Sigma sD1 has richer colors because of the independent layers.
maybe true, maybe not. unfortunately that's always going to be an apples vs bananas argument. lenses affect contrast, black/brass/silver aperture blades, coatings, even lens hoods do a bit too. so does the image processor it's attached to, and will most likely be tuned to whatever sensor type it's connected to.
try comparing 2 different car engines for performance, after they've been built in 2 different types of car with different chassis, wheels, suspension etc, and driving it around a track. you can infer which engine is better, but you'll never get an exact comparison. (sorry for the car analogy, top gear is on in the background).
One thing i've just noticed with the patent, compared to the X3 sensor (at least according to wikipedia): the canon patent specifically says that the blue is widest, then green, then red. the diagram of the X3 on wikipedia may be a simplification, but it looks like the colours are all the same width.
aside: a patent has to contain something new and/or innovative compared to another patent. maybe the width-difference is what canon is trying to set apart from the X3?
aside from the aside: these patents were (probably) written in Japanese, yes? can we get the original japanese patent and an electrical engineer who can read japnese from somewhere? anyone know anyone? these patents really are hard to decipher most of the time...
I wonder why they (Sigma) call it (SD1) 46Mp camera, because actually it produces 46/3~=15Mp pictures.
one word: marketing.
it *should* be called a 15mp picture, because that's how many dots you get at the end of the day. they've got 15m red, 15m blue, 15m green sensors, but on 15m unique spots.
but then a canon 15mp sensor has 5m blue, 5m green, 5m red sensors on 15m unique spots (with 15m lenses on top). one dot only takes one colour, and the processor guesses the values of other colours from the neighbour. so at the end of the day you get 15m dots. but the colour of those dots is only 1/3 definitely accurate, the other 2/3 are guess or interpolated.
(hmmm, by the same logic, my monitor is 1680*1050 pixels (1.7mp). but it actually has 1.7m red, 1.7m blue, 1.7m green on 5.3m unique sites. should i call my monitor 5040*1050?)