[opendtv] Re: A detail in the history of video standards
- From: Ron Economos <k6mpg@xxxxxxxxxxx>
- To: opendtv@xxxxxxxxxxxxx
- Date: Fri, 02 Sep 2005 04:01:47 -0700
If by "refined", you mean motion vector resolution, then both MPEG-1
and MPEG-2 provide 1/2 pel vectors (1/2 pel is optional in MPEG-1
and mandatory in MPEG-2).
If you mean more modes available for better prediction, then all the
new modes in MPEG-2 (field prediction in frame pictures, 16x8 and
dual-prime) are field based.
When coding progressive_sequence = 1 in MPEG-2 (for 720p), motion
prediction is exactly the same as MPEG-1 with 1/2 pel vectors.
You can stick your head in the sand regarding interlace, but this
is to nobodys advantage. Look what happened to SMPTE 421M
(VC-1). They tossed in interlace at the last minute and managed
to create an incredibly obtuse bitstream syntax. The SMPTE 421M
(VC-1) specification is twice as many pages as H.264, and does less.
Ron
Craig Birkmaier wrote:
>At 3:16 PM -0700 8/31/05, Ron Economos wrote:
>
>
>>Your statement doesn't make sense. Almost everything that's different in
>>MPEG-2 from MPEG-1 is for interlace. If all video was progressive,
>>we'd still be using MPEG-1.
>>
>>
>>
>
>At the tool level there is some merit to what you say. But MPEG-2
>brought many new things to the table, some of which have never been
>commercialized.
>
>I believe you will find that the interframe prediction tools are more
>refined in MPEG-2 than with MPEG-1. And MPEG-1 is very limited in
>terms of the rasters that it supports, although this is just a
>"paper" limitation. MPEG-2 also includes the systems work that is now
>used to transport MPEG-1, MPEG-2 and MPEG-4 content.
>
>I would suggest a much different outcome if all video were
>progressive. Initially, it would have been possible to improve the
>inter-frame prediction tools as more Mips (wasted on coding
>interlace) would have been available to improve the prediction
>routines. And I suspect that the innovations in AVC would have been
>standardized much sooner as well, if interlace were not a concern.
>
>Software codecs tend to be updated more frequently than the
>algorithms that get mass produced in silicon. Reading between the
>lines of my previous post, it should be obvious that the companies
>that co-opted MPEG-2 have little interest in continuous innovation -
>they wanted, but did not get a standard that would last for decades.
>
>In this case history is NOT repeating itself.
>
>Regards
>Craig
>
>
----------------------------------------------------------------------
You can UNSUBSCRIBE from the OpenDTV list in two ways:
- Using the UNSUBSCRIBE command in your user configuration settings at
FreeLists.org
- By sending a message to: opendtv-request@xxxxxxxxxxxxx with the word
unsubscribe in the subject line.
Other related posts:
