bayer filter pattern array questions

[Dale]

what device dependent RGB does the bayer pattern use?

I know they double the green cells because the eye has more G rods or
cones than RB

does bayer assume rec709 video?

bayer doesn't assume sRGB or ProPhotoRGB since those came much later

why not have a sensor filtration that is CIEXYZ compatible? that way the
camera can be easier to use in device independent color like
http://www.color.org

the preference of G cells in the bayer pattern I think can cause noise
in places like hair, etc

[Martin Brown]

Post by Dale
what device dependent RGB does the bayer pattern use?

It doesn't have to use RGB at all! Another variant more commonly used
for video sensors uses CMYG or CMYW and yet another Sony CCD used RGBE
(E = emerald green/cyan). Kodak has also patented RGBW patterns. See

http://en.wikipedia.org/wiki/Bayer_mask

They are severely limited by the properties of light stable dye filters
and the need to be able to physically manufacture the filter mask.

Post by Dale
I know they double the green cells because the eye has more G rods or
cones than RB
does bayer assume rec709 video?
bayer doesn't assume sRGB or ProPhotoRGB since those came much later
why not have a sensor filtration that is CIEXYZ compatible? that way the
camera can be easier to use in device independent color like
http://www.color.org

Bayer specifies the pattern and a means to demosaic it back to YCbCb or
RGB taking advantage of the predominance of G in the Y channel and
limitations of the human eyes colour resolution capability. You are
stuck with what the chemical dyes for red, blue and green can actually
deliver in practice. Red is very good (but not always light stable),
green is so-so and blue may have a small long wavelength red leak.

Wratten filters being one set that is used in scientific work graphs:

http://www.edmundoptics.com/optics/optical-filters/color-dichroic-filters/kodak-wratten-filters/1326

Whilst insanely priced bandpass filters are possible they do not help
perceived image quality in conventional colour imaging and would be all
but impossible to manufacture on the scale needed for consumer CCDs.

The demosaic into RGB or YCC can be done back to whichever colour space
you want to choose and then embed the appropriate colour profile.

Post by Dale
the preference of G cells in the bayer pattern I think can cause noise
in places like hair, etc

It limits the resolution you can hope for when imaging in monochromatic
red hydrogen alpha light or blue hydrogen beta, but apart from that it
really isn't a big problem. You can design bespoke testcards that Bayer
struggles with and Foveon fans spend all their time obsessing about it.

In fact a bug in most JPEG decoders is much more of a hindrance for
using monochromatic blue or red light for imaging with a Bayer mask.

[Alan Browne]

Post by Dale
what device dependent RGB does the bayer pattern use?
I know they double the green cells because the eye has more G rods or
cones than RB
does bayer assume rec709 video?
bayer doesn't assume sRGB or ProPhotoRGB since those came much later
why not have a sensor filtration that is CIEXYZ compatible? that way the
camera can be easier to use in device independent color like
http://www.color.org
the preference of G cells in the bayer pattern I think can cause noise
in places like hair, etc

Download dcraw.c and you'll get all sorts of data about bayer patterns
in various cameras as well as color space application.