Re: video size when aspect ratio is not 1

On Mon, 14 Sep 2009 09:47:37 +0200, Philip Jägenstedt <philipj@opera.com>  
wrote:

> On Mon, 14 Sep 2009 08:48:59 +0200, Ian Hickson <ian@hixie.ch> wrote:
>
>> On Thu, 3 Sep 2009, Philip Jägenstedt wrote:
>>> On Thu, 03 Sep 2009 13:47:41 +0200, Ian Hickson <ian@hixie.ch> wrote:
>>> > On Mon, 31 Aug 2009, Simon Pieters wrote:
>>> > >
>>> > > The following paragraph:
>>> > >
>>> > > "The intrinsic width and intrinsic height of the media resource are
>>> > > the dimensions of the resource in CSS pixels after taking into
>>> > > account the resource's dimensions, aspect ratio, clean aperture,
>>> > > resolution, and so forth, as defined for the format used by the
>>> > > resource."
>>> > >
>>> > > ...doesn't say whether to scale up or down when taking into account
>>> > > aspect ratio.
>>> >
>>> > Wouldn't that be up to the video format?
>>>
>>> No, video formats only give the size in pixels and the pixel aspect
>>> ratio (or alternatively frame aspect ratio). The only constraint is  
>>> that
>>> the aspect ratio be correct, which forces us to choose how to achieve
>>> that. Assuming one dimension remains unchanged:
>>>
>>> 1. always scale up
>>> 2. always scale down
>>> 3. always scale x-dimension
>>> 4. always scale y-dimension
>>>
>>> We're suggesting #1. From the rest I've only seen #3 used in actual
>>> media players.
>>
>> Fair enough. I've specced #3 (#3 and #4 are simpler to implement than #1
>> or #2, and the extra complexity doesn't seem to gain us much. I've never
>> heard of anamorphic video data with a ratio less than 1.0, so assuming  
>> my
>> experiences are representative, it's the same as #1 in most cases  
>> anyway).
>
> Actually, standard 720x480 4:3 NTSC DVD video has an pixel aspect ratio  
> of 8:9 and there is no shortage of it. While #3 is easier to both spec  
> and implement, both we and Mozilla have already independently chosen to  
> implement (and write tests for) #1, so I'd much prefer if that were  
> specced. Chrome seems to not do anything and I don't know how to produce  
> a suitable test file for Safari.
>

I just want to note that my math is wrong:

"NTSC 4x3: 10/11 (not 9/10)"

Source:  
http://provideocoalition.com/index.php/cmg_keyframes/story/par_for_the_course/P1/

Still, that's < 1.

-- 
Philip Jägenstedt
Opera Software

Received on Monday, 14 September 2009 21:29:19 UTC