During the past few months I have received quite a number of emails from readers asking me about focus stacking so I thought I would write my next blog on this topic.
There are basically three different ways to enhance an image (DOF) "depth of field". Stopping down to a small aperture such as f/16, f/22, f/32, or maybe even f/45 etc. Use a tilt/shift perspective control lens (or a large format camera)...and Focus Stacking which combines a series of images that have different focus settings in order to obtain the entire subject matter (or frame) in clear focus. Stacking images is not actually a new concept and was also done in the old film era days by layering negatives and using a red-tape material to mask/block out certain parts of the photograph. As you can imagine, this was a rather fiddly, awkward darkroom affair and absolutely nothing like today's quick & easy digital software solution.
I personally like to "stack" certain images that I think really lend themselves to this type of approach. However, it's not something that I would want to apply to ALL my photographs simply because the end result will of course then become rather cookie-cutterish and samey looking. As with any enhancement I believe that there is an appropriate time & place for it.
As always, each method used has it's associated pros n cons...there is no magic bullet. Stopping Down to a very small aperture will of course introduce the "diffraction" factor at a certain point. Even with the best quality lenses going beyond the f/16 range it will start to become noticeable, and after f/22 definitely begins to degrade image IQ. Although, having said that, I do routinely stop down to f/22 (and sometimes f/32) when shooting my static still-life and floral images because I feel that the small amount of diffraction trade-off is well worth the additional depth of field obtained. However, for me, f/22 is generally the cut-off compromise point.
A Tilt/Shift Lens can also help enhance DOF a great deal by allowing the photographer to alter the image plane (of sharpest focus) so that it no longer lies perpendicular to the lens axis. For more detailed information on this subject click here. However, also note that using a tilt/shift lens may not always produce the desired results, especially when shooting very complex structures, such as intricate flowers, macros, etc., as there are often several different axis that co-exist together and therefore the overall image does not conform to the single tilt dynamic. Objects with straight edges, lines, or a singular plane of focus are best suited to the perspective-control lens method. The one advantage that a T/S lens does offer with regard to the extended DOF is that it is created 'at source' in a single take...so there is no extra work required, nor the possibility for image IQ degradation during the post-editing process. A good example of a high quality tilt/shift lens is the Nikon 85mm 2.8D PC-E shown here.
Our third option, and main topic of interest here allows us to 'blend' together a selection of images (that all have slightly different focus points) which cover a much larger range of depth...than a single shot (taken at say f/22 or even f/32) would normally offer. Plus, it also enables us to shoot each individual image at a more optimal aperture (ie: f/8) so that 'diffraction' never becomes an issue. Focus stacking is obviously very much suited to macro 1:1 and/or closeup work...where the regular depth of field is extremely thin. At macro 1:1 distance (using a 100mm lens) even at f/16, the DOF is only approx 1mm. Stacking not only allows us to obtain a much greater depth of field...but also one can choose (to a degree) "where" in the frame the DOF will actually be enhanced. For example, on my 'Blood Iris' shot shown here...the "stacked" region is only the front patterned area of the flower, with the background left blurred and completely untouched.
My personal approach to focus stacking is really the same as it would be for any other "macro" or "closeup" shot in general. All the essential accessories are involved (ie: tripod, geared head, focus rail, shutter release, etc.) with the focus rail now playing the major role. Here's a link to my blog article Macro 1:1 for a short overview on the topic which incorporates many of the same basic elements, application techniques, and strategy.
The focus rail unit (my Novoflex Castel-Q which is shown in the pic) simply allows the entire camera/lens combo to be moved, back and forth, in small, smooth, incremental steps...thus enabling one to slightly shift the "focus point" from front to back (or vice-versa) with each separate image. Focus rails have travel-measurement markings just like a ruler (usually in mm) so that one can carefully adjust each individual take in an accurate and exact manner.
