Selective-focus with tilt-shift lenses has been around for many years, but I think its creative potential for both photo and video (beyond miniaturisation!) lies largely untapped. By publishing this tutorial I’m hoping that it gets picked up by those of you with real creative talent, and from there you can take it somewhere completely new.

Introduction

Have you ever wondered how tilt-shift miniature photography works and why it looks so strangely convincing? Have you ever wanted to have a tilt-shift lens but couldn’t afford the hundreds of dollars needed to buy one? Have you ever wanted to take real tilt-shift photos? Then this short introduction to tilt-shift photography and DIY tilt-shift lens building is probably for you.

You might have seen photographs and video in which real scenes appear to take on a toy-set like appearance. This is commonly achieved by the creative use of a tilt-shift lens – however, this is only a small part of what these lenses are capable of. The fine level of focus control allows you to precisely pick out subjects with a narrow depth of field both close to and extremely far from the camera. Fortunately, you don’t need to spend hundreds (even thousands!) of dollars to have a tilt-shift lens in your camera bag. This tutorial goes over how selective focus effects work and are used, and offers a guide to building several different types of tilt-shift lenses all for much less than US$50.

Tilt-shift miniaturisation examples

Creative use of selective focus

It’s even possible to create lenses and adapters that will permit you to create tilt-shift miniature videos and is a frequently used effect in TV show titles and music videos.

In this tutorial, I’m going to go over how tilt-shift photography works, and go over a few introductory designs to put you on the path to DIY tilt-shift lens building.

Thanks

I’d like to acknowledge John Swierzbin for all his help and suggestions for getting this work accurate and correct, as well as Briony Joshi and Florian Kainz for all their help with proofreading and suggestions

How do tilt shift lenses work?

Our current understanding of tilt-shift photography is shaped by the formal mathematics developed by Theodor Scheimpflug, initially for correcting perspective distortions in aerial photographs. Scheimpflug observed that by changing the angle of the lens relative to the camera body, he could change the perspective in the image.

Lens movements

Tilt-shift lenses are often used in architectural photography. Parallel lines that appear to converge due to perspective distortion can be realigned to appear parallel. Sometimes the terms ‘perspective control’ (PC) lens and ’tilt shift’ (TS) lens are used interchangeably; perspective control lenses are often a specialised type of tilt-shift lens, often with shift-only movements.

In the illustration above, the top left box shows a typical shooting scenario where the camera is placed pointed upwards towards a building at an angle so that the entire building is in frame. However, it is sometimes desired in architecture to get a perspective projection so the building appears “front-on” with neat orthogonal projection (right box of the illustration). This can be achieved while shifting the lens and slightly tilting camera body. Lens shift without some rotation is basically an off-center crop within a large image circle. An example application of this is to create seamless panoramas without the need for specialised stitching software.

Tilt movements of the lens change the angle of the focal plane relative to the camera body. In a normal shooting scenario, the plane in which objects are in sharp focus is parallel to the camera and the lens. If the lens is tilted, this place of sharp focus becomes tilted as well. Scheimpflug is widely credited with developing the formal mathematics to describe the angle of this tilt.

Illustration of the Scheimpflug principle

On the far left of the illustration above, we see a camera in a normal shooting scenario. Parallel to and in front of the lens plane, there are the near and far focal planes. Anything that is viewed by the camera in the volume between these planes is acceptably in-focus. Tilting the lens causes these near and far focal planes to no longer be parallel to the lens plane, turning the focal volume from a box into a wedge. The apex of this wedge lies along the line where the lens plane and the image plane intersect. The rotation and position of this in-focus ‘wedge’ is governed by the Scheimpflug rule.

By carefully adjusting the tilt, it is possible to greatly shorten or extend the range of objects you can have in focus. By using a Scheimpflug rotation, as shown in the middle box of the above illustration, it is possible to move the line along which everything is in focus (sometimes called the Scheimpflug line) so that it follows a plane that is not parallel with the camera body. Here, the Scheimpflug line is aligned with the object being photographed, allowing for more of it being in focus than an un-tilted lens will permit.

