Re: [rtcweb] Stephan Wenger's choices

[Ron <[e-mail]>]

On Sat, Dec 28, 2013 at 11:06:53PM +0000, Stephan Wenger wrote:
> Interestingly enough, H.261 *never* had a set of picture parameters that
> bear a relationship with source formats dominant in the market.
> CIF has 288 lines, QCIF 144, both distinctly non NTSC world.  OTOH, the
> frame rate is 1/29.97 Hz fractions thereof, definitely not PAL/SECAM.

Let's raise the curtain on the full matrix here shall we (:

 NTSC (visible) frame:  352 x 240 @ 29.97 Hz
 PAL  (visible) frame:  352 x 288 @ 25 Hz

 CIF:                   352 x 288 @ 29.97 Hz

So yes, as you point out, if you select the poorest performance parameters
from both worlds, then CIF does indeed not use those.  It instead actually
selected the highest performance constraints from each.  What a surprising
thing to do.  I'm shocked.

So you get all the resolution of PAL, with the higher frame rate of NTSC.

> This is the result of a standardization compromise of the worst sort.

... which I'd hardly consider to be the _worst_ compromise they could
have thought to make.  It's not like we're talking 44.1 audio here.

Especially given the hardware it was expected to be running on, where
dropping a few frames to get to the lower refresh rate would be trivial
even if not quite ideal - and where losing a few visible scan lines was
something that almost every CRT already did anyway, so dropping a few
extra was unlikely to be terribly noticeable.

Both format conversions lose a little, but not nearly so much as they
would have lost if they were forced to repeat the A-law/mu-law farce
for something as expensive to transcode on 20 year old hardware as
video.  Sounds like a pretty solid and well considered decision to
me, even with the benefit of hindsight to second guess them.


> Note that this hasn¹t hindered the adoption of H.261 much.  Perhaps
> because, at the time, there were literally no alternatives.

So kind of exactly like what we're faced with here today then ...  ;?

We wouldn't have even got around to thinking about this as an option
if some people didn't firmly claim that their belief about the FUD
surrounding VP8 was showstopper for them, and if the situation wrt
H.264 wasn't such a plainly obvious minefield wrapped in a pratfall
inside a volcano.

If we continue to accept the belief in the FUD about VP8, then we do
indeed have literally no alternatives.  Except perhaps for snatching
utter failure from the hands of disappointment and ignoring a perfectly
viable MTI to declare there should instead be none.

So we've come nowhere since 1988.  Sometimes that happens in engineering.
The smart engineers accept that and make do with what they actually have.


Can we start being smart engineers now?  Pretty Please?

  Ron