How to focus a landscape scene

Introduction

Photo: Fredrik Torger.

This article contains an introduction to how to focus a landscape scene in order to make a sharp high resolution image. It's expected that the reader has at least some minor prior experience of landscape photography and is somewhat familiar with the Depth of Field (DoF) concept and the Scheimpflug principle.

As you will see tilt (and shift) is a key concept dealt with in this article, meaning that it's primarily written for those of you that use a technical camera or tilt-shift lenses for your DSLR or mirrorless camera. The non-tilted cases are however also covered so there's useful information for every photographer that makes planned shots from a tripod.

When making a high resolution landscape photograph in one shot that's sharp all over tilt is indeed a very useful feature. I could myself not imagine to be without it. However, you may have a style which doesn't need sharpness all over, or a style that rarely has near-far compositions that present DoF challenges, or a type of near-far compositions that only can be solved by focus stacking. If you make your images through mosaic stitching tilt is rarely needed either. In other words it depends on your photographic style how useful tilt is to you.

Traditional landscape photography was made with view cameras with 4x5" or larger film sheets, which gave both high resolution and possibility to tilt and shift in all directions.

Shift is typically used to render vertical features such as trees upright, but can also be used for more subtle compositional adjustments.

Tilt is used to optimize placement of the plane of focus so a large depth of field can be achieved without stopping down too much (which would lower resolution through diffraction). The typical use in landscape photography is forward tilt in near-far compositions. In large format film photography tilt is often used because the high resolution of the system would not be fully utilized if simpler methods were employed.

While only a few use large format film today, view cameras and other technical cameras with similar features exist for the digital medium format. This is a popular choice for professional landscape photographers, but due to the high cost of the systems most enthusiasts use 135/35mm DSLR or mirrorless cameras which today are capable of very high resolution. Some enthusiasts do use old-school large format film cameras which (if bought second hand) is a low cost alternative for the low volume shooter to get very high image quality, and you get all the movements.

Tilt-shift lenses have existed for a long time for 135 cameras but as long as resolution was low the tilt feature did not have much use unless doing macro photography or using longer focal lengths. However, today when these cameras have reached 40-50 megapixels the resolution is similar to (grain-free) 4x5" film and then many have discovered the value of using tilt in landscape scenes to maximize resolution also for wide angle lenses. If you ask me I'd say that you actually do not need the latest high resolution camera though, tilt focusing for landscapes is useful for anything more than 10 megapixels.

Range of useful apertures

A digital medium format view camera with Schneider Digitar lenses, optimal aperture: f/11 (typical).

The typical goal when focusing a landscape scene is to render an image that appears sharp corner to corner. This means that we need a large DoF and thus a small aperture (=large f-number), but smaller aperture means more diffraction that blurs the image so there's a trade off. The diffraction blur can be restored to some extent with (deconvolution) sharpening, but if it becomes too large it's difficult to get good results.

If we open up the aperture really wide the diffraction becomes so small it's invisible, but then the optical performance of the lens is typically lowered so the corners become blurry instead.

The "optimal" landscape photography aperture is generally one where diffraction has become slightly visible but is still easily sharpened. The DoF is then not unreasonably shallow, and the lens is stopped down enough to provide good corner sharpness (and low vignetting).

If possible all images would be shot at this aperture, but many situations require even more DoF and then a smaller aperture is needed. Resolution will then be somewhat worse but up to some maximum f-number sharpening works reasonably well.

Typical aperture ranges for different systems:

A common misconception is that optimal aperture depends on sensor size, but it's actually pixel size. The above listing is not strictly made on a pixel size calculation though, but also taking into account what you can expect from lenses available for the systems.

Schneider Apo-Digitar lenses for digital medium format are for example typically optimized to perform best at f/11. The 80 megapixel medium format digital backs are usually mated with Rodenstock Digaron lenses which are optimized for f/8 or even f/5.6.

Many consider "optimal aperture" to be the smallest aperture when diffraction still is invisible before sharpening, but with a reasonably high resolution camera you will then end up with a very shallow DoF which is impossible to work with in landscape photography. Also, when sharpening is taken into account it becomes apparent that some diffraction is perfectly ok.

I recommend that you test your own system and sharpening workflow to see what you think the optimal and maximum f-numbers are, as it's in part a matter of taste. It may vary depending on lens too, many wides have weak corner performance and a better compromise may be to stop down a little bit more for those. Note that with very small apertures lenses lose global contrast as well (due to diffraction), that is the image becomes a bit flat and dull, which you can restore to some extent in post-processing. Overall the digital workflow has given us better tools to compensate for diffraction than we had in the film days. On the other hand we also typically expect sharper images now than we did then.

Hyperfocal focusing technique

An example of a scene where hyperfocal focusing is applicable. For this long lens the tree is closer than the hyperfocal distance at optimal aperture, but farther away than half of it. If we focus at the tree or the distant background the other will be unsharp, but if we focus at hyperfocal distance both will be sharp.

Hyperfocal focusing has got a somewhat bad reputation. I'd say it's because of two reasons 1) too large Circle of Confusion (CoC), and 2) poor precision of lens distance scales. However, both can be managed and then hyperfocal focusing becomes a relevant technique.

The concept is simple: put the plane of focus as close as possible while keeping infinity within the DoF, meaning at its far edge. That way we get the maximum out of the DoF in a scene when we have objects both closely and at infinity. This focus distance is the "hyperfocal distance".

The main application for hyperfocal focusing is maximizing DoF in near-far landscape compositions. In this case you have the infinity at the horizon, often with small fine details like a distant forest, cityscape or mountain, and a close foreground that you also want sharply rendered.

With hyperfocal focusing infinity will be at the far edge of the DoF, which means it will be as blurry as the chosen CoC dictates. Traditional CoC sizes are quite large (since it assumes somewhat limited eyesight and a long viewing distance) and a hyperfocal focusing based on that will thus cause a somewhat blurry infinity upon close inspection.

One can argue that it's more important to render small fine details sharp than larger structures, and then infinity usually needs to be sharper than the foreground. Thus, with the traditional large CoC hyperfocal focusing will not give the desired result. It's then better to focus at infinity and many photographers indeed do that instead.

However, by updating the CoC to something more modern the hyperfocal concept is still useful and can provide a better trade off than focusing at infinity. Setting the CoC to 2.5× pixel size is one good alternative, this way infinity won't be blurry. If the optimal aperture has some visible diffraction onset (nearly always the case with modern high resolution cameras) a CoC set to the Airy disk diameter works well too, or maybe half the diameter if you are extremely picky. With these CoC definitions the DoF edge is defined as where you barely start to see a difference from the sharpness in the plane of focus, or in other words: everything within the DoF will be indistinguishable from the plane of focus even at close inspection (after post-processing sharpening). If you want to make sharp pictures your DoF tables should be based on this definition, rather than the original relaxed viewing distance assumption.

Lens to a digital medium format technical camera (an ALPA) equipped with a high precision focusing ring. Note the detailed distance scale which allows for precise focusing at a specific distance, such as the hyperfocal distance. (Picture by ALPA.)
A vintage external optical rangefinder, useful for estimating distances. Not as precise as a laser distance meter or the high precision focusing ring above, but considerably better than the typical DSLR distance scale.
The second problem with hyperfocal focusing is the lack of precise distance scales on the lenses. Due to the very high precision required it's hard to manufacture accurate and detailed distance scales, and thus they are not found on mainstream lenses.

