Step by Step Instructions for High-Definition, Deep-Focus Still Life, using Photoshop Auto-Align Layers and Auto-Blend Layers

This article goes into excruciating detail about the procedure I’ve developed to create these images.  At the end there are some footnotes, and two appendices.  If you prefer a more general but still advanced commentary, see the related article High-Definition, Deep-Focus Still Life (with Photoshop Auto-Align Layers and Auto-Blend Layers), nearby here in Artist's Statements.  If you would like a relatively simple description, see the related article Deep-Focus, High-Definition Still Life with the Digital Camera – An Introduction, which is grouped with the photographs themselves.  

Making the Exposures – General Considerations

– Make overall exposure with a normal or wide-angle lens.  You may wish to bracket.  Determine how many sections to use to “cover” the whole image, allowing a slight overlap.  Be sure to figure the overlap on the basis of the sizes of the images when focused at their nearest points, when they will be magnified the most.  Observe this well, and back off a bit if your image is getting too tight.  Allow also (if you’re not using a nodal point tripod head) that the image will be bigger still when aiming down (because, in swinging on the head, the camera comes a bit closer to the subject), while a bit smaller when aiming up (and the camera is a bit further).

– Bring the camera closer to the subject, and make exposures of all sections using a longer focal length, at all necessary degrees of focus.  Manual focusing, I think, is unavoidable for this.  Equally indispensable is to be able to see each exposure immediately on your computer screen so as to critically evaluate the focus.  (Your movements of the lens barrel when adjusting focus are extremely subtle, a few tenths of a millimeter at a time.)  [i]  You may wish to bracket these exposures as well.  (In this procedure, bracketing serves as much or more to have a range of different depths of field to choose from than just to make sure you have an ideal exposure.)  I suggest, when making the exposures, to proceed from the nearest points of focus to the furthest, as this will help you to keep them in order later.  I further recommend, if you want optimum definition, that you use ISO50 (the “L” setting in the camera’s menu) for the exposures of the sections – besides having less “grain” than ISO100, it retains noticeably more shadow detail.  (Keep in mind, if you are exposing manually, that the sections will be somewhat darker – at a given f-stop – than the overall image, due to the bellows factor.  This can be considerable when you work close to the subject, easily 1/4 stop or more.)

– Furthermore, set your camera at a consistent white balance, probably either Flash or Personal, so they’ll all start out the same.  If instead you use Automatic White Balance, the exposures will have slightly different white balances, probably not from shot to shot, but almost surely from section to section, and you’ll have to apply a uniform white balance later in Lightroom.  If you are thinking of your image in black and white, make your exposures in color anyway; you’ll have more latitude later in making a B&W conversion, and often this can make a considerable difference.  (Note: I refer always to Lightroom, but most of these functions can be carried out the same way with another RAW development program.)

– As for the exposures made in series at a variety of focuses, just how many you make will depend on your equipment, on the size and the depth of your subject, and on what you want.  As a rule, you’ll need fewer with a large subject (because you’ll be further away) and more with a small subject, while you’ll need more with a deep subject (because more depth-of-field is required) and fewer with a shallow subject.  My subjects are usually rather shallow, yet I find I need a six to ten different focuses per section, even more if the subject is deeper than usual.  In addition, I have to allow for the uncertainty of manual focusing, so I shoot several more to be on the safe side.  Many of these will be discarded later when it’s clear that they are superfluous, but this is much less trouble than having to re-shoot when you find you have “gaps” where the focus is inadequate. [ii] In practice I commonly shoot 20 to 30 different focuses of every section, of which I may wind up using only eight or ten or fifteen.  You may wish to limit this number for your first attempt.

Adjustments of RAW Files

– Import all RAW files into Lightroom, choose the best exposures, and make any tonal adjustments, white balance and other color corrections you want.  For uniformity’s sake, these adjustments of course should all be the same. (However, if you bracketed your exposures and want to use any besides the standard ones, adjust them to agree with the standard.  E.g., if f/11 is the standard exposure, an f/16 shot will match it very well if you increase its exposure by one stop.)  If you used Automatic White Balance to make the original exposures, be sure to make them all the same now.

– You will probably have to Enable Profile Corrections on all files, to eliminate the curvature of straight lines, distorted by the lens. [iii] (This is especially important if you use a zoom lens, or rather any lens that causes such distortion.)  This should always be done to the RAW file, and usually does a perfect job, but take a good look.  Fine-tune if need be, using the Distortion slider on the Profile side, or the Transform sliders on the manual side.

– Be sure also to Remove Chromatic Aberration from all files at this point.  This command is all but useless on a non-RAW file, although Defringe (on the Manual side) may help somewhat.

– Besides correcting any distortion, you must adjust the overall shot so that it’s squared with the borders of the image – that is, using the Transform sliders on the manual side of Lens Corrections.  This will be important later when making the final assembly.  Think of the overall shot as a sort of rough version for the high-definition image to follow as a guide, so it must be just what you want before proceeding.

