At what point does depth of field begin?

[Rooster]

When taking landscape shots I'm told to use manual focusing. Let's say there's a rock ten feet away for foreground interest, a tree 100 yards away and a hill in the distance which I want to be all sharp and in focus. I select f/22 for maximum depth of field. But where do I focus? If I focus on the rock it ensures everything after that is sharp. If I focus on the tree in the mid-distance, is the rock also still sharp? Or does the depth of field only start from my intial focus-point? Does the depth of field extend backwards as well as forward?

 

[TriggerHappy]

What you seem to be after is something called the hyperfocal distance. This is the setting that offers the greatest depth of field. For an excellent explanation and a handy calculator to help get your head around the concept visit depth of field master.

 

[HoppyUK]

That's a very good link posted by TriggerHappy.

To answer your question, as TH says you should focus on the hyperfocal distance (see link) and depth of field extends both in front and behind the point focused on. How much depends on a lot of things and it varies a great deal.

The other thing is that you should avoid f/22 if at all possible. There's an optical aberration called diffraction that afflicts all lenses at high f/numbers and reduces sharpness a lot. It depends on format but with a full frame 5D I would try not to go higher than f/11, f/16 max. By f/22 things will be getting noticeably worse if you want to print large.

 

[tdodd]

There is another approach, recommended by Harold Merklinger, which simplifies things enormously and seems to work well. The technique is described at some length here - http://www.trenholm.org/hmmerk/DOFR.html

In a nutshell, what he is saying is that for landscape style photography you focus at infinity, or the far point of your scene, and choose an aperture that has a diameter physically no larger than the smallest object you want to resolve in the scene. So, regardless of focal length, picking an f/stop value that give you an aperture ~3mm across will resolve blades of grass at your feet, even if you focus into the distance. They may not be absolutely pin sharp but they will be recognisable and anything further into the scene will only be sharper.

e.g. for a 24mm lens, to get a 3mm aperture you'll need to use 24mm/3mm = f/8.

You can scale up the f/stop value for longer lenses and scale it back for shorter ones, simply be dividing focal length by 3, as follows....

12mm lens needs 12/3 = f/4
17mm lens needs 17/3 = f/5.6
35mm lens needs 35/3 = ~f/11
50mm lens needs 50/3 = ~f/16
70mm lens needs 70/3 = ~f/22

Sure you can stop down a little more, to tighten things up further, but be careful not to go mad and end up in diffraction territory. The formula works regardless of sensor format/size.

This approach is great because not only does it deliver results but it means you do not need to memorise tables or carry charts with you. So long as you can divide the selected focal length by 3 you will have calculated an approximately suitable aperture value.

 

[topcatproduction]

Thats a nice simple one to remember!

 

[Chappers]

If you have an iPhone there is a DOF calculator app available called PhotoCalc. It has Sunrise calculator and other bits as well.

 

[HoppyUK]

There is another approach, recommended by Harold Merklinger, which simplifies things enormously and seems to work well. The technique is described at some length here - http://www.trenholm.org/hmmerk/DOFR.html

In a nutshell, what he is saying is that for landscape style photography you focus at infinity, or the far point of your scene, and choose an aperture that has a diameter physically no larger than the smallest object you want to resolve in the scene. So, regardless of focal length, picking an f/stop value that give you an aperture ~3mm across will resolve blades of grass at your feet, even if you focus into the distance. They may not be absolutely pin sharp but they will be recognisable and anything further into the scene will only be sharper.

e.g. for a 24mm lens, to get a 3mm aperture you'll need to use 24mm/3mm = f/8.

You can scale up the f/stop value for longer lenses and scale it back for shorter ones, simply be dividing focal length by 3, as follows....

12mm lens needs 12/3 = f/4
17mm lens needs 17/3 = f/5.6
35mm lens needs 35/3 = ~f/11
50mm lens needs 50/3 = ~f/16
70mm lens needs 70/3 = ~f/22

Sure you can stop down a little more, to tighten things up further, but be careful not to go mad and end up in diffraction territory. The formula works regardless of sensor format/size.

