The statue is a bit more successful than the man with birds, to my eye, as it expresses a better tonal range. The man and birds shot looks somewhat muddy in the areas that should be nicely detailed and highlighted. Both do not reach a maximum black and have a too soft, low contrast appearance.
Here's my B&W rendering workflow:
-----
This procedure is used with Photoshop CS2. I use a variant of the "Channel Mixer" formula, working entirely in Adjustment Layers. Some of it is automated, but some decisions have to be made per picture as well when you get to the finishing stage. The concept is fairly simple ... the devil is in the details.
Here's the fundamental idea:
- Be sure you're working on a well calibrated monitor and know how to use profiles in printing. You can't make a decent print consistently if your system is not calibrated properly.
- When you apply RAW conversion processing, don't make the mistake of presuming that you're going to make a perfect conversion that needs no further editing. Seek to output into RGB as much data as your image file contains, and presume you're going to be shaping and manipulating that data later. This means 16bit@channel output to RGB in PSD or TIFF formats.
The rest of the workflow is at the RGB channel level so applies to DNG, PEF, PSD, TIFF, or JPEG files equally. Of course, if you're working in JPEG files, due to the 8bit@channel nature of the files, editability is less, but that doesn't mean the workflow breaks down.
1) Look at the photograph before you begin and decide what you want to do with it. High key, low key ... decide where the IMPORTANT details are and where you are willing to let highlight and shadow detail go away. This is
the most important step ... You cannnot achieve a goal without knowing what it is.
2) I apply via a script an Adjustment Layer using the Channel Mixer tool with the settings R=20, G=70, B=5 percents. This is a starting point and not necessarily the best mix for all scenes. You can get a feel for how to manipulate this by turning the adjustment layer off and then looking at each channel in B&W seperately for a moment (the Cmd/Cntrl
, 1, 2, 3 keypresses let you do this very quickly and easily). Tweak the Channel Mixer settings to suit where you've made decisions about how you want your photo to appear in B&W ... if you have a lot of detail in the Red channel and not much in the Green or Blue, bias the mix to Red. etc.
3) If the image has several different kinds of lighting in it that changes the ideal mix in different areas, you can either
- insert Curves adjustment layers under the Channel Mixer layer and tweak the curves for each channel independently, with masking to separate the different areas.
- Mask the channel mixer adjustment layer and use a second or third one to change the mix, with masks again to localize the differences.
(I tend to prefer using the Curves technique as I find it easier to compress or expand a tonal gradient with it. In general, I use step three about 20% of the time.)
At this point you should have a close-to-final rough of your B&W rendering. Up to here, most photos will look like what you get from processing B&W film at a photofinisher. NOW it's time to make your image shine. ... Study your image again and identify what needs to be done to reach your goal.
4) I usually do overall sharpening for the full resolution image at this point as it will change local contrasts and edge effects that you want to take into account when doing tonal edits. Select the background layer, make a layer copy (no destructive edits to the base image...) and use CS2's Smart Sharpening tools. Small adjustments applied incrementally work best. Watch the important areas of the image at 100 and 200% scalings to detect haloing and artifact growth. Back off when you see them ... they look unnatural. Different images require different sharpenings...
5) Now, back to tonal shaping. Curves Adjustment Layers with masking inserted
above the channel mixer adjustment layer will operate only on the grayscale tonality. Again, small steps, one area at a time, with selective area masking ... I watch a particular area, get it the way I want, then fill the mask with black and brush in the adjustment with a soft edged brush and a slow fill rate until I get it the way I want. I build up each area of the photograph in this fashion, a little at a time, merging layers as appropriate when I reach certain points to simplify the document and save space.
6) Once you have everything done as well as you can manage, the rest of the workflow to render for the web is pretty fast. Be sure to save your work in PSD format to preserve all the layers (you should be doing that often throughout the editing process...). I do a profile conversion to sRGB, which auto-flattens the layers and uses the full 16-bit data in calculations. Next, from Pentax full-resolution files, I use "Image-> Image Size.." and resample the image to either 620 pixels for a horizontal or 530 pixels for a vertical, let the other dimension fall where it may, and set 72ppi as resolution (helps with some of the applications I use that honor the density and sizing information for on-screen display). At this point, you will often notice that the image has gotten a little darker. A Curves adjustment layer to tweak the tonal curve upwards, reflatten again. Sometimes a minor application of USM (.8 pixels, 30-40%, threshold=2) to resharpen. Then use "Image-> Mode-> 8-bit" to reduce it for JPEG output, and "File-> Save As..." to JPEG, quality 6.
You're done.
The key isn't just to follow the formula. The real work is:
1) Evaluate the image and understand your goals in rendering it to B&W.
2) Understand what each of the steps is meant to do so that you can modify the processing to suit a particular image problem.
Godfrey