– As for the sections, on the other hand, just make sure you’ve applied the profile corrections – that your straight lines are straight (presuming you want them straight), with manual corrections as well if necessary – and remove the chromatic aberration.  (They must be straight, as they will be difficult or impossible to straighten later. If there’s nothing rectilinear in your subject, this will be less of a concern.)  It’s all right if they are somewhat off-square (inevitable when shooting off-axis), since they’ll be transformed anyway in the final assembly.

– If “noise” is a problem, apply Noise Reduction, since it’s much easier and more effective with the RAW files.  Avoid too much sharpening at this point; it’s something to do at the end, after the final assembly. (Note: check that all files are tonally equal, and correct before proceeding.  My electronic flash system is presumably consistent, but I still occasionally find a file that’s off by a sixth of a stop or more.)

– Export TIFF files of all the files you’ll be using.  (In this you may wish to proceed section by section, to avoid having all those heavy TIFFs just hanging around waiting for you to get to work with them.  After all, you will only be using them one section at a time.) I always export at 16-bit. [iv] In the Finder, rename all the files – I use Automator – to include only the file number generated by your camera, maybe a short name, and any other brief information that will help you keep track of what you’ll be doing in Photoshop.  (E.g., if I’m using files taken at different lens openings, I include the f-number as part of the file name.)  You will want room to add to these file names shortly in Photoshop when they will have become layers.

– In Photoshop, choose “File/Script/Load Files into Series”, selecting, one set at a time, the images with slightly different focuses, and choosing “Try to Align Source Images”.  (If you don’t do it now with the Load Files command, you’ll have to do it immediately anyway with Auto-Align Layers.)  Check the alignment at 100% of pixel size, Hiding and Showing layers and examining the extremities.  Alignment is usually perfect, but it must be, or you’ll have worse problems later. If you have proceeded from the nearest points of focus to the furthest, as I suggest, they should load in such a way that the closely focused exposures will be at the top end of the Layers palette, and the more distantly focused ones towards the bottom.  (This may save you some rearranging.)  They should load with each layer named for the file it came from, which should include its file number.

Notations in Layer names

– In the Layers palette, load the selection from each layer in turn and note its size in the Info palette. Incorporate the size, length times width, into the name of the layer.  (The layers will have changed their sizes in the Alignment.)  Once you have done this to all the layers, compare their sizes and make sure the layers are ordered accordingly, moving them up or down in the Layers palette as need be.  The smallest layers (close-focused) go on top, the biggest on the bottom.  (This may seem counterintuitive, but because the close-focused layers start out at a slightly greater degree of enlargement than the far-focused layers, they must be reduced slightly, relative to the far-focused layers, in order to align with the far-focused layers.  The “bigger” or more-enlarged get smaller, and the “smaller” or less-enlarged get bigger.)  Note any layers that are identical in size or very close to their neighbors and indicate them with a color; you may be able to do without the redundant ones.

Prepare for Auto-Blend Layers

– Take the selection of the topmost (smallest) layer, hide all the other layers, and bring guides from the rulers to the extremities of the top layer so as to leave no transparency, even partial, within the guides.  (If there is any transparency, Auto-Blend Layers will interpret borders between visible pixels and transparency as “sharp”, and so skew or mislead Auto-Blend Layers in making its masks.  You’ll be left with part of an edge masked as sharp just because it’s an edge, and not because it’s sharply focused.)  View at 400% or higher to see individual pixels, and check all the corners well.  The sides of the image will probably be slightly cockeyed now, due to the alignment of the source layers, so don’t expect them to be perfectly horizontal or vertical anymore.

– Hide the top layer and show the next one down, which will probably be a few pixels bigger.  Again, viewing very large, check all the corners to make sure no transparency lies within the guides.  This is necessary because there are often several pixel’s worth of difference in what the camera has framed from one shot to the next, even using a very stable tripod, and so even a layer that is slightly bigger may leave transparency on one or two sides (especially when the difference between one layer and the next is very slight).  If so, bring the guides in a few pixels closer to exclude the transparency.  Hide that layer and show the next, and check again that there is no transparency within the guides.  Do it again if need be, until you are sure there is no more transparency.

– Reduce the image so you can see it all, choose the Crop tool from the Tool palette, and set the crop to the guides.  Enlarge it again to see individual pixels, and check the position of the crop at all corners to make sure it is aligned with the guides.  It may easily miss the guides by a few pixels even if the guides are magnetized (apparently more attracted to the edge of some layer).  Carry out the crop.

Auto-Blend Layers

– Select all the layers, and do Edit/Auto-Blend Layers, using “Create Series of Images” as Blend Method, and leaving “Uniform Tones and Colors” unchecked. [v] In a moment or two, Lightroom will attempt to find the sharpest areas of all the layers, and generate masks to show them, hiding the less sharp areas.

– Check the image closely to see how well the command did its job.  Hide and show the various layers, or take their selections, to see easily just what they include and exclude.  You’ll see immediately where it has been quite arbitrary and wrong.  Now the lengthy (and repetitious) part of the whole procedure begins.  (Have something nice to listen to, music or a spoken-word, since at this point your job is completely visual, and a large part of your brain will be unoccupied by the need to do any non-visual thinking.)

