|Subject:||allow the same matrix to be used for different source/destination channel counts in convert()|
Currently you need different matrix dimensions for each combination of input and output channel counts. This is too much hard work. Optionally, allow a 4 x 5 matrix to always be specified, and adapt that to optimize the output. eg. if the red/green/blue rows are the same, produce a grey output if the alpha channel has 1 in the last column, produce an image with no alpha channel if the source has no alpha, and the alpha row only has 1 in the identity position, produce an image with no alpha channel Don't use the flash 0-255 convention in the URL below.