The main thing to look for in a rail is a solid/strong construction and a smooth motion that is free from wobble or jumpy, erratic movement. Ideally, you want the main adjustment knob to be very responsive and to move the camera along in an easy, gradual "glide-like" manner...without any slack or kickback return in the knob mechanism itself. You really don't want to get a perfect focus set...only then to have the unit move even slightly when you take your finger of the knob...because that sharp focus you just dialed in will now be lost. There are a number of focus rail systems on the market these days (including Velbon, Manfrotto, RRS, Kirk, Novoflex) and they range in prices from quite cheap to moderately expensive. As with any photography accessory today, don't expect top quality for much under $500. My Novoflex Castel-Q cost me $438 here in Canada, which included the necessary Q-Plate (quick release) for the Nikon bodies. There's also the option to buy 2 units and then connect them in a "cross" configuration to allow precision in both the X and Y axis...like this Castel-Q Cross However, this will of course almost double the cost, and in my opinion isn't really necessary, as a decent "geared head" will already offer fine adjustment in the side to side (Y) axis. I don't think that it matters if one shoots their set of images from back to front, or the other way around...so long as ALL the desired region is adequately covered. I've experimented with both methods and haven't noticed any differences in the end result. It is worth noting though - that it's important to make sure that each separate shot is "spaced" the same amount apart to ensure the best quality final stack. For example, let's say that I want to capture a total DOF of 10mm in ten images at a ratio of 1:1 using f/16, then in theory each shot should be taken 1mm apart (ie: forward or backward) on the rail to avoid what I call "the smoothies" which are regions of the image that after being blended show no detail...which has occurred because the frame overlap capture was not sufficient enough and hence some areas were not in focus.
Click on the image on the right (which is a 100% crop) and it will take you to a larger photo that I created to illustrate the "smooth areas" where detail has been lost. Obviously, being very meticulous and precision minded is key here, although after a while of doing this I do think that one gets a good "feel" and knack for the process, and perhaps doesn't rely quite so much on specific measurements, etc. I guess it's kinda like driving a car, after a while you just intuitively know how to do it. One can fix these problem areas quite effectively by using the Photoshop"clone tool" to copy detail from nearby regions...which is relatively quick and easy to do. Also note, that these 'smoothie' regions are often not noticeable when the image is viewed on a PC monitor at regular (33%) ratio and perhaps would be OK/fine if printed up to a standard size. However, if print enlargements are required then I think the image IQ could certainly be compromised.
One point that folks often ask me about is the actual focusing method. Does one always move the entire camera system back & forth along the rail to vary the frame focus points, or can it be simply done by turning the lens barrel, like with a normal shot? Well, it all depends upon what type of lens is being used. For example, an IF lens (that changes focal length when focussing) alters the configuration of the lens and therefore the edge artefacts become more unpredictable. Plus this methods can cause many problems because it also changes the image scales and framing. So using the focus rail is always the preferred route here. Also, because there is no actual touching of the lens itself (which therefore eliminates the added potential for any hand-release focus kickback) I definitely think it's the way to go. In short, it's far more precise (and easier) to obtain a tack sharp image-focus when not physically touching the lens at all. Especially when shooting in "LIVE View" and tethered up to your computer LCD screen, etc. The best approach is to just set the desired ratio (ie: 1:1, 1:2, 1:4 or whatever) on the lens...and then simply move the entire rig along the rail to vary your focus point.
From a purely conceptual viewpoint I do think that focus stacking has the potential to produce "perfect" results because theoretically the photographer could/should be able to create selective DOF wherever he/she wants, but unfortunately it's end performance is ultimately dependent upon two somewhat intangible things - human skill, and stacking software capability. After the series of images have been taken we then first need to edit and process them before they are loaded into the stacking software of choice. All the separate images should be of equal exposure, luminosity-balance, and sharpness to maintain an even keel result. I post edit each NEF accordingly in Capture NX2 and then make a TIFF copy for stacking in Photoshop. Please note - When shooting stacks compose the frame so that there is ample border space to allow for some final cropping once the images have been blended...as there will invariably be some outer border mis-alighnment and areas that need to be removed...due to the previously mentioned slight visual perspective differences between each photo. So be warned - always remember to shoot with this in mind. There's nothing worse than taking the time to setup a great subject matter, carefully getting the lighting absolutely perfect, meticulously taking a series of images, and then suddenly realizing (once you've disasembled the whole thing) that you didn't leave enough edge room for some cropping to be done and that the final composition now looks really crappy :(
Here is a seven (7) x image stack of a pink Rhododendron closeup with the first shot focused on the water droplet at the front and then working gradually to the back in equal measurements. Click on the photo set to see a much larger view which shows the very slight focus point change between each individual take. At this distance the differences will be very small, but once combined into a single picture will show a much greater “depth of field” than any one shot alone. On this one I still used an aperture setting of f/22 to maximize the DOF overlap to ensure that everything from the front to back was sharp, and to help eliminate any “smooth” loss of detail problematic areas, etc. This is the finished stacked image > Pink Rhododendron
Today, there are a few (stand alone) stacking software programs available. The three most popular seem to be:
Helicon Focus, Zerene Stacker, Combine ZM, plus also Photoshop CS5 which includes focus stacking as one of it’s standard on-board functions. At present I am using mostly Zerene Stacker, simply because I seem to get better results with it and I find it very easy to work with. I’m not really advocating one particular program over another here as I’ve read various reviews from people who seem to get very good images from the other software listed above. However, I have also tried them and for some reason have had rather mixed, inferior results…which was surprising as I’d read otherwise. Zerene Stacker seems to align & blend MY stacks considerably better and also with a lot more consistency – so for the time being I’m sticking with it.
