Deep-Focus, High-Definition Still Life with the Digital Camera – An Introductory Essay

– This article describes my method of making high-definition still-life photographs with extended depth of field, and how it evolved.  Technical aspects are discussed in fairly simple terms.  Another article, High-Definition, Deep-Focus Still Life with Photoshop Auto-Align Layers and Auto-Blend Layers covers much of the same ground, but delves more deeply into certain aspects of the procedure.  A third article, Step by Step Instructions, goes into the minutiae of the procedure.  (Both of these articles are to be found under Artist's Statements, the next-to-last item on the main menu).  Some comments, indicated by bracketed, small-case Roman numerals, are relegated to footnotes, to be found at the bottom of the page.

– Since the late 1980s until about 2003, I made still-life arrangements in the studio and then photographed them with a 4”x5” view camera, using black-and-white negative and color transparency film.  Then for a few years I did no more work with the view camera, and busied myself with preparing new work.  I also did a lot of printing in the darkroom.  By this time, it was becoming problematic to work with photographic film and paper.  I won’t mention all the difficulties, but will offer one example.  In 2003 I ordered a shipment of a favorite enlarging paper (a classic, Agfa Portriga), and was very satisfied.  A few months later I was ready to order some more, and found it was no longer available.  It felt strange to think that I had used some of the very last of the supply on the whole planet.  The end of an era!  It was clear which way the wind was blowing.

– Though I had some sentimental attachment to the old analog ways (most profoundly, for the mystical union of photons and silver halides) [i], most of my regret at the new trend was for entirely practical reasons, one of them economic.  Digital technology was very new, and the only digital backs available for use on a view camera were absurdly expensive, and not as good as they should have been for the price.  What’s available now is far better, but still unreasonably expensive. 

– The increased expense is indeed irritating, but the crucial thing for me in this forced transition has been the problem of image quality – or more specifically, overall sharpness or definition.  You don’t bother to use something as cumbersome as a view camera unless you want this quality in the first place.  And you don’t use anything less when each photograph you make – a carefully arranged still life –represents a great deal more work than most other photographic genres.  Though I don’t share the exaggerated and unfounded taste for bigness – really a false monumentality – current in the artworld for some time now, a 4”x5” film image can be printed quite large with little strain.  I wanted an image which would convey a sense of physicality that went beyond what I get from a 35mm film camera, whose syntax and mode of operation are better adapted to other kinds of photography.  I didn’t want to be forced to revert to an image quality hardly better than what I got in 1969 or 1999 with a 35mm camera. 

– I have used a few small digital cameras, ranging from crappy to mediocre to fairly good, but even the best was not of truly professional quality.  For sharpness, their zoom lenses don’t compare to the good, fixed-focal-length lenses I used for 35mm film.  They also tend to have considerable chromatic aberration – that is, color fringes –towards the corners of the image, most noticeable around distinct edges between bright and dark areas.  What’s more, they cause distortion, very evident with rectilinear subjects.  (These last two defects, color fringing and distortion, apparently typical of zoom lenses, can be easily corrected, at least with RAW files.  But it’s too bad one has to do such a thing at all, especially for a camera that is intended to take pictures by the hundred.  Worse yet, the correction of the distortion reduces slightly the definition of the image.) [ii]

– A much more professional item that I used briefly was a Phase One digital back, which is mounted on a standard view camera in place of the film holder.  Its image is somewhat sharper than one 35mm frame, and it allowed me to make a three-panel piece by shifting the back.  The three panels can be combined quite well later in Photoshop.  Very good, but still limited.

– Finally things have improved to the point where, for a medium-low four figures, you can get a professional digital camera and a good lens that makes an image superior to that of a 35mm film camera, and perhaps as good as that of a 4.5x6cm film camera.  I refer to the Canon 5D, and the better lenses offered for it. [iii]  This is what I’ve been using for this body of work.  But a single image made with the 5D still isn’t as good as that of a 4”x5” view camera.  (Let’s not even talk about the 8x10” view camera.)  “We’re getting there”, I thought, but I still wanted more.

