RE: RE: RE: RE: RE: Re: Re: TTML and aspect ratio

Yes, as I said, it's the answer to the wrong question. You can put the region wherever you want it, but it appears that the real problem is that in your workflow you are unable to generate multiple TTML files, one for each video aspect ratio; if the TTML is presented on a different aspect ratio than the one it was authored for, the content may not show up registered to the same video pixel; as I said previously; we don't have either a conditional mechanism to style regions, or a means to position a region relative to the centre of the video.  Moreover if we were to design such a conditional system, it would have to anticipate all the ways a 16:9 frame can be squeezed into a 4:3 window.

So my point is, even if we were adopt your request to specify the aspect ratio in metadata, it would not fix your problem any better than the current solution of specifying it in the root tt element.

-----Original Message-----
From: David Ronca [mailto:dronca@netflix.com] 
Sent: 04 February 2013 13:16
To: public-tt@w3.org
Subject: RE: RE: RE: RE: Re: Re: TTML and aspect ratio

The dimensions of 1080p are 1920x1080.  Based on your math, the top-left will be 192,108.  But if you present CEA-608 captions on a 1080p video, this is not correct.  CEA-608 defines the width of the caption area as 133.3% of the picture height (caption area height defined as 100% of picture height).  The coordinate 192,108 is outside of the CEA-608 defined caption area.  The CEA-608 caption area is physically mapped to the absolute region L= 240 T=0 W=1440 H= 1080, and the position 10%,10% corresponds to the absolute pixel position 384,108.

From: Sean Hayes <Sean.Hayes@microsoft.com>
Date: Mon, 4 Feb 2013 19:00:49 +0000
RE: RE: RE: Re: Re: TTML and aspect ratio

So assuming the video is  1080 x 609; then the top left of the drawn region will be 108, 60.9 The absolute pixel position of the first character will depend somewhat on the font selected, but it will be offset from that origin.

However that answer is not really picking up on my point that the region should be cantered; and to do that in an aspect ratio agnostic manner requires a central anchor point on regions, which we don't currently have - but which has been proposed.

Received on Monday, 4 February 2013 22:33:51 UTC