JPEG/MPEG standard development

[Larry Rowe]

I talked to the head of the US committee many of whose members participate
on the JPEG and MPEG standards committees. I didn't realize that these
standards are being developed by international committees which explains
why they have been slow and not particularly visible. I learned a lot
about the state of the draft standards documents that you might
find interesting.

First, JPEG -- the still image standard. A draft standard document has been
written and it is being distributed by the X3 Secretariat folks in
Washington DC. Call Lynn Barra at 202-737-8888 (hit 11 on touchtone phone
to get operator and ask for Lynn). She can send you a copy of the draft
specification for $8.00. The person I talked to, Charles Touchtone at IBM
in Tampa, said that they have to go through three ballots, each of which
takes 6 months, so it will be about 18 months before the standard is
official. However, as many of you probably know, companies are producing
products based on the draft specification today.

Second, MPEG -- the motion video standard. This committee is much further
behind the JPEG committee. They are currently working on a draft document,
but nobody outside the committee has seen the current drafts. Supposedly,
the specification will have three parts: video, audio, and system. The
system part specifies how video and audio packets are combined. At present,
a draft exists for the video part, an outline exists for the audio part, and
nothing has been written for the system part. So, it will be quite a while
before we see a complete specification. Touchtone said that the video part
was based on a CCITT standard (ptime 64?) rather than the JPEG standard.
This is interesting because the video compression hardware that I'm familiar
with is based on JPEG (e.g., Next video board).

I'm going to try to follow what's happening on these committees, so I'll
keep you informed.
Larry Rowe

[Michael Korcuska]

In article <12...@milton.u.washington.edu> woo...@milton.u.washington.edu (Woobin Lee) writes:

[A whole bunch of interesting and informative stuff on video compression,
including Larry Rowe's post on JPEG/MPEG standards]

So, you two seem to know quite a bit about this compression stuff and digital
video in general. Here at Northwestern we're going to make a move to some
digital video system at some point next year. The question is, where can we
learn about these things? Are there books/journals I should track down?

If people want to mail me suggestions, I'll be glad to post the summary.

-------------------------------------------------------------------------------
Michael Korcuska The Institute for the Learning Sciences
korc...@ils.nwu.edu Northwestern University, Evanston IL
-------------------------------------------------------------------------------

[Woobin Lee]

In article <93...@pasteur.Berkeley.EDU> la...@postgres.UUCP (Larry Rowe) writes:
>
>Second, MPEG -- the motion video standard. This committee is much further
>behind the JPEG committee. They are currently working on a draft document,
>but nobody outside the committee has seen the current drafts.

I'm not in the committee, but I have a copy of MPEG draft.
I don't think it's confidential, as I didn't sign anything to get it.

>Supposedly,
>the specification will have three parts: video, audio, and system. The
>system part specifies how video and audio packets are combined. At present,
>a draft exists for the video part, an outline exists for the audio part, and
>nothing has been written for the system part. So, it will be quite a while
>before we see a complete specification.

I don't know much about audio, but from the contents of the draft material,
I think the audio part is closest to the finalization.

>Touchtone said that the video part
>was based on a CCITT standard (ptime 64?) rather than the JPEG standard.
>This is interesting because the video compression hardware that I'm familiar
>with is based on JPEG (e.g., Next video board).

MPEG standard is very similar to CCITT, such as in motion estimation,
intra/non-intra decision, DCT/IDCT, and quantization schemes. One major
difference is the addition of Bidirectionally Predicted frames in MPEG
to Intra and Predicted frames of CCITT.
The MPEG standard is intented for full-motion video with storage/transfer media
capable of supporting 1.5 Mbps (CD ROM, DAT, HD etc).
The NeXT board incorporating C-Cube chip is intended for still image
compression. The JPEG standard is for still images, and therefore does not
exploit temporal redundancy as in MPEG or CCITT.

>I'm going to try to follow what's happening on these committees, so I'll
>keep you informed.
> Larry Rowe

I think the committe meets once every two months, and the last meeting was
(or will be?) held in Berlin. I haven't got any report from this meeting yet.

By the way, does anybody know where I can get a copy of CCITT H.261 standard
and their reference model (RM8)?

Woobin Lee -----------------------------------------------------------------
||
Image Computing System Lab || woo...@u.washington.edu
University of Washington ||
Seattle, WA 98195 || (206) 543 - 1017