– A concurrent development in digital technology was the growing ability to “stitch” or combine a series of contiguous or overlapping images into one seamless whole.  This was conceived and incorporated into digital cameras and software largely, I suspect, with the amateur sector in mind – people who wanted to take panoramas of scenic places and meld them together. [iv]  As one who has made many multiple-frame panoramas, deliberately working with the frames and interruptions as integral parts of the composition, and eschewing the smooth perfection of the one-piece item as less interesting, I could only look down on this rather easy and obvious photographic gambit.  (A photographer friend of mine once astutely observed to me – referring to the banality of this by-then-universal amateur practice, when one-shot panoramic film cameras had become a fad – that it was “impossible to take a bad panoramic photograph”. [v] Of course you will have a dramatic view of something, but inevitably its effect is more prominent than the subject matter itself.) 

– Unafraid to risk banality in the interest of science, I experimented along these lines all the same, just to see what could be done.  I took some interiors, some views of a terrace with a landscape, and of a wooded area, and around our house and yard, all panoramically.  The camera cued me (obliged me) to overlap by 35-40%, far more than I did with my multi-frame procedure.  When I carried out the Stitch or Auto-Blend Layers command to combine the images, the results were only occasionally about what I’d have expected.  More often they were quite wild and unpredictable, especially when I’d been photographing obliquely – that is, aiming the camera higher or lower than a perfect horizontal, which is hardly unusual.  There were extravagant distortions (perhaps suggestive of expressionistic styles) that might have been appealing to another photographer, but I wasn’t looking for this.

– But when I started to photograph some new still-life arrangements again in the studio, the idea quickly came to me that I could take the pictures in sections and piece them together again, if necessary “by hand” in Photoshop, by using the Transform command to make the necessary realignments, and masking as need be.  But Auto-Align Layers, I found, could also be used to very good effect to this end if an overall image is used as a base to which the separate sections will conform (rarely the case when stitching together panoramas).  Auto-Align is followed immediately by Auto-Blend Layers, which applies masks to the sections (and does a bit of lightening and darkening) to combine the sections as smoothly as possible.  This usually gives a result that requires only local retouching to correct small discrepancies or discontinuities.  (Though it’s not necessarily quick and easy, at least not as I do it.)

– Now I could, with some work, get a definitive image file much bigger than what I could make with just a single frame.  Simply put, first I made a full-frame exposure of the entire subject, using a normal lens on the Canon 5D, as perfectly aligned as I could manage.  Then I determined how many sections to use to photograph the subject, and how much to overlap the sections.  Then I brought the camera closer to the subject and made the exposures of the separate sections, using the good 100mm macro lens, shifting the aim of the camera but keeping it in about the same position in order to maintain consistent perspective.  Finally I imported into Lightroom, made my adjustments, exported TIFF files of everything, and made the assembly of the large file in Photoshop, using the overall exposure as a base, and using Auto-Align and Auto-Blend as described above.  But as such this still presented problems, directly due to my breaking the subject up into sections while simultaneously needing to maintain consistent perspective. 

– What may seem by now an obsession with consistent perspective deserves a little explanation, unavoidably technical.  Perspective is strictly a function of point of view.  If the point of view changes very much from section to section, the different perspectives of the separate sections will not agree in the final assembly, even if Auto-Align and Auto-Blend do their best.  Discrepancies at border areas will be too great to fudge or finesse with retouching so that they will not be obvious.  Therefore, I have to keep the camera in more or less the same position.  But because I must change only the aim of the camera to photograph the sections, another problem is introduced.

– Since I work rather close to the subject, whenever I shift the aim of the camera up, down, left, or right, I encounter a problem of depth of field, or depth of sharp focus.  My subjects are generally flat and not too deep, and when I photograph them frontally there is little problem with depth of field –it’s usually easy to get a good compromise by focusing somewhere between foreground and background.  But shifting the aim of the camera off the center of a flat subject means – because the plane of focus of the lens is no longer parallel to the plane of the subject –that there will be a greater range of camera-to-subject distances.  This means in turn that depth of focus suffers.  (With a view camera, one can easily overcome this problem by shifting the back, keeping back and lens parallel to the subject, instead of changing the aim of the camera – a great advantage. [vi])

– The obvious way to deal with this problem would be to use a smaller lens opening to increase depth of field – one of the rudiments of photographic technique.  However – and this fact is not nearly so well-known, as the difference is subtle – at smaller apertures the image is less “sharp”, we could say less incised.  Sharpness is sacrificed for the sake of depth of field. [vii]  Besides this, the increased range of camera-to-subject distances is too great to be overcome by stopping down the lens, even to the limit.  So this was no solution.  (I’ve already explained why one can’t avoid the problem by shifting the camera this way and that to “map” the various sections of the subject, thus preserving the parallelism that depth of focus depends on – never mind how clumsy it would be to re-adjust the tripod for each section.) 