This approach is great because not only does it deliver results but it means you do not need to memorise tables or carry charts with you. So long as you can divide the selected focal length by 3 you will have calculated an approximately suitable aperture value.

Come on Tim, that's a bonkers theory. No wonder it has not found its way into popular use.

Depth of field can only ever be a compromise theory based on acceptable levels of sharpness and while I agree that modern hi-res digital raises questions about aceptable circles of confusion size, that is not the answer. A more practical version of that theory is just to focus on infinity and use an f/number as high as you dare, and some people do just that based on the asumption that it is usually the far distance that is most important to get absolutely pin sharp in most landscapes.

But the problem there is that you are throwing away sharpness beyond infinity, which is obviously useless, and in so doing losing potential sharpness in the foregound. This in turn forces you to use a higher f/number than you would otheriwse need had you focused on the hyperfocal distance, with corresponding difficulties created by diffraction and long shutter speeds causing camera shake.

The best comprmise solution IMHO is to focus on the hyperfocal distance and, if this doesn't deliver an acceptable level of sharpness at the extremes of the focusing range, then close the lens down by another stop. Or two if needs be.That way at least you are using the lens' maxiumum depth of field potential, at the lowest diffraction-free f/number and highest possible shutter speed.

 

[tdodd]

You call it bonkers but have you actually tried it? I have. It works. Feel free to follow your own path. It's no skin off my nose.

 

[HoppyUK]

You call it bonkers but have you actually tried it? I have. It works. Feel free to follow your own path. It's no skin off my nose.

Tim, no skin off anybody's nose. Yes I've tried it, it's not new. It works like the other 'focus on infinity' method of working, ie far distance pin sharp and loss of foreground sharpness.

Sometimes I work that way. It's fine if the foeground is of less importance, but the thinking behind hyperfocal distance is to get maximum sharpness range at a given f/number, which Merklinger's method clearly does not do.

It's only an opinion bud

 

[Chaz Photos]

For landscape I can see no reason to focus on infinity at all when your focus point will give you aprox 1/3 in front of focus point and 2/3 behind it.

 

[dalex_257]

You call it bonkers but have you actually tried it? I have. It works. Feel free to follow your own path. It's no skin off my nose.

I concur with Tim - Harold has a few books on the focusing subject and though I somehow managed to miss this particular article, his take on focusing view cameras (and tilt/shift lenses) made it a lot easier for me to get consistent results.

I'm going to try this one in action at the first opportunity as it sounds really easy to do and calculate.

Edit: having seen Richard reply, I see what you mean about loosing foreground sharpness. Still need to try it out to see how it works with traditional hyperfocal approach

 

[tdodd]

It's only an opinion bud

Yes, it is only an opinion, and there will be cases where a general rule perhaps does not suit best, but as a general rule Merklinger's proposal is not a bad one, and it is very easy to learn and remember. Unless you have something in the scene very close to the camera it really does work pretty well.

To be honest, I would be happier to modify it slightly, to focus half way into the scene, as that makes more sense to me. The guideline for aperture = focal length / 3 is still quite a handy one to know, at least as a starting point. It should hopefully restrain you from simply stopping down to f/16 or f/22 by default, when really you often don't need to go anywhere near that far.

I can remember "focus half way into the scene and set aperture = focal length /3" far more easily than I can remember a myriad of hyperfocal distance settings, and even though I have software on my phone, I never refer to it. I don't really feel the need.

 

[HoppyUK]

I concur with Tim - Harold has a few books on the focusing subject and though I somehow managed to miss this particular article, his take on focusing view cameras (and tilt/shift lenses) made it a lot easier for me to get consistent results.

I'm going to try this one in action at the first opportunity as it sounds really easy to do and calculate.

Edit: having seen Richard reply, I see what you mean about loosing foreground sharpness. Still need to try it out to see how it works with traditional hyperfocal approach

Hyperfocal distance focusing and Merklinger's method are two different ways of working, according to your prioities. That is the difference.

