| mathematically foresightful engineering constructed standards exceeding specification |
Decades ago the NTSC National Television Standards Committee developed a television that could work for then-current 1940's tube-type electronics technology, but foresaw far greater possibilities: color [1951] found itself a niche in the '50's design methodology, and promised yet greater; but today's [1980's] HDTV proposals appear to have gone awry, supplementing further advance thinking with mere multi-programmability - no stereo-eyes-ed vision, no perfect resolution.
The difference is that the future of NTSC still exists - and we ought consider what NTSC really was, and is, and is to be.
Basically NTSC defined a transformation between a broadcastable time-linear signal and a viewable 2-D motion-display [2.5D] of reasonable proportions approximating the picture 'desirability' and 4:3 [horizontal:vertical] aspect ratio of common-theatre 16mm film. But as technology has advanced, the theoretic-possibility has more rapidly furthered: we could but don't have wavelet processing implemented - it is available as praxi-theory. We still don't have stereo'eyes'ed processing and viewing that's 2D compatible. We still don't have wide-screen 2:1 aspect [now common in theatres not cinerama] or wider - understand, space is 4pi radians and the horizon is 2pi radians, averaging pi:1 aspect - or, for sitting steady with foreward view only space is 2pi and the horizon is pi, and eyes yield about 75% vertical height [limited by brows and cheeks], about pi:1.5, or that is, simply 2:1 aspect. [The tradition of 16:9 was based on generalizing HDTV to best-fit the proportioned average of all known aspects previously used, by film historians who thereby know aspects, albeit the state of the art depended on available aspect, and this method of selection neglected your basic linear measure distribution rule of throwing-away the two extremes, and weighted emphasis on the central values (the extremes are statistical possibilities not contributing to the central lobe mode average, but more to the estimated variance) - curious also, 16:9 = 4:3 squared, but has an irrational diagonal: NTSC had height:width:diagonal=3:4:5, all rational: the advance should have chosen one-under-9:16:irrat., 8:15:17 for its rationality]
[We may also giggle over the new-generation of technolog'ers who dote on elliptical frames-- when we consider that the horizon may be curved, irregular, and slightly tilted, a simple rectangular is a better assured view than an elliptical, of a distorted approximately elliptical interest]The original NTSC concept was 6MHz with 1.25Mhz vestigial, and FM at 4.5Mhz [over video carrier at said 1.25MHz] ... the original intent was that FM, being a constant amplitude, could be instantaneously subtracted from the luminance [except for its audio-modulated very narrow quadrature] for almost 4.75MHz picture. [The technology then-available was not even PLL-ready but they foresaw better] When color came along it was set between 2.3-3.0MHz and 4.2MHz, with its carrier at 3.579545MHz - still without best-fit PLL's.
We may note further that the vestigial-side-band chosen by NTSC was itself a deep-compromise, for they could have chosen full-band luminance, and then decoded the quadrature carrier as more [double] pixels - but that would have required double the receiver demodulator electronics, and that in the era before digital technology and integrated [micro] circuitry. But, we can [consider] pick-up the effort, compatibly, and extend it now: we might take the symmetric band around the carrier as 4.75MHz+4.75MHz = 9.5MHz, with the quadrature representing those 'in-between' pixels as either, double the resolution, or as, the binocular view [compatible with sexichrome - that is, 6 colors = 3 colors for each of 2 eyes - recall that B&W Black&White was also called monochrome, one color, typically white] ... since the 'stereo-eyes-ed' image consists of two views spatially displaced, the resolution is effectually almost doubled, as well as 3D [3.5D in motion] - this is different from one picture with 'Z-depth' sent as an additional modulation: thus there are options yet to consider ... it's just more compatible with the current technology to consider this continuity of design-methodology.
This DDTV 'double-vision' is also naturally wide-screen proportioned - the interleaved horizontal pixel placement allows up to 4x [range] free-stretching without loss of fine-resolution: in fact the natural stretch is root-three, horizontal:vertical, because that makes for equilateral triangular placement of pixels: optimal viewing pixel density.
I've suggested that the current standards for NTSC and HDTV be made/kept more compatible - the time base be digital near 18MHz, 5x color-carrier, and FM 1/4, and add stereo 'Z-depth' inside the vestigial zone [alternating-phase quadrature for stability] by adding video-low-frequency reduplicated into the quadrature to retain some compatibility for the oldest TV receivers until they quit altogether. My System 18/3,4,5 [band, FM, color] proffers a digital time-base closely approximating these parameters - compatible with NTSC, and with receivers in operation today. The horizontal retrace interval can be reduced to a few pixels, with digital 'parity' check - we can extract more from what everyone already uses: multiprogrammability is then, just an add-on.
Near-future digital HDTV should also have signal compatibility that gets most of the picture through the NTSC receivers: There are digital coding schemes that retain a portion of the original signal, and then the lower bits may be the noncompatible codes.
[Further details (below) discuss half-pixel edge-combing and treble-thread-gemming: smoothing and sharpening (1984-1996)]
[America seems to have forgotten not what it had possessed, but what had been foreseen ... Yesterday's done: Let's not waste the perception we've attained: This insight and its method is part-and-parcel to our future in space-navigation: our design methods must always illumine the right answer, fitting specification as well as future development and previously discerned possibility: we must have not merely the art, but the science of upgrades]
ITEM by item, NTSC can be improved directly: some without cost to the consumer. And some improvements (can) apply to HDTV. HDTV is, as yet per the cartel's proposals, a distortion, by certain reasoning.