– I solved the problem by resorting again to Auto-Align Layers and Auto-Blend Layers.  I said before that this combination was conceived largely with panoramic assemblies in mind.  But one aspect is a boon to the serious still-life photographer (or to anyone who wants sharp focus both near and far):  Auto-Blend Layers offers the “Stack Images” option as a blend method, besides the more popular “Panorama”.  (With “Panorama”, it mainly tries to make the joined images contiguous.  But with “Stack Images” it chooses image areas according to sharpness.  Read on.)

– To use it, one makes separate exposures of a subject, each one with the point of sharp focus changed with respect to the previous, in order to focus sharply on all parts of the subject, at all camera-to-subject distances. The separate images are then superimposed in Photoshop, and then aligned with Auto-Align Layers.  Simpler yet, the working file itself is created with Load Layers into Stack, and with Attempt to Automatically Align Source Files chosen, which accomplishes the same thing.  Alignment is usually perfect. [viii]

– Then, finally – the really amazing and useful thing – Auto-Blend Layers finds, area by area, the layer with the sharpest detail, and selects them, masking out the parts which are less sharp.  We are thus provided with a depth of field that would have been impossible otherwise.  This function is far from perfect, but it’s a good start.  (All the same, some photographer friends habitually call these new functions “the Miracle”.)  Invariably there will be some adjusting of the masks, and for me it is considerable.  (The program seems, when distinguishing sharp areas from less sharp areas, to resort to a sort of camouflage pattern in the contours it chooses, as if trying to conceal borders with an irregular outline, especially where difference of sharpness is slight.)  I look forward to a greatly enhanced capacity in this feature.  It would save me a lot of work.

– I should make clear that at this stage – using Auto-Align and Auto-Blend as just described – I am only working with one section at a time. The final assembly comes later, after definitive, deep-focus files have been created of all the separate sections.  The bulk of the job lies in perfecting these files that eventually will be joined to form a whole.  It consists mainly of routine adjustments of the masks, comparing neighboring areas of the image to find where Auto-Blend didn’t do as well as it would in a perfect world.  I make a selection of the area whose masking I want to adjust, and open the mask of the sharpest layer, while closing the mask of the less-sharp layer.

– When all the sections have been brought to deep-focus perfection, I make copies of the files and unify the layers.  Then I make a new file by opening a copy of my main, overall, frontal exposure, which serves as a base layer.  On top of it I import the sections.  I determine how much I need to enlarge the overall single image to match the size of the sections.  (For me this is typically to 200% -250% of the initial size.)  I then increase the canvas size accordingly, and enlarge the overall image to fill it.  I arrange the sections where they belong, positioning them as closely as possible to the underlying image.  I do Auto-Align Layers, and the sections will align themselves to the overall image.  If the result is good, I then do Auto-Blend Layers, which will mask the sections, trying to avoid areas where discrepancies will be noticeable.  This is usually very good, but often needs a little improvement, done mainly by swapping masked and unmasked areas, just as with the perfecting of the separate sections.  Any remaining problems may then be retouched or patched over as needed.

Allen Schill

October 2013

©2014 Allen Schill.  All rights reserved in all countries.  No part of this document may be reproduced or used in any form without prior written permission from the author.  Any one is welcome to link to it freely, but I would like to be notified.

Footnotes:

[i]  Digital photography’s lack of photographic “grain” has never bothered me as I had expected it might (and I have always been fond of grain as part of the aesthetic of the medium).  A digital image, seen on-screen at 100% of pixel size (or more) doesn’t appear as a multitude of tiny squares, which I would expect to find ugly.  It looks instead like a fine-grained photographic film seen magnified.  At normal magnifications it looks as analog as any print enlarged from film.

