Escaping from QT

[François Chaplais]

Le 16 sept. 09 à 21:18, stephen barncard a écrit :

> Animated gifs are terrible for anything with much resolution.
> 256 colors, indexed.
> Animated gifs are good for logos and block images but not  
> gradients.  And
> they still stall in Rev if anything else is going on.
>
> Compuserve invented and owned the patents on the .gif format up  
> until a few
> years ago. Early on Compuserve tried to collect on every use of the  
> every
> gif.  It was a major reason for the superior open source PNG format  
> to be
> created.
>
> On the horizon, working today: ANIMATED PNG<http://en.wikipedia.org/ 
> wiki/APNG>
>

I have to add that considering a video as a simple sequence of images  
is a very inefficient concept. Most videos, most of the time, only  
feature slight variations from one frame to another. Hence the idea  
of including "key frames" with details and intermediary variations in  
typically smaller smaller chunks of data. Detecting abrupt changes  
from an image to the following one is a challenging task. The  
following example shows that this notion of "abrupt change" may be  
somehow subtle.

 From what I remember of the draft of the future-MPEG-to-come in the  
late 90's, it was supposed to include information about "video  
objects". For instance, if you look at a football match (whatever  
flavor) chances are that you will see interesting action in the  
"foreground", i.e. players with fast movements which demand great  
detail, and a more or less static background (supporters) which is  
less interesting and hence, does not require high fidelity. The then- 
future MPEG was supposed to separate the two components and encode  
them differently. I do not know if this was ever implemented.

cheers
	François