Dan Margulis Applied Color Theory

Achieving Neutrality with Mixed Lighting

   Date: Sat, 03 Apr 2004 10:47:44 -0000
   From: “lwb_guitar”
Subject: Acheiving neutrals with mixed lighting etc.

Hi,

OK, been reading for a while and have managed to work up the courage to make my first post ...

I’m a graphic designer, working mainly with artists and arts organisations. My main line of work is producing publications which document gallery exhibitions, and thus most of the images I deal with feature a lot of white wall space.

Of course, a lot of the time the white isn’t white. A lot of otherwise respectable and venerable institutions have no clue how to light exhibits properly for photographic documentation, and a lot of otherwise excellent photographers have no clue how to measure the colour temperature of the lights in the space, or deal with daylight creeping into a tungsten-lit space or the three different types of fluorescent strip that have gotten mixed in with the lighting over the years.

I’d be interested to hear any strategies people on the list have for dealing with this kind of situation. I’ll detail one that’s been successful for me recently (after years of painstakingly painting between differently colour balanced adjustment layers):

First I get a rough neutral using a curve adjustment layer — i.e., kind of ‘splitting the difference’ between the various shades of white. Then I flatten and convert to Lab and make a Hue/Saturation adjustment layer and entirely desaturate the image.

The trick then is to blend that desaturated layer into the background layer using the advanced layer blending options: basically blending if the a or b channels of the background layer contain a certain amount of chromacity, and splitting the sliders to create a smooth transition. In plain words, whatever’s _close_ to neutral is replaced with the entirely neutral desaturated layer, and whatever’s more saturated shows through.

It occurs to me that this is a long winded way of doing something that could be done with one simple move in the saturation curve of the HSB mode that PS no longer supports (I looked into Curvemeister for this reason, but I’m not keen on the interface. Still making the decision whether to shell out for it). My method is a little more powerful, though, as it allows me to neutralise the a and b channels separately, so I can tailor my blending to where the colour casts lie.

Anyone have to deal with this type of images regularly? Comments, suggestions?

Be gentle with me!

Robert Johnston
___________________________________________________________________________

   Date: Sat, 03 Apr 2004 11:41:06 -0000
   From: Stephen Marsh
Subject: Re: Acheiving neutrals with mixed lighting etc.

OK, been reading for a while and have managed to work up the courage
to make my first post ...

Hi Robert, yes the list can be daunting at first, but it soon feels like a comfortable, favourite old piece of clothing after a short while. <g>
 
I’m a graphic designer, working mainly with artists and arts
organisations. My main line of work is producing publications which
document gallery exhibitions, and thus most of the images I deal
with feature a lot of white wall space.

This is an easy problem to solve, compared to many other common edits that are required when colour correcting.

Of course, a lot of the time the white isn’t white.

As this list’s focus is on what happens after an image is captured - we only have to explore correction methods in this thread and do not need to stray into the image capture. I am a firm believer in the person upstream knowing the next stages requirements and taking them into concern, and that getting the initial photography correct is usually better than post processing (but this all depends of course).
 
It occurs to me that this is a long winded way of doing something
that could be done with one simple move in the saturation curve of
the HSB mode that PS no longer supports

Robert, in some ways our different approaches are similar - but I think the one that I often use may be of some help in that it is perhaps less complex. See below...

(I looked into Curvemeister
for this reason, but I’m not keen on the interface. Still making the
decision whether to shell out for it).

As a beta tester on this product - I can say that the author is very receptive and responsive to suggestions (Mike beat Adobe to the punch in incorporating some of my curves ‘wish list items’).


Anyone have to deal with this type of images regularly? Comments,
suggestions?

My method goes something like this, if after a fairly fast but accurate cleanup of these areas (large flat areas like walls):

Sometimes I may use a duped layer, but this can also be done with adjustment layers. One thing I like with de-saturating the duped layer is that it is often more ‘tactile’ and one can easily see what one is doing to the data when blending.
 
In RGB, one needs to add a hue/sat adj. layer - or in CMYK one will also need a channel mixer (and curves if being really ‘accurate’ in forming a neutral ratio). This is blended in color blend mode. One simply de-saturates the entire image - but we only really care about the near white walls, which should vary greatly in tone from the other areas of the image.