If you want the maximum range of acceptable sharpness possible, use hyperfocal. But the extreme of near and far distance will be right on the limit and with high definition cameras that might not be good enough.

If you want maximum possible sharpness at far distance, focus on infinity, but you'll lose some sharpness at closer range however you calculate it. The only way to recover that is by using a higher f/number than you would need with the hyperfocal method.

Edit: crossed post with Tim. The problem with focusing half way into a scene, or to 'focus a third in' which is the other common maxim, is what is half or one third of infinity?

 

[tdodd]

The truth of the matter is that so long as you focus somewhere between the hyperfocal distance and infinity (or the farthest reaches of your scene) you will usually be in pretty good shape, so long as your aperture is chosen wisely.

 

[Jimmy_Lemon]

I go for the highly technical 10-40mm F11ish focus on what I want in focus (normally foreground).

Seems to work for me

 

[HoppyUK]

I go for the highly technical 10-40mm F11ish focus on what I want in focus (normally foreground).

Seems to work for me

Don't be so sensible That probably works fine, but it's a rubbish debate

 

[DiddyDave]

Having played with my (sold it now) 12-24mm extensively for landscapes, and by utilising the DoF online calculators too, it became obvious that if I picked f8 (this lens' sharpest resolution) then it really didn't matter where I focussed as 'sharp' started only a couple of feet away

If I wanted a blade of grass a foot away nice & sharp & the far distance too, then focussing about 6ft away did it just fine

The DoF on ultra-wides by f8-f11 almost negates the need for focussing entirely

DD

 

[Pork]

I go for the highly technical 10-40mm F11ish focus on what I want in focus (normally foreground).

Seems to work for me

Glad you said that, as thats exactly what I do, and seems to work most of the time.

 

[Garry Edwards]

It's worth remembering that although all of these calculators/methods work up to a point, all that they can actually do is to help you to produce an image which just one part - the plane of sharp focus, or the point at which you focussed, is actually sharp and where nearer and further parts of it become progressively unsharp. Whether or not the unsharp bits are acceptable depends on a lot of factors.

It might be worth trying to get hold of one of the 'split filter' attachments, actually a 1 diopter supplementary lens cut in half. This allows you to focus on infinity to get infinity sharp, any close foreground subjects e.g. a flower can also be sharp being they are being photographed through the diopter attachment. SRB used to do them, don't know whether anyone else still does.

 

[Phillip Mac]

All this is why the great landscape photographers us very large format 10x8 it allows leaning of the lens forward to solve all of these problems

 

[DiddyDave]

All this is why the great landscape photographers us very large format 10x8 it allows leaning of the lens forward to solve all of these problems

Not sure I get that one Phillip

We're talking (or at least I think we are) about using very wide angle lenses for max DoF - AFAIK larger formats don't do very wide lenses (nothing like a 10-20mm 35mm equivalent) and they inherently have a smaller DoF at any given aperture too - that they can tilt them to increase their own DoF is because they have problems otherwise - they are not 'better' at landscapes for DoF reasons

DD

 

[dalex_257]

All this is why the great landscape photographers us very large format 10x8 it allows leaning of the lens forward to solve all of these problems

It does not solve this particular problem, it solves a problem where your DOF area is not enough or is not the right shape to cover all your subjects. This however does not automatically mean that simply tilting your camera will extend the DOF from your feet till infinity.

 

[HoppyUK]

Not sure I get that one Phillip

We're talking (or at least I think we are) about using very wide angle lenses for max DoF - AFAIK larger formats don't do very wide lenses (nothing like a 10-20mm 35mm equivalent) and they inherently have a smaller DoF at any given aperture too - that they can tilt them to increase their own DoF is because they have problems otherwise - they are not 'better' at landscapes for DoF reasons

DD

It is however the principle behind tilt & shift lenses. Canon makes five of them, starting at 17mm.

