Create a custom error function

[sagi21805]

the error function will use the 3 r,g,b channels, and will take the individual error of each channel and then multiply it by the part of that channel over all (weighted error)