[whatwg] Technical answers (was Re: several messages regarding Ogg in HTML5)

[Manuel Amador (Rudd-O)]

> We'd be better off discussing technical issues that would inform a
> decision;  questions like
> a) what compression ratio/quality is needed?  Is h.261 'good enough'?

I think you should be thinking in terms of user experience, not just 
compression ratio.  If I put my "user" shoes, all I know is that with my 400 
kbits per second at home, I can watch YouTube videos without having to buffer 
first, whereas with Google Video, oops.  Don't fall into the trap of consumer 
net access because those links are INCREDIBLY overcrowded, once you start 
authoring HTML5 content with multimedia, you're going to see a massive 
suicide increase in Comcast's and Verizon's call centers.

> What about audio?

Ten to one is customarily stereo, CD quality in contemporary alternatives like 
AAC and Vorbis.  Extremely demanding people could expect 2:1 with lossless 
compression.

> b) do we need alpha support?

I think it would be nice.  You could honestly build amazing interactive worlds 
with alpha-enabled video.  None of the video compression technologies that I 
know of support alpha -- but I'm not much of an expert in video compression 
anyway.

> c) what audio compression level is needed?  Is IMA 4:1 good enough?

Most certainly not.  Have you heard how IMA ADPCM sounds and what humongous 
files it creates?

> Could we live with uncompressed PCM?

Sure we could.  In our hard disks as an intermediate step between ripping and 
encoding.  Not online, not if you want Senator Stevens to get his internets 
on time.

> d) what container format features are needed?  Incremental files?
> Seekability and indexing?  (These tend to be in conflict, by the way).

I think it's much more reasonable to prefer seekability and indexing to 
incremental files.  After all, users want to seek much more often than they 
want the benefits of incremental files.

> e) what audio channel count is needed?

N.  Mono, stereo, 5.1, 6.1, cater to Ambisonics if you need to.  But the 
minimum requirement for basic online audio should be stereo, and beyond that, 
you would have to go with one of the encumbered traditional Dolby or THX 
speaker distributions.  Oh, it's not just the channel count that is a problem 
here, but how you encode that info on the wire, and how the speakers are 
distributed, and how you downmix in the presence of a less-than-ideal 
listening environment.  Take a look at Ambisonics, it's rather appropriate 
because it offers N routes between stereo/mono/Dolby 5.1, et al, while 
encoding the audio in a speaker position independent way.  And I think the 
patents expired long ago.

> f) what access protocols should be supported?  One assumes http;
> what about rtsp/rtp?  authentication in rtsp?  which rtsp?  shoutcast?

Anything that could be hyperlinked may be supported, but a minimum of http: 
should be there.  Ideally you'd want FTP and RTSP.  Shoutcast is, as far as I 
understand it to be, over HTTP so there you go.

>
> I am sure there are many other questions...

Not that I can think of before having dinner :-)

-- 

	Manuel Amador (Rudd-O) <rudd-o at rudd-o.com>
	Rudd-O.com - http://rudd-o.com/
	GPG key ID 0xC8D28B92 at http://wwwkeys.pgp.net/

Abandon the search for Truth; settle for a good fantasy.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part.
URL: <http://lists.whatwg.org/pipermail/whatwg-whatwg.org/attachments/20071211/4e042292/attachment-0001.pgp>