- From: Calogero Alex Baldacchino <alex.baldacchino@email.it>
- Date: Mon, 24 Nov 2008 23:46:55 +0100
--------- Original Message -------- Da: "Maik Merten" <maikmerten at googlemail.com> To: "WHATWG Proposals" <whatwg at lists.whatwg.org> Oggetto: Re: [whatwg] media elements: Relative seeking Data: 24/11/08 08:45 > Eric Carlson schrieb: >> QuickTime has used this method this since it started supporting VBR >> mp3 in 2000, and in practice it works quite well. I am sure that there >> are degenerate cases where the initial estimate is way off, but >> generally it is accurate enough that it isn't a problem. An initial >> estimate is more likely to be wrong for a very long file, but each pixel >> represents a larger amount of time in the time slider with a long >> duration so changes less noticeable. > > Well, I do believe this works fine for audio (which usually hasn't a > wildly fluctuating bitrate if you e.g. average over a second or two), > I'm mostly concerned about video. An example for an outrageously off > estimate would be the trailer for "Generic space-pirate movie". > > The first few seconds would be mostly a static green/red/yellow/whatever > screen ("This pirate movie has been rated ARRRRRR!") - this part would > be coded with like 100 kbit/s or less. The next few scenes (this is a > trailer, after all) would mostly show exploding ships, genetically > engineered mutant parrots attacking space-adventurers and a few cuts > into random love scenes - so this part can be multi-megabit/s. After > this the bitrate would dramatically decrease again as the last few > seconds will just show "Summer 2010". > > Does QuickTime also handle such content gracefully (e.g. display a > position slider that doesn't jump around wildly)? Am I overestimating > the problem? > Maik The slider should just indicate a relative position (i.e. a percentage) between 0 and the (currently known) duration of the content, which may be estimated with a variable average time, perhaps retarded at the beginning, and varied according to the bitrate variation with some euristic, to make the computation more accurate (or maybe a few consecutive evaluation, at fixed and rapid intervals, could be averaged to get a better value, before updating anything), so no "crazy horse jumping" should happen. Silvia Pfeiffer has proposed a 'length' attribute to indicate the overall duration in the markup, and I think its value could help to improve accuracy, even when wrong. -- Email.it, the professional e-mail, gratis per te: http://www.email.it/f Sponsor: CheBanca! La prima banca che ti d? gli interessi in anticipo. Fino al 4,70% sul Conto Deposito, zero spese e interessi subito. Aprilo! Clicca qui: http://adv.email.it/cgi-bin/foclick.cgi?mid=7919&d=20081124 -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.whatwg.org/pipermail/whatwg-whatwg.org/attachments/20081124/3aaebb0b/attachment.htm>
Received on Monday, 24 November 2008 14:46:55 UTC