Change color mode to LAB.
J
Good question Aaron, the fact is that Luminosity mode in RGB is not quite the same as L in Lab, but it still does a damn good job at putting yellow into your image!
Aaron.
The Luminosity blend DOES, in fact, preserve in the result the luminosity of the blend color. Possibly you are not reading the luminosity correctly. In Photoshop, luminosity is DEFINED as follows:
0.3R +0.59G + 0.11B
Read the RGB component values at any point, apply the above weighting factors, and add the weighted values to get the resultant luminosity. If the blend mode has been set to luminosity, that result will be the same as the luminosity of the blend color at every pixel site. In the case you cite, with the blend color a 50% gray everywhere, the resulting weighted values (the "luminosity") will add up to 128 everywhere.
While all these answers are essentially correct, I think maybe they’re missing the OP’s point? Although he isn’t very clear, telling us only what *doesn’t* happen but not what actually *does*.
So just to get it out of the way: Layer opacity slider, visibility icon (the eye), no additional layer with luminosity data on top – all check? If so, proceed to above. Unless, of course, something’s really wrong.
Apologies for the unclear explanation, and thank you for the answers. I was looking at the HSB values for the pixel, and expecting the brightness to be set to 50% and the hue and saturation to be unchanged. Clearly, that’s not what’s going on.
OK, so given the formula above for luminosity, how can I predict what the RGB values will be for a given pixel in the example I gave (actually, I’m looking for the general-case equation, but the one example is probably sufficient to derive it)?
Based on some experiments, it looks to me as if what’s happening is that, given an original pixel (R, G, B), PS computes an offset K such that the pixel (R+K, G+K, B+K), where each component is clipped to the range (0-255) will have a luminosity close to 50%. Can anyone confirm that that is the case?
Incidentally, where did you find that formula for luminosity? I did a whole bunch of searching, and I never came up with it.
Much appreciated,
Aaron
Aaron,
The weighting factors for R G and B were determined long ago via subjective responses in tests on thousands of people. there have since been other recommended weights. No set of weights is valid over the whole range of color values. In an imperfect world you compromise. Nowhere is it explicitly stated, but Photoshop has stuck with the older weights. You can verify that for yourself using the Info palette. Other applications may use other luminosity definitions. Also in Photoshop, Brightness is taken to be the highest of the R G and B color values. I think you were equating brightness and luminosity. They are decidedly different parameters. Nor is Luminosity the same as the L parameter in LAB (most often called "Lightness" but often "Luminance"). You won’t find that LAB Lightness is preserved by the Luminosity blend and certainly not Brightness. Using the Info palette, you can also see that the Luminosity blend mode retains the hue and saturation of the base while adopting the luminosity of the blend color.
Actually, unless there’s some additional option in the Info panel of CS3 (I’m actually working with PSE6 at the moment), what’s shown there is HSB, and the saturation is definitely not conserved. The hue does seem to be, unless one or more of the color components gets clipped.
This is what confused me originally – I was expecting Luminosity to be at least on the same axis as Brightness, whereas it is in fact somewhere on the Brightness/Saturation plane.
Thanks,
Aaron
Aaron,
You are right. Saturation is not preserved in a Luminosity blend. But the blend color’s luminosity and the base color’s hue ARE preserved. Saturation adjusts to whatever is necessary to achieve that.
You can readily verify that the resultant hue is the same as the BASE hue by setting the info panel to HSB and comparing before and after hues.
Verifying that the resultant luminosity is the same as the BLEND luminosity is not as easy. You have to read out the RGB values and calculate the before and after luminosities, using the weighting factors I cited.
Resultant hue and luminosity will EXACTLY match base and blend values, respectively.
George
Agreed, that’s what I calculated. Knowing the formula for luminosity makes it make sense. Actually, a little algebra says that the constant K that’s added to each component is just the desired luminosity minus the original luminosity, though things get a bit more complicated if you start clipping component values (because they’ve reached 0 or 255).
Aaron
Aaron,
"…Actually, a little algebra says that the constant K that’s added to each component is just the desired luminosity minus the original luminosity…"
The resultant luminosity of a pixel is the sum of its weighted color channel values. In general, blending-induced incremental changes in these components are not even close to being equal. Adding a "constant K" to each of these components is not making sense to me.
George