Though £2k does seem a lttle steep to solve a problem that, frankly, is not really a problem, as some bloke called Diddy Dave says above

 

[mrtoad]

When taking landscape shots I'm told to use manual focusing. Let's say there's a rock ten feet away for foreground interest, a tree 100 yards away and a hill in the distance which I want to be all sharp and in focus. I select f/22 for maximum depth of field. But where do I focus? If I focus on the rock it ensures everything after that is sharp. If I focus on the tree in the mid-distance, is the rock also still sharp? Or does the depth of field only start from my intial focus-point? Does the depth of field extend backwards as well as forward?

lenses used to have marking on the rim...you just set infinity on one side and check the other that its close enough
if not up the aperture
the resulting value at the centre was the hyperfocal distance
my take is go for the nearer objects as they lead into the shot
forget infinity unless you have really good optics, no atmospherics, and are going to view on a microscope

take a few and then choose later...:shrug:

 

[Garry Edwards]

All this is why the great landscape photographers us very large format 10x8 it allows leaning of the lens forward to solve all of these problems

Agreed, thanks to that nice Mr. Scheimflugg it's simple to shift the plane of sharp focus almost as required, DOF doesn't change but the effect does.
I think it's a great pity that tilt/shift lenses are both seriously overpriced and nowhere near as effective as front standard tilt

 

[Suvv]

lenses used to have marking on the rim...you just set infinity on one side and check the other that its close enough
if not up the aperture
the resulting value at the centre was the hyperfocal distance

and wasn't it easy

nowadays I have to calculate the hyperfocal distance
and then try to guess how far away that is
and then find something at that distance to focus on

 

[StewartR]

Hmmm. Lots of disinformation here, as usual when DOF is discussed.

For landscape I can see no reason to focus on infinity at all when your focus point will give you aprox 1/3 in front of focus point and 2/3 behind it.

This notion of the DOF extending 1/3 in front and 2/3 behind is simply rubbish, but that doesn't stop it being peddled around as if it were the gospel truth.

A little while ago I had a slow day in the office so I did some calculations. (Well actually DOFMaster did.) Suppose you have a 55mm lens on a 350D at f/8. The hyperfocal distance is 20m. With the subject at various distances, the DOF is as follows:

* Subject at 1m: DOF from 0.95m to 1.05m (48% in front)
* Subject at 2m: DOF from 1.82m to 2.22m (45% in front)
* Subject at 3m: DOF from 2.61m to 3.52m (43% in front)
* Subject at 5m: DOF from 4m to 6.65m (38% in front)
* Subject at 6.67m: DOF from 5m to 10m (33% in front)
* Subject at 10m: DOF from 6.67m to 20m (25% in front)
* Subject at 15m: DOF from 8.57m to 60.2m (12% in front)
* Subject at 19m: DOF from 9.73m to 395.4m (2% in front)

Note that there is only one point at which the DOF is 1/3 in front of the subject and 2/3 behind the subject, and that is when the subject is at exactly 1/3 of the hyperfocal distance.

Note also that the maximum DOF is achieved when the subject is at the hyperfocal distance.

However ...

What you seem to be after is something called the hyperfocal distance. This is the setting that offers the greatest depth of field.

I think the notion of of hyperfocal focussing is better in theory than in practice. Yes, the idea is, of course, the you get the maximum DOF. But what that means is that the far distance is, by definition, right on the limit of acceptable sharpness. But that word "acceptable" contains all sorts of hidden assumptions, which may not be valid for your particular situation. If you have details in the far distance which you want to ensure are sharp (eg a range of mountains) then it is safer to ensure that your focus is considerably beyond the hyperfocal distance.

 

[StewartR]

I select f/22 for maximum depth of field.

Not a good idea at all. Read up about diffraction. With modern DSLRs, shooting at f/22 is a pretty good way of guaranteeing that nothing will be really sharp.

 

[HoppyUK]

This notion of the DOF extending 1/3 in front and 2/3 behind is simply rubbish, but that doesn't stop it being peddled around as if it were the gospel truth.