1.a. UN-KELL TV. Current picture definition presumes direct fidelity - unrehearsed imagery. The contrary fact is that edge-combing the image can restore 100% (double) of the resolution lost in the kell-factor (both horizontally and vertically). Think of un-kelling as, combing one's hair so-as to align each strand onto the pixel centers: this can be done fairly inexpensively in the camera iconoscope and image processing electronics, and requires no change in any receiver. The effect is an immediate (100%) sharpening of the image on top-quality receivers: Where the closest strands blurred across adjacent pixels, the combed strands appear distinctly spaced. The average combing adjustment of hairs is a quarter pixel width - maximum, a half - with no degradation visible at proper viewing recess. The cost is for doubling the camera's resolution, and for the un-kell-processor.
1.c. HDTV's bonus factor disappears against this improvement of NTSC, because the proposals have gone for DCT-processing: a giant-step requiring upgrading entire systems, cameras, processors, and receivers, but accomplishing nothing more than un-kelling. The advantage of numeric signal coding is nice but (as yet) incompatible.
2. CLAIMS for HDTV's bigger, better picture have been spurious: the pioneering features of new technology have been idly neglected for filibustering marketing tactics: promises of more talk and less agony.
2.a. ASPECT (ratio) is proffered at a humorous 16:9 - somebody squared NTSC's 4:3 for fun - more precise than one can feel: 7:4 and 9:5 are within 1.6% an 1.3%, and either is simpler to write and remember. And compared to paper: the old 4:3 was exactly the ratio of 8.5×11 in. paper minus a 0.5 in. border (10:7.5), so, if we likewise-consider 'legal' paper at 8.5×14 in. minus that 0.5 in. border (13:7.5) we find it within 2.6% but, then compatible... (13:7.5 also features a 30° diagonal to fully 3 significant digits-- that's where the short side is exactly half the long side, the midlength side, √3/2, and the aspect, 1.732).
Aspect (ratio) should reflect the viewing proportions. We live on a plane (plane to the horizons), and except for traversing a chasm, we rarely look down or up for reading or examining significant details: usually there's only the ground and the sky (or floor and ceiling) below and above; and dramatic actors (and sports heroes) array and move most expertly on that plane. On this mathematically simple basis we should more appreciate either a full-turning view of 360° horizontal and 115° (average) vertical, or a full-face (non-turning) view of 180° horizontal and 86° (average) vertical. The ratio of angle in a circle to solid angle in a viewing sphere is 2pi:4pi, so, average height is 4pi/2pi=2, and the aspect is pi:1. The eyes scope maximally about 180° horizontally and 135° vertically, for an aspect ratio of pi:1.5, or 2:1, and can also turn to look almost at those bounds: so the ratio is main not just peripheral....)
Again, small percentages are not felt: the aspect ratio should be no less than 2:1, and no more than 3:1. The shape is not crucial - though some argue for elliptical aesthetics, the excess in the corners of a rectangle do not contribute much to width or to height; and an ellipse tilted slightly left and right tends to fill a rectangle.
2.b. RESOLUTION ought also to be spatially variable, as most detailed viewing is centered on the screen. It is mathematically possible to vary the pixel (spatial) aspect ratio continuously between 1:2 and 2:1, and maintain a maximal density for approximately spherical pixels, by varying only the horizontal offset-interleaved pixel spacing. A range of 4x is more than ample for stretching NTSC's 4:3 aspect ratio to 2:1 for HDTV, while improving central resolution, and losing little 'compatibility'.
2.b.i. The focal-center position should be frame-by-frame selectable under processor control (with directorial emphasis up to override: such as in "gauze shots"), in six detentes from, detailed middle with coarse wings, to coarse middle with detailed wings ... plus flat; plus a compatible mode, flat over NTSC's 4:3 midrange, with extended wings.
2.c. STEREO-EYES-ING television has been virtually ignored by all but psychedelics, yet numerous technologies exist for producing a 3-D or binocular image: flood-focusing (in the near future) has focus as well as 3-D view, sexichrome (six colors viewed three-per-eye through white 3-D safety goggles) interchanges with HDTV, and, the mundane synchro-shutter-glasses, or even a stereo-eyes-ed 'SITMAN'. An NTSC raster-contour-scan z-depth-carrier needs only to occupy the 1.25MHz vestigial zone.
2.d. COMPATIBILITY, as an objective, is often over-specified: compatibility with NTSC's black-and-white can allow, freeing-up luminance signal space by compressing the horizontal sync-pulse, extremely. (But color needs that very stable timebase).
Digital signal constellations QAM16 and QAM32 (hull expansion) are not directly compatible: the HDTV receiver must detect and switch modes: incompatibility is unnecessary since compatible QAM32/16 exists: it is a QAM16-like base with offset-intersticial QAM32-like constellation of fitted-near-points, that can be alternatives with some margin of energy certainty, and provide for a more graceful degradation of SNR. QAM64/16/4 is more efficiently compatible in this 2-dimensional application.
Also, I'd really like to encourage definition of near-compatible digital signal, in lieu of QAM, so that present TV receivers can discern the picture (albeit without the quality of NTSC) ... as 'communications' is a significant portion of TeleVision.
Encouragement must be given and received, for continuing the research into signal pre-processing, technology up-grading, and striving for compatibility.
[America seems to have forgotten not what it had possessed, but what had been foreseen ... Yesterday's done: Let's not waste the perception we've attained: This insight and its method is part-and-parcel to our future in space-navigation: our design methods must always illumine the right answer, fitting specification as well as future development and previously discerned possibility: we must have not merely the art, but the science of upgrades]
For the farther future of digital high definition video, a note on fully interleaved video scanning is appended.
A premise discovery under the title,