Luminance-neutral tritoning - is it feasible?

[nsecomb]

Background:

I started using Lightroom (1) about 18 months ago, and given my personal preference for monochrome (mono) over colour, I quickly appreciated LR's split-toning functionality for quickly seeing how a mono might look under various personal tint preferences.  That said, looking at the end results (on screen), I thought it'd be nice to have a bit more control in terms of what tones have what degree of tinting, especially in the shadow tones where colouring seemed a bit splotchy at times.

About 12 months ago I started using Photoshop CS3 (PS3), and have periodically tried to master Tritoning and Quandtoning to mimic my favourite tints from LR, but with no success to date.  My latest effort was over the fortnight preceding this post, and as I'll explain below, I've run out of ideas on what isn't working and have concluded there must be some fundamental aspect of PS Duotones or colour science that I haven't grasped, and hence I am hopeful that some of you may be able to provide an insight or advice that help me achieve my tinting objective.

Objective:

Having spent hours massaging a mono image so that its contrasts, tones and overall character are just as desired, apply a Tritone that will provide hues (tints) for the mono image but which is also luminance-neutral.  By luminance neutral, I mean that the tones in the image will not be changed, i.e. if you were to pick 5 points in the image which have L values (from Lab readings) of 5, 28, 47, 63 and 94 in the mono image, then after the Tritone is applied those points will have the same L values, but a and b values will have changed (the hues).

To use the refrain from Top Gear, how hard could it be?

Method:

Make a test strip of 21 grey shades that are very roughly 5% grey increments from L = 0 to L = 100, with a=0 and b=0.  

In Duotone mode, for a Tritone with Black, Colour 1 and Colour 2, develop the curve settings across all three colours so that the L value for each of the 21 test strip areas is the same as their pre-Tritone L values.  Having achieved that (not an easy thing in it's own right thanks to autoshaping of the tone curves in between the 10% ink points... gee point curves in the duotone ink curve interface would seem handy!), save the luminance-neutral Tritone settings and apply those settings to a mono image.

Now I'd have thought that if you have an Lab value of 22 0 0 in a test strip, and the luminance-neutral Tritone settings changed it to 22 x y, then when you used the same luminance-neutral Tritone settings on an actual mono image, a point with 22 0 0 in the image would be changed to 22 x y.  But nooooo... (hear that faint snap sound... that was a small piece of my sanity being destroyed), it got changed to something like 34 w m.     What the...

Issues:

1) I am at a complete loss to explain why a set of Tritone settings applied to a monochrome image would cause luminosity shifts, yet the same settings applied to a monochrome test strip results in essentially no luminosity changes.  Suggestions/insights/experience as to why the outcome for the strip and the image are different would be most welcome.

2) Has anyone found a way to do controlled tinting in PS, i.e. you can control the amount of tint for luminance values across the range of 0 to 100, that doesn't change the tones in the image (only hues are added)?


regards,

Nigel Secomb

[Guillermo Luijk]

I think you are being too maniac about preserving at any cost the L value of the pixels. The L component of the Lab perceptive model is just that, a model, so it is not the absolute and only truth about the perceived luminance. Secondly, once you have your toned B&W image, nobody prevents you from going to Lab and apply a L curve to obtain the desired final perceived luminance, so it doesn't matter too much.

Regarding the way to obtain beautiful tones from pure B&W images, I think anything that the duotone, tritone or n-tone tools do can be 99% done by properly using RGB curves in PS.

Tone Hacker allows to capture the RGB curves that provide the tones found on any B&W toned image, it's a good learning tool. It can also generate the .acv PS curve file so that you can replicate those tones on any other B&W image with a similar histogram distribution, and it preserves the R+G+B combination, so it preserves luminance when applied. The only condition to get the right tones is to work in the same colour profile as the original tone image was described in.



Regards.

[JeffKohn]

Tone Hacker looks interesting, I haven't seen that before.

What I've been doing is slightly roundabout, but works OK. I found toning action I liked tones of, but it changed the luminosity way too much for my taste. So I wrote a little action in Photoshop that runs the toning action on a duplicate copy of my image, then copies it back onto the original image as a layer in "color" blend mode so that I get the toning without the luminosity changes.

[nsecomb]

Dear GLuijk,

Since my original e-mail I've developed a basic Tritone preset that allows me to achieve the practically luminance-neutral hue changes I'm after.  The outcomes are the same way regardless of the histogram.

The mystery, which I've moved on from now that I have curves that work as desired on actual images, is this:  the luminance-neutral presets that I have developed for the test-strip and images respectively, do not cause luminance shifts when used on what they were developed for (test strip or images), but do cause luminance shifts when used for the other test bed (images pre-set on test strip or vic versa).

For anyone who tries to go down the same path I did, i.e. try and find a generic luminance-neutral Tri-tone (or Quad or Duo) preset based on a test strip, my advice is don't.  Instead, find an image with the full gamut of grey values and develop your presets based on how (if at all) the hue and curve settings in your Duotone set affect the image.  By all means, if you have fashionable test strips that you want to tint, develop a pre-set for test strips, just don't expect it to work the same on actual images.  


regards,

Nigel

[nsecomb]

Thanks Jeff, I had thought of using the Tri-toned version as a layer and blending using colour, but I've only made very basic actions in the past and hadn't gotten around to trying to develop an action along those lines.

I'll give it a go as it'd be nice to keep the image in 16 bit, rather than going down to 8 bit.


kind regards,

Nigel