[Matroska-devel] suggestion: method to enable auto lossless pan&scan

Jorge Fuente-Alba the_madman at softhome.net
Wed Jun 8 05:59:47 CEST 2005


Hello, this is a suggestion and my first email to a MKV mailing list. 
Searched the archives and didn't find anything like this, so here goes:

I was looking for a particular aspect ratio and found some of the 
widescreen advocacy websites, and that reminded me that mom still hates 
letterboxing in her movies, and I guess there are a bunch of people out 
there who do too....
So the idea is to add a stream of two coordinates per frame, that point 
to the center of the area of  focus in each frame, so a video player can 
do automatic pan&scan and have the frame follow the action.
You'd have both a proper pan&scan and a widescreen version without 
duplicating image data... the resulting stream should be smallish, too.
Using that info a player could "fill the frame" and follow the center of 
action. Sounds doable? useful? this would also help when viewing 4:3 
footage in widescreen and wanting to fill the screen (tilt&scan, not a 
big problem.. yet).

About creating said stream, people who want pan&scan would like the 
support to be there. We already see MKV videos with multiple languages 
translated by hand and multiple audio tracks, this would be another 
feature among those. Subtitle apps are halfways there with the interfase 
to create this kind of streams. While splines or bezier curves would be 
easier to extract and edit, the would be harder to implement, so here 
goes for simple coordinates. (splines and keyframes would probably be 
useful during the creation of the stream though)

With help from IRC I made a more detailed description:

there should be enough data blocks so that the sum of all d across the 
stream equals the number of video frames.
each block should contain
x: integer value, representing the horizontal coordinate of the center 
of the focus area (should be inside the movie frame)
y: integer value, representing the vertical coordinate of the center of 
the focus area (should be inside the movie frame)
d: integer value, representing the number of consecutive frames the 
focus center applies to.

Thanks for your time :)
and thanks for mkv!!



More information about the Matroska-devel mailing list