2

The context

Photo retouching workflow. I'm learning to retouch my shots using dodge, burn etc. This is a question about workflow and helper layers, not about achieving a specific image A as an output from image B. I'm looking for a way to display information in a better way, if it is at all possible (I'm aware it may not be).

Prime example: fixing a portrait. When you shoot a portrait it comes with imperfections in the skin. Those may be due to the skin itself (a child that just scratched an itch...) or the result of how light was cast on the subject (harsh shadows that should not be there...). Whatever the reason the imperfection is, ultimately, a shift in hue, saturation and/or luminosity. To better/faster understand the problem at hand I want to isolate those three informations to apply a better/faster correction.

The problem

When I'm working on an image I know I can display only color information creating a solid color layer, fill it with grey and set it to luminosity blending mode. I use this to detect hue and saturation shifts, for example because of scars, tan lines, and so on. The same can be done to display only luminosity information and detect light shifts by setting the blending mode of the layer to color.

The point is: while the luminosity information is straightforward and easy on the eye (it naturally comes as a greyscale) I struggle with the way the color information is presented: to me, pure color information is much less readable.

I would like to add further steps to my workflow so that I can display hue or saturation as greyscales, they way luminosity naturally displays as a greyscale.

A process that works in Photoshop would be most useful but this would ideally apply to other raster image manipulation programs as well.

Let's say this is our base image:

Photo01

This happens if I create a solid color layer, fill it with a grey, and set it to color blending mode, resulting in very readable information:

Photo02

This is with the same layer set to luminosity blending mode, the hue and saturation are (both) there, but very hard to read (to me):

Photo03

The actual question

Is there a way to create a group of layers I can switch on and off that will let me display hue or saturation as a greyscale bitmap to help me detect problems in the image I'm working on?

  • 1
    Hi. Welcome to GDSE. I'm not sure I understand your question. True greyscale images don't have any hue or saturation, and can have no colour information. Can you please upload some images to illustrate what you are trying to achieve? Thanks. – Billy Kerr Sep 9 '18 at 12:07
  • Please add these images and extra information to your question. Thanks. – Billy Kerr Sep 9 '18 at 12:39
  • There are several concepts mistakes in your question, which leads to a work process that maybe is not the right one. I think the best is if you look for an image with an aspect similar to the one you want to achieve and insert it in the question. I'm sure the answer is very simple, but the question is not. – Danielillo Sep 9 '18 at 12:59
  • What do you mean by "easier to read". I'm not sure I understand what you are actually trying to achieve, even with those images. What is the ultimate goal? – Billy Kerr Sep 9 '18 at 13:30
  • Reworked the whole question, hope I have clarified things up now. – meh Sep 9 '18 at 13:36
0

Saturation, not the usual RGB saturation, but more like chroma divided by the max possible chroma at the same luminosity at least is produceable as grayscale image by using layer blending modes and Photoshop's image adjustments.

Hue is an open question due its discontinuity (=jump to zero at 360 degrees).I haven'seen the solution.

If you want to stick with the exact RGB saturation, you do better by skipping the playing with blending modes and use the HSB/HSL filter which is available as free plugin here:

https://helpx.adobe.com/photoshop/kb/optional-file-format-plugins.html

It's a part of Electric Image & HSBHSL Optional Multiplugin

HSB/HSL filter calculates RGB color in HSB or HSL color model (you select) and extracts Hue to R channel, Saturation to G channel and Brightness or Luminosity to B channel. All numbers are scaled to 0...255.

If you need the saturation as grayscale image, you simply watch the green channel only (=close R and B in the channels panel), select all, copy to the clipboard and paste to a grayscale image. Probably you get the right size and mode automatically, if you start a new image based on the clipboard.

Here's the filtering dialog and the generated RGB image:

enter image description here

Here the RGB channels are copied and pasted to a grayscale image:

enter image description here

Be sure that noisy images, like yours, give highly unreliable hue at low saturation and low luminosity areas.

Comparing hues meets an extra difficulty due the discontinuity at red. I mean 0 and 255 are nearly the same hue, but quite far away from each other as numbers and as extracted greys.

Of course that's no problem for a programmer who understands math, but calculating only the difference isn't enough. What makes me suspect the usefulness of whole procedure, is the fact that programming is needed to see the sameness of nearly reds. As well the programmer could calculate all needed relations directly from RGB numbers.

NOTE: HSB/HSL filter works ok, the problem is the discountinuous definition of hue, the color circle. If you reveal something of the intended usage of the extracted images, someone might be able to tell how to work around the discontinuity of the hue without math programming.

  • First of all thank you for a solution to my problem and a clear explanation of the hue caveat. Obvious once you think about it. One would have to write a filter that, say, will shift the discontinuity to a hue that is of no matter, for example green when looking for skin imperfections. The example image is just that: something random used to show what I meant, it is by no means a picture that represents what I would call a good shot. :) Could you add to your answer the way to display saturation with layer blending modes? – meh Sep 11 '18 at 7:25
  • Yes, it's always nice to know and experiment with new workflow tools, may give me good ideas even if I end up not using it right away! – meh Sep 11 '18 at 13:38
  • Might also be worth working in lab mode. – joojaa Sep 11 '18 at 18:25
  • @meh I must admit that this case has revealed a hole in my knowledge. I had different calculation for the saturation than the generally used. The lightness (=the average of brightest and dimmest RGB channel has been replaced with a BW mix of the RGB channels. The resulted quantity isn't the usual RGB saturation altough it presents the same visible thing. These are the usual formulas rapidtables.com/convert/color/rgb-to-hsl.html. I must remove the claim the saturation can be presented as BW with blending modes and adjustments in Photoshop. – user287001 Sep 11 '18 at 22:32
  • @joojaa the thing "detecting and classifying skin color variations" should be worked in CIELab (=Lab in Photoshop), but the questioner didn't ask it. – user287001 Sep 11 '18 at 22:35

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.