You are correct about the spectral overlap with CFA filters (and this goes to color separation as well in the Foveon layers). If they were very narrow width filters, the camera wouldn't be able to "see" colors other than the primaries... .
[{POST_SNAPBACK}][/a]
The overlap has to do with the efficiency of the filters in relation to different wavelengths. The primary color 'green' for example is not one specific wavelength, like 530nm, but a range of frequencies with 530nm being in the centre of that range.
One should not assume because a green filters allows through, at a reduced intensity, some light in the red and blue range, that the total proportion of light blocked is necessarily less than 2/3rds, on average. The precise figure will vary from scene to scene and no doubt will be heavily influenced by ISO selection, which is probably why, after a brief Google search I could find no statistics on this.
But what I
did find is a very revealing statement (if true) which implies that the light loss in the standard Bayer CFA could be at least 2/3rds and possibly even more.
Speaking of the subtractive primaries, Kodak used an image sensor in the DCS620x camera that used a cell with two yellow filters, one cyan filter, and one magenta filter, where yellow = red + green, cyan = blue + green, and magenta = blue + red. This gave the same color fidelity as a standard Bayer filter array, but allowed twice as much light to reach each light-sensing element, thereby allowing shorter exposure times. Sony makes cyan, magenta, yellow, and green array image sensors for use in TV cameras, such as the ICS429.
For anyone who's interested, the full article is at [a href=\"http://www.quadibloc.com/other/cfaint.htm]http://www.quadibloc.com/other/cfaint.htm[/url]