Jazz

I am making a project for uni which tracks colours and I am trying to filter-out colours which I don't want to be included.

I have used the jit.op object to remove the background so i am left with the person and a black background. The person has different coloured pieces of felt attatched to his hands and feet and my patch so far is tracking the colours fairly well. However, I need to be able to track more efficiently for this project.

I am using the jit.findbounds object to select the minimum and maximum colour range I want to track, is there a way to boost just the 'band' of colour I want to track, making it easier to detect?

Also, I am trying to filter out low amounts of colour that conflict with the colours being tracked. I know it's possible to make a low-pass filter using jit.op but im a bit confused by the help files and examples in the max/msp folder.

Any help I could get regarding the above would be very much appreciated.

I have used the jit.op object to remove the background so i am left with the person and a black background.  The person has different coloured pieces of felt attatched to his hands and feet and my patch so far is tracking the colours fairly well.  However, I need to be able to track more efficiently for this project.

Also, I am trying to filter out low amounts of colour that conflict with the colours being tracked.  I know it's possible to make a low-pass filter using jit.op but im a bit confused by the help files and examples in the max/msp folder.

jitter-colour-filter

brcosa, or scalebias will increase the 'contrast' and boost the 

colors, as will operands of jit.op, a lowpass flter would be 

something akin to jit.op @op > @val 20, which would not pass any 

value less 20 out of 255 (assuming char data)

You could also do a jit.unpack to split each of the color planes into 

On Dec 27, 2006, at 7:17 PM, Ben Harvey wrote:

> I am making a project for uni which tracks colours and I am trying 

> to filter-out colours which I don't want to be included.

> I have used the jit.op object to remove the background so i am left 

> with the person and a black background. The person has different 

> coloured pieces of felt attatched to his hands and feet and my 

> patch so far is tracking the colours fairly well. However, I need 

> to be able to track more efficiently for this project.

> I am using the jit.findbounds object to select the minimum and 

> maximum colour range I want to track, is there a way to boost just 

> the 'band' of colour I want to track, making it easier to detect?

> Also, I am trying to filter out low amounts of colour that conflict 

> with the colours being tracked. I know it's possible to make a low- 

> pass filter using jit.op but im a bit confused by the help files 

> Any help I could get regarding the above would be very much 

brcosa, or scalebias will increase the 'contrast' and boost the  
colors, as will operands of jit.op, a lowpass flter would be  
something akin to jit.op @op > @val 20, which would not pass any  
value less 20 out of 255  (assuming char data)

You could also do a jit.unpack to split each of the color planes into  
individual greyscale matrices.

>
> Hi,
>
> I am making a project for uni which tracks colours and I am trying  
> to filter-out colours which I don't want to be included.
>
> I have used the jit.op object to remove the background so i am left  
> with the person and a black background.  The person has different  
> coloured pieces of felt attatched to his hands and feet and my  
> patch so far is tracking the colours fairly well.  However, I need  
> to be able to track more efficiently for this project.
>
> I am using the jit.findbounds object to select the minimum and  
> maximum colour range I want to track, is there a way to boost just  
> the 'band' of colour I want to track, making it easier to detect?
>
> Also, I am trying to filter out low amounts of colour that conflict  
> with the colours being tracked.  I know it's possible to make a low- 
> pass filter using jit.op but im a bit confused by the help files  
> and examples in the max/msp folder.
>
> Any help I could get regarding the above would be very much  
> appreciated.
>
> Many thanks

Did you look at the CV objects. They have a load of object recognition and tracking externals, which work great in the new versions:

Did you look at the CV objects.  They have a load of object recognition and tracking externals, which work great in the new versions:
http://www.iamas.ac.jp/~jovan02/cv/


Also I might suggest using blue or green fabric and dropping out the red channel and boosting the blue and green in jitter, which would eliminate alot of the person.

that CV.Jit motion tracker is really top notch, I played with it again last night and I highly suggest it.

Jitter Colour Filter