Christopher

I'm new to Jitter and am making a patch that tracks all vertical motion on the left and right side of an incoming image via [jit.qt.grab 320 240]. If I use [jit.matrix] to down-sample to a 2 x 240 matrix it simply drops all but the 1st and 160th pixel columns. Any motion in between these 2 columns doesn't register. Using [jit.qt.grab 2 240] seems to work well, except I also need a 320 x 240 image resolution for other parts of the patch, and it seems you can only run one [jit.qt.grab] object at a time. Any other solutions/approaches?

down-size-problem

On May 23, 2007, at 11:26 PM, Christopher Keyes wrote:

> I'm new to Jitter and am making a patch that tracks all vertical 

> motion on the left and right side of an incoming image via 

> [jit.qt.grab 320 240]. If I use [jit.matrix] to down-sample to a 2 

> x 240 matrix it simply drops all but the 1st and 160th pixel 

> columns. Any motion in between these 2 columns doesn't register. 

> Using [jit.qt.grab 2 240] seems to work well, except I also need a 

> 320 x 240 image resolution for other parts of the patch, and it 

> seems you can only run one [jit.qt.grab] object at a time. Any 

I might suggest iterated averaging using jit.demultiplex to sum your 

rows into a smaller matrix. Here's one such patch from the old 

http://www.synthesisters.com/hypermail/max-msp/Sep03/6502.html

And here's one specific to your tasks that shows how to overcome the 

power of two issue for the avg operator (which zero padding doesn't 

> I'm new to Jitter and am making a patch that tracks all vertical  
> motion on the left and right side of an incoming image via  
> [jit.qt.grab 320 240]. If I use [jit.matrix] to down-sample to a 2  
> x 240 matrix it simply drops all but the 1st and 160th pixel  
> columns. Any motion in between these 2 columns doesn't register.  
> Using [jit.qt.grab 2 240] seems to work well, except I also need a  
> 320 x 240 image resolution for other parts of the patch, and it  
> seems you can only run one [jit.qt.grab] object at a time. Any  
> other solutions/approaches?

I might suggest iterated averaging using jit.demultiplex to sum your  
rows into a smaller matrix. Here's one such patch from the old  
mailing list archives.

And here's one specific to your tasks that shows how to overcome the  
power of two issue for the avg operator (which zero padding doesn't  
really deal with).

Down size problem