All-
Oh yeah...and reduced latency is a real win as well.
Phil
Peter Marshall wrote:
>
> On Tue, 23 Feb 1999, Tyler Johnson wrote:
>
> > For videoconferencing, I believe you can get pretty much equivalent
> > results from MPEG2, although quite frankly, we are seeing users pleased
> > with much lower quality than that.
>
> I thought that one of the advantages of a non-temporally
> compressed stream was that it introduced significantly less latency
> into the stream. This is a decided advantage for some video
> conferencing applications. DV looks like it might be a fairly
> inexpensive way to tap into some hardware that does realtime
> compression with very little introduced latency.
>
> > Further, we are actively discouraging the use of motion JPEG-like
> > codecs for videoconferencing because they use no temporal
> > compression (I suppose DV is like mezzanine MPEG2, can someone
> > confirm that).
>
> I'm not sure that I understand the term "mezzanine MPEG2". DV is
> basically a 5:1 compression (which I have seen referred to as
> mezzanine level). DV has been described to me as using "essentially"
> the MPEG2 algorithms for image compression but it keeps each frame
> independent. Basically a stream of index frames. But if it was that,
> then wouldn't it be standard MPEG2 -- or are difference frames
> required in that standard? So there must be a bit more in there than
> that (or perhaps it is just all the signalling around it that makes DV
> DV rather than MPEG2.)
>
> > Further, this doesn't really represent a standards-development
> > direction including call signalling, tracking, etc.
>
> I think that there are a number of directions that this work could go
> in once the basic idea is tested and understood. One might be to work
> in the DV codec into some other signalling and tracking standard.
> We shouldn't need too much of existing schemes to integrate in a new
> codec.
>
> I think that there may be some application for this technology in
> high-definition, low-latency, high-fidelity video conferencing. One
> example that I've been working on is to use the facility for music
> instruction.
>
> > So that leaves video production. We do use some DV for video production,
> > but are focusing on SMPTE 259M and SMTPE 292M for that. DV isn't high
> > enough quality for many applications, being very compressed, and it
> > doesn't support HDTV. Even the news folks, which are the bottom of the
> > heap in my book, are using souped up DVCPro running at 50mb/s (100 mb/s
> > for a two-way conference).
>
> I'm not sure where the loss of quality is, but for most TVs and TV
> viewers, DV quality video (if delivered cleanly) would be a
> significant step up in quality. So not much of all of that SMTPE 292M
> is getting through! FYI: consumer DV runs natively at about 40mb/s.
>
> > So I guess I am curious about just how folks see this being used.
> > Obviously it's a big bitstream to stress the routers with, so I think
> > that's legit. Is someone envisioning an application or virtual network of
> > this? I might like to participate if so.
>
> I think that a big bitstream might well be part of the project. But
> it is also trying to use a technology that provides significantly more
> resolution in an frame editable stream, from consumer-oriented
> equipment (at consumer oriented prices). It may well be that in the
> next couple years, real (and reasonably priced) MPEG2 cameras will be
> available. They'll probably blow DV out of the water. Then we can
> substitute them for the DVs and have an even more flexible system.
>
> Peter Marshall, Advanced Networks Applications Specialist
> CANARIE Inc, c/o ARDNOC, #560-55 Metcalfe St, Ottawa, Canada K1P 6L5
> 613-781-0667 fax: 613-234-7488 http://tweetie.canarie.ca/~marshall/
This archive was generated by hypermail 2b29 : Thu Mar 23 2000 - 03:35:20 CST