I often find that a true de-saturation looks too artificial, one ‘expects’ to see some colour variation and transition of hue - but not to any real ‘perceptual’ degree, it is often more a ‘hint’ than a more noticeable hue variation...hope this makes sense. So the adjustment layer may only be blended at 70% or whatever to retain some small degree of hue variation in the mostly neutral scene.

It is then a simple case of using the layer option blend if sliders to only contribute this de-saturation of the lighter tones of the de-
saturation data to the lighter tones of the underlying layer. One can also add a ‘density mask’ (based off the 10 channel content or ~ selections etc) as a layer mask to this layer or adjustment layer data if not happy with the sliders or if one needs to quickly paint out the effect or lessen it in some spot areas.

One has to pay attention that coloured areas of a similar luminance range to the ‘near white walls’ will become desaturated, but it is easy to overcome this with the various commands and tools in Photoshop to mask out these areas.

This obviously presumes that you are working in a safe editing space RGB and you know the destination CMYK so that you can supply a neutral file...or that if in CMYK you have the right profile or ‘magic numbers written down on paper’ for neutral CMY values at various % points from white to black.

Be gentle with me!

No need Robert, you have asked a very on topic question and have demonstrated that you are more than capable of coming up with a very flexible and powerful solution on your own.

I look forward to your future posts.

Stephen Marsh.
___________________________________________________________________________

   Date: Sat, 3 Apr 2004 09:44:52 -0500
   From: Dragonfly Imaging & Printing
Subject: Re: Re: Acheiving neutrals with mixed lighting etc.

On Apr 3, 2004, at 6:41 AM, Stephen Marsh wrote:

I am a firm believer in the
person upstream knowing the next stages requirements and taking them
into concern, and that getting the initial photography correct is
usually better than post processing (but this all depends of course).

Absolutely Stephen. Those were my first thoughts as well.

With digital capture and RAW post-processing, many common problems will be a thing of the past. Exciting times!

It might also be possible to use a Gretag Eye-One device and their Share software to read the actual Lab values “in situ”.

John Toles
Dragonfly Imaging & Printing
http://www.dragonflyprinting.com/
http://www.dragonflygallery.ca/
___________________________________________________________________________

   Date: Sat, 03 Apr 2004 15:43:59 -0000
   From: “lwb_guitar”
Subject: Re: Acheiving neutrals with mixed lighting etc.

—- In colortheory@yahoogroups.com, “Stephen Marsh” wrote:

... This is blended in color blend mode.

That’s a really helpful tip. It hadn’t occurred to me that the overall luminosity can change when desaturating an image ... I was just using ‘normal’.

I often find that a true de-saturation looks too artificial,
one ‘expects’ to see some colour variation and transition of hue -
but not to any real ‘perceptual’ degree, it is often more a ‘hint’
than a more noticeable hue variation...hope this makes sense. So
the adjustment layer may only be blended at 70% or whatever to
retain some small degree of hue variation in the mostly neutral
scene.

Absolutely. If you have obvious signs of artificial lighting (spots on walls etc.) then the viewer _knows_ that it’s a tungsten-lit image and expects to see a little red/yellow casting. In a way it’s the viewer’s preconception that you need to match, and not the ‘actual’ scene if that makes sense. I often take the opacity of the desaturated layer down to 70% or so. Other times, when casts are really severe, I’ll keep it all the way up at 100% and then, after the fact, warm up the entire image with a small move toward red/yellow. Warm is always more palatable than cold, anyway.

One can also add a ‘density mask’ (based off the 10 channel content or ~
selections etc)

Can you explain what you mean by 10 channel content or ~ selections? I’m ashamed to say I haven’t bought any of Mr Margulis’ books yet otherwise I’m sure I’d be familiar with these terms ...

This obviously presumes that you are working in a safe editing space
RGB and you know the destination CMYK so that you can supply a
neutral file...or that if in CMYK you have the right profile
or ‘magic numbers written down on paper’ for neutral CMY values at
various % points from white to black.

I’m working in RGB (Colormatch), so it’s easy to gauge neutrality with the info panel.

