The Problem With The Focus-Recompose Method

A Post By: James Brandon

This post was written as a follow up to my last article on using Single Point AF. In the comment section of that article there was a bit of debate over which focus point to use and when. One camp was stating that there was really no point in using anything other than the center focus point. A lot of people thought it was silly to use anything else because the other focus points aren’t as accurate. Some examples of comments for this argument include…

Just use the center focus point and recompose. After all, how often do you shoot action shots anyways?
Focus-Recompose is so much easier than messing around with changing focus point. Why risk missing the shot?
Nice tip but it’s much easier to just use the center focus point, keep your finger down, then recompose the scene before shooting
Center focus point only. Recompose. Always.

So, is there any truth or accuracy in these statement? Well for one, any time your subject is moving that could be considered an action shot. So I shoot action all the time. As far as the focus-recompose debate goes, I strongly disagree with the focus-recompose method and in this article I’m going to present my case against it. This debate may not start a flame war or anything like that (well, maybe), but I do think it’s a good idea to dive deeper into this topic and find out which technique is better. Center point focus and recompose….or full access to all your camera’s focus points? Let’s go.

Why The Focus-Recompose Method Often Fails

I think you’ll agree that I’m pretty awesome at drawing stick figures with my Wacom Tablet, so I thought it might be easier to explain why focus-recompose often fails by creating a little diagram and then explaining this diagram in steps. So here we go…

In the diagram above, ‘A‘ represents pointing the camera up towards the subjects face and placing the center AF point over the subjects eye to achieve focus. Most photographers using the focus-recompose method know that nobody wants to see an eye or face right smack dab in the middle of the frame, so they then recompose the scene by moving the camera down to get the entire body in the frame or to simply move the subjects face off center. This new camera angle is represented by ‘B‘ in the diagram.

I’m certainly not a mathematician but one of the few things I remember from geometry is the Pythagorean Theorem and the common sense that the length of A is longer than the length of B, and that if you were to lay the A line down on top of the B line, you would see the difference in length between the two.

If you stand 4 feet from your subject and point the camera up at the subjects face, then you are no longer 4 feet away from what you’re focusing on. If the length from your camera to your subjects chest is 4 feet and the length from your subjects chest to their eye is 2 feet, then the length from your subjects eye to your camera is 4.5 feet. Are you getting this!? That means that if you focus on your subjects eye, move the camera down to their chest to recompose, then your focal plane is now half a foot behind your subject! The difference between the two lengths is show in figure ‘E‘ in the diagram, with ‘C‘ being the actual focal length when recomposed and ‘D‘ being actual distance to the subject.

So What’s The Big Deal With Half A Foot?

Some readers out there might just be wondering what the big deal is about 6 inches. So here’s a screenshot from an app I’ve got on my iPhone called Depth of Field Calculator. This app will calculate your depth of field based on the settings and information you provide. If you can’t quite see the screenshot just click on it to see a bigger version.

The app lets us know that if we are 4 feet from our subject with a 50mm prime wide open at f/1.4, then our depth of field is only 0.16 inches. In other words, our focal plane will begin at 3.92 feet away from the subject and focal will end at 4.06 feet. Now if you remember from the diagram above, if we focus on the eye and recompose to where the center of the camera is pointed at the subjects chest with their head in the top third, then our 0.16 ft plane of focus is actually 4.5 feet behind our subject. Therefore, we have an out of focus image. Want proof? I thought you’d never ask!

The Proof

The demonstrate my point, I slapped a 50mm prime onto my 5DMII and went over to the dart board in my office. With my camera on a tripod I placed the center focus point over the ’20’ on the dartboard, achieved perfect focus and then took one picture. Without changing anything I recomposed the scene by moving the camera down until the center focus point was over the center of the dartboard and took one more shot.

Camera info: Canon 5DMII with Canon 50mm lens, f/1.4, Shutter Speed 1/125th, ISO 250

Here are the resulting images…

And here are both images zoomed in to 100% to see if they are sharp….

Conclusion

Focus-Recompose is easy, intuitive, quick and self-defeating. While this method will work in some situations, there’s no way to know how well it will work without calculating your depth of field before each shot to see if you have any wiggle room with your depth of field. So don’t be afraid to use your other focus points to avoid focusing and recomposing. I won’t disagree that the center point is the most accurate but the other focus points are hands down a better option than recomposing without refocusing.

Agree or Disagree? Let me know in the comments, but do your best to keep it civil. There are people trying to learn here and we don’t need another article turning into a Canon/Nikon debate!

Are You On Google+?

I know this is off topic, but man have I fallen for Google+. If you’re still in need of an invite code, use this link to get in. If I run out of invites from that link I may still be able to get you in by email. Once you’re in be sure to add my profile to one of your circles and drop me a line. Maybe we can get some people together for a G+ hangout/discussion over this topic. Cheers!

Some Older Comments

khu July 20, 2013 11:41 pm

If your focus is at 10 feet and recompose. Your focus is lock at 10 feet and your subject is still 10 feet from your camera.
Steve July 4, 2013 04:53 pm

The proof is flawed. No right thinking photographer would focus on something then put it at the edge of the frame.
JK May 10, 2013 07:28 am

How does one know that the focus plane is actually flat? And not spherical? Makes a lot of difference. Spherical makes more sense to me, but I don't know. And more over, if the field of depth is minimized as in the proof example, should the rotation center not be in the exactly at the image sensor to proof a point anyways? In the proof example the rotation center is at the tripod rotation point. This results in a rotation AND a translation of the image sensor.
Tony May 1, 2013 03:19 am

In your example using the dartboard you are correct / and incorrect. If you use a 50mm prime lens at it's widest aperture (or most any other lens for lens for that matter) you are making your depth of field as small as it can be. Thus, in your example you are correct because recomposing your shot is shifting the plane and you have literally no room for movement. Thus you achieve a softer focus than what you were shooting for. However, if you were to open your aperture up to (say) F-8 and take the same shot you will see that your focus will remain sharp. Maybe not tact sharp, but that is where you need to consider what your end goal is. If your end goal is to be tact sharp on a specific area (IE Eyes) of your subject you are better off composing your shot first and then selecting the focus point that aligns with your subject. On the other hand if your end goal is to achieve a shot that is the type of shot many everyday camera users go for ( "good") focus evenly across the shot, this can be accomplished by increasing your depth of field. In this case you will be fine by focusing initially on a specific spot and then recomposing because you field is deeper. It really depends on what your ultimate goal is for the shot.
swapnil deshpande February 21, 2013 07:31 pm

dear sir,
loved this article.
i use a 600mm on a cropped sensor (50d canon). i use backfocussing technique keeping AI servo mode and centr point focus. same prob is seen with eyes being outta focus :( .
my question is which can be a faster tech? back focussing or conventional.
would u advise to us e conventional tech and shfting focus points?? wud tht be a faster n better method?
pls advice.
thanks
Brent January 25, 2013 12:21 pm

While I don't doubt your test with your lens, I just did a test with my Sigma 30mm f/1.4. In this instance there is no "plane" of focus that tilts with the camera. I have a "dome" of focus or a radius. I tested by using Magic Lantern firmware to display animated contrast focus in live-view while pointing the camera directly at the floor. If the center is in focus some closer proximity is required to shift the edges into focus, which then softens the center. Another test is to angle the camera at about 45-degrees and observe the same sharpness animation in live view. If my lens had a focal "plane" it would show a straight line of focus cutting across the carpet. In this instance it is a curve. This means that any recomposing of the shot may leave the eyes in focus as they are still equidistant from the sensor.
I'm anxious now to test other lenses to see if any I have access too have a plane or "dome" of focus.
Alex Ruiz E. December 20, 2012 07:57 pm

Here's what I gather from this heated discussion:
Imagine I focus an object at 4 ft. Every plane that is parallel to the sensor and tangent to a sphere with a radius of 4 ft from sensor will be in focus. If my subject doesn't move but I move the camera around to recompose, the plane in focus will move accordingly and the element I focused initially on (say the eyes) will be out of focus.
How much out of focus depends on the aperture I'm using, and so this debate really only matters for very shallow depths of field, and is negligible for higher apertures.
Matthew Miller December 1, 2012 06:53 am

I know this is an old post, but: the problem is that the diagram is badly drawn. The vertical red and blue lines shouldn't both be vertical — at least one needs to be tilted to match the tilted camera.
Archie Macintosh November 29, 2012 09:54 pm

Marianna is quite right - the best way to focus depends on what you're doing.
Street photographers often use 'zone focus' - for example, manually focussing the lens at, say 8', and setting the aperture to f/8 so that a 35mm lens on a crop-frame SLR will be in focus between about 6' and 12'. Then you leave things set like that - there's hardly time to use autofocus, let alone mess about moving the AF point to the 'right' place in the frame. You only have time to compose.
I guess a photographer doing a fashion shoot for Vogue with a Hassy on a tripod might adopt a slightly different approach!
Marianna Zelichenko November 29, 2012 08:14 pm

I've had this problem every once in a while, usually with portraits and such. However, I would say it really depends on the type of photography. I do a lot of work on the streets and sometimes I just need to be really fast and don't have time to select the right AF point.
Andre Recnik November 28, 2012 10:50 am

The focus plane is parallel to the sensor plane (subject to field curvature), so the diagram is not quite correct - the first focus plane should be tipped a little to make it perpendicular to the line from the camera. Then it will make more sense.
However this is a minor issue, and the math and application is correct. Even though the distance from the camera to the eye hasn't changed in reality, the location of the focus plane has - so you will get these issues.
Of course if you are shooting at a longer distance the effect will quickly approach 0, since the angle will be much smaller. So focus and recompose will work just fine in that case. However working closely up with a fast lens will make this problem very noticeable.
Ben Strasser September 25, 2012 11:27 pm

Jim, Regarding your article on focus and recompose let me point out that your images of the dart board were taken at different distances. If this is your concept of recompose then surely the focus point would change. When I think of focus and recompose I associate it with a zoom lens and not a prime lens. Focusing on let’s say the eye, holding the shutter button half way down and recomposing with the zoom will not change the focus point if #1 the lens is parfocal and #2 you do not move. In your example you said you used a 50mm prime lens so the only way you could recompose would be to move closer or farther away from the subject thus changing the focus point.
a different Don September 24, 2012 12:11 pm

This sphere-plane debate reminds me on the conflict over the flatness of the earth, though perhaps a back-to-front version of it.
Is this not correct? A single plane of sharp focus is intended by the manufacturer to be flat but if on the diagram you drew a series of these planes corresponding to each degree of tilting the camera a diagramatic arc would be produced.
In the real world your camera will still only ever see a single flat plane. The DOF around that plane is necessarily also flat, and if the model's eyes are not within it they will not be sharp.
Discuss.
greg September 6, 2012 09:55 am

@DON
you're right.
Before recomposing "UM 20" was placed at the centre of the lens, that's why it's sharp. After recomposing "UM 20" was placed at the edge of the lens, that's why it is soft.
Ray August 29, 2012 05:00 am

When I first read this article, I was convinced that the logic was wrong for all the reasons previously described by many others in the posts above. The discussion about focal planes versus spheres finally caused the light to come on and now I get it. For those still struggling to understand, I would like to offer a few additional words of explanation using the stick-man drawing at the top of the article. It should be remembered that the focal plane is perpendicular to the direct line from the camera sensor to the focal point. Referring to the line designated as "A" in the drawing, remember that the focal plane will remain a constant DISTANCE from the sensor to that focal point. Whatever that distance measurement happens to be, when the camera is rotated down consistent with line "B", the focal plane measurement remains the same DISTANCE as before and therefore corresponds with the red vertical line rather than the original blue vertical line. For that reason, the focal plane will have shifted further back from what was originally intended.
Juanita April 21, 2012 06:50 am

YES! This is the missing link I've been looking for - for the last 2 years!!!! I've been using both methods (unintentionally - just randomly depending on the scenario). Although they look fine on the LCD of the camera, I would get home and be sooo aggravated in the post processing when I saw my image focus was "soft". I never paid really close attention to exactly how I was focusing. I was afraid that my lens or camera was faulty or getting overworked as time went on... I am totally going to pay attention on my next shoot. As of late I have been getting quite good at flipping the points around with my thumb - so that is already my preferred method. THANKS!
David January 31, 2012 12:46 am

Did you try the test the same shot at F4 with wide angle lens?
Geoff January 2, 2012 04:30 pm

James thanks for pointing that the focal plane is a PLANE not a sphere, for 22 years I've focused and recomposed for some reason today I questioned it and found this page.
It's pretty clear that the sphere is a common misconception - could you maybe do do another article to publicise the focal-plane-is-(nearly)-a-vertical-plane thing?
Quoting your post for anyone passing by ;
"..for those stating that the focal plane is in the shape of a sphere – it is not.... walk up to a wall and photograph it and most shallow depth of field you can attain. If you shot the wall straight on, you’ll see that the entire wall is in focus, no matter how shallow the depth of field. If the focal plane were spherical, the center of the wall would be in focus and the wall would gradually become blurry towards the edges"
Bleed November 6, 2011 03:59 pm

Has anyone here tried pressing AF lock after you set your focus? maybe you can experiment this function. I Believe the camera companies has answered this issue. Try it.
Bill M. October 10, 2011 10:00 am

Participated in a photo walk recently using models and F&R method. Acceptable shots were more luck due to depth of field when using manual mode. Have to admit, never moved focus pints before. They are a major PITA on my K-x, which is prohibits moving focus point in Manual mode and defaults to center point, but is still an option on my old DRebel. Looks like it will be back to the Rebel unless someone can offer some advice. Maybe move to Av on K-x?
Paul September 18, 2011 08:26 am

Interesting, something I hadn't considered too much before!? I tend to use F&R when I need to work quickly
samar September 12, 2011 02:37 pm

As far as the maths is considered(as per the diagram) there is no change in the distance of the infocus point after u recompose(assuming u just rotating from a fixed point) and hence eyes of the subject should still be in focus, but in real life it might not deliver the sharpest of images since the subjects would be moving and also our hand would not be tripod stable. I guess this can be little compensated by using a higher F if possible. I feel focus-recompose method might be useful when u need to take a shot fast as manual focusing and multiple focus point can take a while. I feel focus-recompose might be useful for a far off and slow moving subject. like a old man waking along the beach.
Kimberly September 10, 2011 03:47 am

my apologies, I haven't a CLUE why it posted twice! and NO I DIDN'T SUBMIT TWICE. I corrected a couple typos - and stopped the upload of the first post. I thought that the 2nd one posted only, then i just came back and they are both there. . sorry -
Kimberly September 10, 2011 03:44 am

Yes, this is a rant - but for goof purpose. And , my intention is NOT to offend a SINGLE person here - but, I need to say some things... so read on if you so choose so. And im well aware that its super long, it wasn't intended to be, but thats how it ended up and im not going to edit what I have to say, this needs to be purged...
I don't understand why so many people get butt hurt over any article here which are meant to HELP. - GROW UP. Quit being jerks, have a little appreciation and if someone may be flawed, say it like an ADULT and try to be a little POLITE you don't need to go out of your way to be an A$$HAT to prove a point.
This is SUPPOSED to be a place where we as creatives and photographers get together and have a certain level of respect for those who take time out of THEIR lives and away from THEIR families and THEIR craft to help us... They are doing US favors - and NO-ONE EVER says their articles are THE ONLY WAY TO DO SOMETHING. It's up to YOU to do what's best for and tweak it to make it work for YOU - and your shooting situation, and how it may apply to the type of gear you have.... Frankly i'm not wealthy, and I love a sharp image, but for me to make it what its all about, is a sharp image,i'm unfortunately not ABLE to go drop the dough for a Hassle... I need to make my Canon 50D, my Canon 7D, my Panasonic Lumix LX3 all work to their capabilities.... (and hopefully a full from 5D Mark III in maybe 4 years when I could afford it - and of course Leica, but thats another story...) :)
Sarcasm Alert: *Gawd, if only I could get through my own thick skull that if I thought so highly of MYSELF and of my perfect techniques as some of you folks here, then my photography would be perfect too and I wouldn't need this lousy, flawed place. So, if you ARE so perfect, why ARE you are and aren't at your own blog/site putting your own spin on things?
And props to those of you who are ladies and gentlemen who appreciate or offer their suggestions or ideas on these articles. I learn so much from peoples inputs, even the nasty ones, whether it be how NOT to conduct myself, or how to ENHANCE what was said in an article. I just don't want to read the attitudes. I am 100% FOR freedom of speech - its what our country is based upon - however, I don't have to LIKE the negative and holier than thou attitudes. Ridiculous. Please think again before you just spout your obnoxious retort. It will bring you better karma, and gawd knows we all need that....
Carry on peeps! Carry On!
(and yes, this has been building for about a year and a half and have bit my tongue and not said anything to those of you who know exactly who you are) - so sorry for the blow up if I personally offended someone else because I spoke my mind on the lack of courtesy and respect that happens around here so often. Reminds me of those nasty Adobe Forums - those are the WORST, I just don't want this place to get that way with the readers turning into dictator as opposed to the admins, and forum heads, and gurus, and 'celebrated experts' 99% of them have heads so big that they can't get see the forest through the trees - and these people at the Adobe Forums actually PRIDE themselves on being jerks especially the MAC guys.. its just sad... Just go over there and read any of those articles in any of the forums...
Another reason why I love those who write the articles here, no pretense, no airs, no ignorance, and totally here ONLY to help us with complete approachability.
thanks for reading - and again I APOLOGIZE if I personally have offended ANYONE - as that was not my intention. I think all of you know this...
Please be nice to one another and respect others 'words' as really that is our only truth there is - but its up to us to seriously respect our passions, which here is all types of photography. It's what MY Life is all about and I wouldn't trade it for anything in the world.
Kimberly September 10, 2011 03:43 am

Yes, this is a rant - but for goof purpose. And , my intention is NOT to offend a SINGLE person here - but, I need to say some things... so read on if you so choose so. And im well aware that its super long, it wasn't intended to be, but thats how it ended up and im not going to edit what I have to say, this needs to be purged...
I don't understand why so many people get butt hurt over any article here which are meant to HELP. - GROW UP. Quit being jerks, have a little appreciation and if someone may be flawed, say it like an ADULT and try to be a little POLITE you don't need to go out of your way to be an A$$HAT to prove a point.
This is SUPPOSED to be a place where we as creatives and photographers get together and have a certain level of respect for those who take time out of THEIR lives and away from THEIR families and THEIR craft to help us... They are doing US favors - and NO-ONE EVER says their articles are THE ONLY WAY TO DO SOMETHING. It's up to YOU to do what's best for and tweak it to make it work for YOU - and your shooting situation, and how it may apply to the type of gear you have.... Frankly i'm not wealthy, and I love a sharp image, but for me to make it what its all about, is a sharp image,i'm unfortunately not ABLE to go drop the dough for a Hassle... I need to make my Canon 50D, my Canon 7D, my Panasonic Lumix LX3 all work to their capabilities.... (and hopefully a full from 5D Mark III in maybe 4 years when I could afford it - and of course Leica, but thats another story...) :)
Sarcasm Alert: *Gawd, if only I could get through my own thick skull that if I thought so highly of MYSELF and of my perfect techniques as some of you folks here, then my photography would be perfect too and I wouldn't need this lousy, flawed place. So, if you ARE so perfect, why ARE you are and aren't at your own blog/site putting your own spin on things?
And props to those of you who are ladies and gentlemen who appreciate or offer their suggestions or ideas on these articles. I learn so much from peoples inputs, even the nasty ones, whether it be how NOT to conduct myself, or how to ENHANCE what was said in an article. I just don't want to read the attitudes. I am 100% FOR freedom of speech - its what our country is based upon - however, I don't have to LIKE the negative and holier than thou attitudes. Ridiculous. Please think again before you just spout your obnoxious retort. It will bring you better karma, and gawd knows we all need that....
Carry on peeps! Carry On!
(and yes, this has been building for about a year and a half and have bit my tongue and not said anything to those of you who know exactly who you are) - so sorry for the blow up if I personally offended someone else because I spoke my mind on the lack of courtesy and respect that happens around here so often. Reminds me of those nasty Adobe Forums - those are the WORST, I just don't want this place to get that way with the readers turning into dictator as opposed to the admins, and forum heads, and gurus, and 'celebrated experts' 99% of them have heads so big that they can't get see the forest through the trees - and these people at the Adobe Forums actually PRIDE themselves on being jerks especially the MAC guys.. its just sad... Just go over there and read any of those articles in any of the forums...
Another reason why I love those who write the articles here, no pretense, no airs, no ignorance, and totally here ONLY to help us with complete approachability.
thanks for reading - and again I APOLOGIZE if I personally have offended ANYONE - as that was not my intenssion. I thin all of you know this...
Please be nice to one another and respect others 'words' as really that is our only truth there is - but its up to us to seriously respect our passions, which here is all types of photography. It's what MY Life is all about and I wouldn't trade it for anything in the world.
mike September 10, 2011 01:12 am

@James: Yes, very good idea to make an article about plane vs sphere. I'm surprised how many misunderstandings there seem to be around that subject. I guess a pitch could be how well different lenses actually handle this. Another thing i think deserves attention in this context is how bad most lenses are in the edges, even expensive ones. I know some photographers step back and then crop to get around the problem (in that case it helps with higher resolution).
James Brandon September 10, 2011 12:06 am

Traian - Jasmine actually chimed in on my discussion about this article over on G+. She said that she only uses F-R when she has to. In other words, she uses the furthest focus point the edge as possible, and then if she still needs to move the subject higher in the frame she recomposes. I never said F-R will NEVER work, just that it is a flawed technique that under certain/most conditions (and especially for portrait shooters) it will render the image out of focus at worst or soft at best.
Sav - I'm not confusing myself, I'd argue that you're confused. I've had several people come at me with the same argument and once I go through everything with them one on one they realize the math is right, and I didn't move the camera incorrectly while recomposing. I wouldn't go through all the trouble of researching the math and the facts behind this just to go up and take a picture as proof without thoroughly thinking it through before pressing the shutter.
Think about is Sav. The math is correct, there really isn't a way to argue with that. Your saying I moved the camera wrong during my proof shots. If the difference between point A and B is .5 ft at the distance in the example, then camera movement is a mute point anyway. Hand holding the camera and tilting vs. tripod and tilting is not going to have anywhere near a .5 ft difference between techniques.
For everyone else still having issues with the focal plane. I hope you're all realizing that the focal plane is NOT spherical, it's flat. Looks like I've got another article to write :-)
Nelson September 9, 2011 06:57 am

Completely disagree. (at least in logic)
While your geometry is correct, the distance to the object you are focusing on has not changed if you recompose correctly. If you simply tilt the camera on a tripod, you have in effect changed the distance to the image sensor. Look at the geometry of how typical tripod mounts are setup, the hinge point is probably about 4-5 inches below the sensor. So tilting down brings the camera closer, effectively changing the distance. The out of focus "20" may be lensing aberration. Figuring out the true focal point of lenses, especially with compound lense, at low focal lengths gets messy. I think you've over simplified this.
Either way the end result is I agree with the focus-recompose issue. Its not effective. Its better to compose and manually focus or use some focus selection (face recognition or eye tracking) built into the camera.
Lewis September 8, 2011 04:16 am

I feel like this is only an issue when you are really close to our subject. The Picture of the dart board used to demonstrate the idea looks like it is just a few feet from the camera and the 20 is probably about a foot from the center. And with respect to the other example, a Nikon 50mm f/1.4 lens has a maximum angle of view of 46deg. Doing a little trigonometry that's 3.4ft from 4ft away so you couldn't focus on the eyes and then recompose 2ft lower.
That being said I find it easier to use the alternate AF points while while composing my images if I can. I usually shoot with a D70 though which only has 5 AF points so I often have to recompose to some extent.
TWhite September 8, 2011 02:40 am

Hey does focus and recompose work with a large family also or a wedding party. Would you just focus on the eye of the person in the middle ???
Sav September 8, 2011 01:43 am

You're confusing yourself and the readers with this flawed argument. But nice try with good intention. Agreed with Don, stas, and PJMixer, there's no mathematically inherited disadvantage but a movement issue. And your photo clearly showed that. Thanks.
Traian September 7, 2011 11:04 pm

I remember watching Jasmine Star photographing the wedding at CreativeLIVE in 2010. A lot of people following the live stream were wondering why she was bobbing her head all the time while taking photos, and her answer was obviously: "focus-recompose". I do not remember her complaining about missing the focus (though she was using Canon 5Ds in relatively low light), and her photographs (and success) are proof that this approach is not so error prone as the article would suggest.
Yes, there are situations where you need to think twice before focusing and then recomposing. Two other famous photographers raised the same issue as the article above during their CreativeLIVE seminars, and they tried to increase the awareness of the potential problem of focus-recompose. This does not make it a wrong approach, it just means that it has certain limitations.
Used carefully, focus-recompose is a method that gives good results, especially in situations where the scenario is fast changing and out of the photographer's control, as weddings tend to be.
Ardi September 7, 2011 11:24 am

thanks for sharing.. i will keep this in mind as best methode when using large apparture opening..
PJMixer September 7, 2011 02:41 am

I ran out of time reading all the good comments, but I wanted to say thanks for the valuable info. The math wasn't 100% accurate as Don first pointed out, but further explanations of being able to rotate the camera around the same plane was the kicker and in some situations where I think I have fallen victim to the focus-recompose problem, especially with wide apertures. Thanks all, I'll be trying to use my focus points a little more often when shooting < f/2.8.
Stas September 6, 2011 05:18 pm

I am afraid your explanation is flawed.
You got the math right - the distance to the face of standing person is greater than that to his chest, and when recomposing, the chest would be slightly out of focus.
BUT, it's the face we wanted in focus in the first place, and distance to it doesn't really change when we recompose (I ignore for this purpose the small movement of camera itself).
Now if we are shooting with very low depth of field (like good portrait lens wide open), the depth of field is very small, and even that small movement of the lens can move the focus plane a couple of cm. Not much, but often enough to throw eyes out of focus ...
Bill Duncan September 6, 2011 06:39 am

Amazing the amount of traffic this question has achieved. Focus recompose works, always has. Used it with the old film SLRs and still use it today. Rarely have time for fiddling with changing the point and leave it in the centre most of the time. The distance for the line drawn to the eyes haven't changed.
mac September 5, 2011 08:44 am

DON is absolutely right: you want A to be in focus, but you are reframing so that B is central - but the whole point is that you DON'T want B to be in focus; you want A to stay in focus. So 'focus and reframe' is really useful if you don't want to mess about with adjusting focus points all the time. I never use any other method.
Luc September 5, 2011 08:07 am

I don't get this. I agree with Don's comment as I don't see that line B has anything to do with focusing along line A. I think that any failures are due to either subject or camera moving slightly during the focus-recompose but not to the "geometry" that the author tries to make here. My 2 cents, of course the focus point selection method is not bad in any way, just might be slower.
Chance September 5, 2011 05:52 am

Hello,
I think that perhaps the biggest factor in the change of focus in the dartboard example is that you are at f/1.4 and the 20 is now on the edges. That wide, the edge sharpness of almost any lens will be much less than the center sharpness. It might be helpful to try both methods, and then compare. For example, the focus recompose, and then using a different focus point without recomposing. I have a feeling the two will be similar.
That said, I have always felt that the focus recompose method does have that flaw of changing the distance to the subject. Not because of the Pythagorean theorem, but because the pivot point of the camera makes the front lens element arc in a circle. So the distance changed is a function of the length from the pivot point of the camera (or rather, the shooter, since we often are not tilting the camera, but our heads, backs etc) to the front lens element.
(pivot point -> lens element)*(cos(degree of rotation) - cos(final degree of shot))
So if we assume the distance from the pivot point to lens element is 6 inches, and we tilt the camera 30*:
6* (cos(30) - 6cos(0)) == ~ 0.8 in. So we are now about 4/5 of an inch behind our subject. it doesn't make a huge difference at longer distances, but closer up, it can have a pretty big effect.
Bill September 5, 2011 12:33 am

The best message I take away from this discussion is to have and use a depth of field calculator. Without the understanding the DOF, any method of focus/compositions could have a high failure rate, especially with large apertures. It's cool to have that "fast" lens, but not cool if you images are not in focus. A depth of focus calculator can be a real eye opener and one of the cheapest pieces of gear in your bag.
MotionFreeze September 4, 2011 06:54 pm

For the first time this simple article has explained why i was not getting the sharpest images. Thank you so much James.
I wonder why canon or nikon has not taken out a camera that has tracking for the auto focus points. Just imagine..... you focus with the center point, and when you recompose the shot the focus point changes to keep the subject focused. That would be a great tool.
Lavanya Photography September 4, 2011 03:53 pm

Dear James,
I am sorry but I will disagree. Have a look at this :
http://www.facebook.com/photo.php?fbid=134617283269089&set=a.134616876602463.25380.121557097908441&type=1&theater
in this image i focused on the right eye of the subject (left eye for we readers) and then recomposed it.
I majorly use this focusing and recomposing method for candid wedding photography and till date i have achieved what I wanted. Even after refocusing obtain the sharpness at the part where I wanted the sharpness to be.
similarly have a look at this :
http://www.facebook.com/photo.php?fbid=134617346602416&set=a.134616876602463.25380.121557097908441&type=1&theater
I focused on the left eye of subject (right eye for we readers) and i got it.
Same thing is here.. http://www.facebook.com/photo.php?fbid=174239282640222&set=a.174225665974917.46155.121557097908441&type=1&theater
i focused with center AF on the guy in front and then i recomposed WITHOUT LEAVING THE SHUTTER BUTTON to include second guy.
Lavanya Photography September 4, 2011 03:52 pm

Dear James,
I am sorry but I will disagree. Have a look at this :
http://www.facebook.com/photo.php?fbid=134617283269089&set=a.134616876602463.25380.121557097908441&type=1&theater
in this image i focused on the right eye of the subject (left eye for we readers) and then recomposed it.
I majorly use this focusing and recomposing method for candid wedding photography and till date i have achieved what I wanted. Even after refocusing obtain the sharpness at the part where I wanted the sharpness to be.
similarly have a look at this :
http://www.facebook.com/photo.php?fbid=134617346602416&set=a.134616876602463.25380.121557097908441&type=1&theater
I focused on the left eye of subject (right eye for we readers) and i got it.
Same thing is here.. http://www.facebook.com/photo.php?fbid=174239282640222&set=a.174225665974917.46155.121557097908441&type=1&theater
i focused with center AF on the guy in front and then i recomposed WITHOUT LEAVING THE SHUTTER BUTTON to include
Mark September 4, 2011 07:13 am

Your math diagram works if you are head and shoulders shorter than your subject and you are shooting at close range, otherwise the diagram is not reliable. the further you are from your subject the DOF increases exponentially and inversley as your aperture changes. (larger aperture shorter DOF) I always try to stay away from shooting wide open apertures unless I am trying to achieve selective focus to blur objects in the fore or aft of my subject.
Bhushan September 4, 2011 04:11 am

Hello All,
Good point put forth..
I agree with Don. The DOF can be considered as a Sphere as don suggested.
But then James missed a point, that our sensor or the film in the camera is not sphere its still a plane.
When we focus our subject by center focus point and move the camera (however well you do to rotate it about the axis to keep the distance between the camera and the subject same) the image on the sensor moves out of focus. thus pushing the subject out of focus in the resultant image.
In other words Focus and recompose will work flawlessly if your sensor is concave (but then you will be stuck with a single focal length for the rest of your life).
so i conclude the article holds good. we need to use all possible focal points.
John Douglas September 4, 2011 03:47 am

Wow. Amazing how much ignorance and contention can be generated on a topic that, in the real world, seldom causes problems. E.g., assuming you are shooting with a 50mm lens on a full frame camera, moving the subject from the center point to one of the intersection points for a perfect "Rule of Thirds" composition would swing the camera 8 degrees. The cosine of 8 degrees is 0.990. If you want to see how much the plane of sharp focus moves behind the subject when you recomposed, just divide the distance to the subject by 0.990. For 10 feet, that works out to just over 1 inch. Of course, this assumes an absolutely flat plane of sharp of focus, and while lens makers strive (some more than others) to achieve a flat plane of sharp focus, there's always a little diversion from true flat, though on a decent lens it won't be much. In any event, neither the lens variations nor the small angle of swing will make any detectable difference except under extreme conditions.
If you truly must have "tack sharp" focus on a subject off center in extreme situations, but want to preserve your creativity by staying with "focus and recompose," then just oopt for a Hasselblad H4D:
http://www.hasselblad.com/media/2234814/when%20true%20focus%20makes%20a%20difference.pdf
The Hasselblad will compensate for the focus shift for you. Oh, BTW, that linked document has a better drawing illustrating the issues than James's drawing that kicked this off (no offense, James). Of course, if you go the Hasselblad route, you'll be out a lot of $$$$$$, but you will be "tack sharp." Assuming that's your goal in life, you should go for it.
mike September 4, 2011 02:06 am

One thing that i guess was missed in the images above, and perhaps missledd a lot of readers is this:
The crop of the left picture (named "Focus") shows the image in the dead center. The right picture (named "Recompose") shows the crop of the image at it's far edge. I would argue that the majority of the blur there is attributed to the fact that the lens has a terrible distortion in the edges at f/1.4.
To make justice to the theory, it would be much appreciated if James could post a series of images that actually used the the focus point in the edge to and compare that to the "focus-recompose" version of the shots. How can anyone judge the theory if only one technique is shown?
Happy shooting!
mike September 4, 2011 01:22 am