Most lenses don't have very accurate (or detailed) distance scales; typically there's just numbers printed on the barrel with no indication where to line it up to precisely set it to that distance, and the hyperfocal distance is often somewhere between the last printed number and infinity. Impossible to set. Traditional view cameras are no better as they don't have any distance scale at all. Only some highly expensive technical cameras have reliable distance scales that allows for accurate hyperfocal focusing directly on the focus ring (ALPA most known).

If we do not have a distance scale we can trust we need something in the scene that is at hyperfocal distance, and some way to know that it is at that distance. Some carry a laser distance meter (common among professional architectural photographers where distances and DoF are even more critical than in landscape photography) which indeed is a useful tool. When using it for a while you get better at estimating distances when being without it too. Much smaller and without batteries is a vintage external optical rangefinder such as Leitz Fokos, not as precise but certainly better than the typical DSLR lens distance scale or estimated distance. Fotoman still makes this type of rangefinder new, otherwise you find various brands of them in the second hand markets.

However, even if we have a laser distance meter it's not too likely that we have an object to focus at that is exactly at the hyperfocal distance. Fortunately the non-linear behaviour of the DoF means that even if we overshoot the hyperfocal distance with as much as say 50% we still gain considerable DoF in the foreground compared to focusing at infinity. This is best demonstrated by an example: 50mm at f/11 with CoC 1× airy disk means hyperfocal at 14.8m, if focused correctly DoF goes from 7.4 to infinity, if overshooting 50% to 22m we get DoF from 8.8m, and if focusing at infinity we get it from 14.8m. This property allows us to focus at some other object a bit farther away, and also makes it feasible with some practice to use estimated distance instead of measured.

A forest scene with too close and high foreground to solve with tilt. Hyperfocal candidate? Well, there's no infinity visible, and if the background is slightly soft it may actually add depth. Focus has been put on the tree, closer than the hyperfocal distance.
Foreground obviously past hyperfocal distance, so we can focus at infinity. If the camera lacks reliable infinity stop it's however safer to focus a little bit closer.

In other words, practical hyperfocal focusing usually means to focus at some object in the scene that sits somewhere between the hyperfocal distance and infinity. Since the far horizon is typically in more need of high resolution (due to its small details) it's better to focus farther than the hyperfocal distance than nearer.

In the very rare case when there is only empty space at the hyperfocal distance you could point your camera elsewhere and/or move to another position and focus at any object at proper distance and then return to composing your image having the focus already set.

How often you will actually use hyperfocal focusing will depend on your shooting style. I use the hyperfocal table often (I have memorized the common entries) to see if tilt is worthwhile, but rarely actually focus at the hyperfocal distance. As a matter of fact I had real trouble finding an example picture to put in this section. What usually happens is one of the following:

If I had a technical camera with a high precision focusing scale so that I could dial in hyperfocal distance and trust it I'd probably use it more though.

If you end up with that the scene is best focused at infinity beware that most camera systems lack absolute reliability of infinity marks / stops. This means that it's often wiser to focus at something a bit closer than infinity but clearly farther away than the hyperfocal distance to ensure sharpness at infinity. Otherwise there is a risk to focus slightly "past infinity" which causes the whole image to be less sharp than it could be.

The remaining case when you do not need sharpness at infinity but need close focus, like in the forest scene example, I usually do by experience; focus on a main subject in the scene and estimate a suitable f-number. A DoF table or a DoF smartphone app helps in making the estimate.

Note that while most mainstream lenses lack a detailed distance scale (as it wouldn't work due to sample variation in manufacturing), you could theoretically add one yourself to your specific lens, at least on a manual focus lens which has direct drive on the focusing ring. You would then calibrate the infinity and near-limit position on the ring through repeated test shots, and then paint the scale in-between. It's hard to do such a paint-job though so I consider it theoretical. More feasible however is to calibrate and mark the hyperfocal position (for some typical aperture) with a single mark on a piece of tape. I have not done this myself though as I would rarely need to use it.

Tilting the plane of focus

Textbook example of a near-far composition of a landscape scene where tilt is useful.

When the lens is tilted so is the plane of focus (PoF). While the lens is usually tilted around its nodal point (axial tilt) the PoF axis will be at the hinge line which is below the camera lens if tilting forward. The hinge "line" is an axis which the infinitely wide PoF turns around. In the typical diagram drawn from the side it's thus seen as a point rather than a line, which can make the name a bit confusing. I'd prefer to call it "hinge axis" but I'll use the established terminology.

The lens can be tilted backwards (hinge line above the camera) or left/right (called swing in view camera terminology) or diagonally, but the basic use case for landscape photography is tilting forward, it probably makes up for more than 90% of the cases.

Tilt/swing is used in more varied ways in macro photography. Here some swing to fine-tune focus of the three flowers.

In macro photography, for example when photographing flowers, the use of tilt and swing is generally more varied. The goal in that case is not to fit the whole scene within the DoF (impossible in macro) but to place the PoF so the most important parts becomes sharp.

In this article I will mainly discuss the common landscape use case with forward tilt. This is also a case when calculations and tables can be useful, while in macro photography tilting is more about working from experience and use direct feedback on the live view or ground glass (the all-around tilt focusing technique described separately will help though). Another difference is that in macro it's common with large tilt angles (sometimes more than 10 degrees) while landscape photography usually is in the range 0.5 - 3.0 degrees (smallest with wide angle lenses). With large tilts the DoF gets too shallow to make it possible to bring a whole scene into focus.

Scheimpflug and hinge line

The geometry of tilted focusing.

The Scheimpflug line is easy to visualize, as its the intersection between a plane extended from the sensor (image plane) and tilted lens. For larger tilts you can look at the camera from the side and estimate where the Scheimpflug line is.

However, as the PoF has the hinge line as pivot point (that is it rotates around that when focus distance is changed) that is better used as a reference. The hinge distance does not change when we change focus distance as the Scheimpflug distance does. However since the Scheimpflug distance for the common landscape focusing scenario is usually considerably longer than the focal length the hinge and Scheimpflug distances are relatively speaking almost the same, and you can use them interchangeably for small tilts.

The hinge line does not have as visible geometry as the Scheimpflug, it's an intersection between the lens plane moved forward one focal length along the tilt axis and the image plane moved forward to the lens (not shown in the figure).

I'm myself used to thinking about the Scheimpflug line due to the simple geometry of it, and I still use that out in the field when doing estimates. However, as mentioned the Hinge line is the true anchor point of the PoF so I use it as reference, and it's also the standard in the established literature.

"Upper half wedge" depth of field

Close to the camera the DoF is very narrow and then it grows towards infinity, forming a wedge if looking from the side. The typical use case for a landscape scene is to place this wedge so that the far (lower) DoF edge follows the ground, and the near (upper) DoF edge passes above the tallest feature in the scene, this way everything will be sharply rendered.

Relating this to the field of view (FoV) one can say that the DoF wedge is placed in the upper half. The lower half of the FoV goes into the ground and what we can't see we don't need to have in focus.

Tilted PoF and resulting depth of field placed as a wedge in the upper half of the lens field of view. The idea is that the lower DoF edge should be slightly below the ground level and the upper above the tallest feature in the scene, except possibly clouds which are so low contrast that they may be out of focus. The gain compared to hyperfocal focusing is an area close to the camera, useful for typical near-far compositions. (Example shows 47mm lens f/11 1 degree forward tilt on a 48mm high medium format sensor.)

