Let's talk about the rgb sensitivity in raw.

Started Dec 23, 2012 | Discussions thread
crames Regular Member • Posts: 192
Re: How RAW RGB Channel Data is derived from RGB Photosite Image Data

Iliah Borg wrote:

Anders W wrote:

Iliah Borg wrote:

When you say that exposure plays a part here as well, is what you have in mind the problem I try to outline in the post to which I link below?


Or do you have more/other things in mind? If so, could you please elaborate a bit?

From the spectral responses one can see that same wavelength combinations can produce different colours if the power of each wavelength is multiplied by the same factor.

Well, this is what I have difficulties seeing.Take your example with a light source that emits 65.4 mW at 450 nm and 41.6 nm at 550 nm. The channel responses would be:

R: 65.4 x 0.02 + 41.6 x 0.07 = 4.220

G: 65.4 x 0.13 + 41.6 x 0.90 = 45.942

B: 65.4 x 0.66 + 41.6 x 0.04 = 44.828

Now, if we increase the exposure by a factor of two, we would have

R: 2(65.4 x 0.02 + 41.6 x 0.07) = 2 x 4.220

G: 2(65.4 x 0.13 + 41.6 x 0.90) = 2 x 45.942

B: 2(65.4 x 0.66 + 41.6 x 0.04) = 2 x 44.828

So that the channel balance remains exactly the same.

You need to take at least 2 things into account, first, convert to Lab; second, look at how metamers are handled. Optionally, the third thing is to calculate average Lab values taking noise into account.

To the first one, let's take a simple matrix profile with gamma = 1, sRGB chromaticities, and use colour calculator

icclu -ff -ir -pl -s100 /Users/iliah/Library/ColorSync/Profiles/Simplified\ sRGB\ g1.0.icc

4.220000 45.942000 44.828000 [RGB] -> MatrixFwd -> 66.969664 -34.828540 -9.382103 [Lab]

8.440000 91.884000 89.656000 [RGB] -> MatrixFwd -> 88.535226 -43.881210 -11.820709 [Lab]

You can see "a" and "b" values are not even close.

Is is not to be expected that the a* and b* CIELAB coordinates will change with changing luminance? What is constant in your example is hue and saturation (location on a chromaticity diagram). They really are the "same color," only with a different lightness.



edit: note to Joofa - CIECAM02 saturation is identical

Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow