Re: [rtcweb] On video codec for rtcweb

[Cameron Byrne <[e-mail]>]

On Mar 23, 2012 5:49 PM, "Gregory Maxwell" <[e-mail]> wrote:
>
> Roman Shpount wrote:
> > I am not arguing for making H.264 a base codec. I would actually prefer
> > VP8. All I was saying that despite lack of interoperability on
> > encryption, ICE, and identity, there are still significant benefits of
> > supporting the same codecs when communicating with legacy equipment,
>
> This is all true, but there won't always be a gateway or SBC in the media
> path and users may not want one even if all it did was transcode. The case
> with no media gateway can _only_ work reliably by virtue of compatibility
> being baked into the specification— in the other cases interoperability
> is 'simply' a matter of different tradeoffs.
>
> An interesting observation on the cost of transcoding: a number of
> mobile providers are now transparently transcoding their users' http
> video downloads on the fly, not for the purposes of changing the codec,
> but solely to make them smaller. This can do terrible things to the video
> quality (beyond just the reduction in bitrate: MP4 wasn't designed to
> stream from a live source). Perhaps this is an argument for more pervasive
> end-to-end encryption— but the point remains that CPU power, especially

Excellent points both about encryption and codec. This carrier transcoding
mess is a real challenge and should be architecturally thwarted.

Cb

> CPU power in high density data-centers, is still relatively cheap.
>
> _______________________________________________
> rtcweb mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/rtcweb