I’ve written previously about what exactly contributes to the ‘medium format look’. However, I think to some degree we also need to both define what constitutes the hallmarks of smaller formats, but more importantly figure out where each format’s strengths lie. Having now shot what I’d consider ‘enough’ with a complete MF system wth lenses ranging from ultra wide (24mm, or 18mm-e) to moderate tele (250mm, or 180mm-e) I think I’ve built up a much more complete picture. No doubt this will change if the recording medium size increase further – with the 54x40mm sensors, for instance – but I think it’s fairly safe to extrapolate based on the differences between subsequent smaller formats.
Depth of field
This is probably the most immediate giveaway: for a given subject magnification, aperture and angle of view, larger formats will appear to have shallower depth of field. Remember that the real focal length and aperture are what determines the degree of background blur; the larger the format, the longer the lens required to cover that angle of view. Even though lenses tend to get slower as the format increases – faster apertures for longer real focal lengths are harder to build – in practice, what we notice is larger formats have some degree of separation between subject and background/foreground planes, where smaller formats may not. However, this may require larger output sizes or resolutions to discern; 1MP web jpegs are almost certainly going to be missing something. It gets a bit more complicated, though: for an ideal lens, the transition between in and out of focus becomes significantly more abrupt as the focal length increases. With a shorter real focal length, that transition might be very gradual – and thus appear to have almost no visual separation at all.
Angles of view, projection and optical formulae
That said, things get more complicated still: longer lenses are easier to fully correct for, which means optical formula limitations that degrade our perception of sharpness of transition (think of an image ‘sliced into planes’ ) such as coma, longitudinal and lateral chromatic aberration are more easily minimised. When these aberrations don’t ‘eat into’ the edges of the focal plane, we are given the impression of shallower depth of field because the boundary of confusion is much tighter – a subject is either in focus, or it is not.
On the reverse side, when engineering for shorter real focal lengths, all sorts of other problems must be taken into account. For instance, if the design requires a relatively long image distance compared to the actual focal length (e.g. Nikon F back flange, ~47mm, focal length, 21mm) then a telecentric design must be used, which contains another set of elements to ‘straighten out’ and extend the ray path so that it might focus the projected image at the correct distance. Even if not, and we are using a system with a short flange back distance – symmetric designs aren’t always ideal because you’re going to land up with very extreme ray angles towards the edges. Since digital sensors are not truly planar but more like a waffle grid, interactions at the edges of the individual photosite can produce both vignetting and purple fringing towards the edges of the sensor. Offset microlenses over the photosites can compensate for this to some degree, but the potential for aberrations to start affecting transitions remains high. This is one of the reasons a lot of fast wide lenses lack the sort of clean transition you’d otherwise expect.
But, we’re still not done yet. The wider the lens, the greater the projection distortion we’re going to have to deal with: think of squashing a printed balloon onto a glass window, or translating a globe to a flat map. We can represent a map of the world on a 2D surface, but it requires some cutting and/or stretching (which is how we land up with things like the Mercator, Hammer, Panini etc. projections). Stretching is necessary should we want to present a continuous surface, which obviously causes some distortion to the image being projected. All lenses have some degree of field (focal plane) curvature, which becomes harder and harder to correct for as the focal length becomes shorter and the angle of view wider.
The wider you get, the worse this becomes. Once you pair that with the inherent difficulties of shorter real focal lengths and physically smaller lenses (i.e. fewer elements, less image circle to work with) – you can probably see why it’s easier to design a 50mm lens with a 60mm back flange distance (~1:1.2, 645 format) than a 30mm one with a 47mm back flange (~1:1.6, 35mm-e), even though they cover the same angle of view. As this ratio gets larger, design becomes exponentially more difficult for reasons previously described. Every time you want to correct for something – telecentricity, field curvature, distortion etc. – you introduce additional elements which must themselves in turn be corrected for.
In the earlier article, I also touched upon some of the underlying media differences: no longer do we have the same recording medium in different areas, but physically larger sensors will a) not just have greater spatial resolving power for the same output resolution, but b) generally have larger photo sites and greater spatial resolving power. This leads to significant gaps in dynamic range, noise and in turn, color resolution. On top of that, the potential for greater spatial differentiation leads to much better tonal separation and subtlety: you’ve got a lot more steps to describe the same transition.