The figure shows an example, in this case optimal aperture is used and the upper DoF edge angle is a little bit lower than the field of view, meaning that even at infinity a small part of the top of the frame will be out of focus. If that part is only covered with sky this is rarely a problem though.

The actual gain compared to hyperfocal focusing is close to the camera, marked with yellow in the figure. This may look small in the figure, but in an actual picture this close range may cover half the frame.

Horizon placement

When composing a landscape photograph it's quite rare to place the horizon in the center. Most often we apply some vertical shift (or tilt the whole camera if we cannot shift) to get a more pleasing composition. Having the horizon at 1/3 up or down is classic. The following figures show the effects of this.

If we look down more (horizon on the upper half) there is more to gain compared to hyperfocal focusing (yellow area).
When we shift up (horizon on the lower half) we still have some gain in the near field but a major part of the upper image will be out of focus, which may be a problem (unless it is covered with an empty sky).
If we when looking up compensate with increased f-stop the result is very similar to hyperfocal focusing.

As seen in the figures if we adjust the composition to show more foreground (horizon placed in the upper half of the picture) we gain more from tilt, and the opposite if horizon is placed low. Geometrically obvious when you see it in a diagram, and the takeaway message is that if the horizon is low hyperfocal focusing technique is likely to work just as well or better, and if horizon is high (or there is an empty sky in the upper parts) tilt is likely to provide the best solution.

If the camera is tilted instead of applying shift to change position of the horizon the PoF direction needs compensation, but DoF is not affected much at all. That we apply shift instead of tilting the camera is to render vertical features like trees and buildings upright, we don't gain or lose anything significant concerning DoF. The hinge line is indeed moved a bit (backwards if camera is tilted down), but much too little to have significant effect.

Wedge span at infinity

For a specific aperture and tilt the tilted DoF edges hit infinity at the same distance from the PoF regardless of focal length or PoF direction. The PoF is always exactly in the middle between the DoF edges. This span can be called "in-focus window", "wedge span" or "DoF height".

Tilted DoF has the property that for a certain tilt and aperture it spans the same distance on the image plane at infinity, regardless of focal length and PoF direction (this is a slight approximation, the exact solution does require PoF direction, but the difference is so small it's meaningless in any practical application). Rather than calculating the actual angles of upper and lower DoF edges it's easier and more useful to simply point out where in the image they are headed.

At infinity the hinge distance, which is 1 - 10 meters for useful tilts, can be ignored since it's negligible compared to the large distance to the background. In practice this means that objects that are far away can be assumed to fit within the DoF if they are within the "in-focus window", or "wedge span" which I will call it here.

PoF direction is affected both by the amount of tilt and the by turning the focusing ring (or focusing wheel if a view camera). The usual workflow is however to control focus at the near foreground with tilt (since that affects hinge distance which has the strongest effect on the foreground), and adjust PoF direction with the focusing ring.

Hinge parallax

For objects at closer range we may need to consider that the origin of the depth of field wedge is at the hinge line, not at the lens. The problem is that what looks like the highest point in the image from the lens may not be the highest point if we lower the viewpoint to the hinge line — it may be a closer object. It's never an object farther away, so we only need to consider objects in front of what seems to be highest.

The figure below illustrates the issue. There we have two trees, a larger one at 30 meters and a smaller one at 15. If we look through the lens the larger tree farther away reaches higher in the image, but if we lower the viewpoint to the hinge line (the ground or a bit below) the closer tree reaches higher.

We can also note that if there's obivously only one tall object that needs coverage we can ignore the hinge parallax, as regardless of parallax the midpoint and span will stay the same. The problem can only occur if there is a close object in front of the tallest object that reaches higher if we lower our viewpoint to the hinge line.

When we use "wedge span at infinity" table values to fit the wedge over objects at closer distance we may need to estimate the effect of the hinge distance offset / hinge parallax. The yellow lines show the DoF coverage if we pretend that hinge distance is zero, and green the actual.

So how to compensate? First we need to make sure that there is a problem at all, and the easiest way is simply to try to look from as close to the hinge line as possible, lay down on the ground if required. With some experience we can in our mind project a line from the hinge line towards the tallest object and figure out if we need to look at a closer object or not, without having to change our viewpoint.

If we then find out that there indeed is a closer object that is higher, we simply use that as reference. To find out the required wedge span, do the normal way and measure from the ground to top, and focus at the middle of it. If the object is distant measuring the required span is usually easiest to do on live view or ground glass. If the object is close the base of it may actually be outside the field of view and in these cases it may be easier to measure the span using angles, in this case a depth of field calculator app with angle measurement feature is useful. If you only have a table showing wedge spans at infinity on the sensor plane you will have to make an estimate.

It's quite rare to have problems with hinge parallax, and in cases you have it's quite common that the scene is better focused without using tilt.

Upper half wedge tilt focusing technique

There are many tilt focusing guides out there, with generic and similar workflows. This guide is not so much different, but is geared towards the typical landscape use case and has the use of tables as an integral part of the workflow.

Landscape scene tilt focusing (upper half wedge) workflow:

  1. Start with a rough "upper half wedge" setup.
    • Alternative A (when no or unreliable distance scale): 1) Guess tilt (use hinge distance value from table), 2) focus at the middle of upper half at infinity or a lower position on a closer object.
    • Alternative B: 1) focus at hyperfocal distance (for estimated f-stop) without tilt, 2) guess and apply tilt (use hinge distance value from table).
  2. Fine-tune tilt on near-point.
    • When changing tilt the PoF direction is affected. The other way around, when changing PoF direction, the hinge distance is actually not affected but as near-point is bit ahead of the hinge line the sharpness there can be affected. In other words fine-tuning tilt for near-point affects focus on far point and the other way around. By starting with a rough setup close to where we want it in the end this undesired side-effect is minimized.
    • Try to get away with as little tilt as possible as it gives a larger DoF wedge span at infinity. As the lens first sees the ground a bit ahead of the camera, the hinge distance can often reach below the ground level (that is a smaller tilt can be used). The longer the lens, the narrower the field of view, and the farther ahead the first visible ground is and then the hinge line can be farther below ground level.
  3. Estimate or measure required infinity wedge span and get corresponding f-number from table.
    • Consider the hinge parallax issue, that is if you may have a closer object that becomes higher than your tallest target when the viewpoint is at the hinge line rather than at eye or lens level. If so you should use that object as reference instead.
  4. Pick a far-point at midpoint of wedge span and fine-tune focus.
It's also wise to check the hyperfocal tables to see if the scene is possibly solved better with hyperfocal focusing, the rule of thumb there is: horizon low in the image → hyperfocal, high horizon → tilt.

Example scene focused with the "upper half wedge" tilt focusing workflow.
In the figure to the left is an example scene, let us walk through it. It's shot with a digital medium format camera with a 33 megapixel 48x36mm sensor using a 47mm lens. The optimal aperture for landscape scenes for this system is f/11. For a full-frame 135 system (36x24mm sensor, half the size) with similar pixel count the corresponding would be 35mm lens at f/8 (same field of view, same relative amount of diffraction).

The scene is laid out as follows - the grass turfs in the foreground are 2.5 meters from the lens (measured diagonally straight at them), and the water level is about 2 meters perpendicularly below the lens. The closest trees to the right are about 70m away and to the left about 180m. The distant background is the closer part at 500m and the farther at 2000m.

We make a rough composition, shift down 10mm since we want the horizon high in the frame.

Then we start the tilting workflow. First a rough setup of the hinge distance by looking in a table. Since the grass is somewhat high and sharpness of the close range water is not critical we don't want to put it too far under the camera. 1.5 degrees yields 1.8 meters, a little bit higher than the water level. Then we make a quick focus at the middle of the highest tree to the right, since that is the close to the final direction of the PoF.

Now we look at the ground glass with the loupe and fine-tune foreground focus by minor tilt adjustment. The result is a slight reduction down to 1.4 degrees. Since the table gives us a close starting point there are only minor adjustments which means that the PoF direction (which is already close to its final direction) is affected very little. This is a key aspect of an efficient workflow, because if we need to do large tilt adjustments the PoF direction will turn a lot as a side effect making it much more difficult to focus.

Hyperfocal and tilt table for 47, 90 and 120mm focal lengths with the values discussed in the example highlighted. The yellow shows 1.8 m hinge distance for 1.5 degree tilt of the 47mm lens. The green shows 16mm wedge span at infinity for 1.5 degree tilt at f/13. The gray shows 4.1m hyperfocal distance for 47mm at f/20 which gives focus from 2.1m.

With the tilt finalized we then look in the frame and estimate our required wedge span at infinity. The smaller the better, because then it's larger chance we can shoot at optimal aperture.

The topmost of the frame contains empty sky so we don't need focus there. The tree to the right is the tallest feature and there is no closer object that would become higher considering hinge parallax. The required height is shown in the figure and corresponds to 16mm of the total 48mm height. Looking in the wedge span table we see that in the 1.5 degree column we would need f/13 to get 16mm, that is only 1/3 stop worse than optimal aperture. Since our actual tilt is a little bit less than 1.5 the wedge span gets a little bit larger so we get some margin.

We take a quick look at the hyperfocal table to see what it would require, which is about f/20 to get the near foreground into focus, that is significantly worse (1 and 1/3 stop) concerning diffraction, so we can conclude that tilt provides the best solution, as we would expect with this open scene with horizon placed in the upper half.

Finally we fine-tune the direction of the PoF by turning the focusing wheel. With most camera systems it's not that easy to see and get an exact direction, so it's good to have a sense if it's better to aim a little bit higher or a little bit lower than the exact direction. In this case it's better to hit a little bit lower rather than higher to optimize sharpness of the distant features and the house behind the trees. Lower means focusing towards infinity (move lens closer to the image plane).

Then we are ready to shoot! Or at least get on with light metering. In this scene a gradient filter was used to avoid getting an underexposed foreground.

All-around tilt focusing technique

A scene where it's obvious where to put the plane of focus, and we do not need to worry much about the wedge thickness.
Camera tilted down to see the ground, with chosen near-point and far-point. As the far-point is off-axis we need an iterative workflow. The shown start position is with less-tilt-than-you-need and far-point brought into focus, iteration 1 is after a small increase in tilt and far-point brought into focus again. Repeat iteration 2 and in 3 both near-point and far-point is in focus.

Iteration 1b at the bottom shows what happens if we after the start position decide to bring near-point fully into focus by applying as much tilt as we need: overcompensation occurs and we will not find a solution.

While the rotation axis for the PoF is always fixed at the hinge line when changing focus distance, it typically is somewhere inside the FoV when changing tilt, as seen in this figure. When the axis comes close to the near-point it becomes hard to use tilt-controlled focus-peaking, because the near-point is close-to-sharp over a wide range of tilt.

To avoid this one should place the far-point close to the tilt rotation axis, but that is not always feasible so one has to be capable of focusing in this condition.

In the dedicated upper-half wedge workflow described previously we focused at the specific properties of the suitable scenes, how you should place your plane of focus and why. In this section we will look into a generic all-around tilt focusing technique that only goes into how to place the PoF after you have decided where you want it to be.

You have probably seen the base technique described before: pick a near-point and a far-point where you want the plane of focus to cut through, and with the points as guides iteratively move the PoF into position. There are however both shortcuts and challenges buried in this seemingly simple technique.

A shortcut can be had if we can relate to how the camera tilts the lens, so let's start with going through that.

Today the most common tilt design is "axial tilt", which means that the lens is tilted around its nodal point, and thus we do not affect lens-to-sensor distance (that is focus distance) when the lens is tilted. In theory that is, in practice no cameras have an exact match with the nodal point, but it's close. Tilt-shift lenses for DSLRs have this design and most digital medium format technical cameras too.

Some of the older large format view camera designs have "base tilt" which means that the tilt axis is below the lens and thus lens-to-sensor distance is modified when tilting, which can make them harder to work with. Photographers used to it may prefer that anyway. Some view cameras allow tilting the back (=sensor) instead of the lens which can be more precise in terms of keeping the lens-to-sensor distance as the film plane is easier to identify than the lens nodal point.

Here comes the shortcut — with axial tilt you can draw a line on the ground glass (or live view) at the rotation axis (many view cameras actually have a dashed line there), and everything along that line will have the same distance to the lens regardless of tilt. This tilt axis can be used to significantly simplify the tilt-focusing workflow and completely remove the need for iteration.

Tilt focusing with on-axis far-point:

  1. Pick a far-point along the tilt rotation axis, and pick a close near-point. Start with zero tilt.
    • To avoid mistakes in finding the tilt rotation axis (usually horizontal center of the lens) it may be wise to finish tilting before applying shift.
  2. Focus at the far-point using the focusing ring.
    • Make a rocking motion to find the focus peak.
  3. Focus at the near-point using the tilt knob.
    • As the far-point is along the tilt rotation axis the lens-to-sensor distance for it won't change when tilting, meaning that it will not be pulled out of focus.
  4. If the tilt rotation axis is not exactly in the sensor plane or lens nodal point, or if the far-point is not exactly on-axis, fine-tune by performing steps 2 and 3 again, possibly more than once.
This is a very simple and user-friendly workflow. As most cameras are not perfectly axial the refinement step is typically required so we don't get rid of the iteration completely, but it's usually easy to do as the points are already very close to sharpest focus and gets closer for each iteration (and not farther away which can happen if the far-point is too much off-axis). Some DSLR tilt-shift lenses can be quite far from rotating at the nodal point, which means that the refinement step must be be performed a few times before getting both points in focus, but still having the far-point along the horizontal center in the image is generally better than having it towards the edge.

Some view cameras have refined this by making the tilt axis asymmetrically placed (high in the image rather than in the center) so you don't have to place the far-point along the horizontal center. This is called "asymmetrical tilt". However, you still have the line where you have to pick the far-point position, so if you don't have anything in the scene to focus at there you either need to aim your camera towards that direction (and thus limit your compositional flexibility), or place the far-point off axis. While there actually are cameras that have a customizable tilt axis position these are rare (non-existent?) in the digital world so we need to master the case with the far-point off axis as well.

When the far-point have to be placed off axis the workflow can become somewhat challenging. There are two problems to handle: 1) when we change tilt to focus on near-point we will pull far-point out of focus, and 2) the PoF often rotates close to the near-point making tilt-controlled focus-peaking very difficult. To battle this we need an iterative workflow, which has to be performed with great care.

