Re: Assumption in the SVG specifications

At 15:13  +0200 31/08/06, Chris Lilley wrote:
>
>DS> whereas the 'animation' tag seems to ask a completely orthogonal question
>DS> i) is it coded using vectors or rasters?
>
>In fact the image element also asks that, in 1.2.

agreed.

>
>DS> The first three questions enable the client to know 'does it use
>DS> screen real-estate?', 'do I need to connect to an audio output
>DS> device?' and 'do I need to manage the content handler providing a
>DS> visual presentation that changes over time, might the user want
>DS> (perhaps) at least a play/pause control?'.  They are all questions
>DS> about how the content is *presented*.  The animation tag seems to be
>DS> asking about how it is *coded*, by contrast, and not surprisingly,
>DS> this raises questions.
>
>Once could phrase it as 'do I need to set up a nested viewport' and, 
>for Fullbut not Tiny, 'do I need to clip content outside that 
>viewport'.
>
>Raster images, in general, do not have content outside their width 
>and height. By definition.

Hold tight.  All content has an intrinsic width and height -- the 
bounds of what it draws -- and a desired drawing region.  whether one 
scales to that region or clips to it (or clips before asking this 
question) is still not dependent on the drawing technology.

I don't see that all vector formats use nested viewports, either.

Dang it, QuickTime used to include a 'curve' codec which could be 
used to draw pictures, or (as a sequence of I-frames, effectively) 
movies.  It was just another 'image' codec for the rest of the 
machinery, as far as the implementation was concerned.  Are you 
really saying that when I embed "todays-entertainment.mov" from 
LaughForToday, Inc., I have to call them up and ask them to promise 
either to use the curve codec always or never, or my SVG will be 
incorrect sometimes?  This is truly bizarre.  (Since I must use 
'animation' if they use the curve codec, but 'video' otherwise).  The 
user shouldn't even be able to tell if the same piece of content was 
rendered to a sequence of (say) PNG images or left encoded as curves 
(unless he watches the bit-rate and CPU consumption).


>They are also not timed elements (this means that MNG, animated GIF, 
>etc would go on a video element, incidentally).

agreed, they have time-variant behavior.

>
>DS> I have to say I am quite surprised that the answer to the question
>DS> "what tag should I use to display an illustrator image" is not
>DS> "image" but "animation", but that the answer would change if I
>DS> rendered the image *to the same visual quality and effect* to a PNG
>DS> file and then embedded that (the user should be able to detect no
>DS> difference at all).

OK, now I am checking more.  SMIL 2.1 says something else:

animation
Animated vector graphics or other animated format

img
Still image, such as PNG or JPEG


 From which it is clear at least that non-animated images are, um, 
images.  The ambiguity on animation vs. video remains.
-- 
David Singer
Apple Computer/QuickTime