On the other hand, by using an anti-Scheimpflug rotation, the opposite effect can be used to artificially make the amount of the object in focus smaller by having the Scheimpflug line intersect at an almost perpendicular angle with the object (far right of the above illustration).

How does tilt-shift miniaturisation work?

Anti-Scheimpflug rotation can be further exploited to ‘miniaturise’ subjects – that is, to make them look like that they are part of a tiny toy set. Many photographers use this to great effect, such as Olivio Barbieri and Kieth Loutit.

On the left of the figure above we can see the region of space that is in-focus in front of the camera (I’m going to refer to it as the ‘in-focus volume’) being used to isolate an object. This shallow focus technique with model sets gives us the distinctive toy town look (people who film scale models for visual effects commonly try to maximise their depth of field to avoid this effect!). However, when trying to photograph life-sized objects (such as a building) such tight control of focus is typically not possible. Objects that lie completely past half the hyperfocal distance away cannot be isolated in the same way.

By shooting from an elevated position and using a tilt-shift lens, it’s possible to place the in-focus volume so that it extends past the hyperfocal distance and can be used like a narrow-beam in-focus volume to narrowly pick out large structures such as buildings or vehicles (see right hand side of the miniaturisation figure above).

The size of this wedge can be greatly varied by subtle changes in the tilt and shift of the lens, as well as tilting the entire camera and lens together. It’s also worth noting that composition is an important factor here – anything that lies in the in-focus volume will be clear and unblurred, so it’s important to line up the objects in your scene so that you don’t end up highlighting something unexpected (e.g. a corner of a building in the foreground that just intersects the in-focus volume).

How the human visual system uses focus cues to estimate the size of a scene is not very well understood at this stage (a good discussion of this is in section 4 of Held et al.). In the tilt-shift miniaturisation scenario, it can be argued that when we use the in-focus volume to pick out a large object, we are tricked into thinking that the in-focus volume only works with small objects and so the subjects appear small.

Faking it?

Tilt-shift miniature fakes make it easy to create selective focus images without the need for specialised camera equipment. The method of creating the ’tilt-shift effect’ in Photoshop or similar is well documented and it can also be easily applied to video.

*Before and after tilt-shift faking mask applied*

But can post-processing achieve the same effect as using a real tilt-shift lens? For miniaturisation a study, “Using Blur to Affect Perceived Distance and Size” by Held et al. from the Computer Graphics department at UC Berkeley, demonstrated that careful choice of the placement of the blur gradient in the image will create a convincing miniaturised image. This result was verified by showing a series of test ‘faked’ images to volunteer subjects; they found a significant connection between the amount of blur and the perceived ‘scale’ of the miniaturised objects. Moreover, another recent paper by McCloskey et al demonstrated that the pattern of out-of-focus blur caused by lens tilt is a linear gradient. As stated in Held et al., “…it stands to reason that a tilt-and-shift image could be similar to a sharply rendered image treated with a linear blur gradient.”

Placement of the blur gradient. A more effective result is achieved (left) when the blur gradient is aligned with a plane that recedes away (e.g. the ground plane)

There are a number of small tweaks you can make to improve the look of a tilt-shift faked image:

If possible, use a real lens blur filter instead of a simple Gaussian (or similar) blur
The blur gradient should run along what you choose to be the ground plane. A poor choice of gradient alignment won’t be as convincing – see the above figure
If you are using a soft blurring mask, don’t just pick a narrow band of sharpness and have the rest of the image uniformly blurry. Ensure that as you move away from the line of sharpness the blurriness keeps on increasing right out to the edges of the image
If you are going for a toy look, increase the saturation and contrast of the image a little

One interesting approach that is a compromise between real and fake tilt-shift image capture is Smallgantics, developed by Bent Image Labs. Here several image planes receding away from the camera (i.e. z-depth planes) are captured simultaneously; each plane has significant elements identified, and these elements and planes are blurred by different amounts to place the image tilt-plane in space. The major advantage of this technique is that it is arguably of better quality than a just one image plane, and allows the tilt plane to be arbitrarily placed and moved at any time, allowing for significant creative freedom.