A little while ago I had a slow day in the office so I did some calculations. (Well actually DOFMaster did.) Suppose you have a 55mm lens on a 350D at f/8. The hyperfocal distance is 20m. With the subject at various distances, the DOF is as follows:

* Subject at 1m: DOF from 0.95m to 1.05m (48% in front)
* Subject at 2m: DOF from 1.82m to 2.22m (45% in front)
* Subject at 3m: DOF from 2.61m to 3.52m (43% in front)
* Subject at 5m: DOF from 4m to 6.65m (38% in front)
* Subject at 6.67m: DOF from 5m to 10m (33% in front)
* Subject at 10m: DOF from 6.67m to 20m (25% in front)
* Subject at 15m: DOF from 8.57m to 60.2m (12% in front)
* Subject at 19m: DOF from 9.73m to 395.4m (2% in front)

I think the problem with these assumptions, rules of thumb, maxims or whatever you want to call them, is that some people regard them as absolute truths when they are nothing more than generalisations. But there is usually some truth in them.

As for 'focus a third in' if you take the average of the percentages above, you get 31% - almost exactly one third. In the absence of more specifically accurate information, 'focus one third in' isn't such a bad plan.

 

[StewartR]

As for 'focus a third in' if you take the average of the percentages above, you get 31% - almost exactly one third.

I'm guessing you didn't study maths at college, Richard?

You simply cannot take an average of a random set of numbers like that and expect it to mean anything. You just can't.

In the absence of more specifically accurate information, 'focus one third in' isn't such a bad plan.

Yes it is. Let's see why.

Here's a random example landscape. It's not one of mine - it happens to be in the most recent thread in the Landscapes Forum as I write, which is my way of making a random choice.

So what do we have? Some rocks and bushes maybe 2m from the camera, more foreground bushes stretching out to maybe 20m away, some trees maybe 2km away, and some mountains in the distance that could be 30km away or even more.

You know what comes next, don't you? If we want to focus "one third into the scene", where is that exactly? If those mountains are 30km away, then that rule suggests we should focussing on a spot 10km away. I can't think of any other meaningful way to interpret "one third into the scene", can you?

On the other hand, proponents of hyperfocal focussing would have a rather different recommendation - the hyperfocal distance for this image (40D, 17mm, f/22) is 0.7m.

I think both rules are wrong. Focussing 10km away runs the risk that the foreground might not be sharp enough, because it's close to the limit of the zone of "acceptable" sharpness. Focussing 0.7m away runs the risk that the background might not be sharp, for the same reason.

 

[HoppyUK]

I'm guessing you didn't study maths at college, Richard?

I can read a profit & loss spreadsheet. Does anything else matter?

You simply cannot take an average of a random set of numbers like that and expect it to mean anything. You just can't.

Why can't you? Bearing in mind that I am not trying to defend these 'rules of thumb' I'm just trying to explain them...

If you look at the optimum focus point for maximum depth of field in an image, you will find that it is always less than 50%, and very rarely less than 25%, so settling on 'one third in' in the absence of more specific information is probably not a bad idea.

Here's a random example landscape. It's not one of mine - it happens to be in the most recent thread in the Landscapes Forum as I write, which is my way of making a random choice.

So what do we have? Some rocks and bushes maybe 2m from the camera, more foreground bushes stretching out to maybe 20m away, some trees maybe 2km away, and some mountains in the distance that could be 30km away or even more.

You know what comes next, don't you? If we want to focus "one third into the scene", where is that exactly? If those mountains are 30km away, then that rule suggests we should focussing on a spot 10km away. I can't think of any other meaningful way to interpret "one third into the scene", can you?

On the other hand, proponents of hyperfocal focussing would have a rather different recommendation - the hyperfocal distance for this image (40D, 17mm, f/22) is 0.7m.

I think both rules are wrong. Focussing 10km away runs the risk that the foreground might not be sharp enough, because it's close to the limit of the zone of "acceptable" sharpness. Focussing 0.7m away runs the risk that the background might not be sharp, for the same reason.

The 'one third in' rule doesn't work with infinity subjects. I said so in post #13. You're absolutely right.

