Object position in video


I am playing around since hours with the freeframe patches, but I am too unexperienced to get the result I want. Lets say, a person with a red hat is walking through the picture and I want the x/y-coordinates of that hat. How can I divide the video-input into e.g. 12x9 squares and get the square with the most red pixels for each frame?


didn’t you know Pipet (EX9.Texture) ?