- From: Davy Van Deursen <davy.vandeursen@ugent.be>
- Date: Thu, 28 Apr 2011 08:18:12 +0200
- To: David Singer <singer@apple.com>
- CC: Jack Jansen <Jack.Jansen@cwi.nl>, Media Fragment <public-media-fragment@w3.org>
On 26/04/2011 18:38, David Singer wrote: > > On Apr 26, 2011, at 0:06 , Davy Van Deursen wrote: > >> Hi David, >> >> On 22/04/2011 2:03, David Singer wrote: >>> >>> On Apr 19, 2011, at 23:56 , Davy Van Deursen wrote: >>> >>>> Jack, >>>> >>>> On 15/04/2011 9:30, Jack Jansen wrote: >>>>> Great, everyone seems to agree! >>>>> >>>>> I especially like the way David phrased it (for NPT start-of-media is deemed to be t=0, regardless of any timestamps embedded in the media), but I can't find anything similar to this in the spec. At least, not around section 4.2, where I would have expected it. I suggest we add it (with suitable emphasis). >>>> Not sure I agree on that :). Taking into account our Media Resource Definition [1] and the various discussions we had on for instance which pixels do we mean in spatial fragments [2], temporal (and also spatial) fragments needs to be applied after decoding the media. Suppose we have an MP4 file where the first timestamp is 10s. When we play this file, you will see nothing the first 10s. Thus, applying #t=5 to this file will result in 5s nothing followed by the first frame (this is how I always interpreted the spec ...) >>>> >>>>> This also means that the test cases (which, incidentally, aren't listed on the homepage:-), TC0007-UA, TC0007-S and friends, need to be revised. >>>> You're right, after reading what I wrote above and reading TC0007-UA, I finally understand the statement "for NPT start-of-media is deemed to be t=0, regardless of any timestamps embedded in the media", which means that I actually agree with everyone, I think :). I will fix the related test cases! >>>> >>> >>> Seems like we're agreed, so maybe I should let it rest. But I am not sure what you mean by an MP4 where the first timestamp is 10s. MP4 files time everything relatively, so (unless they have a SMPTE time-code track, and that's a separate case) they don't have timestamps T1, T2 etc. but only T2-T1, T3-T2 etc., against a presumptive zero origin. >> Ok, what I actually mean is that the first frame of the movie appears after 10s. For instance, within an MP4 container, you can obtain this effect by using an edit list. For example, if you play the mp4 file at [1] with QuickTime, the first frame appears after 2 seconds. > > I would say that the media (the MP4 file) starts at zero with a deliberate two seconds of nothingness, myself. *Why* it would be authored that way is not clear (well, maybe the music builds to a crescendo, and finally the video starts and our illustrious chairman walks on...). > > So, the first timestamp is still 0; just nothing visual appears for 2 seconds... Agreed, which means that TC0007 and friends are redundant test cases. Best regards, Davy -- Davy Van Deursen Ghent University - IBBT Department of Electronics and Information Systems - Multimedia Lab URL: http://multimedialab.elis.ugent.be/dvdeurse
Received on Thursday, 28 April 2011 06:18:40 UTC