Applying the lessons — post-processing workshop 1

I'm going to try to put some of the lessons from the classical portraitists together now, by looking at a 'snap-shot' or candid image taken under natural light, and how it can be worked over in post-processing.

This picture was taken in Birmingham Botanical Gardens on a day out with no special preparation, and no access to fill lighting or even a reflector.

Here is the image as shot:

There are some strong aspects to this image which make it suitable to look at further. The camera angle gives a pronounced angularity to features, making the face distinctive. The angle of the face, chin slightly sticking out, hints at a determined character. This could be fortuitous and misleading, but, in the case of this subject (actually, my wife) it reflects character accurately. Blonde hair, blue eyes and this facial bone structure typically attracts the photographer to shooting a portrait which emphasises natural beauty, idealising the sitter into little more than an object of beauty. This may be fine for a glamour or model shoot, but it misses the point of portraiture, which is to represent the sitter, not to create a sensuous image for its own sake.
But, equally, there are some issues with this image which have to be thought about. The background is visually confusing, and does not say a great deal about the subject: although a reasonably attentive gardener, this sitter is also a translator, a musician, a traveller, and a speaker of six modern languages. A studio composite might attempt to bring these together into a single image, but this particular background, in as much as it communicates at all, misleads.

Also, this particular image is just a little bit too candid. the artefacts of lighting throw the skin into unnaturally sharp relief. Likewise, the wind blown hair introduces too many sharp edges, taking the viwer's eye down to the middle right of the picture.

The following picture represents a preliminary study for how this image might be treated:

I say this is a preliminary study for an important reason. It's tempting to launch straight into Photoshop and work to produce the finished image in one go. But doing so almost always leads to the postprocessor applying his 'standard' workflow. This would lead us in the direction of the Dr John Ashcroft portrait, where the artist applies all the usual rules to produce an image which, three centuries later, could be a picture of any doctor. For preference, we would rather go in the direction of the 'Portrait of a collector' which, despite our lack of information on who the collector was, has the power to fascinate us with the man's character.
When I'm shooting an image for advertising, I will do some preliminary post-processing, and then pass it to the Art Director, who will perform sometimes quite radical surgery on it in order to make it work for the advertising campaign. For a portrait, I will often post-process a 6 megapixel JPEG file, before returning later to the full sized Raw image.
Let's look at the image in more detail:
The first thing I want to do is to get the focus right. In the old days I would have used a capture sharpening routine, such as Photokit Capture Sharpener. However, I now use a true deconvolution filter which actually restores focus.
This is the result:
The upper image is the JPEG - Small - Fine downloaded from my D3. The lower image is the result of refocusing. I tend these days to refocus the skin once, and then do two more passes over the eyes only. I want to be able to actually read off the reflections ? this is something you can do in real life as you stare into someone's eyes. Eye reflections always appear sharp, even when we are so close to the face that everything else is blurry.
The next thing I want to think about is the skin. On a bad skin day, skin can be blotchy, reddened, and full of imperfections. A week later, the same person's skin may have cleaned up. What's more, because of the way that digital cameras capture images, the harshness of skin is emphasised, whereas in real life, our eyes rove across the face, returning again and again to the eyes and, to a lesser extent, to the mouth, teeth and nose, and to the aura of backlighting around the hair.
I now typically use a plugin called Portraiture to speed up this the process of fixing skin. There are many other processes you can use, but I like Portraiture because it is essentially a skin cloner, rather than a blurring routine. If you don't want to shell out for Portraiture (Nikonians get a discount), you can accomplish the same thing using the clone tool at a relatively low opacity and flow rate. You then keep going over the skin, so that texture is retained, but wrinkles and other irregularities are smoothed. Portraiture also sorts out the larger areas of shadow which give skin a structurally 'tired' look. You can sort this out with the dodge and burn tool and low levels. If you are going to do this, I would recommend doing everything on separate layers, so you can blend them in more or less.
In all this, this trick is to keep looking at the picture. No plugin or technique, when applied automatically, will achieve the results you actually need.
Here is the skin 'before and after'

It's not very nice to look at this close, and without context. But what you should see is that skin texture is preserved, so that the result is not blurry skin, but skin that looks in better condition. This is important, because not all portraits benefit from the 'soft focus' look.
For the preliminary study, the final phase is to mask out the distracting background. This is easy to do, but takes a little time. It is vastly simpler if you use a graphic pen rather than a mouse.
Photoshop now includes a convenient 'extract' tool. We used this extensively when it first came out, but we've reverted to old fashioned masking. To create a mask, double click the background layer in Photoshop, to turn it into a regular layer, and then click on the create mask tool. which is at the bottom of the layers palette. Set the paintbrush tool to black, and start painting away, until you've painted away all the annoying background. I usually prefer to paint too far in, and then change the tool to white and paint out. The trick when masking is to ensure that your mask is never sharper than the image is. If you have a sharp mask on a softer image (and all photographic images are softer than sharp masks), the result is an unnatural 1970s TV style cutout. If you end up with this, use gaussian blur on the mask to soften it.
Finally, create a new background layer, and fill it with something. A gradient will usually work better than a flat colour, and using a colour which is similar to the original background will help conceal any shortcomings in your technique.
My preliminary study now looks like this:

Actually, there's one more stage I haven't told you about. When I was finished, I went back with the history brush to introduce a bit more of the original skin tone into the picture, limiting the effects of the skin treatment we did with Portraiture.
Where now with this image? Assuming I'm happy, there are two ways forward. I can go back to the Raw file, extract it at a size as close to my final print size as possible, and do the process again. This doesn't take very long the second time through, provided you have kept a record of your stages. Or, alternatively, I can note the pose and recreate it in a studio environment, with more controlled lighting, and perhaps a pre-designed background.