ed dowie

I am currently working on a project where I am using Jitter for colour-tracking. I have 4 different colour balls which are being tracked with jit.findbounds. Whenever I set up the system, I have to spend a while configuring the different colours (depending on the light of the room) so that Jitter knows which colours to track. Does anyone know a way to create a white balance, so that I simply have to show the camera a white object, and it'll work out what colours is should be looking for?

white-balance

below is one quick&dirty way to do something similar.

Thanks ever so much for the patch Nesa! (sorry about the delay in replying, I've had problems with my internet). I have to confess that I don't understand what this patch is doing at all!! I've been doing quite a lot of head-scratching trying to work it out. If you don't mind, could you give me an idea of what is happening conceptually in the patch? I'm sorry for my ignorance, I'm very new to Jitter to be honest.

Sorry for not explaining it back then, I was in a bit of hurry...

This patch assumes that pure white on computer screens should be presented 

as red, green and blue channels at maximum: 255 in this case.

So, if you are grabbing an pure white pixel, you should get rgb values of 

255 255 255, but if the image is a bit greenish, you would get something 

like r200 g255 b207 (only green at maximum), or bluish: r203 g201 b255, etc. 

So to make that pixel white, you can scale non-maximum channels so that they 

are at maximum(greenish example: r128, g255,b128 means that you have to 

multiply red and blue by two to reach 255...or you could divide green by 

So, the patch first grabs the image, takes a slice of it at the 

center(jit.submatrix), then it analyses the image using jit.3m and gets the 

current maximum values of red, green and blue channels.

After that it calculates scaling values, and sends them to jit.scalebias 

that does the channel scaling on the image.

1) Adjust graphical gate object depending on your prefered grabing 

now you should see captured image, with a green square in upper window. this 

green rectangle shows what part of matrix will be analysed, so you should 

put your white object in front of camera, and make it cover entire green 

4) turn the gate on for a moment, to calcuate scaling values(while white 

thing is covering green square), and turn it off

you should see the results in lower window afterwards.

This patch assumes that pure white on computer screens should be presented 
as red, green and blue channels at maximum: 255 in this case.
So, if you are grabbing an pure white pixel, you should get rgb values of 
255 255 255, but if the image is a bit greenish, you would get something 
like r200 g255 b207 (only green at maximum), or bluish: r203 g201 b255, etc. 
So to make that pixel white, you can scale non-maximum channels so that they 
are at maximum(greenish example: r128, g255,b128 means that you have to 
multiply red and blue by two to reach 255...or you could divide green by 
two).
So, the patch first grabs the image, takes a slice of it at the 
center(jit.submatrix), then it analyses the image using jit.3m and gets the 
current maximum values of red, green and blue channels.
After that it calculates scaling values, and sends them to jit.scalebias 
that does the channel scaling on the image.

Step by step tutorial:
1) Adjust graphical gate object depending on your prefered grabing 
object(usually dx for xp, qt for osx);
2) Open the graber;
3) Turn qmetro on;

now you should see captured image, with a green square in upper window. this 
green rectangle shows what part of matrix will be analysed, so you should 
put your white object in front of camera, and make it cover entire green 
square;

4) turn the gate on for a moment, to calcuate scaling values(while white 
thing is covering green square), and turn it off
you should see the results in lower window afterwards.

I see you're based in Leicester, so am i, so i've dropped you an email to see if you want to meet up and possibly form a local patching circle?

Thanks so much for the explanation Nesa (as well as the patch!), that's really helpful!

White Balance?