[foms] Proposal: adaptive streaming using open codecs

[Philip Jägenstedt]

On Tue, 16 Nov 2010 05:54:02 +0100, Frank Galligan <fgalligan at google.com>  
wrote:

> On Mon, Nov 15, 2010 at 6:19 PM, Silvia Pfeiffer
> <silviapfeiffer1 at gmail.com>wrote:
>
>> We also have to look at what makes
>> more sense in the players and what makes sense for small content
>> providers.
>>
> Right that is what I want to do. But before we just went off and made an  
> api
> that was forcing how files should be physically laid out I wanted to make
> sure that is the best option. We have a chance to design
> adaptive streaming so it will work best for content providers as well as
> clients.
>
> Let's assume that interleaved chunks are easier to implement then
> non-interleaved streams for the client (again I'm not convinced this is  
> so).
> Now lets look at it from the small content providers point of view. If we
> have one video and one audio the content provider should be happy  
> (negating
> the issue of manging all the chunks). If we have 6 video streams and 3  
> audio
> streams for one presentation that would be 18 combinations of interleaved
> chunks the content provider is going to have to host vs 9 for
> non-interleaved streams.

Is this actually the case? Wouldn't you actually match low audio quality  
with low video quality, and end up with at most 6 combinations, where the  
each audio track is duplicated twice? In actual deployed solutions, how  
many different bitrates is anyone actually bothering to provide? My guess  
is 3 (low, mid, high) but I truly have no clue.

> That number is going to grow very quickly for
> interleaved chunks once you start adding more streams like text streams.  
> You
> can alleviate the growing number of file combinations for interleaved  
> chunks
> by adding special software on the server side but I don't think this is
> feasible for small content providers. This is why I think for small  
> content
> providers the best physical layout of files for http adaptive
> streaming would be non-interleaved streams.

I don't think captions is a real problem. Duplicating them doesn't really  
impact the bandwidth and it's easy to deliver separately from the  
audio/video tracks anyway (using <track>).

-- 
Philip Jägenstedt
Core Developer
Opera Software