Re: [rtcweb] H261/MPEG-1 video quality

Dear Stephan,

thanks for highlighting the differences between H.261 and MPEG-1 Part 2.

Regarding just coding efficiency, I guess that apart from the refined 
coding tools in MPEG-1 much of the difference in practice would stem 
from outdated or buggy implementations: I simply cannot find a modern 
H.261 encoder with, e.g., rd optimization on mode decisions or trellis 
quantization (ffmpeg has these options, but crashes when enabling them 
for H.261). H.261 has clearly been neglected in this regard, making it 
difficult to assess the quality that actually can be achieved.

Best regards,

Maik

Am 15.11.2013 02:37, schrieb Stephan Wenger:
> Folks,
> Please don’t consider H.261 and MPEG-1 part 2 as being in the same
> league in terms of coding efficiency or network friendliness.  They
> clearly are not.
> H.261 is what many call the first generation video coding standard.
>   MPEG-1 (and MPEG-2) are second generation.
> MPEG-1 has half-pel motion compensation.  H.261 has not.
> MPEG-1 has B frames.  H.261 has not.
> MPEG-1 has (arbitrary sized) slices that can be used for MTU size
> matching (although they are not commonly used for that purpose).  H.261
> has not.  Instead, H.261 has the Group Of Block picture segmentation
> mechanism, that is clearly more optimized for parallel processing than
> for MTU size matching.
> MPEG-1 allows for significantly larger motion vectors (necessitated by B
> frames and the resulting longer prediction interval, but can be used
> even in P frame only coding).
> MPEG-1 has arbitrary picture sizes.  H.261 allows QCIF, CIF, and 4CIF
> (in “still image” mode, designed for low frame rate application; could
> run at high frame rate though).
> H.261 was ratified (in its first version) in 1988, and in the for all
> practical purposes final version in 1989.  Most people believe that all
> related patents have expired.
> MPEG-1 was ratified in late 1992.  Its “bug fix” successor MPEG-2 (which
> adds interlace support) was ratified less than a year later.  There are
> at least two major disputes going on today regarding technology
> allegedly infringed by a compliant implementation of MPEG-2.  Based on
> my technical understanding, one of these technologies is not in any way
> related to interlaced.
> Draw your own conclusions.
> Regards,
> Stephan
>
>
>
>
> From: Adam Roach <adam@nostrum.com <mailto:adam@nostrum.com>>
> Date: Thursday, 14 November, 2013 at 15:22
> To: "rtcweb@ietf.org <mailto:rtcweb@ietf.org>" <rtcweb@ietf.org
> <mailto:rtcweb@ietf.org>>
> Subject: Re: [rtcweb] H261/MPEG-1 video quality
>
> On 11/14/13 17:16, Adam Roach wrote:
>> # At 74 seconds and 4.7 MBytes (i.e., 37.6 Mbits), this encoding works
>> out to 508 kbits/second total.
>
> Whoops, I messed up my math. It's 148 seconds long, not 74 (Quicktime
> seems to divide it by two for some reason, although the javascript
> decode does the right thing). This works out to 254 kbps.
>
> /a
>
>
> _______________________________________________
> rtcweb mailing list
> rtcweb@ietf.org
> https://www.ietf.org/mailman/listinfo/rtcweb
>