Re: RAW colortheory

Lee Varis

On Mar 17, 2005, at 10:02 AM, John Ruttenberg wrote:

If I try to recover blown
highlights (like clouds) by turning down the exposure, I get a distinct
magenta (maybe just pink) cast in the highlights I'm trying to recover. Why?
Most digital cameras (and the 1DMkII is no exception) use what's known as a Bayer pattern chip. The CCD/CMOS chip is covered with tiny red, green and blue filters in an alternating pattern that provides twice as many green sensors as red or blue. The green channel in an RGB image carries 60% of the luminosity of the image so its not surprising that the chip should favor green over red or blue. Typically these "green" sensors are the last ones to "clip" when saturated with light -- the red and blue sensors max out first. There is only so far you can go when recovering blown highlights in a digital capture and if the red and blue sensors are clipped only the green pixels will carry any tonal variety - hence the overall magenta or pink color. The highlight recovery can only operate on the darker than white pixels left in the image to render them darker so what you end up with is magenta colored "detail" that is actually carried in the green channel. This is also the reason why specular highlights often have a magenta fringe around the "clipped to white" region - this is the edge of green pixels that haven't clipped yet (they are darker than max white).

You can mitigate this effect, once you've opened the RAW file into Photoshop, using a Channelmixer adjustment layer, monochrome checked with the green channel at 100% - set blending options to "blend if" (using the slider for the underlying layer) everything that is darker than the lightest parts of the image - this will de-saturate the highlights and maximize what little detail is available in the green channel


Lee Varis

Join to automatically receive all group messages.