Real tilt-shift image. Note the perspective distortion on the wheels as well as the depth-of-field blurring

Real tilt-shift images, however, introduce a number of subtle artifacts that are difficult to reproduce via faking. In the above ‘miniaturised’ image, the lens has been tilted and shifted ever so slightly. While a simple treatment of a sharp version of the image could reproduce the blurring, the slight lens shift introduced by the DIY lens design introduces perspective distortion (note the elongated wheels) which add significantly to the cartoonish miniature effect. These perspective distortions are difficult to reproduce in post-processing. More significantly, tilt-shift lenses can be used for more than just miniaturisation (i.e. selective focus effects) and most of these applications are much more difficult (or even impossible!) to reproduce in a post-production setting.

Building a DIY tilt-shift lens

Building a custom DIY tilt-shift lens is an fun introduction to the world of homemade lens hacking and building. Why build your own? The first big factor is cost – DIY tilt-shift lenses typically cost around US$20-50, whereas professionally built options start out at ~US$1300 (notes on more affordable commercial options are in Appendix B). The designs below involve a minimal amount of hacking, and should be relatively simple starter projects for the amateur lens enthusiast. The second factor is simply the pleasure that is derived from building and shooting with something that you have made yourself.

Choosing a lens

Modern SLR lenses are complex pieces of optical engineering. They are often built from several lenses which correct for distortion and various optical artifacts. The lenses are designed so that they must be at a precise distance from the sensor or film to form a sharp image. As shown in above, this distance is commonly known as the Flange Focal Distance, which is measured as the distance between the lens mount and the sensor or film in the camera. Different lens and body combinations have different distances; an excellent table of lens system parameters (which includes the flange focal distance) can be found on wikipedia here.

As shown in the left box in the figure above, it’s entirely possible to use an unmodified standard lens from the same camera system for DIY tilt-shift work. This is often performed by simply holding the lens in front of the camera and tilting the lens, and is commonly referred to as freelensing. On it’s own it can produce good results, as shown in the examples below. However, the edge of the lens will bump into the lip of the lens mount, limiting tilt movements. This can be minimised by choosing a lens with a smaller throat diameter.

Ideally you’d like a lens with a narrower diameter than the mount of the camera, so it can sink into the body a little if needed to allow for a greater range of movements. Sinking anything near the mouth of the camera body requires a great deal of care, so again be careful not to collide the lens with the camera mirror. Moreover, leaving a gap open to the air between the lens and the camera body will easily allow contaminants and dust on to the film or sensor, as well as introducing significant light leaks into the image.

Freelensing examples, taken with a 50mm f1.8 lens. More examples here

This approach has some immediate drawbacks. If you are using a full-size sensor or film, any lens movements will cause serious vignetting (i.e. when the image cast by the lens moves off the sensor). Even if you are using a smaller sensor (such as an APS-C) then the lens movements are still somewhat limited before vignetting occurs.

The fix for this is to remove the back of the lens so that a greater range of lens movements are possible at the flange focal distance (see Modifying a Lens, below), or to use a lens that has a larger flange focal distance (see Plungercam 2, below). The latter approach can be achieved for 35mm cameras by using a medium format lens (the right-hand box of the above figure), which typically have a flange focal distance at least double that of standard 35mm lenses. Better still, the lenses are designed to create an image for a film/sensor that is much larger than 35mm, so a much greater range of lens movements are possible without vignetting. This is holds equally true when you are coupling lenses for a larger sensor/film with a smaller camera body – for example, using medium-format lenses with a standard-format camera body, or, more interestingly, standard 35mm with a mirrorless camera body (example here).

Body type	Lens type	Works?	Notes
Mirrorless	Mirrorless	Maybe	This might work, but the lens would need modification
Mirrorless	35mm	Yes	This seems like a good combination
Mirrorless	Medium format	Yes	This seems like a great combination – a huge range of lens movements would be possible
35mm SLR (crop sensor)	Mirrorless	No	The micro 4/3 flange focal distance tends to be way too short
35mm SLR (crop sensor)	35mm	Maybe	This might work – you’d need to verify this by experimenting via freelensing
35mm SLR (crop sensor)	Medium format	Yes	This seems like an excellent combination
35mm SLR (full frame)	Mirrorless	No	The micro 4/3 flange focal distance tends to be way too short
35mm SLR (full frame)	35mm	Maybe	This might work – you’d need to verify this by experimenting via freelensing
35mm SLR (full frame)	Medium format	Yes	This seems like a good combination
Medium format	Mirrorless	No	The micro 4/3 flange focal distance is way too short for this
Medium format	35mm	No	The 35mm flange focal distance is too short for this
Medium format	Medium format	Maybe	This might work – you’d need to verify this by experimenting via freelensing

