New Luminosity Curve - a word of warning

With PL8 introducing a Luminosity Curve, I was finding that, most times, so far, I could simply swap the RGB curve for a luminosity curve without too much problem and, often, it would improve the overall appearance by removing the extra saturation it can bring.

That is, until I came to prepare this image for a teaching session…

Beautifully balanced with a nice warmth to the wood panelling on the roof of the bridge.

However, after changing the same curve to the Luminosity channel…

… partly due to having reduced the colour temperature in the first version to 3646°K, in order to help with the over-saturation and colour cast from the lighting, photographed at my habitual 5600°K, which looked like this…

So, I thought, why not reset the WB to 5600°K on the new version with the Luminosity curve? Well, this is what happened…

… a horrible “sickly” salmon pink cast.

It took changing the Temperature to 3237°K and, most of all, changing the Tint from 0 to 24 to get rid of it.

So, it’s not just about desaturation, the relationship between the curves seems to be a bit more complex.

Could this be a bug, or just an unavoidable result of changing from RGB to Luminosity?

1 Like

The Luminosity Curve is supposed to change brightness without changing colour saturation, whereas the RGB curve changes colour saturation when you brighten or darken the image with the curve.

I think this explains what you are seeing in your examples. I don’t know how accurately the colour can be maintained with large Luminosity Curve changes. This may also explain your unwanted colour casts.

Of course, changing colour temperature will change the colour of your image.

I am happy to be corrected, just my take on the subject.

You might read Understanding Digital Camera Histograms: Luminosity and Color

George

@Joanna

I think it’s just difficult (if not impossible) to counteract strange color casts from streetlights and the like, probably due to the discontinuous light spectrum. – And by using the Luminosity Curve tool you made it more visible.

You may be able to “restore” individual colors using the HSL tool.

Indeed. That was my point, to highlight the differences that the new curve makes to coping with such diverse lighting.

I wouldn’t like to try and work with a JPEG file, where I couldn’t change the green/magenta tint.

Eventually, I took the WB pipette from the grey concrete of the pillars, which I knew were as neutral as I was likely to get

DxO calls it the luminance or luma curve. Luminosity is a different beast, but related.

Luminance is a weighted value that takes human vision into account. Luminosity doesn’t do that and (our perception of) colours can shift due to such differences.


Above, we see the results of pulling the “white point” towards the center of the tone curve’s scope, once for the RGB curve and once for the LUMA curve.

Depending on how each RGB triplet travels in 3D, perceived colours can change in the planes perpendicular to the black-white axis. → sRGB in 3D

Please note that the differences are less visible in well exposed images that have a histogram curve centered around its middle.