Re: Temporal fragments of media with time stamps

On 15 jan 2010, at 16:56, Bailer, Werner wrote:

> Dear all,
> 
> I had an email discussion with Richard Wright from the BBC, who brought up the following point: If media files have embedded time stamps (as it is possible e.g. in MPEG-2), temporal fragments should use them, instead of just using the offset from the start. This could for example make a difference when time stamps do not start at 0 at the beginning of the file or in case frames have been dropped during recording (i.e. the frame count from the start does not match any more).
> 
> The current specification is not clear, as it does not state whether it considers embedded time stamps if present:
> 
> - For the wall-clock time code I assume it does, otherwise it would have to get the date/time from some other source. 
> - For the normalized play time I would expect to always specify a time offset relative to the start of the file.
> - For the SMPTE time codes, it could consider time stamps if present.
> 
> Maybe you have already discussed that issue, then the specification should state how it is handled.


We discussed this early in the process. This thread <http://lists.w3.org/Archives/Public/public-media-fragment/2008Oct/0003.html> has part of that discussion (even though the initial message was about something slightly different).

But: I can't remember whether we actually reached a decision about embedded timestamps. One of the ideas was indeed to follow a scheme like you suggest, but I seem to recall there were also people who weren't in favor...

Anyone's memory better than mine? 
--
Jack Jansen, <Jack.Jansen@cwi.nl>, http://www.cwi.nl/~jack
If I can't dance I don't want to be part of your revolution -- Emma Goldman

Received on Friday, 15 January 2010 21:37:03 UTC