Deep
I’ll put this down now while it’s still fresh in my mind. Focus and matters of depth of field are purely optical affairs, right? So computers and software ought not by rights to get involved (other than for blurring focus, which is trivial). But over the last couple of years I’ve been introduced to two mildly obscure programs that have solved real, practical photography problems that I’d never even thought of tackling. And one of them recently made possible a shot I wouldn’t otherwise have attempted….
Factory worker pressing discs of tea, Yunnan
The laws of optics set very definite limits for how much of a scene-in-depth can be sharp at the same time. Stopping down the aperture is a solution up to a point (and only a real point, as in a pinhole camera, comes close to endless depth of field). But a tiny aperture does the image quality no favours, as diffraction reduces sharpness. And the smaller the scene, as in a close-up, the less of it can be brought into sharpness.
A long time ago there was a specialised technique for overcoming this in macro shooting, and it involved basically a slit-scan set-up (Stanley Kubrick used this in 2001: A Space Odyssey, incidentally). The idea was to have a precisely collimated thin sheet of light shining on the subject — such as an insect — at exactly the plane of focus. Everything else completely dark. Then, the subject was racked closer, different parts of it passing through the band of light, in a time exposure. The result was front-to-back sharpness of something never normally seen that way. But it was a highly specialised and painstaking method.
Then along came multi-shot digital photography, the general idea being if you take a set of frames in which the framing is identical (camera fixed on a tripod), but some particular quality varies, such as exposure, then software can combine the best of each frame into a single image. The algorithms tend to be complex. The best-known multi-shot technique is exposure blending, in which a high-contrast scene is brought into the range of a single image by taking a series of different exposures. HDR (High Dynamic Range) imaging is one widely publicised method.
I remember thinking when writing a book on HDR that it would be more amazing if the same thing could be done with focus instead of exposure. Take a series of images of a deep scene by starting with the lens focused at the nearest point and ending with it focused at the furthest. But this is no small matter for a software engineer. The images have to be compared and the sharpest version from each point has to be given priority. More than that, the scale changes when you re-focus, so all the images have to be re-scaled to match. Try doing this by hand on a layered stack of images, erasing bits at a time, and you’ll quickly get an idea of the problem.
Well, the problem has now been solved, in one of those delightfully designed programs in which, while you can make adjustments, you normally need do nothing but press a single button. Using the Ukrainian software Helicon Focus, I was able to shoot and process the image above, and with no pain. The Pu’er tea from this region of China (southern Yunnan) is typically pressed into cakes the shape and size of a discus. In this factory, the traditional stone weights were replaced by a hydraulic steel press. The pressed cakes were stacked in shelves in front of the press, and I decided to shoot through. Then realised that I could close in with a telephoto lens and still get everything sharply focused by taking a series. Rather than stop right down and suffer quality loss, I set the aperture for ƒ11, did a quick dry run to check the near and far limits of focus, and shot the entire sequence of 14 frames in 20 seconds. That was it, Once the Raw files were processed, the Helicon software did its bit in just under 2 minutes — and these were 16-bit, 24 mpx files, so a total of 336 MB loaded.
Focus blending from a sequence of differently focused images
As usual, this entertaining but focused example made me think more about the subject in general. Focus — where and how much of it — has at various times caused conflict of opinion in photography. A major driving forces behind lens development has always been sharpness, and the ideal for almost every lens ever made has been a high resolving power. No surprise there, and not much ether for getting s much of the image as sharp as possible. This, after all, was what cameras could do.
Among the best-known evangelists for sharp, deep focus were the West Coast photographers who in 1932 formed Group f/64 (named for obvious reasons) and who included Edward Weston and Ansel Adams. They were in no doubt that this was what photography should be about. Weston wrote, “I just think — today — that there is nothing more beautiful as a sharp, long scale, glossy photograph.” The group formed as a reaction to the soft-focus, dreamy style known as Pictorialism. Ansel Adams was particularly indignant: “In the early ‘30s the Salon syndrome was in full flower and the Pictorialists were riding high. For anyone trained in music or the visual arts, the shallow sentimentalism of the ‘fuzzy-wuzzies’ (as Edward Weston called them) was anathema…. We felt the need for a stern manifesto!” (for examples, see the Ansel Adams website, such as ‘Cathedral Peak and lake’ at http://www.anseladams.com/index.asp?PageAction=VIEWPROD&ProdID=197)
Ironically, Pictorialism began with a scientific attempt at visual accuracy, strange though this may seem. Peter Henry Emerson, whose 1886 Life and Landscape on the Norfolk Broads was a strong early documentary photography, developed a theory of selective focus, based on those of the German physicist von Helmholtz. He believed that photography had a duty to reproduce how we see, which for him meant sharp focus on one point, and the peripheral vision softer. Well, anyone attempting to reproduce the experience of vision in a single framed image is doomed to some degree of failure, because our peripheral vision suffers not so much from simple blurring as from what neurobiologists call ‘spatial imprecision.’ Some of Monet’s paintings get close to this. Emerson later retracted his theory, but this didn’t prevent a whole movement growing up around a misunderstanding of his ideas. The so-called Pictorialists aimed at soft and hazy images with no harsh blacks or whites, printed on textured paper for an even more sentimental effect. Or some samples of Emerson’s work, see http://www.getty.edu/art/exhibitions/emerson/, and for the kind of Pictorialism he didn’t care for, see http://www.leegallery.com/seeley.html
Modernism had no time for this, hence the Group f/64 manifesto. ‘Form follows function’ underpinned Modernist practice, and for photography this was taken to mean that the pure, correct approach was to let the camera and optics lead the way. If sharp focus and clarity of imaging were the special characteristics of using a camera, then that is how images should be, not manipulated into something else.
With a view camera, movements held the key to maximising sharpness. Rotating the front standard holding the lens, or the rear standard holding the film, changes the plane of focus from perpendicular, as below…
How tilt changes the plane of focus (which intersects with the image plane and lens plane)
The geometry of this follows what’s called the Scheimpflug principle, more easily illustrated than described in words. So, by tilting the lens forward strongly (or the rear standard backward, or both together), the plane of focus is tilted so that something a few inches from the camera can be in the same focus as the horizon. Ansel Adams used it in images such as ‘Mount Williamson’ (http://www.anseladams.com/index.asp?PageAction=VIEWPROD&ProdID=118), and it’s also at work in the picture below…
Dr. Johnson’s name in the accounts book at Pembroke College Library, Oxford
Or you could tilt the front and back in the opposite directions to this and get the shallowest depth of field possible, like this…
Scribe working on the St. John Bible, the Scriptorium, Wales
This kind of extreme selective focus goes in and out of fashion, and a few SLR lenses have built-in tilting (including an independent manufacturer – http://www.lensbaby.com/. It had a vogue about ten years ago, and a new, slightly different one currently. This latest fashion centres around shooting large-scale scenes — cityscapes, landscapes — with an unexpected shallowness of focus that makes them look like miniature sets. Here’s an example from the Lensbaby site — http://www.lensbaby.com/gurus-sweet.php. As fashions go, it’s fairly gimmicky, so don’t expect it to last. I wonder what Emerson would have thought of it.
And the second program I mentioned at the beginning? Restoring focus. Surely impossible? No, the answer is a computing-intensive method known as deconvolution. The principle is that if you know exactly what the lens did to de-focus the image in the first place, you can reconstruct by reversing the process. This goes way beyond sharpening, and at the moment there are very few retail programs available. One that I’m happy to recommend, as I’ve used it for a few years now myself, is FocusMagic— at http://www.focusmagic.com/