Another point of view that i think is worth to mention in this context:
If the used lens (Canon 50mm lens, f/1.4) in the example above, would have been stopped down to f/4, two things would happen:
1) The image would be much sharper and less distorted, as the lens in question gives terrible sharpness and a lot of distortion at f/1.4.
2) The DOF field would increase and to a big part void the issue this whole article is all about (@ 2 meter: f/1.4 => 8cm DOF, while f/4 => 24cm DOF). With a 24cm DOF, focus-recompose would be just fine in the example that the OP posted above!
This is exactly the reason i don't use f/1.4 and that shallow DOF. it simply won't bring the best out of my gear in terms of image quality, and it limits my freedom to frame the subject in a natural way.
If you are in doubt about the image quality i address here, please visit this site and check the sharpness and other data for this particular lens at f/1.4 vs f/4 (the same holds true for most lenses, at least those within my budget):
http://www.dpreview.com/lensreviews/canon_50_1p4_c16/page4.asp
I have stated that focus-recompose is fine for me, and maybe that is because the way i use my equipment. I try to get the best out of it while having margins and freedom to shoot the way i want. Going wide open aperture is not a good idea, unless you have very low light, or perhaps are forced to do so because you can't separate a subject that is too close to the background. For most lenses, stopping down a few stops brings the best out of your lens. Perhaps one time out of a hundred, i find myself reaching lower than f/3.0 for my prime 50mm (it's my favorite lens, I use it a lot!). In other words, the techniques proposed in the article are fine at a rate of about 1/100 for me.
Again, there is no wrong or right, just different ways of using our equipment depending on our artistic goals. This is my take on it, but that does not make me right or wrong. I think EVH put that down well: 'fuck the rules.' If it sounds good, it is good." -Eddie Van Halen, Guitar World, January 1981
Happy shooting!
Doug McKay September 4, 2011 01:20 am

Do I get a prize for being the one millionth comment? :) This is a touchy subject even in the days of film and manual cameras. In the early days of point and shoot and in the first digital cameras the focus and compose method was actually in the manuals. It.works between 10 and 30 feet on most camera set ups.
If you keep the subject to focus only in modern cameras you could be missing out on a load of other information the camera uses for other calculations for exposure and color. That is not to say you could be mostly manual and you may have learned to use the effect of slightly exposure results in a pleasing way.
The way I have found that works best with people I instruct in use of their cameras is to imagine that you focus length is very much like a broom handle for older folks and a light saber for younger people. It is an adjustable fixed distance which you set either by settings on your camera or by your fingers moving the lens. What ever is at the very end of that distance is in focus, how much more is dependent on the aperture opening and the amount of light.
Jumbo John September 3, 2011 08:58 pm

Interesting use of trigonometry, but your basic premise is wrong, I'm afraid.
This is quite simple. Once you have focussed on the eye in a portrait and then re-compose without releasing your 'trigger finger', the distance between the lens and the eye doesn't change. Hence it stays in focus.
I don't know what you did in your 'proof' photos, but you can't have done this.
QED
Jumbo John
Bristol, England
L. C. Hingada September 3, 2011 03:34 pm

When you focus on the eyes,you want them to be in focus
and you do not care about the other parts of the body after you recompose!
jody burnem September 3, 2011 02:29 pm

To combat the out off focus problem with the focus recompose method I just always use the motion tracking feature. With it turned on ,use the center focus point, then focus and recompose. The camera will lockon to your focus point and when you recompose it works just like your subject moved and not you. It works realy great this way and the only times i have trouble with it is in very low light.
cris nicole September 3, 2011 10:13 am

i'm so glad this was posted! i suddenly had some issues with focusing after years of shooting with no problem. the only difference was that i started using autofocus for the first time (and i of course figured that should mean my focus would always be on point). i did all kinds of research and thought maybe it was just my camera--D200. i finally realized that my issue was simply that i was using the focus-recompose method. thanks for this article. i hope it saves others the agony i went through trying to figure it out (although i learned a ton doing the research).
Cliff Lewis September 3, 2011 05:19 am

James is right. The camera has a plane of focus, not a sphere of focus. It is easy to test this yourself--I just did. I took three pictures of fine text, with my camera mounted on my tripod. I set my zoom lens at 50mm and opened the aperture as far as possible (f4.5). The focal plane of the camera was about 47 inches from the subject. For the first picture I focused on a point straight in front of the camera. The text at the top of the picture was as much in focus as the text in the center. For the second picture I raised the camera on the tripod to focus on the text at the top, then lowered the camera in the same plane so that the center was in the same position as before. Again, both top and center of the picture were in focus. For the third picture I rotated the camera about the tripod mount to focus on the top part of the text. Then I rotated it back to the central position. Both top and center text were much more out of focus. The effect is subtle at f4.5, but it would be much greater at a larger aperture. So if you are going to recompose you need to move the camera straight without rotating it.
Nick September 3, 2011 05:06 am

Have to say that James's (post at 2:49) is correct.
There is clearly confusion around focal plane and focal distance and for the purpose of this debate we should ignore focal plane and also exposure, which can be set at spot / centre weighted / evaluative etc and performed manually.
All lenses, good and bad, have distance marked on the barrel. If you achieve sharp focus at 3 metres, then anything that distance from the camera will be in focus. If the exact focus point acted like a laser cutter on glass and you moved the camera all around you (keeping the sensor in the same position, you would create a virtual ball around yourself. A rather dramatic explanation I know, but this does endorse why people are mentioning 'arcs'.
Turning to the practical use of focus and recompose, you are photographing a model on a beach and you want her on the third or off centre at least. There is only 15 minutes to get the shots in different locations before the sun sets and she disappears forever. You are shooting at f2.8 because you want the sea out of focus and her sharp. By locking focus on her and then rotating the camera, say 30 degrees, she will remain in focus. This will save:
- wasting valuable time setting the focus point as you change location.
- cropping the image and losing quality.
The key point is that you maintain the same distance from the camera to your model. Referring to my post yesterday and the original image, assume you are snapping a person 10 metres tall and you are standing 1 metre from the subject with eye level on the belly button. You will focus on the eyes approx 5 metres away, so when you focused and recomposed on the tummy 1 metre away, the focus would be way off. Clearly, the eyes are far further away from the tummy and F & R will not work.
However, as I said yesterday, the method you use has to depend on the situation and location and really there is nothing wrong with either approach, providing you know whet you're doing and / or have the time. The aperture is critical if there is any margin for error. Using a 1.4 lens will create sharp and out of focus areas on a face if shot close enough, even on a tripod, whether you focus and recompose or not. These lenses are usually used for specialist effects, i.e. the Canon fast 50 and 85mm lenses, which are not considerably better than their lower priced relatives in terms of image quality.
This has been a really interesting debate - thanks.
Ed Hamlin September 3, 2011 04:30 am