Sets of Layers – Saving – 4GB limit

– If you have more than just eight or ten layers, you may wish at about this point to divide them into sets, such as “foreground”, “middleground”, and “background”, just to be able to manage them more easily.  Of course you will need to Save your new document from time to time.  Right now, after your Auto-Blend Layers, is a good time to Save; the file will be at its heaviest.  If you have very many layers, your file may even exceed TIFF’s 4GB limit, and you will have to divide your file into two parts – or even three.  (You may wish to work with less cumbersome files just so your computer will work more smoothly.)

– To do this, choose Image/Duplicate and make copies of your file. In one copy, assuming a division into two parts, eliminate the layers focused on the foreground and middle-foreground, and Save it.  In the other, eliminate the layers focused on the background and middle-background, and Save it as well.  (If you haven’t been able to save the entire file yet, don’t close it until your newly subdivided files have been safely saved.)

Intermediate layers for very large files

– In making these divisions, however, it’s very convenient to keep an intermediate layer and its mask present in both upper and lower halves of a file – at the bottom of one half and at the top of the other.  That way, I can use that layer mask as a sort of holding area for the selections that must ultimately be applied to a layer not present in the part of the file I am working on.  I just pass the selections down from the upper layers in one part, and pass them up from the lower layers in the other part.  When I’m ready to switch to the other half of the file, I open it up, and replace the transitional layer with the now-modified one I’ve been using as a holding area (dragging the layer from one file to the other with the Caps key held down). Then I work on that layer mask, distributing its open parts to the layer masks of the proper layers.

– To establish this intermediate layer, drag the bottom layer of the foreground and middle-foreground to the top of the file containing the middle-background and background.  (Or else, drag the top layer of the latter to the bottom of the former.)  If you’ve divided the file into three parts, copy the bottom layer of the foreground to the top of the middleground file, and the bottom layer of the middleground file to the top of the background file.  Save again, and your files are ready to elaborate separately.

Adjusting the Masks – Swapping masked and unmasked areas

– Essentially you proceed, area by area, and adjust the masks to better show the sharp areas and hide the less sharp.  You will see, in broad areas more or less the same distance from camera to subject, large areas that are unambiguously the sharpest.  Bordering these will be transition zones – with protuberances from the large areas, or isolated blobs – where the sharpest layer might be the current one, or another one above or below it in the Layers palette (quite possibly far above or below).  The whole thing suggests somewhat a topographical map and somewhat a camouflage pattern.  Your job now is largely to make it look as much like atopographical map as possible, one that corresponds to the actual contours of your subject.

– Look at these areas at 100% and compare the current layer to one that may be sharper.  Do this by showing or hiding the layers in question, or by momentarily deactivating the layer mask of a layer above the layer you think you may want to hide in favor of another which is sharper in that area.  If the current layer is not the sharpest available, load its selection, then restrict the selection to just that part of the layer mask that you want to adjust right now. (Most likely you’ll use the Lasso, but any marquee tool will do.  At times you’ll want to simply subtract from the selection, rather than intersect with it, to leave the part you want.)  Leave feathering at zero, and anti-aliasing unchecked in the tool’s option bar, as you want very neat and sharp borders to your selections, like the ones Auto-Blend Layers made for you.  Hold down the Option and Shift keys while you apply the tool, or else choose “Intersect With” in the Options bar.

– Make sure you have the layer’s layer mask selected in the Layers palette, and fill your special selection with black.  That area will disappear from your image, as it is now hidden by the layer mask – you will have only transparency there.  (If that part of the image goes black, it means you had the image part of the layer selected, instead of the mask. Undo the fill immediately and switch to the mask.)  Keep the selection.

– Now switch to the layer mask of the layer which is sharper in the area of your selection, where you want to open up the mask.  Expand the selection by two pixels with “Selection/Edit/Expand”, and fill that part of the mask with white.  I use an Action for this step, consisting of two parts: “Expand Selection 2 Pixels” and “Fill with White”.  (The masks ordinarily generated by Auto-Blend Layers overlap by three pixels.)  The newly chosen area of the image will now be visible.  If you hop back a few steps in the History palette, you can easily compare the before-and-after.

– (Sometimes these two pixels will carry you a little too far, especially in areas that have an abrupt transition from near focus to far, and you’ll have to adjust the mask by trimming a bit from and transferring it back to the layer mask it came from.  As you proceed, take note of the areas where the mask is still too sharp; you’ll have to soften these a bit at some point.  It may be a good idea, in these cases, to open the mask of the underlying layer – focused at somewhat greater distance – a few pixels too far, and then, later on, opening the mask of the upper, more closely-focused layer, by painting with white with a very small but soft brush.  Don’t try to be quicker by using a bigger, soft brush to soften the edges of the masks – a zone which shows two layers, each at more or less 50% opacity, will just look fuzzed rather than sharp, and this may be noticeable unless it is quite small.  But save these fine-tunings for a later stage of the job, or at least for after you are sure the masks are basically all right, so as to avoid needless reworking.  Keep in mind, also, that if you subsequently sharpen the image, any discrepancies will become more evident.)