The compatibility chart above outlines the rough combinations of lens and camera bodies for DIY tilt-shift lenses. Any lens you choose to use for tilt-shift work can be easily tested by freelensing – leave the body cap off the camera, and slowly move the lens towards the camera body to see if a sharp image is formed. If you can form an image without the lens colliding with any camera internals, then the lens is probably suitable for DIY lens building.

How to initially choose a lens for suitability

Roughly evaluating if a lens is going to be suitable for building into a DIY tilt-shift lens is a good first step. This is easily done by first setting up your camera body on a tripod, in an area that has objects both near and far away. Set your lens to infinity focus, take the body cap off the camera and place the back of the lens next to the camera body (being careful not to bump the mirror!). Looking through the viewfinder, move the lens forward and see if you can form a sharp image of near or far objects. If you can’t form a sharp image, then the lens is probably unsuitable. If you can, then it may be worth slightly tilting and shifting the lens when it is in a sharp focus position to determine what kinds of movements this lens might permit. From here you can also estimate what movements cause vignetting.

Retrofocal lenses

Many wide-angle SLR lenses use the Angenieux retrofocus lens design, which places a large negative lens element at the front of the lens allowing for a greater back focal distance than would otherwise be possible. This larger distance allows the rear lens element to be a good distance from the film/sensor, and in SLR cameras allows for room for a mirror. Some commercial tilt-shift lenses also use the retrofocal design, with an extra long back focal distance and larger imaging circle, which is exactly what we’d like to emulate by choosing coupling a larger-format lens type with a smaller-format camera body. For example, compare the rear-element to rear-flange distance on a normal 24mm lens with a 24mm tilt-shift lens.

Effect of lens choice on tilt angles

Different focal length lenses will respond differently to lens tilt (see tilt analysis, above). Generally speaking:
As lens focal length increases:

The angle of the plane of sharp focus decreases (for any given lens tilt angle)
The offset of the plane of sharp focus increases (this really only is significant for objects close to the camera)

As the f-stop increases:

The angle of the in-focus volume increases

This means that short focal length lenses are more sensitive to tilt, and faster lenses produce a shallower depth-of-field volume. Some of these relationships are not obvious, and a more detailed look into these relationships can be found in appendix A.

Modifying a lens to fit

When using a lens from the same format as the camera body – e.g. a 35mm lens with a 35mm body – it is necessary to physically modify the lens to make it more mechanically suitable. This kind of modification is quite destructive and requires disassembly of an existing lens to make it work. I’d highly recommend using an old and inexpensive lens for this. As shown in the diagram above, lenses matched to the same format (here, a 35mm format lens with a 35mm SLR) will collide with the camera body when tilted, as indicated by the region in green. To solve this problem, many older lenses can be disassembled to remove their mounts and aperture setting mechanism. In this case, I’ve taken the back mechanism off a 28mm Minolta Rokkor. Once the lens back is removed, the range of movements is much greater.

To disassemble the aperture mechanism you will need to identify where this mechanism is and remove it. Most lenses that I have come across have the aperture mechanism held on by screws, and only minimal force is required. Some, however, will require forceful disassembly, and careful use of a hacksaw may be needed to achieve this.

Zoom lenses can be adapted for use as tilt-shift lenses, but there are a few things to note. In certain zoom configurations, these lenses tend to have a large space between the back lens element and the flange. By keeping this distance and removing the rear lens housing, it is possible to allow for large lens movements as the back lens element will not collide with the camera body. However, the back mechanism for zoom lenses tend to be much more complex than for prime lenses, so extra care must be taken when disassembling the lens back.