James couldn't agree with you more on the thought process for focus and recompose. I only use it when a focus point is not critical, very rare though. I do have some thoughts though. First off your example is based on a very shallow DOF (depth of field) hooting at large apertures 1.4 - 4.0 is pretty shallow and focus from eye to eye can be off using the method, Once you reach 5.6 and above the leeway you have starts to increase and keeping things in focus is a bit easier if you know the DOF.
Here's one last thought, I often shoot in difficult focusing circumstances, http://believefotografie.com/believe/fire/ Smoke, low light, deep shadows can make it difficult to use a aperture that gives a deep DOF. In the Fire series some initial shots I used focus - recompose. The majority of those images a very soft/out of focus. Thus, when I can get a good focal point I never recompose I always use a point that is the true story in the image.
The real issue is what is you DOF and will the primary point of the image be in focus, If not, then rethinking the use of focus - recompose is a must.
Jason Malwitz September 3, 2011 03:47 am

Why is it that every time someone decides to bash the focus/recompose method its always an extreme example like this one offered? For portraits in good light, shooting wide open or nearly wide open at very close distances, absolutely it is better to toggle focus points. Won't argue the math at all. But try using one of those outer points in a hazy backlit situation or a dark church or anything less than ideal lighting situations and you will have many more focusing issues than one would have with focus/recompose. What is really required here is a good solid understanding of dof with various focal lengths and camera to focus point distance.
Jack September 3, 2011 03:34 am

hell so many talk..... this stuff is decades old.. as old as the AF.
Rick September 3, 2011 03:28 am

Your right if you such big apertures but if you use f8 or more theres is not a problem
CanonMaiden September 3, 2011 02:58 am

You can sit and do math all day long, but it's true. Especially in portraiture, focus/recompose often fails. Period.
mike September 3, 2011 02:41 am

Sorry, should read:
DOF is relative to the distance, so if DOF is ~0.5cm at 0.5m it will be ~5cm at 5m. Considering that, the distance to the subjects makes little sense for the test. I moved in close just to get my 2 x A4 papers to fill the frame.
mike September 3, 2011 02:31 am

@stewart zollinger:
In this case the distance was about 50 cm which gives a DOF (50mm @ f/1.4) of about 0.5cm, small enough to reveal out of focus in the edges, if there should have been so.
DOF is relative to the distance, so if DOF is ~0.5cm at 0.5m it will be ~50cm at 5m. Considering that, the distance to the subjects makes little sense for the test. I moved in close just to get my 2 x A4 papers to fill the frame.
Helen September 3, 2011 01:43 am

I absolutely agree you cannot use this focus/recompose method with moving objects, namely horses/showjumping.
Simplest way here is to lock your focus on a static object, then press your focus lock button. Wait for the subject to come into frame and it will be in focus when you're there shooting off 6-10 fps in burst mode to get the 'winning' shot.
However, for portraiture, absolutely spot focusing, or even for still life/product photography.
And for the nifty fifty users (me included), it's terrible at focusing wide open.
You have to learn to shoot at f8 upwards for a tack sharp portrait otherwise you get all sorts of distortion.
Stewart Zollinger September 3, 2011 01:29 am

@mike
What size was the paper and how far away were you focusing? If it was the minimum focus distance, with a 50mm lens at f/1.4, with an A4 sheet (landscape), you would only see very minimal blurring towards the edges, regardless of where the focal plane is flat or curved.
A focal plane is curved, lenses try to minimise that curve, as you said though, no lens is perfect.
I would suggest that depending on the parameters - subject distance, recompose distance, your camera, your lens, whether or not your lens/camera is front or back focusing (or even dead on), focal length, aperture - you may find that using the centre focus point and recomposing could make your composition's focal point more or less sharp.
Mike September 3, 2011 12:39 am

@stewart zollinger
Yes, that is exactly my point, mind the paper size.
I just tried with my 50mm f/1.4 and there is no doubt what so ever that there is a focal plane and not a focal sphere. However, no lens is perfect (unless you have a NASA budget), so the image will be more blurry at the edges.
I think that pretty much wraps up the theory part with a very practical example as shown above: When you rotate the camera the focus _plane_ is shifted and brings what you focused on slightly out of focus. For those who still don't believe it, just do the test above and you will see that this is correct.
But then again, i really don't care much about that when i'm shooting; center-focus and re-compose is what i do, and i've seen no reason as of yet to change that.
Cheers!
AJ September 3, 2011 12:26 am

You realize your graph is wrong don't you? You changed the point of body you focused on!!! Well unless the midbelly is your focal point. I assume you want to focus on the front eye in both methods? Then what is changing is the focal point inside your camera NOT the location on the person as you show in your example. Your A & B points are inside the camera with the C point the front edge of the eye unchanging. Your entire argument is worthless here Try again doing it properly and you will find micrometers of difference (the length of the difference between the focus points on your sensor) not half a foot.
Margie September 3, 2011 12:23 am

You are 100% correct! Focus recompose sucks. It rarely gets you a TACK SHARP image. Great article!! Excellent stick figures too!
rebecca September 3, 2011 12:06 am

Oops, bad link...http://www.digital-photography-school.com/focusing-quick-tip-single-point-af-methods?utm_source=feedburner&utm_medium=email&utm_campaign=Feed%3A+DigitalPhotographySchool+%28Digital+Photography+School%29
rebecca September 3, 2011 12:05 am

Well, that explains my issues with focus being off. I just read this and I am using the method linked below now. I'll see if it gives me crisper photos.
http://mail.verizon.com/webmail/driver?nimlet=showmessages&view=openmail&id=8234#mail_Emails_INBOX
Alex September 2, 2011 11:49 pm

@Mike: exactly
The actual discussion is about focal plane vs. focal sphere.
Lenses are constructed to return a focused **plane**.
And then one needs to understand a bit of geometry.
Thanks for bringing it down to the point!!
Larry September 2, 2011 11:48 pm

The two photo examples only prove that whatever you focus on will be sharp in your photo. You should show two examples using focus/recompose and using a secondary focus point in camera.
I have been using focus/recompose for years and eyes/faces are sharp because I focus there, even with my 85 f1.2 (which I only use at that setting at night). To say "your subject is not in focus after you focus/recompose" is bogus. My most important part of my subject is the eyes and they are tack sharp.
OddShooter September 2, 2011 11:31 pm

Canon's 50mm at 1.4 is notoriously blurry around the edges, so you should have also taken a picture with another focus point. That way we could see if the blurriness is because of a changed focus or because the lens is crappy around the edges.
Stewart Zollinger September 2, 2011 11:21 pm

@mike
To magnify this as much as possible, let's take an extreme example (feel free to correct me if I'm wrong with my calculations...)
- Canon 5DMarkII 16-55mm at 16mm has a minimum focal distance of 28cm
- Canon 5DMarkII at 16mm has an angle of view of 96degrees
The piece of paper would have to be 61cm wide on the table, right?
Are you trying to say that that the text on the piece of paper would be sharp across the entire width, even at it's maximum aperture?
kelvin September 2, 2011 11:15 pm

Digging further on this site shows another interesting post to read - Hyperfocal Distance
Cécile September 2, 2011 11:08 pm