These differences must be considered together, mostly because of simple economics: smaller/worse/cheaper vs bigger/better/more expensive. There aren’t any really bad MF lenses because the underlying cost of the main component (sensor) determines the pricing level anyway; however, there aren’t that many good entry level kit lenses for similar reasons. In practice, I believe what this means is that there’s a sweet spot for every format/system; however, this isn’t immediately obvious to most people as few have shot a variety of systems under an equally wide range of conditions. My personal feeling is that there really isn’t a one-size-fits all both because of underlying system characteristics (e.g. small vs large sensors and attendant body features such as AF) and because no one system can give you a sufficiently wide variety of rendering styles. By sensor size, here’s where I think the strengths and weakness lie:
Good for: Casual documentary, stealthy documentary, social record. Small size and ubiquity means you are inconspicuous; small sensor means you have almost unlimited depth of field and minimal focusing issues. Fast to use because there are few parameters to control (e.g. changing aperture makes no difference anyway, focus distance only when below a certain range).
Limitations: No DOF separation, restricted dynamic range and poor overall image quality, few options other than a wide FOV.
Good for: I’m struggling a bit here. They’re a bit better than the compacts, but not so much that you’d be happy with one of these for primary output. Rendering of wide options tends to be quite uninspiring and just a bit messy looking: there’s enough depth of field differentiation to make transitions nervous, but not so much DOF that everything is in focus. The biggest strength I can see is for highly compressed telephoto perspectives where everything is in focus – the crop factor means you have much greater DOF than you might think for a given aperture, plus light gathering ability doesn’t become so much of a problem.
Not good for: Similar limitations to compacts and phones, but more lenses available.
M4/3″ and APSC
Good for: Given the lens selections, this isn’t a bad all-round solution for most people. I still haven’t found any wides whose rendering I really like, though – for the all the reasons we previously discussed. I think the main strength is again telephoto work: excellent small lenses with fast apertures (thanks to crop factor) offering enough separation (and not so little you have problems getting enough of the subject in focus) but still the option to get everything in focus if you need to.
Not good for: Wide angle work; a consequence of lens design and simple physics: the wider the field of view, the more spatial information (i.e. resolution) you need to avoid a gritty look.
Good for: General shallow DOF work, moderate wide to moderate tele, supertele with shallow DOF. I think most people have realised by now that to get a wide range of distances in focus with high resolution 35FF bodies, you’ll either need to stop down past the diffraction limit, be shooting pretty wide and/or with subjects constrained beyond a certain distance, or use perspective control (i.e. tilt). Yet because of the completeness of the attendant systems, 35FF remains the all-round choice format: everything from 8-800+mm in a single lens without TCs, plus perspective control and macro.
Not good for: Getting everything in focus outside wides or what tilt shifts are on offer; subtle separation at wide/normal angles of view
MF digital (44×33, 645-54×40)
Good for: Superwide to moderate telephoto; getting some sort of DOF separation at all distances and focal lengths even with moderate lens speeds
Not good for: Getting everything in focus without the use of tilt shift adaptors or putting the back on a technical camera, telephoto work, anything fast moving
Good for: Full perspective and DOF control, a very ‘flat’ look no matter what the angle of view – minimal distortion and clean projections
Not good for: Anything where you don’t have space for a tripod and several minutes to set up each shot, digital work with the full format area
There are three ways of approaching this: either select the format that will give you the greatest coverage of the kinds of things you normally shoot; restrict yourself to one format that produces results of the kind you want to produce; or, have multiple systems (or a system that allows for some interchangeability of lenses and sensor sizes, e.g. Nikon 1/CX, Nikon APSC and Nikon FX: three bodies, one set of lenses). I’ve gone down the multiple system route because I found the the adaptations simply don’t work that well – it usually isn’t a limitation of optics, but the whole thing just becomes very clunky in use. But of course, we also must remember that doing something differently with unconventional formats almost always yields unusual (and sometimes quite aesthetically pleasing) results…MT
Ultraprints from this series are available on request here
Visit the Teaching Store to up your photographic game – including workshop and Photoshop Workflow videos and the customized Email School of Photography. You can also support the site by purchasing from B&H and Amazon – thanks!
Images and content copyright Ming Thein | mingthein.com 2012 onwards. All rights reserved