– If you unexpectedly see white instead of image, it means you had the image part of the layer selected, rather than the mask.  Watch out for this; if you white-out a part of your image and don’t catch your mistake right away, you’ll have a hassle replacing those pixels.  (Choose the checkerboard pattern to represent transparency – in Preferences – so that there will be no doubt as to whether you’re looking at transparency or white.)

– Find another area where you want to swap masked and unmasked areas of two different layers, and again, select the layer mask of the layer you want to hide in that area, take the selection of that entire layer mask, intersect it with a Lasso selection to confine it to the area you want to adjust, fill that part of the mask with black, switch to the layer mask of the layer you want to show in that area, expand the selection by two pixels, and fill that part of the mask with white.  Never forget the Lasso intersection step, or you’ll be filling the whole mask instead of just a part. This is easy to overlook when you are working close up.

– Proceed like this throughout the entire image until you are satisfied with the result, examining the image closely to find abrupt differences in sharpness.  Your eyes will get very sensitive to subtly different degrees of sharpness.  You will be able to see at a glance, without even enlarging the image, that a given area has been chosen badly – for example, an area in the background of your image where one of the foreground layers has been chosen instead. In those cases, you’ll be able to swap your masked and unmasked areas without enlarging the image to check closely.

– When you are working in an area of deep shadow, it can be very difficult to distinguish sharp areas from less sharp.  For this I create a sort of “floodlight” at the top of the layers palette:  a Levels adjustment layer, set at around 0 – 1.8 – 255, to brighten the whole image, and I turn it on when necessary.

– Sometimes your “target” layer for a given area will be approximate in terms of which is really the best layer, but you can clean up such areas later, when they are incorporated into the masks of other layers and you are working more closely, that is, on the finer distinctions of sharp and not-quite-so-sharp.  Contours will emerge in the layer masks that suggest that topographical map.  You will “see” your image in depth, and be able to judge very well where the layer masks are liable to be less than optimal: the bands that emerge will have protrusions that don’t conform to the contours of the subject, thus indicating where you need to adjust some more.

– You may see, as you go along, that one or more layers can be discarded because they are too similar to their neighbors to be worth keeping in the file.  If you entered the sizes of the various layers as part of their names before cropping, as suggested far above, you will be able to identify easily the layers that might be superfluous by comparing their recorded sizes. For example, a layer named “Img7005 - 47.84 x 31.90cm” and another named “Img7006 - 47.85 x 31.89cm” are so close you can certainly get rid of one or the other.  If you aren’t sure, save those layers to a new document of the same pixel size, so you can be reimport them easily if need be.  This will reduce the file size (as well as its complexity), which is always welcome.  (Keep track of the layers that have been discarded; if they are really unnecessary you can discard the corresponding RAW files as well.)

– You will also reach a point where you can merge one or more layers, further reducing the file size.  I find it most convenient (and least confusing) to work from the top down:  once I’m sure my uppermost layer is what I want, I apply the layer mask, load the selection of the pixels of that layer, switch to the layer mask of the next layer down, fill that part of the mask with white to open it up in that area, and merge the upper layer with the lower (Merge Down), maintaining the layer mask.  Then I will Save the file, and perhaps try to boil it down some more by merging more layers. (Again, if I’m not entirely sure, I’ll save the layers I’ll be merging to a new document before doing Apply Layer Mask and all the rest.)  As I proceed I keep track of what layers comprise a merged layer by renaming the layer accordingly.

– If I have divided the file into two or more parts, I work with the parts until I can eliminate enough layers to allow me to reunite the whole file.

– Perhaps the trickiest part of this procedure (probably best to leave until near the end) is the adjustment of the masks that follow abrupt transitions between near and far parts of the subject.  Often you are left with a noticeable, blurry fringe around the contours.  If it happens to be inconspicuous because it’s in a shadowy area, consider whether it might show if you subsequently decide to brighten the image (or just the shadows or the blacks).  Where it is evident, you will have to do some judicious retouching – cloning in from a nearby area – to recreate substance and texture convincingly.

– Once you have done all you think needs to be done (or all you can stand), make a copy of the file, still in layers, and do Merge Layers to get a copy of the file on a single layer.  You now have a file that is sharply focused throughout in a way that wouldn’t have been possible otherwise.  Take a good break, because now you will have to do the same to all your other sections. By now, if not before, you can discard the TIFF files from which you generated this Dagwood sandwich.

– Let’s assume you haven’t thrown in the towel, and have gone on to make deep-focus files of the other sections.  (If you’ve even read this far, you may be someone with the tenacity to actually do such a thing.)  Besides keeping definitive, multi-layer versions of all sections, you have single-layer versions.

– Open the TIFF of the overall photograph that will serve as a base for the final assembly. Double-check that it’s ready to use.  For me, with my boxy, rectilinear compositions, this basically means squared to the sides of the image. With this image open and active, choose again Script/Load Files in Series, and use Browse to choose the single-layer files you’ve worked so long to make.  Make sure “Try to Align Source Images” is unchecked, since you’ll be doing that later.

– The overall image should be the bottom layer in the Layers palette.  If it is a background layer, make it a regular layer, and call it “base”.  Arrange the layers of the sections above as you like.  Put them in a set, separate from the base layer, so you can show or hide them all at once, or change their opacity.

– Compare the scale of the overall image to that of the sections by putting the opacity of the sections at 50% or so. For example, if a given part of a section measures 20cm and the same part of the overall image measures 10cm, the difference in scale is 2:1. If instead they are 24cm and 10cm, the difference in scale is 2.4:1 or 12:5.  You must enlarge the overall layer to match, as closely as possible, the scale of the sections.  In the case of our examples, this would mean to bring the overall layer to 200% or 240%, respectively, of its original size.

– Increase Canvas Size to whatever this larger size needs, plus a little bit more to accommodate eventual transformations at the extremities of the image.  E.g., if the overall is to be increased to 240% of its original size, increase canvas size to 250%.  Put a set of guides along the edges of the overall layer before you enlarge it.  By now, if not before, Save the new file.

– With Transform, enlarge the overall layer as planned.  Put new set of guides along the edges of the newly enlarged layer.  Lock the layer in the Layers palette so that it can no longer me moved or transformed.

– With the opacity of the section layers still at 50%, rearrange the sections to superimpose them on the base layer as closely as possible.

– Bring the section layers back to 100% opacity.  Make sure the base layer is locked.  Select all the layers, base as well as sections.  Do Edit/Auto-Align Layers.

– See how well the sections now conform to the base layer and to one another.  Quickly Hide and Show sections, or reduce their opacity to 50%, to see better what you’ve got.  You will easily notice the separations between sections because there are no masks yet. If something is drastically out of kilter, back up in the History palette to just before Auto-Align Layers, shift the positions of the sections a bit, and try again.  If you see only minor, local discrepancies, it’s probably as good as you’re going to get.

– Deselect the base layer, leaving only the sections selected.  (Make sure they’re at full opacity.) Do Edit/Auto-Blend Layers.  (I recommend that you leave “Uniform Colors and Tones” unchecked as before, when you were preparing the separate sections.)  After a moment, Photoshop will spit back out your file with the section layers masked to hide the joins as much as possible.  If everything looks basically all right, save the file again.

– Inspect the result closely, and note where there are problems to resolve, big or small.  I always make a new layer for notations – circles highlighting local discrepancies, and arrows to indicate the direction (and a number to indicate how much) a given portion needs to be moved to conform to the overall image.  For example, an up arrow near the border of a section, with “2mm” written beside it, means that that portion of the image needs to be shifted up by 2mm.  A down arrow on the other side of the border, with “1mm” written beside it, means that that portion of the adjacent layer must be shifted down by 1mm.  These indications help me a lot to make adjustments at high magnifications, when I may not be able to see the entire area concerned.  They may be relative to the base image, or simply to an adjacent layer if that layer doesn’t need to be adjusted at that point.  (Such an indication does not mean that the whole section needs to be shifted or transformed – that would spoil other areas where presumably things are well-adjusted.  It means that a part only needs to be adjusted or patched over.)

– Sometimes you can get a better result just by shifting the borders between sections (adjusting their layer masks).  To do this, simply proceed as with the several layers at different focuses:  load the selection of the mask you want to close down, narrow the selection to just the area you want to affect, and fill the mask with black; then switch to the mask of the layer you want to open up, expand the selection by 2 pixels, and fill with white.

– Small discrepancies can usually be resolved by a bit of retouching.  If any significant portion needs to shifted (while in other respects the section is well-positioned), do not copy from the file you’re working on, which has already undergone a transformation in Auto-Align Layers.  Instead, import a patch from the single-layer file it came from, reduce its opacity to about 50%, and adjust it by hand with Transform so that it conforms to the whole image. (This is to avoid unnecessary transformations, which cumulatively degrade the quality of the image.) Mask as need be.  Once I’ve got them right, I accumulate such patches in their own layer, along with my various small bits of retouching.

– Occasionally there is a slight tonal discrepancy between one section and its neighbor.  This can be resolved with a bit of dodging or burning applied to one or both sections, but it should be done without any active selection.  Otherwise, an edge will be revealed if you do any further shifting of borders.

– I also make a new layer (that I call Edges) in which I imprint the outlines of the sections, to use later to closely inspect the borders where there may be defects not so easy to notice.  To make this, I create the new layer at the top of the document, load the selections of the various sections one after the other, and apply a stroke that’s perhaps 16 pixels wide, centered on the selection outline, and filling it with a bright color.  I leave the layer invisible. To use it, I load its selection, and follow it throughout the image, viewing at 100% of pixel size, retouching as I go along.  (I may do something similar to easily find the edges of retouched areas and patches, so I can double-check without straining my eyes unduly.)

– Finally, you will probably need to do some retouching in the outer portion of the image (where the sections extend to different degrees towards the edges of the file), or crop the image to the good part, or both.

– When everything looks good, save the file one more time.  Make a copy, and Merge Layers in the copy to produce a final, single-layer file.  These are your master files.  Back them up, and import them into Lightroom.  Now you’re ready for printing, or whatever.

Printing

– For me, the photograph isn’t finished until it’s been printed.  A digital file is only that – sure it can be shown on screen, put on line, etc. – but I visualize my images from the beginning in terms of actual, physical prints (slap me if I say “output”), so the print is the true realization.  To me, a digital file (or a negative or a transparency) is to a print as a musical score is to a performance.  (Now that I think of it, Ansel Adams said something similar.)

– Since about 2008 I’ve been printing with a Canon Pixma 9500.  On the whole I’m very happy with it – although I wish it printed a little bigger, and didn’t require such big margins at the far ends of the paper.  But I saw immediately that my test prints were considerably darker than what I saw on screen, especially in the shadows.  To “match” what I saw on screen (which I liked), and printing from Photoshop as I did at first, I had to apply an Adjustment Layer of 0-1.15-255 or so to brighten it, sometimes moving the shadow cut-off slider up a few points as well.  With Photoshop I would create a file the size of my printing paper, and position the image in that format, determining the margins right within the file.

– Then I had problems printing from Photoshop (not worth explaining), and I switched to Lightroom, which was a good thing.  I was impressed with Lightroom’s utility and its ease of use, especially for adjusting tonality with the several sliders.  This was far better than using Adjustment Layers in Photoshop.  (In one respect it is similar to using Curves in Photoshop, but much more intuitive.)  This turned out to be a boon for me. My images tend toward the crepuscular, with large, deep shadows which nevertheless have detail I like to preserve, so my case may be particular.  But by judicious use of the Basic sliders I can achieve a tonality that I could never get, at least not nearly as easily, with Photoshop.  With the Shadows slider and the Blacks slider I can control those dark areas that tend to fill in, and with the others establish the proper relation of highlights and middle tones.  I have to make a few tests for each image, but it’s convenient to use a test file, itself catalogued in Lightroom, and work with groups of four or five images at a time.  (For each test, I adjust the source file and export a TIFF, which I then import into a test file, reducing it to fit among other test images.  When the test is good, the source file is ready too.)  The great thing about digital printing is that, once you’ve got the settings just right, there’s not much more to do than load paper and click Print.

– If you print black-and-white from Lightroom using images that started in color, don’t miss Lightroom’s possibilities of black-and-white conversion.  The difference between LR’s basic B&W conversion (in Basic) and its Auto B&W (in HSL/Color/B&W) is considerable.  You may also often find very useful the Custom B&W Mix, which is better and much easier than Photoshop’s Channel Mixer ever was.

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.  Anyone is welcome to link to  it or to quote brief passages, but I would like to be notified.

Link to relevant section of AllenSchill.com:  Deep Focus Still-Life Photographs

Footnotes:  (see also Appendices I and II, on tripod issues, below footnotes)

[i] Dedicated Exposures – the finer points

– If desired, you can also incorporate exposures made at different exposures, or different ISOs or “film speeds”, in addition to (or instead of) the usual aperture bracketing (e.g., f/11, f/11 1/3, f/11 2/3, etc.).  This might be done to increase tonal range in the case of a subject with unusually deep shadows and/or strong highlights.  (With RAW files, something similar can be done by exporting TIFFs made at different exposure settings in Lightroom’s Develop module, but the tonality of a exposure dedicated to the shadows or to the highlights will be superior to that of a TIFF made with the Lightroom “Exposure” boosted or reduced by a stop or two (or with other exposure settings changed very much).  Boosting especially tends to generate shadow noise, although this can be mitigated with Lightroom’s excellent Noise Reduction.  (Lightroom is fantastic, but with all the new possibilities, it sure creates plenty of work for you!  In a way it was easier when our possibilities were limited.)

– (An aside to a footnote:  this technique had its forerunner when I made files of 4”x5” negatives and diapositives by combining scans made at different scanner exposure settings.  In Photoshop, I would make a sandwich of the main exposure, the highlight exposure, and the shadow exposure, then mask them as needed.  It was very useful with files with particularly strong highlights and shadows.)

– Be careful of mixing exposures made at different flash outputs in your deep-focus assembly (i.e., avoid bracketing with the flash output), because sometimes the color temperature of the light changes depending on the output.  If you are shooting with a standard white balance, you’ll see a difference you’ll have to correct in Lightroom, and it can be tricky to get this quite right.  With my flash unit, an output of 25% is warmer than the 50% by about 150°K and four points of magenta.

[ii] Critical Focus, and other problems of exposures made in series

– I have never felt as sure about my focus with the digital viewfinder as I did with a view camera or a 35mm reflex camera.  Likewise, the camera’s preview isn’t very helpful for evaluating focus critically.  If I enlarge the image all the way on that little screen it still doesn’t seem sharp, even if it is.  If instead you can see your exposures at full size immediately on a computer screen, this will be very helpful, but it’s not convenient for me.

