calufrax@blake.acs.washington.edu (Chad Fogg) (01/06/89)
The following excerpt is from an article published on page
193, of PC MAGAZINE (January 31, 1989 Vol.8 No. 2). It has
many relevant points to subjects raised recently on the net -
mainly by implication that is.
(* -clarified at conclusion of this posting)
-----------------------------------------------------------------------
GE/RCA's DVI (Digital Video Interactive) technology
directly addresses the bandwidth* problem. DVI uses delta and
ADPCM* techniques, but its special forte is _Asymmetric_
compression. During the off-line preparation of a CD-ROM,
the DVI process requires about three seconds of a powerful
Meiko parallel-processing number-cruncher's time to compress
each image frame at a staggering 120:1 ratio. But at
playback, everything happens in real time: the powerful (12.5
MIPS)* DVI chip set decompresses the image on the fly. The
result: up to 72 minutes of full-screen, full-motion digital
video (at a resolution of 256 by 240, with 16 million
colors)* can squeeze onto a single standard CD-ROM.
Fully digital video can be manipulated with ease -as the
gasps of amazement at DVI demonstrations attest. Truly
interactive TV sets lets you use a joystick to "look around"
in any direction and "walk" at a speed you control.
"Synthetic" video lets you overlay photographic textures and
colors onto 3-D models of rooms and furniture. And DVI
offers much more, including on-the-fly "exact reproduction"
compression for high-resolution graphics and a 10-frame-per-
second "edit level" mode for video development.
DVI's major obvious drawback is that you need a DVI
system to decode its disks. DVI is now available only as a
three-board set for standard 16-bit PC AT slots: one for
video, one for audio, and one to control a CD-ROM drive and
joystick. Currently in beta testing, the boardset is expected
to roll out to the public for $7,000 to $10,000 (depending on
the configuration) in the first quarter [of] 1989. Development
software and support are likely to run another $15,000. It's
fair to guess that developers of high-end interactive video
applications will be major DVI pioneers. But increasing
integration could bring prices down fast.
-Stephen Manes
--------------------------------------------------------------------
some clarifications:
*Bandwidth problem -CD-ROM's slow data transfer rate (~150
KiloBytes per second)
*ADPCM -probably Analog to Digital, Pulse Code Modulation
*MIPS -Million Instructions Per Second
*256X240 is somewhat less than the broadcast NTSC maximum
resolution of 336 (Horiz) by 484 (Vert) lines. However,
much vertical detail is lost to the psychovisual effects of
TV. NTSC analog resolution specifics tend to be vague,
whereas digital parameters are precise. As well, DVI probably
has less than 30 decompressed frames per second for the given
resolution.
-C. Foggmarcel@mlogic.UUCP (Marcel Samek) (01/07/89)
In article <494@blake.acs.washington.edu> calufrax@blake.acs.washington.edu (Chad Fogg) writes: >*ADPCM -probably Analog to Digital, Pulse Code Modulation Actually it stands for: Adaptive Delta Pulse Code Modulation M. -- Marcel A. Samek | Media Logic Incorporated | 2501 Colorado Blvd. Suite 350 ARPA: mlogic!marcel@sm.unisys.com | Santa Monica, CA 90404 UUCP: ...sdcrdcf!mlogic!marcel | (213) 453-7744
rbean@hpindda.HP.COM (Rich Bean) (01/07/89)
> *ADPCM -probably Analog to Digital, Pulse Code Modulation > -C. Fogg I think ADPCM might be: Adaptive Delta Pulse Code Modulation Delta means it encodes the difference between successive samples, which usually takes fewer bits than the samples themselves. Adaptive might have something to do with changing resolution to match the input stream, using the fewest bits possible ... but I'm not sure about that. Rich Bean rbean@hpindbu
regisc@tekgvs.GVS.TEK.COM (Regis Crinon) (01/10/89)
In article <7070001@hpindda.HP.COM> rbean@hpindda.HP.COM (Rich Bean) writes: >> *ADPCM -probably Analog to Digital, Pulse Code Modulation >> -C. Fogg > >I think ADPCM might be: > Adaptive Delta Pulse Code Modulation > >Delta means it encodes the difference between successive samples, which >usually takes fewer bits than the samples themselves. Adaptive might >have something to do with changing resolution to match the input stream, >using the fewest bits possible ... but I'm not sure about that. > >Rich Bean rbean@hpindbu ADPCM stands for Adaptive Differential Pulse Code Modulation. "Pulse Code Modulation" refers to the fact that the input signal has been sampled and quantized into N (generally 8 ) bits per samples. "Differential" means that what is quantized is the difference between two successive samples rather than the samples themselves. More precisely, "Differential" can involve the difference of the current input sample with the output of a prediction filter. "Adaptive" means that the predictor coefficients ( the prediction filter ) can be readjusted by measuring statistics of the incoming samples. "Pulse Code Modulation" and "Delta Modulation" are both well known digital signal processing techniques; So are "Differential Pulse Code Modulation" (PCM) and "Adaptive Differential Pulse Code Modulation" (ADPCM). See any textbook on Digital Communication. I have never heard of "Adaptive Delta Pulse Code Modulation". Regis J. CRINON Digital Signal and Picture Processing Group TEKLABS TEKTRONIX INC. -- crinon
jr@bbn.com (John Robinson) (01/13/89)
In article <4470@tekgvs.GVS.TEK.COM>, regisc@tekgvs (Regis Crinon) writes: >In article <7070001@hpindda.HP.COM> rbean@hpindda.HP.COM (Rich Bean) writes: >>> *ADPCM -probably Analog to Digital, Pulse Code Modulation >>> -C. Fogg >> >>I think ADPCM might be: >> Adaptive Delta Pulse Code Modulation >> > > ADPCM stands for Adaptive Differential Pulse Code Modulation. > "Pulse Code Modulation" refers to the fact that the input > signal has been sampled and quantized into N (generally 8 ) > bits per samples. ADPCM is an international (CCITT) standard for use in telephony. The usual digitized voice, using PCM, requires 64,000 bits/sec of bandwidth. ADPCM halves this to 32,000, doubling the capacity of networks and phone switches. I am not sure that this can apply to DVI as-is, though the general approach might well. The compression can be even better through the use of a variable encoding rate rather than constant as telephone networks require. -- /jr jr@bbn.com or bbn!jr