DIY tilt-shift lens examples

Your choice of DIY lens design should be driven largely by what you’d like to do with it. Below is a table that enumerates some of the choices, but it is by no means a limiting factor – experiment with the lenses, and you’ll be pleasantly surprised by the range of effects that are possible.

Shooting style	Attributes	Lens
Free-form	Compact, portable
Free-form	High-quality, large tilt-shift movements
Fixed	No lens modification required but parts must be 3D printed and assembled
Fixed	Compact, portable
Fixed	High-quality, large tilt-shift movements

Tilt-o-matic

The tilt-o-matic is an example of a 3D printed adapter to couple a larger format lens with a smaller format body. In this particular case it’s coupling full-frame 35mm-format Canon EF lenses with APS-C mirrorless format Fuji-X camera bodies, but could be applied to any 35mm-to-mirrorless coupling with the right specification for lens/body mounts.

The design allows for three movements. Near the lens end there is a shifting rail. Below that is a single-direction tilt slider; the rotational radii of the tilt is centered at the position of the sensor to help maximize the overlap of the lens image with the sensor. Close to the body is a rotational rail to arbitrarily turn the lens tilt and shift mechanism, and three combined allow for a wide range of tilt and shift positioning.

The adapter sits between the camera body and the lens; adjustment knobs allow the sliders to be moved as desired and then fixed into place.

Bendycam

Bendycam is a selective focus DIY lens that is suitable for free-form shooting. While looking through the viewfinder, the photographer has to adjust the lens into position by hand. This allows for very fine control of the placement of the selective focus plane; however, because each shot requires placing the lens by hand, each shot is more-or-less guaranteed to be unique. Additionally, this design also allows for significant controllable shift of the lens as well as tilt.

Bendycam is based on the design originally proposed by Johnnyoptic, who has built a series of beautiful DIY tilt lenses with an elegant, compact take on the bellows mechanism which uses bicycle inner-tubes to couple lenses to a lens-mount.

As shown in the bendycam assembly diagram above, the bendycam itself is made of only a few parts. A is simply a camera body cap, on top, which has the centre drilled out of it (on eBay for US$3). B are a few cable ties; if a single tie isn’t going to be long enough then you may need to chain a couple together (about US$1 from a hardware store). C is a short section (roughly 10cm, or 4 inches) of bicycle inner-tube – you want to get as wide a diameter as possible (you can pick up used inner-tubes for free at pretty much any bicycle shop). D is a used 35mm lens with the focusing mechanism removed (see lens modifications) – in this case, a 28mm Minolta ($10, from eBay).

The first step is to preset the lens focal length to infinity, as this may be difficult to change once the lens is assembled. Next stretch 2-5cm (1-2 inches) of the inner-tube around the back of the lens, taking care not to tear the rubber (top-right of the assembly diagram above). Take the body cap end, and insert this into the free end of the inner-tube close to the end. Align it so that it is parallel with the rear of the lens (bottom-right of the assembly diagram above). Take the cable ties, and apply these tightly around the section of the inner-tube that covers the body cap. If needed, you can also use more cable ties (or a pipe clamp) to secure the inner-tube section that wraps around the lens as well.

The bendycam is surprisingly intuitive to use (not to mention extremely fun!), and it’s likely that first-time users of it will not have come across anything like it before. Focus is controlled by simply moving the lens by hand; I’d highly recommend just squashing the lens close to the camera body and moving it outwards while tilting the lens in different directions to experiment with how it works (again, be aware of where the lens is and don’t bump it into the mirror!).

Although this is a manual lens, the level of control offered for focus in free-form shooting is, as far as I have found, unparallelled. While initially this lens takes a bit of getting used to, it offers the ability to generate some pretty unique-looking results. Most of all, it’s fun to use, and there’s something satisfying about taking a shot and knowing that you’ll never be able to make it exactly the same way again.

Plungercam mini

Assembly of plungercam mini is relatively simple. On the left of the assembly diagram above, the four basic components of the lens assembly are indicated. A is simply a camera body cap, on top, which has the centre drilled out of it, bottom. These can be found on eBay for US$3. B, on top, is a 5cm (2 inch) pipe coupling, easily available at a hardware store, and cost about US$2. Cut off a short length of it, and make sure it is long enough to house the lens comfortably (bottom). In practise, for the lens I was using, I found that 3cm wide (just over an inch) was sufficient. C is a 5cm (2 inch) turnkey clamp; again, easily available from a hardware store, and a pack of two will set you back about $2. In the diagram, D, like with the bendycam is a used 35mm lens with the focusing mechanism removed (see lens modifications).