– As I’ve said, I see no alternative to manual focusing.  (The Auto-Focus of the Canon 5D works very well, but offers only nine points near the center of the image; I’d like to be able to choose any point in the viewfinder as the auto-focus target.  As things stand, I have to focus by eye on anything outside the center, so to be systematic I just do it all that way.)  I’d love to see a “focus bracketing” feature that would allow me to avoid the eyestrain and guessing of hyper-careful manual focusing.  It would allow one to establish near and far points of focus and then a number of regular intervals in between, and let the camera adjust its focus accordingly and make the exposures.  This could be done either numerically (e.g., by entering 35cm and 40cm as near and far points), or automatically, with two autofocus readings, and stipulating a number of exposures between them.  (At the same time, it could make bracketed exposures, as the camera already does.) It would also complement the new possibilities offered by Auto-Blend Layers.  Canon (and all the rest), are you listening?

– A related problem (and another reason it’s useful to have a surplus of images) is that, even with a very consistent electronic flash, I occasionally get an exposure that’s off by enough to notice, possibly even 1/4 stop, which may have to be excluded.  (Sure I can adjust them to conform to the others, but it is a nuisance, and it’s a problem I may notice only after I’ve begun to work very closely on the files.)  Another is the rare shot that shows, amazingly, a bit of camera shake.  This I get even with a fast electronic flash, and treating the camera gently to avoid vibrations.  I usually use a two-second delay on the shutter to allow the camera to stabilize after I’ve touched the button.

[iii] Zoom lenses and distortion

– I hardly ever used zoom lenses in the past, so I was surprised to get to know the Canon EF 24-70mm f/2/8L USM.  I had thought distortion to be mainly a problem with wide-angle lenses.  For many years I used a very good 24mm Rokkor lens for my Minolta 35mm SLR, and it did not distort straight lines (and if it doesn’t distort them, it won’t distort anything).  However, Rokkor-Minolta also made a 21mm lens which did distort – there was a slight fisheye effect – and I supposed it was a question of the limitations of optics and lens design.  What was possible at 24mm wasn’t possible at 21mm.

– The EF100mm f/2.8L Macro IS USM seems not to distort at all.  Which sounds right to me, as macro lenses usually are “flat-field” lenses, or (more to the point) they are intended especially for photographing flat subjects without distortion.  I expected the 24-70mm to show some of the usual barrel distortion (a light version of fisheye distortion) at the shorter focal lengths, and in fact it does.  The surprise was the distortion at 50mm and 70mm – the opposite type, pincushion distortion, in which the sides of the image are bowed inward, not outward.  Not extreme, but enough to notice when the profile corrections are applied. At 24mm, 28mm, and 35mm I had barrel distortion, but less and less.  I estimate that at around 45mm there is no distortion, after which the pincushion distortion develops.

– The real mystery of lens design, to me, is why these zoom lens for digital cameras distort as they do.  Is it possible that the zooms for film cameras were the same, and that I don’t know it because I never used them?  It wasn’t possible to correct such distortion in analog days, so I can’t imagine that photographers would have tolerated it.

[iv] Bit Depth

– Honestly, I see scarcely any difference between prints made with 8-bit and those made at 16-bit, but the latter should up better in any final tonal adjustments you make just when you’re preparing to print.

[v] Uniform Tones and Colors in Auto-Blend Layers

– “Uniform Tones and Colors” is not necessary for still-life work in the studio.  The function is there, surely, to deal with the reconstruction of panoramic images, in which there tends to be a lot of tonal adjustment to be done between one image and its neighbors.  It is intended only for when no adjustment of the masks is anticipated.  But if you expose a single still-life subject under highly controlled conditions, the tonality of a given part of the image should be quite uniform among the several focusing distances, and thus these automatic adjustments should not be needed.

– Uniform Tones and Colors makes permanent changes in the layers that comprise your sandwich.  If you don’t adjust the masks, probably it’ll look pretty good.  If you adjust the masks, you’ll see irregularities emerge that will be may be difficult or impossible to correct by retouching.  (Worse yet, you may not see that there is any problem until you’ve already worked a lot on the file.)

– If you leave your tones and colors as they were, you’ll be free to adjust the masks as you wish, and if you see that any burning or dodging is necessary, it will be easy to do as a final step.  But: do it without any active selection (such as that of the layer mask).  This way, if you have to adjust the mask further, there will be no sharp demarcations to reveal between manipulated and non-manipulated.

