixels) are three of the many generally used production formats.
* The 720p data matches part of that contained in the DTV standard’s table 3.
* Under 1080i, the pixels-per-frame does not match anything in the DTV standard’s table 3.
The other data under 1080i and 1080p is consistent with the DTV standard’s table 3.
Going beyond the DTV standard, there are other issues due to the accepted marketing definition of what can be called HD:
1. In the Consumer Electronics Association’s (CEA) effort to simplify marketing of consumer HD products, they recommended HD television be defined ONLY by the number of rows of pixels: at least 720p or 1080(i or p). The number of pixels-per-row is not considered. Many displays, and some cameras, which qualified as HD under the CEA’s recommendation, compromised the horizontal pixel count — such as 1440×1080 instead of 1920×1080. There have been industry pundits who questioned whether we can perceive, and therefore need, the full 1920-pixel horizontal image resolution. Visual acuity, image size, and distance from the screen determine human visual perception (see sidebar “Visual Acuity”). Since the HD-formats’ pixels are square, if the viewer is located the optimum distance for HD vertical image resolution (where he can resolve all the vertical resolution), he can also resolve the full number of horizontal pixel specified.
2. Since qualification as HD only requires 720 rows of pixels, progressively scanned, a computer display with a pixel count of 1024×768 (a 4:3 aspect ratio, if the pixels are square) can be, and has been, marketed as an HD display, even though interpolation will be required in the horizontal direction to make the image fit, while the image will be letterboxed by 24 black rows, top and bottom. Even the best interpolation results in image quality compromises (look at the small text at the bottom of commercials to see the variation in text quality; it also shows up on pixelated computer screens when the native display resolution is not used).
3. While terrestrial broadcasting yields the best HD image quality the consumer can receive (not counting the Blu-ray and HD DVD formats), the bitstream is greatly data-compressed using MPEG2 (from around 1.485Gbps — billion-bits-per-second — down to a maximum of 19.39Mbps — million-bits-per-second, including audio and metadata), it is well known that the direct broadcast satellite (DBS) providers compress the bitstream even further. DBS providers are starting to use MPEG4 (H.264 AVC) which can yield picture quality better than MPEG2 encoding at less than half the MPEG2 data rate. Either way, this means that the image quality seen by the creator in production and post-production (if a properly calibrated legitimately HD display is used) does not get to the home viewer. Watching what happens to standard definition and high definition commercials during an HD broadcast helps bring this point home. Content creators should take that into account.
4. None of this addresses the additional fact that while there are color display/camera specs for the red, green, and blue color primaries and the coordinated color of white (D6500), there are two sets of primary coordinates — one for standard definition, the other for HD (they’re close but not identical). Beyond this, consumer sets are universally not properly set up, and only a minority of viewers spend the money to a) select sets that can be properly calibrated and b) pay to have a knowledgeable professional perform the calibration (there are not that many people, even self-proclaimed experts, who can do this job properly). Only when this is done can the creator have some hope that the consumer sees what was created.
5. Another display characteristic that needs to be properly adjusted is gamma (the rate at which the display goes from black through gray to white, with a linear gray-scale input; most often acknowledged as correct is a gamma of about 2.2). For the camera, lens quality (including color fringing and geometric distortion, particularly near the edges), image sensor size (linearly proportional to pixel size), inherent image noise level (larger pixels usually results in lower image noise in dark images), and others need to be looked at to determine if the camera’s performance is acceptable for the specific application.
This does not mean movie image characteristics should be compromised with consumer-product limitations in mind, but that in monitoring the work during creation and evaluation, it is necessary to ensure the displays are properly designed and calibrated.
David J. Weinberg (Tobias Audio, Silver Spring, Maryland; 301.593.3230; WeinbergDa@cs.com) is an engineering consultant and technical journalist on audio, video and film technology. He provides audio and home theater engineering consultation and professional on-location digital audio recording services to companies, radio stations and individuals. He brings to his work an MSEE, a First Class Radiotelephone license and over 40 years of continued study and active involvement in the audio, video and computer industries. He is Vice Chairman of the Audio Engineering Society’s DC section, and a manager in the Society of Motion Picture and Television Engineers’ DC section. David has authored articles on various phases of audio for video and film, is Associate Editor of Multi Media Manufacturer (www.MultiMediaManufacturer.com), Contributing Editor to Widescreen Review (www.WidescreenReview.com), plus serves as Membership Officer for the Boston Audio Society (www.BostonAudioSociety.org) and Editor of its journal: The B A S Speaker.