The body cap should fit tightly inside the cut-off piece of pipe coupling; as the first assembly step, put it into place (top-right of plungercam mini assembly). Next, slide the lens into the other end of coupling and the turnkey clamp on the outside of the coupling, and the lens is assembled.

Usage of the lens is relatively simple. When approaching the subject, loosen the turnkey clamp so that the lens easily turns in it’s housing. Looking through the viewfinder, you should easily be able to hold the camera with one hand and freely manipulate the lens with the other (see left of plungercam mini usage). In the case where you are shooting video, you can set the lens to the desired position and re-tighten the clamp so that the selective focus remains the same while you are shooting.

Plugercam 2

Plungercam 2 has much the same design as plungercam mini, except that it doesn’t require lens disassembly. This is achieved by using a medium-format lens – in this case, a Zenza Bronica 75mm.

As shown in plungercam 2 assembly, above, there are only a few components required. A is a cheaply available t-mount (sturdier than a body cap, and capable of holding more weight, but doesn’t allow the lens to get extremely close to the body) – about US$10 on eBay. B are a pair of 5cm (2 inch) and 8cm (3 inch) pipe clamps (US$2 each from a hardware store), C is a 5cm to 8cm (2 inch to 3 inch) pipe coupling and D is the lens itself.

The narrow end of the pipe coupling is first shortened to as short as is comfortably possible so it can still hold a pipe clamp around it. The t-mount is inserted into this narrow end (again, it should be a fairly tight fit), and the pipe clamps are placed on the narrow and wide ends of the coupling. The lens is inserted into coupling and the clamp is tightened to hold the lens in place. Note: a more practical alternative is to use a 8cm (3 inch) turnkey clamp on the wide end of the pipe coupling, but this is not demonstrated here.

Shooting with Plungercam 2 is similar to plungercam mini; one hand holds the camera, and the other loosens the pipe clamp for the lens end and adjusts the lens position until the desired level of selective focus is achieved (see plungercam 2 usage above). Again, the lens can be fixed in position by adjusting the lens as needed and tightening the clamp, which is ideal, for example, for shooting tilt shift video.

Plugercam Classic

The whole plungercam series of lenses was originally inspired by a series of designs by Dennison Bertram and Keith Loh, from which I built the original Plungercam classic. Like bendycam, the advantage of this type of lens is that the shooting is purely free-form; because the springy bellow mechanism pushes back against the operators hand, each shot is more-or-less guaranteed to be unique.

Like the other lenses, assembly is relatively simple. A is an EOS body cap that has had the centre drilled out, and this will be the attachment to the camera body. B is a rubberised toilet plunger (new – not used :P), C is a three-inch diameter standard pipe clamp and D is a used, Zenza bronica medium format lens ($12 from ebay!).
I lopped the head and tail off the plunger; the smaller end had a diameter to match that of the body cap, and the larger end that of the lens. You can see where the male EOS adaptor was attached in the top right of this figure. I ended up needing to glue the adaptor into place to make a secure fit – standard PVC glue worked just fine.

Once the glue has dried (the longer you can leave this, the better – I gave it 12 hours), the lens slides into place on the wide end of the plunger and is held into place by the pipe clamp (bottom right of this figure). The plunger is relatively stiff, and I feel fairly confident that the lens mechanism holds up under its own weight without flopping about dangerously.

Using plungercam classic is a bit daunting, given the enormous size of the lens. I typically used it by holding the camera with both hands, and used my fingers wrapped around the pipe clamp to bend the plunger into place. The plunger provides a fair bit of resistance, so it’s really only properly suited to shooting in well-lit scenarios where a relatively fast shutter speed can be used.

Other DIY types

You aren’t limited to using old camera lenses. A number of people have successfully reused enlarger lenses to build custom tilt-shift lenses; some great examples can be found here, here and here. I prefer to use old lenses to save on cost, and also because I don’t feel so bad if I damage them.

