carstenw wrote:
Still, there are only 16MP locations where the light is being measured (albeit three times), so for the purposes of outresolving glass, it should still be 16MP, not 30MP or any other number.
Well, no actually, if you are comparing to Bayer - that was the point of my post. A 16MP Bayer imager does not spatially resolve at the Nyquist spatial frequency you'd expect for its pixel count because of the anti-alias filter and the demosaicing process which softens the image. The Foveon sensors do resolve all the way up to Nyquist since they don't have an anti-alias filter or have to demosaic. Again, check out the link to the Sigma/Canon resolution tests earlier, or the DPR tests of the earlier Sigma cameras.
So a 16MP Bayer sensor will be significantly softer than a 16MP stacked pixel sensor and the Bayer sensor will mask lens softness that the stacked pixel sensor can detect. As far as lens resolution limitation goes you either need to consider a Bayer sensor to have fewer MP than stated when compared to a stacked pixel sensor or alternatively the stacked pixel sensor to have more.
I do suspect the relative resolving power differences between the Foveon imager and the Bayer imagers has moved a bit in the past few years - I believe more and more Bayer cameras have been using weaker anti-alias filters and improved demosaic algorithms. So the gap is probably a bit narrower than way back when.
Ken