[ii]  For a while there was an annoyance with the use of the G10, the best of these – but really only the fault of the RAW-developing program, Digital Photo Professional.  This was the default sharpening – level six out of ten – with which the images were visualized, too strong and none-too-well carried out.  Sure, you could reduce or fine-tune the sharpening, but I found no way to impose a custom default.  The best I could do was to reduce the sharpening to zero, export a copy of the image, and then do the necessary with the far better and highly controllable sharpening in Photoshop.  But this problem has been superceded with Lightroom, which is far better at controlling sharpness, noise, tonality, and much more.  Now I use DPP only for quick viewing when I first import new images from the camera’s memory card, and apply useful file names in the Finder.  After that I import the RAW files into Lightroom and do just about everything there.  I only use Photoshop for retouching and heavy work with masks and assemblies.

[iii]  For example, the EF100mm f/2.8L Macro IS USM, or the EF 24-70mm f/2/8L USM, which are pricey but superb, and not the cheaper but decidedly mediocre EF50mm f/1.4 USM.

  I regret the significant and repeated expenses of this transition, even more than I regret the sudden obsolescence of all my fine equipment, including an excellent 4”x5” enlarger.  I have more to say about the economic and commercial aspects of the changeover to digital, but this is not the place to say much.  The problems of dilettante photographers do not particularly interest me, as there is already too much mediocre photography even among presumed experts.  More important, and most unjust, is the fact that it’s very hard for a serious person without much money to obtain equipment of a quality that matches his or her ambition.  In the past it was relatively inexpensive.

[iv]  Or, to pick and choose the most favorable facial expressions of the subjects of a group photo, avoiding the yawns and grimaces, the blinks and the wayward glances. 

  It would be interesting, or at least amusing, to go the opposite way, to deliberately make a decidedly bad group portrait.  (Around 1960 the popular TV show Candid Camera did a bit that made fun of people’s propensity to fawn over the photos of other people’s new babies:  they did a photomontage of several baby pictures, combining the worst traits of all of them into the ugliest baby you ever saw, a sort of Neanderthal monkey.  Still:  “oooh, he’s so – cute.”)  Go against the current, I say.

[v]  Melvin Dennis, New York City, about 1990.

[vi]  Shifting the camera back is limited only by the circle of coverage of the lens.  A view camera lens will generally focus an image considerably larger than the area occupied by the film.  The photographer can frame the central part of the image (excluding the extremities of the subject), and then, by shifting the camera back (in which the film holder will be positioned) up, down, left, or right, while keeping the lens in the same position, photograph the extremities without altering the point of view.  Thus consistent perspective is maintained among the separate exposures.  (The only disadvantage of this is that the sharpness of the lens begins to fall off a bit near the extremes of the circle of coverage.)

[vii]  If you doubt this, take test shots in your studio at f/11, f/16, and f/22 (compensating not with ISO settings, which would alter the “grain”, but with flash output of, respectively, 25% of capacity, 50%, and full power – this might be 800 WS, 1600WS, and 3200WS).  Look at them well on your computer screen at 100% - the f/22 will be noticeably less sharp in its detail than the f/11, even though it will have greater depth-of-field.

  If you know how depth-of-field works, it seems counterintuitive – a small aperture means better depth-of-field, which means sharper, right?  But sharpness, or resolution, is also a function of other aspects of optics such as (obviously) lens quality and accuracy of focus.  A lesser-known factor is that lenses are generally at their sharpest at medium openings, which allow the lens to gather a lot of light to form the image, and which use “the best part” of the lens.  A lens that goes from f/2 to f/22 will give a sharper image (at the plane of sharp focus) at f/8 than at f/22, even if the focus is perfect in both cases.  The f/22 will have more depth-of-field, but will not be as sharp at the plane of sharp focus as the f/8.  The image formed by a lens at a small aperture begins to show the traits of a pinhole camera image, which has virtually infinite depth-of-field but which is not really sharp anywhere.  (The pinhole might be around f/128 or f/256.)

  These differences are subtle, and are not what make a picture good.  But as long as I am sure of what I’m doing on the aesthetic side, I like to do whatever I can to make them technically all they can be.

[viii]  Focusing subtly changes the size of the image produced in the camera – focusing on near objects makes the image a bit larger, and focusing in the distance reduces it –so some re-alignment is necessary for them to be perfectly superimposed.

Using Format