Tilt focusing with off-axis far-point:

  1. Pick a near-point and a far-point, in this workflow we assume that the far-point cannot be placed along the tilt rotation axis (off-axis far-point). To finish faster you can apply tilt here in the start, but make sure that you apply less than you need.
    • For example if you think you are going to need around 2-3 degrees, apply 1.5 degrees.
    • Use a hinge distance table and/or estimate.
    • If you can make a quite precise estimate of hinge distance you may start closer to the expected target.
    • If you are uncertain start without tilt, but then it may take a few more iterations to finish.
    • You must know which direction to tilt, that is if you want the hinge line be above or below the camera. It's almost always the latter and then you tilt forward.
  2. Focus at the far-point using the focusing ring, make it as sharp as possible.
  3. Apply a little bit more tilt to make the near-point sharper, but do not pull it into focus, unless it's close to sharp already.
    • We need to be careful in this step due to that the far-point is off-axis. This leads to that when we tilt for the near-point the far-point will be pulled out of focus as a side effect.
    • If the tilt is still a fair bit from optimal the near-point may not actually become visibly sharper unless you turn a large amount — don't do that! That will cause you to apply too much tilt. Add only a little tilt, say if we start at 1.5 degrees and think we need 2-3, then apply 0.3-0.4 degrees or so. It should be enough to make the far-point become visibly blurrier.
    • If the near-point is still very blurry it can actually be better to look at the sharp far-point instead and tilt (in the correct direction) so it gets just a little blurrier.
    • Looking at the far-point and see it become tilted out of focus is also typically easier and clearer for picking a suitable small tilt step rather than looking at the near-point and watch it become sharper.
  4. Repeat steps 2 and 3 until both are sharp.
    • If you don't finish, sanity check the tilt to see that you have not tilted too far. If near-point gets blurrier and blurrier you have tilted too far.
To succeed it's important that you know which direction you should tilt and constantly tilt in that direction in small steps. For the far-point focusing you don't need to know if you should focus nearer or farther, just rock the focusing ring/wheel and bring it into sharpest possible focus (as you start with less tilt than needed you will be focusing nearer for each iteration, that is increase the lens distance).

This method is quick on a ground glass as you can quickly switch between near-point and far-point just by moving the loupe. On a DSLR or digital back live view it can be a bit slow though depending on how fast the back is at scrolling between the two positions.

You may notice that the near-point is problematic as it's reasonably sharp within a quite wide tilt range. In other words it's difficult to know when you have reached the focus peak, and if you attempt focus-peaking by rocking the tilt knob you will notice that seeing the peak is almost impossible in some situations. Fore example, you may have to turn more than 3 degrees to see any significant difference in a situation you need to be within perhaps 0.2 degrees to get good focus. The reason for this is that with many scene arrangements that require tilt (like camera pointed down towards the ground using a longer focal length) combined with an off-axis far-point causes the PoF to turn close to the near-point when you change tilt, so it stays close to focus (=hard to focus-peak). This is not always the case, but it does occur often (fortunately it's not the case in the upper-half wedge use case, so tilt fine-tuning is often easy to do there). When you change focus distance the hinge line is as always the fixed pivot point for the PoF so its behaviour is more predictable.

Focus-peaking the far-point with the focusing ring is generally much easier, and it also typically reacts quicker to a small tilt change as it's generally farther away from where the PoF turns. Therefore it can be better to keep the loupe/live view at the far-point for the first iterations, but be sure to check so you don't get too far. When it's quick to switch view between far-point and near-point, as with a loupe on the ground glass, you can always use the far-point when you add a small step of tilt (as it's easier to see when you've added a suitable small step), and then verify sharpness of the near-point.

How to know when to stop, that is how to know if the near-point is at peak focus or just almost there? With a fully zoomed digital live view on a good screen it may be possible to see, but on the ground glass you may need peaking to see if you are there, especially if your loupe doesn't have very high magnification. Rocking the tilt knob often doesn't help much due to soft peaking, but there is a trick: when you think you are at the peak, look at the near-point and use the focusing ring instead of the tilt knob and do a rocking motion to see if it's truly at the peak. Since changing focus distance will (typically) produce sharper peaking you can more easily see if the near-point is as sharp as it can be. However, this is just for verification so make sure to restore focus at the far-point before continuing.

Note that depending on scene arrangement it varies where the PoF turns with tilt, so it may happen that it turns close to the far-point and tilt focus-peaking on near-point is actually quite easy, but I would say it's far less common, at least for cameras with axial tilt. Should it happen the workflow becomes easier of course. The above description is about tackling the difficult cases.

Do train on this method, for example by pointing down the camera towards the floor at home and bring it into focus. Test placing the far-point both on-axis and off-axis and notice the difference in workflow. Use a longer focal length so you get to work with larger tilts. It does require some training to get a feel for this and know when to stop, especially with the far-point off-axis.

In many other tilt guides the focusing technique is described vaguely and rarely describing the tilt rotation axis aspect, which can lead to the misconception that one should bring the near-point into focus each iteration even when the far-point is off-axis. This rarely works well, as you then will apply way too much tilt during the first iteration, and when compensating with focus distance in the next it will be so closely focused that you may not be able to get the near-point into focus at all no matter how you turn the tilt knob.

Key to success when dealing with an off-axis far-point is a sane starting point, knowing which direction to tilt, and applying only small amounts for each iteration.

If you have a special camera, such as a Sinar P, it may be worthwhile to look up the manufacturer's instructions of how to tilt, as the camera may be designed to be used with a specific workflow.

Combined tilt and swing

In view camera terminology tilt in the horizontal direction is called swing, and is usually controlled by a separate knob on that camera type. On 135 DSLR tilt-shift lenses and many digital medium format technical cameras the tilt axis is locked in one direction, but some allows rotating the whole tilt mount so you can tilt diagonally, that is combine the vertical tilt and horizontal swing in one tilting motion.

The cases when tilt and swing are combined are quite rare. When it happens it's quite often in "impossible" scenes, that is when you cannot have everything fully within the DoF but must compromise and you have some diagonal element in the scene, for example closer foreground to one side of the frame. In that case one typically first focuses with only tilt, and then with an estimate add a smaller amount of swing.

If you actually have a fairly flat diagonally sloping surface to focus on you will want to make a precise combination of tilt and swing. One way to practice on this is to set up the camera at home, use the tripod head to tilt it both down and to the side with 20-30 degrees each and try to get the floor in focus.

It's most intuitive to do with the cameras that can turn the tilt mechanism, then you simply start with turning it so the hinge line will be at the same angle as the surface, and then tilt with the normal workflow, and possibly fine-tune in the end.

If you have a view camera with separate tilt and swing you need to use the on-axis far-point method or else it will be almost impossible to place the PoF.