Shooting with DIY lenses

A good starting point for beginners is to initially focus in on one point in front of the camera. Tilt the lens, and see if you can move the focal volume around to select two or more objects that are separated by a reasonable distance and get them all in focus.

A practical note to keep in mind is that once the mechanism is removed, the aperture needs to be set manually. I personally like to shoot with the aperture wide open, which keeps the already shallow depth of field even shallower and often introduces all sorts of artifacts when light hits the lens at an oblique angle. A couple of examples can be seen in examples below.

Bendy Bokeh

An interesting thing to note is that because the aperture is at an angle to the sensor, image artifacts take on strange-shaped distortions. An interesting example of this effect can be seen in this image by Johnnyoptic here, where you can see the lens flare turning into jellyfish shaped highlights instead of circles. This also means that as objects move away from the focal plane, the blurring can take on a strong directional bias (e.g. being stretched out horizontally or vertically). An example can be seen below.

Note the strange directional bias of the blur of the out-of-focus areas in the image, especially on the power-lines on the top-right

The shape of the aperture strongly affects the characteristics of the bokeh (out-of-focus areas) in the image. Point sources of light in out-of-focus areas take on the shape of the aperture. Tilt-lenses cause the aperture to look like an ellipse relative to the image plane, so out-of-focus areas in the image are also going to be blurred in an elliptical rather than a circular fashion. This gives real tilt-shift images a subtle, unique look that is sometimes difficult to replicate in photoshop.

For shots involving miniaturisation, I find it useful to set the lens focus at infinity. In fact, for most type of shots (except for relatively close macro shots), I find that leaving the lens at infinity works really well.

DIY tilt-shift lenses skew the optical path quite significantly inside the camera, and as such this can throw off the auto-exposure metering in digital SLRs. In practise, for daytime shots, I’ve found that under-exposing the image by about 1 stop tends to correct for the errors in metering quite well.

Results

A bigger gallery of plungercam shots can be found here.

Appendices

Appendix A: lens relationships

As mentioned earlier, the modern analysis of tilt-shift lenses was developed by Scheimpflug in the early part of last century. He developed a number of equations to describe what happens to the focal planes of a camera and lens combination when the lens is tilted.

The equations below make a number of assumptions to simplify the discussion, not least of which is that we assume that the lens is reduced to a thin-lens approximation for calculations. They by no means produce very accurate numbers, but given the approximate nature of the measurements in the DIY-lens building they are meant mostly just to illustrate the rough characteristics of the lens as various parameters are changed.

When the lens plane is tilted relative to the image plane, they intersect along a line. From this line, a focal wedge is projected, bounded on one side by the near focal plane, on the other by the far focal plane and in-between by the optimal plane of sharp focus. This wedge is offset from the centre of the the image plane by an offset, J:

J = f / sin (lensang)

where:
J = offset distance
f = focal length (in mm)
lensang = lens angle tilt to image plane in degrees

As shown in tilt offset analysis, as the focal length increases, the offset of the intersection also increases, decreasing as lens tilt increases. A long offset is useful when one wishes to isolate a distant object in-focus while still remaining on the ground (such as a building). This permits the user to use the focal wedge as a kind of focus-isolating flashlight that is held at a long distance from the camera.

The angle of the plane of sharp focus is governed by the equation:

sharp focus angle = atan ((objdist/f) * sin (lensang))

where:
objdist = distance to object being focused on (in mm)
f = focal length (in mm)
lensang = lens angle tilt to image plane in degrees

From the above equation, it can be seen that that the sharp focus angle is inversely proportional to the focal length.

Proving that the f-stop affects the angle of the in-focus volume is a bit more involved. In the leftmost box of the above diagram, we measure the back focal distance – approximated as the distance between the lens plane (LP) and the image plane (IP) when they are parallel to each other – as V. The size of the aperture is a, in mm. It’s worth noting that:

N = f/a

where:
N = f-number (e.g. 1.8, 4, 5.6 etc)
f = focal length (in mm)
a = aperture size (sometimes known as diameter of the entrance pupil – in mm)

To find the position of the near and far focal planes, you need to know how much blurriness you image can tolerate before the object is out of focus. This is defined as the maximum size point source of light then can become on the image plane, and is known as c, the circle of confusion. It is usually quite small (e.g. in the Canon 7D is is 0.019mm).