Thanks a lot for your informative reply Stephen,

Robert
___________________________________________________________________________

   Date: Sat, 03 Apr 2004 09:33:48 -0700
   From: Andrew Rodney
Subject: Re: Re: Acheiving neutrals with mixed lighting etc.

on 4/3/04 7:44 AM, Dragonfly Imaging & Printing wrote:

It might also be possible to use a Gretag Eye-One device and their
Share software to read the actual Lab values “in situ”.

With a dot release of Match coming very soon, you1ll be able to read strobe with the Eyeone.

Andrew Rodney
http://www.digitaldog.net
___________________________________________________________________________

   Date: Sat, 03 Apr 2004 09:39:36 -0700
   From: Andrew Rodney
Subject: Re: Acheiving neutrals with mixed lighting etc.

on 4/3/04 3:47 AM, lwb_guitar wrote:

I’d be interested to hear any strategies people on the list have for
dealing with this kind of situation. I’ll detail one that’s been
successful for me recently (after years of painstakingly painting
between differently colour balanced adjustment layers):

The right numbers in your file can be output with regard to light by tweaking the output profile. That is, measure and build a profile for the output device with the light in which the print will be viewed. If you have say an Epson for this output you profile the paper but measure the light under which the print will be viewed with the EyeOne from GretagMacbeth using the ambient head and EyeOne Share. That can be built into an existing profile for the printer IF it was made using ProfileMaker Pro, also from GretagMacbeth. One profile from one set of measured data from the Epson can be built and updated in a minute or two (load the Light source from EyeOne Share as a .CFX file and click on the build profile button).

Makes a lot more sense to alter the output based on it1s eventual appearance than try and adjust the file which will likely need all kinds of tweaks per device.

Andrew Rodney
http://www.digitaldog.net
___________________________________________________________________________

   Date: Sat, 03 Apr 2004 23:35:52 -0000
   From: Stephen Marsh
Subject: Re: Acheiving neutrals with mixed lighting etc.

Can you explain what you mean by 10 channel content or ~ selections?
I’m ashamed to say I haven’t bought any of Mr Margulis’ books yet
otherwise I’m sure I’d be familiar with these terms ...

Perhaps more commonly known as a luminosity mask or luminance mask - but I like density mask as a term as the mask content may not be starting out from luminosity data.

The ‘mistermed’ luminosity mask is often generated via and cmd/ctrl + opt/alk + ~tilde key keycut to load a selection based off the composite channel of the image (one can also opt/alt click the comp. channel icon too). As Dan describes this data - it is a greyscale version of the images tones, it has less contrast than the
L channel of LAB.

The 10 channel concept for masking (Dan has an even more powerful 10 channel concept for colour correction which is more complex) is simply using the channels of LAB/RGB/CMYK as a source for starting a mask, if appropriate. If starting with a RGB orginal, the CMY plates may not help too much - but the K from CMYK and the AB of LAB are very different to the more similar RGB/CMY channels and may be more useful. The CMY channels will be slightly different, but generally speaking they will also have ink limiting and dot gain applied to them which may be less than useful in all cases...it all depends.

I may find myself using apply image to mix a mask from image channel
data from duped files in other modes, using blend modes, opacity,
fades to blend modes after the apply image step, curves etc.

The layer option blend if sliders in luminosity blending ranges are often doing the same thing - so the ‘density’ mask added to the layer or layer
set in question is often only used to fine tune what the sliders (which
are often opt/alt key split to perform smooth transitions in the blend if).

I’m working in RGB (Colormatch), so it’s easy to gauge neutrality
with the info panel.

So all you need is to convert to the correct output profile, and that the output device performs as per the profile description.

Dan has a lot of freely available information even if you do not have his book/s or been to his colour class:

http://www.ledet.com/margulis/ACT_postings/ACT.htm

http://groups.yahoo.com/group/colortheory

http://www.ledet.com/margulis/articles.html

http://www.ledet.com/margulis/PP7_Ch02_ByTheNumbers.pdf (this sample chapter should ‘sell’ you the book, beware it is advanced and things are taken for granted such as how to create a great mask, in the rare cases that Dan uses a mask/selection)

http://ep.pennnet.com/ (Dan’s Make Ready column, soon in the NAPP Photoshop User mag!)

Stephen Marsh.

Adobe Photoshop training classes are taught in the US by Sterling Ledet & Associates, Inc.