I always prefer to use TIFFs over jpegs for obvious IQ related reasons…and so, software resource capability can (and does) become a limiting factor when using the much larger TIFF files. Photoshop CS5 will comfortably handle 5-6 TIFF images together just fine, but when that number increases to more, than say, 10-12 images, then it really starts to puff and has a hard time processing the combined heavy MB workload. I have been informed that CS5 performs a lot better in this regard when running the latest 64-bit platform…but with 32-bit, it does seem to be limited. One can always use smaller JPEG files to help speed things up should this become an issue…but, IF you are going to use TIFFs for your stack, I would recommend that you try and limit it to around 6-8 images maximum. By comparison (even with a 32-bit system) Zerene Stacker can process up to 40-50 TIFF files together in one stack with no problem.
To load your stack into CS5, launch Bridge, select the individual pics, and go to > Tools > Photoshop > Load Files Into Photoshop Layers. Then once the images are in Photoshop select all image layers (ie: highlight them in the layers pallet on the right) and go to >Edit > Auto-Align Layers > OK. Once the pics are aligned, then the second step is > Edit >Auto-Blend Layers > Stack Images > OK. Then SAVE the stacked image to TIFF or PSD file. At this stage, I will then crop away any unwanted peripheral misalignment around the edge-border, and also complete any final additional editing that may be required...such as image clean-up, area lightening, mid range dynamics adjustment, selective sharpening, etc.
Note - IF you are saving the final stack to TIFF for further editing in Capture NX2...save the file as one single image (without layers)...because NX2 will not be able to open a large (layered) TIFF file of more than approx 60MB, or so.
Another (relatively new) piece of kit that has come onto the market is the automated STACKSHOT system made by Cognysis.
I've not had the opportunity to try out this nifty looking unit yet...but I have heard good things about it from others who have. For around $500, the Stackshot will take all the potential for error out of the stacking process by ensuring precision accuracy between image stepping, and producing a very smooth, fluid movement of camera along the electronically controlled macro rail.
Ultimately, I think for serious macro shooters and/or stackers...that this approach is definitely the next generation hardware route to take. Having the ability to ensure very precise, equal steps between images within the stack...should (in theory) produce flawless results.
The last thing I'd like to mention here is shooting stability. It's really important to make sure that your tripod & camera rig is set on a good, solid foundation, surface, etc.,.preferably concrete or a really hard floor. I'm often surprised just how much "movement" can result when shooting stacks on a carpet, or wooden floors that slightly bend, creek, sway, and all the rest of it. Even pay attention to what's going on 'outside' the shoot location...ie: vehicles driving by, trains, local workmen up the road, or people upstairs walking around in the same house or office building. Always be aware of how you transfer your body weight/feet (when you are carefully adjusting the focal rail for each incrimental shot)...as any shift, or abrupt movement can offset the STACK framing and completely ruin the finished image. This is where the meticulous and personal discipline attributes come into play.
Is focus stacking perfect? Answer - no, it's not quite there yet, there are still certain limitations to what can be achieved and stacking software is not technically up to that kind of level at the moment. Plus, it can also be a lot of extra work, time and effort involved. Having said that, it does allow us to create enhanced DOF images (that are just not possible via the traditional route) with stunning results.
Here is the link to my recent stacked image gallery.
Thanks for reading...
© Kev Vincent Photography. Copyright 2012. All Rights Reserved.