  1. Pick a near-point and a far-point which is as close as possible along the vertical swing rotation axis as seen on the ground glass.
    • That is, place near-point and far-point on-axis for the swing axis.
    • Usually the best approximation of this is the vertical center, but it depends on camera model.
  2. Perform the normal tilt workflow.
  3. Pick any side-point to the side of the frame and pull it into focus with swing.
    • As tilt points are on the swing axis we will not pull them out of focus.
    • We may only adjust focus of the side-point by applying swing though, not change focus distance as that would pull the tilt points out of focus.
    • We only need to look at one point for the swing, as the placement of the PoF will be defined in 3D space with three points.
    • Looking at more points along the sides can help in getting a suitable compromise, but as soon as we look at more than three points it may be impossible to get them all in focus as they may not be in the same plane.
  4. Fine-tune against all three points if necessary. Only use very small movements when fine-tuning.
    • As most cameras don't have perfect axial swing there may be some adjustment required of the focus distance and tilt after applied swing, but as we now have three points to take into account it's quite difficult to work with.
    • Sometimes stopping down one more stop may be better than to try to make the perfect tilt+swing PoF placement.

Depth of field tables

Here follows tables for hyperfocal focusing, tilt focusing and mid-range DoF. It's intended to cover what you need in the field for landscape scenes. The CoC is set to 1× the Airy disk diameter, and the tables are thus not dependent on sensor resolution or size. For this to work well the optimal aperture (the lowest f-number you will use) should have some diffraction onset, which indeed is the normal case.

The values in the tables are conservative, that is if hyperfocal distance would be 12.44m the table says 12.5 and half 6.3, the rationale is that the rounded hyperfocal distance should guarantee sharp infinity (always round up). In the mid-range DoF table the near edge is rounded up and the far edge rounded down. In the tilt tables the wedge span is rounded down. Hinge distance is rounded to the nearest though as there is no reason to do otherwise.

Since the DoF in these tables represents near perfect sharpness (the CoC is defined such that everything within the DoF is almost indistinguishable from the plane of focus sharpness even at close inspection, at least after post-process sharpening), it's wise to not follow them too rigidly in the field, that is in difficult situations you can have some parts slightly outside the DoF without losing much quality.

You should know your system and have some feeling how hard diffraction hits. With a smaller aperture you get more DoF (even more than traditional tables since the CoC is diffraction-compensated here), at the cost of resolution loss through diffraction. Depending on your shooting style and scene condition you may want to relax the DoF (let it be a bit blurry at the edges) rather than stopping down to the smallest aperture.

Note that I do not relate to viewing conditions. The traditional thinking is that the CoC should be larger (=larger DoF) to mirror larger viewing distances since then the eye will not resolve into the finer details. This does make sense if you want to calculate what "sharp enough" for a certain viewing condition is, but this is not the way I work when I do high resolution landscape photography. I'm not into "sharp enough" but rather "as sharp as possible". I simply want to maximize the resolution of what the camera can deliver (so I can decide later to print big and/or watch close if I want to) and these tables are designed for that task. If I know in advance I don't need high resolution I shoot hand-held with a small fast camera and don't ever need tables.

The text file dof-tables.txt contains f/11 to f/32 tables for some popular medium format focal lengths and f/8 to f/22 tables for popular DSLR tilt-shift lenses. You can cut it down to suit your own lens collection.

Here is an example table with the focal lengths of the Nikon PC-E (DSLR tilt-shift) lenses:

       f# 8       9       10      11      13      14      16      22
hf/hf2    m
 24mm     6.8     5.4     4.3     3.4     2.7     2.2     1.7     0.9
          3.4     2.7     2.2     1.7     1.4     1.1     0.9     0.5
 45mm    23.8    18.9    15.0    11.9     9.5     7.5     6.0     3.0
         11.9     9.5     7.5     6.0     4.8     3.8     3.0     1.5
 85mm      85      68      54      43    33.7    26.7    21.2    10.6
           43      34      27      22    16.9    13.4    10.6     5.3

tilt(deg) 0.5     1.0     1.5     2.0     2.5     3.0     3.5     4.0
hinge     m
 24mm     2.8     1.4     0.9     0.7     0.6     0.5     0.4     0.3
 45mm     5.2     2.6     1.7     1.3     1.0     0.9     0.7     0.6
 85mm     9.7     4.9     3.2     2.4     1.9     1.6     1.4     1.2

tilt(deg) 0.5     1.0     1.5     2.0     2.5     3.0     3.5     4.0
span   f# mm
        8  19  -- 9.7  -- 6.5  -- 4.8  -- 3.9  -- 3.2  -- 2.7  -- 2.4
        9  24      12     8.2     6.1     4.9     4.1     3.5     3.0
       10  31      15      10     7.7     6.2     5.1     4.4     3.8
       11  39  --  19  --  13  -- 9.7  -- 7.8  -- 6.5  -- 5.5  -- 4.8
       13  49      24      16      12     9.8     8.2     7.0     6.1
       14  62      31      20      15      12      10     8.8     7.7
       16  78  --  39  --  26  --  19  --  15  --  13  --  11  -- 9.7
       18  98      49      32      24      19      16      14      12
       20 124      62      41      31      24      20      17      15
       22 156  --  78  --  52  --  39  --  31  --  26  --  22  --  19

      doff(m) inf         80          40          20          10
dof    f# dofn|pofd
 24mm   8  3.4|6.8      --|--      3.2|5.8     3.0|5.1     2.6|4.1
       11  1.7|3.4      --|--       --|--      1.6|2.9     1.5|2.6
       16  0.9|1.7      --|--       --|--       --|--      0.8|1.5
 45mm   8 11.9|23.8   10.4|18.4    9.2|14.9    7.5|10.9    5.5|7.1
       11  6.0|11.9    5.6|10.4    5.2|9.2     4.6|7.5     3.8|5.5
       16  3.0|6.0      --|--      2.8|5.2     2.6|4.6     2.4|3.8
 85mm   8   42|85       28|41     20.6|27.2   13.6|16.2    8.2|9.0
       11   21|42     16.8|27.7   13.9|20.6   10.3|13.6    6.8|8.1
       16 10.6|21.2    9.4|16.8    8.4|13.9    7.0|10.3    5.2|6.8
(Note that some lenses may need a smaller aperture than f/8 to provide good corner performance, especially when shifted.)

Explanations:

Depth of field calculators

A depth of field calculator app that supports both normal depth of field and the "upper half wedge" depth of field workflow for tilt-shift lenses.

Today when most people carry a smartphone you could use a depth of field calculator app instead of a table card. I actually use both, I prefer to have the table card at hand if the smartphone is short on batteries.

In the app stores there are almost a hundred depth of field apps, most of them free, but few are of good quality. You need the ability to configure the circle of confusion size to what you prefer, and it need to be easy to use in the field and present the numbers you need. Many are obviously not very well-designed when it comes to actually being used in the field to solve real depth of field problems.

If you in addition to normal focusing use tilt there are very few apps to choose from.

I was myself frustrated with this and eventually I made my own app that is available under my company's Lumariver brand. It's highly configurable and supports the landscape focusing techniques described herein, and above all it's designed to actually be used in the field in an efficient manner. I suppose you can consider this a commercial but when you compare the various offers out there I'm confident that you'll find mine to be the most practical. So if you're interested in having a DoF app for your smartphone or tablet please do take a look at the Lumariver Depth of Field calculator app. Even if you don't want the app, its manual is worth reading as it has some insights into the choice of CoC sizes among other things.