– Just to see how it works, I made a document with six photographs of the same landscape scene, composed the same way, but taken at various times of the year and hours of the day, and so with widely different appearances, like snow or sun or fog. (They are of two trees on a hillside, seen from above.  I have an ongoing project to photograph the same scenes from time to time; in three or four years I’ve made from 50 to 200 images each of four scenes near my home, always framed the same way.) I did Auto-Blend Layers, choosing Series of Images (just as for a still life), andchecking Uniform Tones and Colors this time.  (The result is very strange – a mixture of seasons in one image!)  The point here, however, is that the separate layers were adjusted drastically in color and brightness to agree with their neighbors in the blend result determined by Photoshop.  (Auto-Control means Out-of-Control.)  This was obvious when I hid all the layers but one, and disabled its layer mask.  With the mask enabled, there were also evident hairlines that bordered nearly every form.  Such a thing would be impossible to correct by dodging and burning.

Appendix I:  Tripods – Mechanical Issues

– I have a special problem with my tripod – which is not even a tripod, but rather a column with a firm base and a boom arm on which is mounted a tripod head.  It’s very solid and easy to use, but the weak point is (as with humans) in the head: specifically, the part that lets you adjust the angle of the camera up or down.  The left-or-right swivel to is easy, smooth, and fairly precise – that is, when I tighten the clamp to lock the head in position, the camera doesn’t move.  The raising or lowering of the angle of inclination, on the other hand, is a nuisance, with a lot of second-guessing:  I loosen the clamp and adjust the aim of the camera up or down, then tighten the clamp, but the camera always moves down a certain amount. Usually I have to repeat this at least a few times before getting it right.

– Changing the height (or elevation) of the camera is fairly easy, but I still have my complaints:  when I loosen the clamp to move the camera up or down the column, it’s easy and smooth, but there’s no way to keep the boom from rotating as well, so I have to control its angle carefully.  It would be very helpful to be able to retain the boom angle.  Another welcome improvement would be to have index marks imprinted on the column and the boom arm, so that I could record these positions and quickly return to them.

– On this mechanical theme, I would add the nuisance of needing to use an adapter to mount the camera, a Canon 5D, on the head of the tripod, which was made for view cameras.  The adapter is small, and it interrupts the broad contact with the base of the tripod head and its rubber pads, so that the camera – a brick – no longer has the support it ought to have.  Also, the adapter is weak:  after just a month or so of use, the weight of the camera at that point bent it down close to ten degrees.  (I realized that’s why they sell a package of two identical adapters.)  For fear of ruining the mounting hole on the base of the camera, I improvised a wedge of hard cardboard to keep the adapter from bending further.  (Canon, are you there? You could make the camera with a hole the size of the usual mount for heavy-duty tripods.  Or perhaps this is the province of the makers of tripods and overly expensive photographic accessories.)

Appendix II:  Problems Resulting from Tripod Adjustments

– One problem I encounter with this type of camera support in my procedure is that when I raise or lower the angle of inclination of the camera, the camera-to-subject distance changes considerably, by as much as several millimeters each way, and the closer I’m working to the subject, the more it may be an issue.  This is because the camera on the tripod head is positioned off-axis, that is, not on the axis of the boom arm.  Suppose I’m aiming straight down at a subject and I want to photograph it in three horizontal sections, one above the other:  the central part, and the upper and lower parts.  I’ll shoot the central part first, then the upper and the lower.  When I angle the camera upward to shoot the upper part, the camera also moves further from the subject than it was when I shot the central part.  When I angle the camera downward to shoot the lower part, it moves closer than for the central part. This means that I have to be sure I’ve allowed enough overlap between sections to still cover the whole subject, as the lower section will be enlarged more than the central, and thus framed a little more tightly than one might have expected, while the upper part will be framed a bit more openly.  (Once or twice I have raised or lowered the height of the camera a few millimeters to compensate, but this can be tricky, and I haven’t found it necessary.)

– It also means that when the final assembly is done with Auto-Blend Layers, the upper section will be enlarged somewhat to match the size of the central part, and the lower section will be reduced with respect tot he central part.  This might be around +/- 5%, depending, probably a little less than the distortion necessary to correct perspective.  (This might be little or nothing at the near edge, where it will start out close to the size of the central image, but perhaps 10% at the far edge, where it has to be enlarged to agree with the base.)  Since this transformation happens at the same time as the reconfiguration of the separate frames to agree with the overall base image, it’s not a big issue where definition is concerned, but it would still be better if I could limit the transformation to the necessary perspective correction, and avoid other adjustments.

– Even more fundamentally, it means that the perspective changes a bit as well, as the position of the lens itself changes (the front element of the lens, basically).  It is mainly this that leads to the slight discrepancies of the final assembly which can only be resolved with retouching or a patch.

– Ideal for this procedure would be a tripod head that allows the photographer to adjust the aim of the camera without altering perspective or camera-to-subject distance, that is, without altering the point of view.  I realize that such “nodal point heads” exist, but I haven’t had the occasion to look further into the matter.  It seems they are mostly designed for panoramic photography, whether the common type in which exposures made with an ordinary camera will be stitched together, or the special panoramic cameras that use a pan head and take a single, very long-format exposure.  I’m not sure whether they’d be adapted to what I’m doing, or sturdy enough, especially since the camera is quite heavy, and I often shoot straight down.  (My habit, anyway, has always been to improvise, and to try to find solutions using what I have at hand.)

Using Format