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

On Mon, Feb 4, 2013 at 10:59 AM, David Ronca <dronca@netflix.com> wrote:

>  > Required by whom? ****
>
> ** **
>
> It is our understanding of the new FCC requirements, that all devices must
> present a given caption in a consistent manner.
>

Could you provide a reference to "the new FCC requirements"? Is "consistent
manner" defined?


> If the author uses CEA-608 or any format that works on a 4:3 presentation
> area, then by definition, the presentation area is a region that is 100% of
> the video height, and H*4/3 width, and this window is centered over the
> video.
>

That doesn't follow by my reading. If a 4:3 media object is captioned for
4:3 display, and some end user device scales that media object for 16:9,
e.g., by linear scaling in horizontal, then I would expect the caption
regions to be repositioned and resized by the same horizontal scaling.

That is precisely what happens with TTML when using percentage values to
define origin and extent of non-root regions.


>   This is crucial because the actual screen location of the position
> 10%,10% is different for 4:3 and 16:9 presentation area.
>

Correct, but the position of the non-root region with respect to the
related media object is not different when scaling is applied to percentage
length specifications. In fact, if a device scales 4:3 video to 16:9 but
does not scale the non-root caption regions by the same amount, then the
author defined positions of captions with respect to the underlying video
are no longer registered (i.e., no longer in correspondence with the
original video position when the video is unscaled).


> This is independent of the display aspect ratio of the video.  If a client
> knows the presentation AR, and percentage or cell positioning is used, then
> the captions can be scaled and presented to fit in the video window
> properly without any use of pixels or “external context”.
>

Yes it could, but we are still questioning the rationale for this feature
as you have described it. Thus far, I don't see the requirement for
non-scaling the root display region while scaling the related media. If you
can point at specific language in the FCC rules that makes this clear, then
please do so. Otherwise, I have to ask whether your requirement is really a
requirement or not.

G.