So I tried Sigma DP1 Merrill...it could use some changes

xpatUSA wrote:

D Cox wrote:

xpatUSA wrote:

doeyee wrote:

I'm a man of science and logic, not faith, and I do not have faith in a corporation nor a CEO.

I've been known to be a little pedantic myself

They took away red and green pixels in the new Quattro sensor, so they have to invent new pixels through interpolation.

Interpolation is fraud. It may be very common but that does not mean it should be considered acceptable.

I must agree. In a way, the Quattro layout is sorta kinda like a sub-sampled TV signal or perhaps a Y'CbCr 4:2:0 JPEG - the horizontal and vertical color resolution halved, just like layers 2 and 3 are.

But there is a degree of fraud in the other Foveons; just a degree, mind. As most here know, the layer colors are not actually blue, green or red and they do have a very wide spectral response compared to the Bayer CFA. Thus it is that we get fraud of a different kind, due to the extreme matrices needed to get from sensor space to RGB.

But the cones in the retina are not blue, green and red either. The medium and long cones have responses that overlap just as much as the medium and long layers in a Foveon. You don't need mutually exclusive bandwidths to detect colour differences.

With all due respect, I never said we did and I agree with what you've said. But (my turn) the brain appears to be vastly superior in "converting" to what we think we see. Whereas, conversion from camera space thru XYZ to RGB involves some pretty hefty coefficients. Sigma claims that it's no worse than the Bayer system, noise-wise, but we still suffer some blotching from time to time, in spite of improvements in SPP.

I get green and magenta blotching at low ISO settings with my Sony A65 too Ted . . . in certain circumstances. It's not just a Sigma thing. I even started a thread about it. I can't find it now though.

Of the two "frauds" mentioned in this post, I am 100% for pre-Quattro Sigmas because I find Bayer color-aliasing quite irritating.

Perhaps the solution will be to use far more pixels (maybe around 200 Megapixels) and downsize all images in the camera.

You'll get Roland's and Scotty's vote on that!

YES, definitely have MY vote on that!

-- hide signature --

Ted