But I sometimes do use the one third in rule for closer subjects, usually in non-critical situations where I want to make the most of the DoF I've got, without being too critical about it. Maybe a social group where I know pretty much that I've got more than enough DoF to cover it, but I might as well try to optimise sharpness over the main subject area. As a casual guide, it works well enough.

When things are more critical, I have a little DoF calculator disc. The other thing that works is the Canon A-DEP setting - I've tried it a few times and it really does - I just don't know why I don't trust it!

Then again, being practical, how critical is DoF in most landscapes? With a wide lens on a crop camera, at f/11 everything is sharp from the end of your nose anyway Digital photography is liberating in many things that were difficult with film. High ISO makes a huge difference, plus the inherant extra DoF you get with crop format digital, you're at least three stops better off even before you start.

 

[InkZ]

I usually just focus one third in, it seems to work ok for me. I'm not really interested in landscape with no foreground personally.

I'm no expert but it gets me results I'm happy with.

 

[markta]

So presumably to focus at infinity, you compose then set the lens to manual focus, turn the zoom ring to the infinity scale, and say if the calculator says 0.7 metres at f11, you then focus at a part of the ground 0.7 metres in front of the camera/tripod?

 

[HoppyUK]

So presumably to focus at infinity, you compose then set the lens to manual focus, turn the zoom ring to the infinity scale, and say if the calculator says 0.7 metres at f11, you then focus at a part of the ground 0.7 metres in front of the camera/tripod?

To focus on infinity, you focus on infinity. Simple as.

To maximise depth of field from infinity to close distance, you focus on the hyperfocal distance. This varies with both focal length and f/number and you really need a scale or calculator to do it.

The 'focus one third in' idea isn't so helpful when you want sharpness at infinity, because you don't know where infinity starts - it varies with focal length, just a few metres with a super-wide, hundreds of metres with a super-tele. If you do know where infinity is, chances are you have read it off a DoF table that will also tell you where the hyperfocal distance is.

 

[StewartR]

You simply cannot take an average of a random set of numbers like that and expect it to mean anything. You just can't.

Why can't you?

I give up. :bang:

 

[mrtoad]

and wasn't it easy

nowadays I have to calculate the hyperfocal distance
and then try to guess how far away that is
and then find something at that distance to focus on

hey man..its a gas...at least we can 'understand' whats going on and take the damn pictures...

with digital you can mop up the mess with the delete key...and your printer doesnt get to see your mistakes

cheers

 

[jerry12953]

Most people have depth of field preview buttons on their cameras, don't they?

I'm all in favour of using hyperfocal distance to get best focus. Given that its roughly 1/3 of the way in to the image, set that as your focus point, check using DoF preview button, re-adjust if necessary!

It used to be so much easier in the good old days when everything was better.

Well, when lenses had DoF scales on them, anyway, as mrcrow says.

And 10 x 8 plate cameras have such miniscule depths of field you need any trick that's going to get good focus.

 

[jerry12953]

I can read a profit & loss spreadsheet. Does anything else matter?

Har, har, har......

(re copyright discussion recently......)

 

[HoppyUK]

I give up. :bang:

Stewart, the point is that the figures you posted and not random. They are all relatively closely grouped around one third. If you average them all, you get 31%. If you take away the two extremes and average the rest (common method of statistical analysis) you get 32%... :shrug:

If you know the distance of the near and far subjects needed to be sharp, focusing 'one third in' is a good general rule. In the absence of more accurate information, it will not be far out.

Most people have depth of field preview buttons on their cameras, don't they?

But they are not very helpful even on a full frame camera. With the smaller, darker image inherant to a crop-format camera, they are virtually useless.

If you want to do it visually, take a snap and scroll over the LCD image on high magnification. That's not a bad way.

It used to be so much easier in the good old days when everything was better.

Well, when lenses had DoF scales on them, anyway, as mrcrow says.

Most primes still do have DoF scales.

Har, har, har......

(re copyright discussion recently......)

Re recent copyright discussions, you would be significantly better off if you took guidance from your own profit & loss accounts. (Sorry bud, don't mean to be rude, but I think you know what I mean )