Summary

Hyperfocal distance

Tilt

DoF in general

Appendix

Depth of field formulas

Here are the formulas required to make your own tables. Note that f-number is the exact number based on aperture value, that is f/11 is actually sqrt(pow(2,7)) = 11.3, f/13 is sqrt(pow(2,7.3333)) = 12.7 etc. The circle of confusion formula is rather than a fixed number (such as the traditional 0.03mm for 35mm film) an airy disk diameter estimator as I have chosen it to relate to the diffraction blur, the reason for that is discussed elsewhere in this article.

   General:
    fnumber = sqrt(pow(2, aperture_value))
    focus_dist = (lens_dist * focal_length) / (lens_dist - focal_length)
    circle_of_confusion = fnumber / 750  # we choose CoC = Airy disk

   Plain DoF:
    hyperfocal_dist =
       (focal_length * focal_length) / (fnumber * circle_of_confusion)
    dof_near = (hyperfocal_dist * focus_dist) / (hyperfocal_dist + focus_dist)
    dof_far  = (hyperfocal_dist * focus_dist) / (hyperfocal_dist - focus_dist)

   Tilted DoF:
    hinge_dist = focal_length / sin(tilt)
    wedge_span = (2 * fnumber * circle_of_confusion) / sin(tilt)

Diffraction

When light passes through the aperture of the lens the physical phenomenon diffraction causes some scattering showing as slight blurring of the image. The smaller the aperture, the larger the diffraction and thus the blurrier the image. Apart from longer shutter speeds, diffraction is the main drawback of using small apertures.

The exact shape of the diffraction blur from a single ray of light is quite complex and depends on the color of light and exact aperture shape, but can for practical photography be approximated as a circular disk (blur spot) with a diameter of f-number divided by 750. That is at f/8 it's 8/750 = 11 um, at f/22 = 29 um. This is also called the Airy disk diameter.

When the diffraction blur spot is significantly larger than the sensor pixel size there is a reduction of resolution. So the smaller the pixels on the sensor, the larger aperture (lower f-number) you need to maximize resolution.

Beginners are often a bit too afraid of diffraction and as a result get overly difficult problems with too short depth of field when shooting landscapes. While diffraction becomes visible at the pixel level quite early it's a soft onset and proper post-processing sharpening (usually involving deconvolution) can restore much of the lost sharpness. In theory deconvolution can reverse diffraction in full, but in practice there is limit of how much diffraction that can be sharpened well. For a recent DSLR that can be f/16 or f/22.

A less known effect of diffraction is that it has effects outside the Airy disk diameter, since that diameter is only up to the first minimum of an actual pattern that continues outwards. This means that it affects global contrast too, and can cause some bleed from uniformly bright areas into uniformly dark areas. This low frequency effect is usually not visible until very small apertures, like f/22 on a 135 full-frame camera.

Circle of Confusion (CoC)

The scene is in focus only exactly at the plane of focus (PoF), all other parts are out of focus, getting blurrier and blurrier the farther away from the PoF we get. However if the resolution is limited we won't notice that something is a bit out of focus. This is how we get a depth of field (DoF), a depth in space where everything inside appears to be in focus since the limited resolution of our image (or eye) cannot represent the minor blur occurring close to the PoF. As the blur increases it becomes visible and we can see that an object is no longer in focus — we are outside the DoF.

In DoF theory that maximum acceptable blur is called Circle of Confusion (CoC). A tiny spot perfectly in focus will be rendered as a soft circle (or rather a disk) when out of focus, that is the CoC.

Obviously the smaller CoC we choose the shorter depth of field we get in our calculations. In the film era the CoC was chosen to be something related to typical eyesight capability at a typical viewing distance, translated to film diagonal divided by 1500. For 35mm film cameras this becomes 0.03mm, which is about the same blur you get from diffraction when shooting at f/45 (note: Airy disc blur is concentrated more to the center so you need about twice the diameter of diffraction to match the CoC blur, hence f/45 which yields a 0.06mm airy disk, we don't count the low frequency effects of diffraction here though). This is obviously a very relaxed condition, and many even in the film days found this CoC to be way too large. Many have better eyesight than the original viewing condition assumes, and many prints are watched much more closely. While a portrait is typically watched at a comfortable distance to see it in full all at once, a detailed landscape printed big and mounted at comfortable viewing height is in addition often watched closely to see all the details.

The original thinking was also that people would stand farther away from larger prints so larger ones would not need higher resolution, but as in the example just mentioned we know this is often not true in practice. The viewing condition varies depending on subject, print size, presentation and audience.

All this means that if you are into making sharp images with deep DoF with your digital camera this traditional CoC size is way too large. So what to use instead?

A smaller CoC size for the digital era

"Pixelpeep" depth of field (DoF) example. Circle of confusion (CoC) set to Airy disk diameter for the given aperture. The example shows f/11 (Airy/CoC 15um), 100% crops from 33 megapixel 36x48mm digital back (7.2um pixel pitch). Plane of focus is barely sharper than DoF edges, even here on screen at 100%. This example looks the same as it would at f/8 with a 36 megapixel 135 full-frame Nikon D800E (1.4x crop factor). (The difference in brightness is due to uneven lighting may cause the far edge look less sharp than it is.)
Depth of field with traditional CoC size, set to 42um for the 36x48mm sensor, same visual result as with 30um for 135 full-frame. (Excuse me for the very dark far edge, the light did not really reach there. Near and far edges are equally blurred though so one can look only at one of them.)

In modern high resolution digital cameras some say "there is no depth of field", what they mean is that the images are so sharp that when you look at 100% on screen you can see a difference in sharpness even at the slightest distance from the PoF. While being an exaggeration it's true that if we for a 50+ megapixel camera (without anti-alias filter) were to chose a CoC so small that absolutely no difference could be seen in ideal conditions the DoF would be so small that it becomes hard to work with.

There are some common ways to choose CoC that relate to modern digital cameras and their resolution. One is to set it to 2× pixel pitch, another is to set it to the diffraction blur spot (Airy disk).

Both of these CoC will give slightly blurrier DoF edges but still sharp enough to respond well to sharpening and after that look "critically sharp" even at 100% on screen. The intention of this definition is not that the DoF should represent "sharp enough" for some viewing condition, but rather define DoF as "everything within the DoF is indistinguishable from the plane of focus, even at close inspection", which if you ask me is a definition much easier to work with when doing deep DoF landscape photography.

The elegance of relating the CoC to pixel pitch is that a higher resolution sensors will get shorter DoF, and if you watch the image at 100% this is exactly what happens since the increased resolution makes it possible to see that something is out of focus closer to the PoF. When using 2× pixel pitch it's assumed that you don't use so small apertures that diffraction is severely affecting resolution.

Setting the CoC to the same as the diffraction blur spot assumes the resolution is not limited by pixels but by diffraction, which is typically true when your aim is to maximize DoF (note that diffraction is not a hard limit as pixels are though). As you stop down diffraction increases which leads to reduced resolution and thus the CoC should be increased to reflect that. This is elegant, but has the drawback of being dynamically related to aperture.

Some think that those new ways of defining CoC is only for "pixel peepers", and that the old way with a fixed fairly large size also related to viewing distance still is relevant. I would say it depends what you intend to use your DoF calculations for. If you happen to know in advance what viewing conditions there will be using a CoC adapted to that may be relevant. But if you shoot with a high resolution camera with expensive high resolution optics most want to maximize what it can do and be flexible on print size and presentation and have ability to crop heavily for specific uses etc, and in that case a CoC related to the system's resolution capability is more relevant.

