Re: [Gimp-user] Histogramm -- values
- From: Elle Stone <ellestone ninedegreesbelow com>
- To: gimp-user-list gnome org
- Subject: Re: [Gimp-user] Histogramm -- values
- Date: Tue, 03 Dec 2013 08:55:39 -0500
On 12/02/2013 02:31 PM, Wolfgang Hugemann wrote:
human spectral sensitivity has a pronounced peak in the green region,
see http://en.wikipedia.org/wiki/Luminosity_function.
This is why any function deriving the perceived brightness or luminance
from RGB, such as Rec709Luminance (0.212656 * R + 0.715158 * G +
0.072186 * B) weights the green far more than red and blue. This just
reflects the spectral sensitivity of the human eye.
The perceived brightness of purely red image at, say, 100 bit will be
far less than that of a purely green image at 100 bit. Correspondingly,
the latter should result in a leighter equivalent grey than the former.
Gimp proceeds this way when converting a colour image to grey, as I have
just tested:
100 R --> 21 G
100 G --> 72 G
100 B --> 7 G
Gimp 2.8 uses the values from Poynton, which are the right values to use
in a non-color-managed application when sending signals to a D65 Rec709
display device. The correct sRGB values for an ICC profile color-managed
application are slightly different: R 22%, G 72%, and B 6%. See:
http://ninedegreesbelow.com/photography/srgb-luminance.html
This is very close to Rec709Luminance.
Gimp 2.8 operates on the nonlinear sRGB RGB values, so it computes luma
rather than luminance. See Poynton's ColorFAQ #9 and #11
(http://www.poynton.com/PDFs/ColorFAQ.pdf). Calculating true luminance
requires that you first linearize the RGB values. Gimp 2.10 will use
linear sRGB to calculate true luminance rather than luma.
I just suggest that it should offer such a weighting in the histogramm
for a colour image, i.e. the equivalent grey value at Gimp itself would
calculate it.
I concur 100%. I hope Gimp 2.10 includes exactly such a histogram,
except personally I would prefer luminance (calculated using linear
sRGB) rather than luma. Luminance is a mathematically expression of how
we actually perceive relative light and dark, based on what you said
above about how we perceive colors (very sensitive to green, less to
red, even less to blue). Luma is mathematically easier to calculate than
Luminance, but as far as I know that's its only virtue.
Elle
[
Date Prev][
Date Next] [
Thread Prev][
Thread Next]
[
Thread Index]
[
Date Index]
[
Author Index]