Kinda
To a computer, there is no such thing as color, so an RGB image is made up of three monochrome images. With a Bayer sensor, you don't get a true monochrome image (like what we would think of from B+W photography) directly from the sensor because each pixelsite is filtered for a different color. Those filters work pretty much the same as colored filters work in B+W photography. With the Foveon, you get one blue filtered image, one green filtered image and one red filtered image, just as though you had taken three B+W photographs with different filters. If you look at a picture of the sky from a Sigma's red channel, it'll be dark - in the blue channel it'll be light. With a Bayer sensor, one dark pixel will be next to a light pixel and the sky will look like a checkerboard.I guess SG will be over to jump in on this one.
I dont mean to nit pick, but that adds up to 125%.
--
--
--Kinda
image is made up of three monochrome images. With a Bayer sensor,
you don't get a true monochrome image (like what we would think of
from B+W photography) directly from the sensor because each
pixelsite is filtered for a different color. Those filters work
pretty much the same as colored filters work in B+W photography.
With the Foveon, you get one blue filtered image, one green
filtered image and one red filtered image, just as though you had
taken three B+W photographs with different filters. If you look at
a picture of the sky from a Sigma's red channel, it'll be dark - in
the blue channel it'll be light. With a Bayer sensor, one dark
pixel will be next to a light pixel and the sky will look like a
checkerboard.
So yes, the RAW data from a Bayer sensor is essentially monochrome,
but essentially unusable as such, whereas the RAW data from a Sigma
is truly monochrome and very usable. It kicks a$$
--
Good call. It didn't take old SteveSD9 long to come contradict me. Makes me feel more confident of my answer.
http://www.alt-vision.com/documents/Color-CMOS-Cameras.PDF
It is amazing to see how you just quote pages and documents that contradict your claim to support your claim, this one is even more obvious than the X3 pixel page last time.
I coldn't see how it even had any relationship to his point. Slide 11 does have the word "monochrome" in it with respect to how one can sense color using a monochrome sensor with filters over it, but does that bear on the question on whether the raw file represents colors? Here's what it says as one of the color separation methods:
That's a difference mostly in how you are thinking about the data. Is it necessarily a differnce in how the data are stored or represented? I'm not so sure.
At 1.5 bytes (12 bits) per pixel, the 5 MP camera needs 7.5 bytes and the 10 MP camera needs 15 bytes. That's a simpler way to look at it. (Your numbers are based on 2 bytes per pixel, which is overkill, but same idea: all you need is the pixel count).
I've read some of the DNG spec, and I recall that it specifically provides for both fully-sampled (X3-like) and various forms of CFA data. Its not structure is not closer to one or the other.
I think this pretty much explained the missunderstanding of Steve.
