🚀 go-pugleaf

In article <727706283snz@chaz.demon.co.uk>, Ashley@chaz.demon.co.uk
(Charles Ashley) wrote:

> I've seen C-Cube's MPEG board running on a PC delivering MPEG sound and
> video at 150kb/sec.  It looked really good; the chips and board are avaliable
> now I believe.  

One thing to keep in mind is that, although MPEG1 decoders are starting to
appear, MPEG1 *encoding* has been very slow to develop. Because the
potential mass market is in the decoders, most vendors seemed to have
treated the encoding part as an afterthought. Therefore, MPEG1 encoding
technology is well behind the decoding hardware. Today, the only MPEG1
encoding solutions available are either relatively low quality or are
extremely expensive and nowhere near real time (between 50 and 400 times
real time is typical).

This situation is important because the quality of an MPEG1 decoder is
largely dependent on how well the encoding was done. Much of the stuff
shown in demos is the result of painstakingly optimized encoding (manual
insertion of I frames, very thorough motion-vector searching), which will
not be economically feasible for many users. In essence, those great
looking MPEG1 demos that most people have seen at trade shows probably
represent the upper bound for MPEG1 performance. It remains to be seen what
the standard will do in the real world.

C-Cube and a number of other companies are working on real-time and
near-real-time MPEG1 encoders for 1993, but there are sure to be
cost/quality tradeoffs. Therefore, although a decoder may look good in one
demo, it may look substantially worse with differently coded material. The
key is the encoding.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Steve Krause, SRI International, 333 Ravenswood Ave., Menlo Park, CA 94025
Phone: 415-859-4746  Fax: 415-859-4544  Internet: steve_krause@qm.sri.com