Olympus-OM
[Top] [All Lists]

Re: [OM] CCD image sensor cameras

Subject: Re: [OM] CCD image sensor cameras
From: Ken Norton <ken@xxxxxxxxxxx>
Date: Mon, 6 Apr 2020 14:41:15 -0800
> I appreciate the analysis, although others may not. But to summarize... It is 
> the color filter?
> As a former filter designer, I appreciate the effect of a sharper roll off 
> filter. So we are saying that it is the color filter creating the main 
> difference? And that the sensor, whether CMOS or CCD is just capturing 
> photons.

Mostly, so, but nothing is exactly stand-alone. It's the entire A/D
chain that has to work together. If the sensor itself is biased one
way, you have to bias the electronics or mathematics the opposite way
to create balance. Anything you can do BEFORE creating numbers still
has unlimited bit-depth, but is subject to analog noise. Once
converted to numbers, everything you do is throwing away bits.


> That is, a less sharp filter will bleed a bit of one color into another 
> color, loosing color separation. Which can not be recovered in post 
> processing the way a shaper color filter would allow. If true, then different 
> sensors will provide a clear difference in an image, as discussed.

By making the filters less color discrete, there should be less color
aliasing artifacts because color hues are mathematically derived
instead of directly captured. Saturation is a far more complex
determination and some cameras and converters just don't yield
flexible files. An image file that has the color information
"processed in" may already be at the limits of color manipulation. The
pictures from my iPhone look really nice, straight-out-of-phone, but
they are horrible to try get all fancy with.


> Are you also saying that the location of the color pixels helps separate the 
> colors? That is red and blue are further apart in wavelength, while green is 
> in the middle. I don't understand triangle separation.

If you look at the Bayer Filter, it is made up of twice the number of
green sensels than either the blue or red. At every corner there is
one red, one blue and one green represented. Depending on the version
of the converter and camera, the output pixel is based on the
intersection of four pixels (2G,1B,1R) or three pixels (1G,1B,1R). The
three pixel version will always be sharper.

The original concept of the Bayer Filter was four-pixels and the green
ones determined brightness value. It didn't take long before raw
converters got a little smarter than that. Kodak stayed with this
method, though, and the Olympus CCD sensors actually have dual green
channels which stuck around through the initial versions of Panasonic
sensors. I haven't seen evidence of this being true any longer because
the filters are no longer as discrete.

As to determining hue and saturation, Adobe's Version 5 converter
appears to use a wider selection of pixels to come up with a value.
Earlier versions were either 3 or 4 pixels (depending on the camera).
But I'm seeing evidence of a MUCH wider spread being used to suppress
false colors and create more accuracy for the mix-minus calculation.
This also really helps with skintones and plays on the strengths of
the Fuji scrambled-egg sensor design without sacrificing sharpness.


> If this is correct, that the difference is in the color filter, then there 
> could be CMOS sensors that achieve the same color separation/characteristics 
> of a Kodak CCD?

Yes. Phase One. And I'm going to go out on a limb here and say that
Leica SL also has achieved the same look.

AG Schnozz
-- 
_________________________________________________________________
Options: http://lists.thomasclausen.net/mailman/listinfo/olympus
Archives: http://lists.thomasclausen.net/mailman/private/olympus/
Themed Olympus Photo Exhibition: http://www.tope.nl/

<Prev in Thread] Current Thread [Next in Thread>
Sponsored by Tako
Impressum | Datenschutz