Luminous Landscape Forum
Raw & Post Processing, Printing => Adobe Lightroom Q&A => Topic started by: JohanNyberg on June 30, 2010, 03:03:32 am
-
Hello. It seems to me that if I increase the overall contrast in the development module in LR3, the colour saturation also increases. I wonder if this is an illusion, a bug or a feature, or something else. Perhaps my screen trying to do something on its own.
Johan Nyberg, beginner with digital (and colour).
-
Hello. It seems to me that if I increase the overall contrast in the development module in LR3, the colour saturation also increases. I wonder if this is an illusion, a bug or a feature, or something else. Perhaps my screen trying to do something on its own.
Johan Nyberg, beginner with digital (and colour).
This normal in any photo editing program. In Photoshop which uses layers you can set the layer blending mode to luminosity to avoid this. This one of the reasons that you should use Photoshop for advanced editing.
-
Hello. It seems to me that if I increase the overall contrast in the development module in LR3, the colour saturation also increases. I wonder if this is an illusion, a bug or a feature, or something else. Perhaps my screen trying to do something on its own.
Johan Nyberg, beginner with digital (and colour).
Johan,
You might want to read Mark Segal article on curves: http://luminous-landscape.com/essays/Curves.shtml (http://luminous-landscape.com/essays/Curves.shtml)
This is a very good and in-depth article, possibly much more than what you asked.
-
JohanNyberg - Both Stamper and Francoise make the excellent point - the result you are seeing is one of the so-called "Golden Rules" of colour in photography - the higher the contrast the more saturation in RGB results.
-
JohanNyberg - Both Stamper and Francoise make the excellent point - the result you are seeing is one of the so-called "Golden Rules" of colour in photography - the higher the contrast the more saturation in RGB results.
Exactly. Not doing this move more than often results in undesirable rendering. But if one doesn’t like this, just season to taste with the Saturation or Vibrance sliders OR HLS sliders for selective color sat adjustments. One could make a preset that affects both.
-
Thanks to you all. I will read Segal's article on curves.
Johan Nyberg
-
JohanNyberg - Both Stamper and Francoise make the excellent point - the result you are seeing is one of the so-called "Golden Rules" of colour in photography - the higher the contrast the more saturation in RGB results.
Actually, hue, saturation, and brightness are separable as shown by HSB color spaces. As I recall from the Adobe forums, Thomas Knoll has stated that it would be easier in programming not to increase saturation as contrast increases, but he chose the current implementation where saturation increases with contrast, since it is preferred by most photographers.
-
Actually, hue, saturation, and brightness are separable as shown by HSB color spaces. As I recall from the Adobe forums, Thomas Knoll has stated that it would be easier in programming not to increase saturation as contrast increases, but he chose the current implementation where saturation increases with contrast, since it is preferred by most photographers.
I also remeber having read Thomas comments on contrast/saturation. Is there no option (to avoid increased sat) in the newer versions of Photoshop? I don't have my PS computer with me… I might be dreaming
-
I think that you can convert the RGB file to Lab color, and then apply your contrast adjustments to just the L or luminance layer. Then you flatten the layers and convert back to RGB.
John
-
I think that you can convert the RGB file to Lab color, and then apply your contrast adjustments to just the L or luminance layer. Then you flatten the layers and convert back to RGB.
You could, tossing away a decent amount of data and spending a lot of time doing so. You could just do the adjustment on an Adjustment Layer and set the bend mode to Luminosity and produce the same results faster, without all the Lab conversion inducing data loss.
-
You could, tossing away a decent amount of data and spending a lot of time doing so. You could just do the adjustment on an Adjustment Layer and set the bend mode to Luminosity and produce the same results faster, without all the Lab conversion inducing data loss.
Andrew, I stand corrected. I should have hesitated to post on a colour topic, but I was just trying to be helpful. Now I shall stick to my B/W pictures
John
-
You could, tossing away a decent amount of data and spending a lot of time doing so. You could just do the adjustment on an Adjustment Layer and set the bend mode to Luminosity and produce the same results faster, without all the Lab conversion inducing data loss.
Using the luminosity blend mode is definitely easier than converting to and from L*a*b. That conversion would entail considerable data loss if done with 8 bpc, but the visual effect often will be negligible. However, using 16 (or 15+1) bpc will dramatically reduce the data loss and I think that it would be visually undetectable even with the most extreme adjustments. I invite anyone to prove me wrong.
-
However, using 16 (or 15+1) bpc will dramatically reduce the data loss and I think that it would be visually undetectable even with the most extreme adjustments. I invite anyone to prove me wrong.
No disagreement there. Absolutely do this kind of editing in high bit.
-
No disagreement there. Absolutely do this kind of editing in high bit.
Andrew, you must be getting old? You didn't mention Dan once.
-
Andrew, you must be getting old? You didn't mention Dan once.
I am but it beats the alternative.
-
As a former sound technician with some experience and theoretical knowledge of digital processing, I wonder if there are any usable analogies with image processing. I guess that high resolution computation is important with images as well. The mixing console I used employed something like 56 bit precision for 24 bit output. A lot of effort is put into dithering output with shaped noise before truncating to the output word length. It would be interesting to know what LR does. I suppose it all boils down to information theory, of which I have very little, and engineering. And taste.
Johan