If you use the CoC for hyperfocal focusing you should use a small one (related to diffraction or pixel size are both ok), since with a large one the infinity will be slightly blurry which is rarely desired in a landscape scene. However, when you can place the PoF at the most important elements in the scene and want to use DoF calculations to see how small aperture you need to get the rest reasonably sharp those small CoCs may push you too far, stopping down too much or use focus stacking when not really needed. When using tilt this is very evident, with small CoCs it seems like tilt cannot solve any focusing problem (especially in the near field due to the thin wedge tip). However, when you take into account you can put the most important elements in very sharp focus and less important parts fairly in focus, tilt becomes a problem solver in many situations.

If using tables and calculators I prefer to have those based on a small CoC so I know where it's "critically sharp" and then relating to it in a relaxed fashion depending on situation. That is usually more relaxed when focus is on a specific subject in the scene, and more rigidly if there is hyperfocal focusing of the whole scene.

Large format film may need the larger CoC

The smaller CoC size suggested here works very well with available digital cameras and lenses from the smaller sizes up to the largest medium format sensors; despite the reduced CoC size the DoFs get reasonably deep and the aperture range of the lenses from optimal to smallest gives a very workable range — you can often solve scenes at optimal aperture and smallest aperture gives enough DoF for almost any situation.

However, it's interesting to note that the smaller CoC size does not work as well for 8x10" film which was a popular landscape photography format in the film era. The problem is that large format lenses cannot be stopped down enough to provide reasonable depth of fields if using a small CoC size (many lenses are limited to f/64 minimum aperture, which corresponds to only f/9 on a 135 full-frame). If they could there would also often be a problem with very long shutter speeds. This means that if you use those formats you are better off using tables based on the old CoC tradition and accept that at close inspection of larger prints it will be possible to see sharpness fall-off towards the DoF edges.

The smaller 4x5" format was the most popular though and it has a more flexible range, although it still is a little bit more limited than modern digital systems. I would say that it will be a bit tight to use the small CoC sizes suggested here even for that.

I personally think the new digital formats are better balanced in terms of deep DoF than large format film is (especially when taking deconvolution sharpening into account), but there are surely those that prefer the look-and-feel of large format film and the less defined DoF. How you want to work with depth of field is indeed a matter of taste, but I'm confident in saying that with digital the general taste is now different, and a smaller CoC size matches that better.

Depth of field and format size

The depth of field is decided by focal length, focus distance, aperture and your chosen CoC. Format size is not a factor. Still it's often said that larger sensors have shorter depth of field, for example that if you upgrade from an APS-C camera to full-frame you get shorter depth of field. There is however nothing magical about sensor size, it just defines how large part of the lens image circle that we record. What may give us a different experience of depth of field is that we due to the larger sensor area may choose to use other focal lengths, or step closer and thus focus closer.

For example, if we make a headshot with 85mm at f/1.4 on our APS-C camera we may stand and focus at 1.5 meters to get a suitable framing. On a full-frame camera the 85mm lens will have a wider field of view (FoV) so to get the same framing we need to step closer and focus at 1.0 meter, and thus due to the closer focus distance the DoF gets shorter.

The aperture f/1.4 is probably the widest that 85mm lens goes, and you cannot find a product for an APS-C camera that would give you the same DoF and framing, it would be a 50mm f/0.9 lens (divide focal length and aperture with crop factor). So in that sense the larger full-frame sensor gives you the opportunity to shoot with shorter DoF in portraits for example. It's however not the sensor size that makes this possible, but the available lens products. If we look at digital medium format the largest aperture is typical f/2.8 (with rare exceptions) and even if we compensate for the crop we can see that full-frame 135 format wide aperture lenses provide shorter DoF.

In landscape photography we are however rarely interested in achieving short DoF, instead we want the maximum possible. In digital cameras lenses are less limited concerning smaller apertures, most go to f/22 (some even f/32 or more) and those apertures gives us so much diffraction that we rarely want to use them. In other words, we won't hit a lens aperture limit as we may do when we want to have a short DoF (an exception to this is large format film, lenses for 8x10" are often limited to f/64 as the smallest aperture which corresponds to only f/9 on a full-frame 135 system).

So what happens when we want to shoot the same landscape scene with a larger sensor? If we don't move and want the same framing we need a longer focal length. With longer focal length we get shorter DoF, right? Yes, if we don't change aperture. However we can stop down to compensate. If we shot at 24mm f/5.6 with the APS-C camera we shoot at about 35mm f/8 full-frame (24×1.5, 5.6×1.5) and then we get the same DoF and FoV. But doesn't the smaller aperture cause more diffraction? Yes, but it's a zero-sum game, the resolution loss due to the increased diffraction is fully compensated for by having larger sensor area, so in the example a 20 megapixel full-frame sensor will get just as sharp pixels as a 20 megapixel APS-C sensor (assuming a perfect lens).

There is one drawback though, by having to use a smaller aperture to compensate the shutter speed becomes longer. In landscape photography shooting from a tripod this is rarely a problem. However say in a video camera where you typically want fairly large DoF and have a fixed shutter speed it's better to have a small sensor, since you then can get a larger DoF while maintaining a large aperture so that shutter speed gets short enough for video.

Back in the film days a larger film area meant higher resolution, since the resolution was limited by the film. In the digital era this is not as simple, there are at the time of writing APS-C cameras with 24 megapixels at the same time as 12 megapixel full-frame cameras. If you want to max out the resolution you need to relate diffraction to the pixel size (use larger aperture for smaller pixels) and the CoC for your DoF calculations need to be related to the resolution. All this is also a zero-sum game concerning sensor size, if you want a all-pixels-sharp picture it does not matter what the sensor size is, only the resolution matters. Thus you will have a larger DoF challenge with that 24 megapixel APS-C sensor than the 12 megapixel full-frame (since the latter allows for larger CoC and more diffraction without noticing).

The takeaway message for the landscape photographer is that sensor size does not matter, only resolution does. The more megapixels you got the tougher it will be to get them all sharp.

Movements in digital cameras

Large format film cameras usually had very flexible movements, you could shift both lens and back left/right up/down and also tilt (and swing = tilting left to right). With the smaller digital formats and increased sharpness expectations (due to that we love to pixel peep) the analog view cameras have been replaced with more rigid but less flexible digital medium format technical cameras.

It varies between different technical cameras what movements you can do, as they have been designed towards more specific use cases. Some can do almost all the most flexible analog view cameras could, others are more limited.

Tilt-shift lenses for DSLRs can be very limited. Two examples are Nikon PC-E lenses and older Canon TS-E. These have shift and tilt axis locked relatively to each other, typically 90 degrees. That is if you tilt up/down you can only shift left/right.

In landscape photography forward(down) tilt and vertical shift are the two most common movements, often employed simultaneously. If you have only these two movements you will probably still do quite well. This means that if your tilt-shift lens is 90 degree locked you should probably take it apart and remount it with shift and tilt in the same direction (which usually is possible).

Sideways shifts are typically more important in architectural photography, where you may want to place a building to the side of the frame without turning the camera (which would change the perspective of the building). It can be used for subtle composition adjustments in landscape photography too though, for example moving the camera sideways to avoid/change merging of a foreground and background object and shifting to maintain general framing. Once you have it, it may be hard to be without.


Revision history


© Copyright 2012 – 2016 — Anders Torger.
top