In the middle box of the above digram, we can see that if the the light from a point source moved away from the camera, it focuses at a point just in front of the image plane and creates a circle of size c on a sensor. We’ll nominate this back focal distance for the far focal plane as as Vff. Similarly, for a tolerably out of focus object near to the camera (i.e. the near focal plane) in the rightmost box of the above digram, we’ll nominate the back focal distance for the near focal plane as Vnf.

Using similar triangles, we can state that:

(V - Vff)/c = Vff/a

Which rearranges to:

Vff = v/((c/a) + 1)

Similarly:

(Vnp - v)/c = Vnp/a

Which rearranges to:

Vnp = v/((c/a) - 1)

In the top box of the above diagram, you can see the standard configuration for a tilted lens with the lens plane (LP) titled to the image plane (IP), and the plane of sharp focus emerging from the intersection of the two. We nominate the perpendicular distance between the IP and the LP relative to IP as V’, and the perpendicular distance from the LP to IP as V. Given a lens tilt angle of L, we can write:

V=V'cos(L)

Using an alternative form of Scheimpflug’s equation, we can say that:

P = atan (V'sin(L) / (V'cos(L) - f)

The near and far tilted focal planes can be determined by substituting V’ with Vff’ and Vnp, as shown in the lower box of the above diagram; i.e.

Pff = atan (Vff'sin(L) / (Vff'cos(L) - f)

Pnf = atan (Vnf'sin(L) / (Vnf'cos(L) - f)

For any given arbitrary focal length and lens tilt angle, there is an almost linear relationship between f-number and the angle between Pff and Pnf, as plotted in the bottom of the above diagram.

Appendix B: Commercially available options

There are many commercial options available for tilt-shift lenses. Unlike the current crop of DIY options, many of these allow for precise and measured control of lens movements, which can be essential for work such as perspective correction for architectural photography. I’ll briefly go over a small selection of commercially available options here.

View camera

View format cameras describe a class of camera where typically an articulated lens plate connected to a large film/sensor plate (usually bigger than 9x12cm (4×5 inches)) by bellows. The lens is designed to cast a large imaging circle and can be adjusted with a large range of motions. This is a classic camera type, used by many pioneering photographers (such as Ansel Adams), though it’s versatility for perspective correction and depth of field adjustment didn’t come until later. An excellent writeup on view camera movements can be found on the wikipedia page, and an excellent writeup on the geometry, optics and mathematics of lens movements (with special regard to view cameras) by Merklinger can be found here.

Lensbabies

The lensbaby is an inexpensive, easy-to-use option if you’d like to start out experimenting with selective focus shooting but don’t want to build your own. There are several versions available, each performing a number of different and clever tricks with manipulating the optical path in the lens; an excellent writeup on the lenses can be found at dansdata, here. John Swierzbin points out that ‘off-axis aberrations such as coma and field curvature cause the distinctive “Lensbaby Look” in which the image is increasingly blurred as you move away from the on-axis sweet spot. Although focal plane tilt is also present, this effect is generally drowned out by the radial blur from the off-axis aberrations.’

Actual tilt-shift lenses

Both Canon and Nikon offer precisely engineered tilt-shift lenses for use with their 35mm cameras. These often tend to be quite wide and fast, and are an excellent choice if you need reproducible, reliable results, as is often the case when you are shooting professionally. They are, however, not cheap – at the time of writing, the cheapest Canon started at ~US$1350, and the cheapest Nikon was ~US$2000.

Adapters

Another option is to use pre-made tilt-shift adapters, which sit in-between the camera body and the lens and provide tilt and/or shift. The classic Pentacon Six series of lenses, which already have an easily available series of adapters to mount them to modern-day SLR’s, also have tilt and shift adapters. More serious adapters are available for medium format cameras, and apparently there are also commercially available adapters for micro four thirds and other mirrorless cameras too.