Username: Password:

Making "Ballroom in B&W": Adventures in Using Calculations & Contrast Masking


 

I think this image is decent fodder for some blog discussion, primarily because of its imperfections and the degree of difficulty in obtaining a usable print. And I very much like the final print. Perhaps all the more because of the blood left on the field of battle in getting there. (You can find the regular portfolio version of the image here: http://www.photocritiq.com/set?setid=690#photoid=27032)  

Problematic is a pretty good adjective for this image’s origin file. That said, what’s useful to note about this photograph’s “development” is that: 

 

  • 80% of the heavy lifting in post processing was done in the raw converter. Unusual and still somewhat startling for me. I have no doubt that the ability to edit in linear gamma made the difference in executing a usable print since most of the work was done in shadow and highlight areas. 
  • I used calculations to fine tune the monochromatic conversion, after initially using a global Black & White adjustment layer in Photoshop. Calculations are the wild and woolly underpinnings of Photoshop and few people use them in their unadorned form…but there are nuances of image control available in calculations (or Apply Image…) that do not exist in the more polite user interfaces of the program. 
  • I used a localized contrast mask to slightly compress the upper third gray tones in the final print file. To my eye, contrast masks, (in moderation), often bring subtle details to fuller life by dampening edge contrast but without loss of sharpness. 

 

Here’s the unadjusted raw file with EXIF information: 

 

 

The issues here are obvious. The scene’s dynamic range exceeds the capability of the sensor and detail has been lost in the overloaded pixels of the highlighted floor. Shadows are pretty deep, despite the + ⅓ of a stop exposure bias. Lens flare takes an ugly turn and because this was taken through glass, at an angle, there are small reflections everywhere. In short, a mess. But, I liked the shot. The perspective of the drawn shadows and draped fabric in the overbearing flood of light is very compelling to me. 

So I worked on the file in Adobe Camera Raw (ACR) and managed to make the image worse than what I started with. Stop, put aside. A week later, I tried again and reached new lows in post processing. Stop, put aside. Another week later, my confidence in editing refreshed, I tried a third conversion. Better but nothing that I’d care to print. Stop, put aside. Now I am thinking about the damn image in my spare moments, trying to visualize what I want in the print. Two weeks and three other photo shoots go by and I try a fourth time, having erased all vestiges of the prior efforts. This time I get something workable but no dunk shot. It looks like this: 

 

Big moves in highlight recovery and fill light to compress the tonal range. Color temperature cooled a bit to offset yellows. Blacks boosted to keep some punch in the image. It all looks  pretty prosaic now but it took me a long time to get here. 

Here are the adjustments made in parametric and point curves. Changes in the parametric quadrants echo the basic adjustments, i.e. compressing tonal range. The point curve, however, was used to restore some of the lost contrast. 

 

And here are the HSL adjustments to pull back the sky’s “blues”. :

 
 
As a final note on the raw conversion process, I also applied color calibration adjustments that are customized to my “unique” sensor. These factors were obtained by photographing a GretagMacBeth ColorChecker Chart and then running Thomas Fors excellent calibration script, which can be found at http://www.fors.net/chromoholics/index.php
 

Once in Photoshop, edits were comparatively forthright. Here’s the layers palette for work file: 

 

More light was tone-mapped to the interior. I boosted midtone contrast, and then converted to black and white, using the infrared preset as a starting point. I used a multiply blur on the lower-than-midtones and then went to work on the flare spots with the intent to mitigate rather than eliminate their impact. 

Because I was still lacking detail in the highlight areas, I elected to use calculations in multiply blending mode and localized to a bright-lights luminosity mask. This did a nice job of bringing back some detail to the burned areas. 

The work file was finalized with a mild dose of smart sharpen and local contrast enhancement (unsharp mask at a wide radius), with the blend-if sliders set to the midtones. 

I duplicated the work file, flattened it, and set the file format to TIFF for the print version. 

 

After setting up my soft proof, I used a 50% opacity contrast mask on the upper third of the image to further even-out the gray tones. The “recipe” for a contrast mask looks like this:

 

And the print file layers, with output sharpening, looks like this:

 

which outputs an enchanting 18 X 12 print. 

 

 

 

 

 

 


 

 

by Patricia A. Minicucci on January 5, 2008, 6:18 pm Tags:

Comments:

by Aaron Falkenberg, Sun January 13, 2008, 12:14:27
Wow Pat, I had no idea just how intensive the transformation of this image was. I remember when I first discovered contrast masks ala Luminous Landscape. It was one of those "light blub" moments.

Very interesting use of both parametric and point curves. I've generally used one or the other, but I see there may be benefits to using both.

Three cheers for color calibrating your camera! What effect do you think it had on the black and white conversion?

Was the tone-mapping a seperate process? I'm only familiar with the term from within Photomatix.

"Calculations are the wild and woolly underpinnings of Photoshop." So True. I still go back to it if I have a particularly tricky conversion because of its link to blending modes, but for 90% of things I find the Channel Mixer and some good old dodge+burn gets the tones exactly where I want them. The new "Black and White" in CS3 is ok; being based on the brightness or darkness of individual colors, it's more "visual" for those who aren't quite sure what to do.

Once again, tremendous workflow. What kind of paper did you print this out on?
by Patricia A. Minicucci, Sun January 13, 2008, 15:33:36
Aaron: Thank you for the thoughtful comments.

Re the point curve in ACR, I think of it as an instrument for small, precisely-targeted changes, while the parametric curve uses something more akin to blunt force. It is like using a lockdown curve in PS, allowing just a small selection of pixels to be tweaked. That said, I am always amazed by what a difference it can make.

The camera color calibration factors (in its latest iteration) makes an enormous difference in my sensor's ability to offer up richer colors generally and blues in particular. I've used Thomas' script for years, for three different cameras, but this latest version has really hit a sweet spot, at least for the 5D. I've keep a spreadsheet with all of the different calibration results generated over the years and it is interesting to see how updated versions of the script and ACR change the factors when using the same base image. The swings are not major but a refining process has been at work and the results show it.

I do not know the answer re impact on B&W conversions but I expect there is some. I will tell you that when I know I want an image to be B&W, I often juice the Vibrance scale in ACR to help tonal differences stand out. (I used to use steepened A, B curves in LAB mode.)

Re tone mapping, I use Photomatrix's plug-in at very low strength (like 8 on the scale of 100) and frequently fade and mask the output for selected areas, although I did not mask this image. It does a nicer job of subtle, complex tonal shifts than a straight ahead curve, to my eye.

I used to use calculations for almost all B&W conversions. The exception was when the color channels revealed distinct textures for certain elements. Then I'd use John Caponigro's method of breaking out the lightness and color channels as distinct layers with individual masks. Sort of a channel mixer on steroids. I've grown to like the B&W adjustment layer tool, though and it is weaning me away from calculations. It's weakness is that it often assigns a too-broad color category to variable tones. I've found this can be overcome by using more than one B&W adjustment layers with masks for selected tonal areas and knock-out set to "deep". For images with distinct color channel textures, though, you cannot beat Caponigro's technique. There is a tutorial PDF and action script for it here: http://www.adobe.com/digitalimag/ps_pro_primers.html

I printed this on Harman's Gloss FB AI, using ImagePrint's RIP and grey profile. Expensive but delicious.

Please login or register if you want to add a comment.