Moving the focus point.... is it a Canon thing? I need to check but I think I cannot do it on my Pentax K20. if someone has advice , they are welcome. In argentic I did not have any alternative (maybe i am wrong but at least not some i have heard of). Thanks for the article and it seems to be a hot topic on other forums too!
mauro September 2, 2011 10:46 pm

Outstanding! You've just explained what we found out in our pictures!
Rich September 2, 2011 10:40 pm

Focus-schmocus. This will all be a moot point soon. There is a camera in development that lets you adjust your focus after you shoot.
http://articles.cnn.com/2011-06-22/tech/auto.focus.camera_1_auto-focus-camera-lenses-images?_s=PM:TECH
Mike September 2, 2011 10:09 pm

To those who still think that the focal plane is a sphere and not a plane, try this:
1) Mount your camera on a tripod and aim for a paper with text on a table
2) Set minimum aperture and go as close as you can (minimizing DOF)
3) Use the center point to focus
4) Shoot and enlarge on your PC
Now, if the the focal plane indeed was spherical, only the center point were you focus would be in focus, the rest out of focus - but as you can see, the entire paper is in focus (mind some lens distortion, chromatic aberration, vignetting etc).
I hope this will remove any doubt about the spherical vs plane question.
Though i never use anything else than the center point and re-compose (as i don't see any real-life benefits with the alternative), i have to agree that the OP is correct in theory.
Stewart Zollinger September 2, 2011 09:31 pm

The article raises a valid question - one I had event thought about myself - but it's very, very flawed and comes to the wrong conclusion.
Have to agree with Don on this. If you pivot around the centrepoint of the sensor, the camera to subject distance (subject being your original focus point) remains EXACTLY the same, therefore your focus distance remains EXACTLY the same.
The "resulting images" shown in the article are to do with edge sharpness versus centre sharpness, nothing else. To do an accurate representation of your two shots, the framing of the full shot needs to be identical.
James, try answering the following two questions -
- Point the camera at your subject's eye: What is the distance between the sensor and the subject's EYE?
- Point the camera at the subject's chest: What is the distance between the sensor and the subject's EYE?
If these distances are the same (which they are) then the focus will remain the same.
In the real world, we don't pivot around the sensor, but even taking that into consideration, you would have to have a very short subject distance, a very wide lens and a very large aperture.
BlueFightingCat September 2, 2011 09:19 pm

Hi everyone,
Interesting conversation. Here is an article I found that should answer all your questions. Essentially the focus recompose method does cause some error. However its quite negligible. It only becomes an issue in rare cases. This this article:
http://www.bobatkins.com/photography/technical/focus_recompose.html
Hope this helps.
BFC
Chris September 2, 2011 08:44 pm

Interesting, but fundamentally flawed. The simple truth is that if you focus on the eyes and hold that focus you can recomposed to your hearts content - you will still be focused on the eyes and surely that's what you want. Using Pythagorus changes nothing. If you focus at 4.06 feet (or whatever) that's where it will stay and recomposing will not alter that dimension unless you move the camera other than on it's axis.
I note your 'proof' with the dartboard and suggest your camera needs a sharp blow with something heavy (a hammer perhaps) because it's clearly not holding focus as it should. I tried a similar experiment at a more critical 2 foot and maintained the original focus regardless of what I did to recompose. Obviously my AF is working properly.
vamapaull September 2, 2011 08:24 pm

I completely agree with Krzysiek
elmarie September 2, 2011 07:26 pm

Thanks you so much for the article. I used this method quote a lot, and quite often I cannot understand why my photo is not in focus. This explains it and actually it is so logically if you think about it.
Adriana Morett September 2, 2011 07:20 pm

Hi Darren, I've been a big fan and follower of this website for years. Just don't use to comment, but this topic really brought my opinion and here it is: I've been using the focus-recompose method for quite some time now, and I have actually seen many of my photos out of focus (I like using very open apertures) and I want them tack sharp. So you know what? I AM GOING TO TRY THIS. It's simply a matter of choice, but it's always better if it's an EDUCATED choice, so THANK YOU very much for this tip. In a few weeks I'll come back here and comment on my results.
Whatever the result of my "experiment", I will have learned something new. Thank you for that.
Ike September 2, 2011 07:01 pm

I agree with the author - focus & recompose isn't accurate enough for pin sharp photos. Once you get the hang of moving the focus point it becomes 2nd nature. Worst case if the action is really fast you're at least a bit closer and can crop to regain composure if needed. I'm reasonably new to photography but it's working so far.
Roberto September 2, 2011 07:01 pm

@Don (and those who agree with him):
.
Sorry, but you are wrong. Think about it. Of course you are not re-focusing at B. You've focused at A. Now, the FOCAL DISTANCE from the camera sensor to whatever the camera is pointing at is A. And A is longer than B. So, when you've recomposed, the camera is actually focused behind the subject.
MikeStrom September 2, 2011 06:46 pm

A thing that came "out of focus" here is the fact that most lenses produce better images when stopped down a bit. Going wide open is not a good idea for most lenses. So, perhaps this discussion belongs in the "low light photography" department? =)
Matt September 2, 2011 06:38 pm

If your DOF is an issue with focus & recompose, then you need to learn how to close down the aperture and produce better photos.
Nobody except Flickr hipsters viewing 640px wide photos loves the "one eye in focus, the other not, and the nose is a huge giant ball of blur". Those photos look like crap and PRINT like crap.
Stop dofsturbating, people.
In the "50mm on a FF at f/1.2, four feet away" example from this article, the aperture should have been closed down to at least f/3.5.
Rick V September 2, 2011 06:30 pm

Javier, Do the camera optics really preserve focus along the plane parallel to the sensor plane or along the sphere at the focus distance? This seems to be the source of confusion in the posts.
Sean September 2, 2011 06:29 pm

Cool story, bro! Actually, with anything other than my 50 1.4, I've ne'er had this as a problem. And I use this method all the time. If we only had me old Spotmatic focus aid screen in my 7D... Manual focus of 1.4 would be easier... But, usually, my OOF shots are because they weren't in focus right before the recomposing...
MikeStrom September 2, 2011 06:24 pm

By the way, for those who really want to dig into the subject, there is an exhaustive explanation here;
http://en.wikipedia.org/wiki/Depth_of_field
Phil September 2, 2011 06:24 pm

Javier has explained it right. The people that still don't understand it are thinking that the focus is a sphere when it is actually a plane. I can't explain it any better than it has already been explained, people just need to sit and think about it, draw it out on a piece of paper or take some photo's as an experiment.
The only way the Focus and Recompose would work, accurately and whilst wide open would be if someone focused on the eyes with the camera in the horizontal plane, then lowered the camera still in the same plane on a tripod to recompose at chest height. No rotation of the camera, just up and down. But that's a pain in the @$$ and it's easier to just focus using focus points.
MikeStrom September 2, 2011 06:21 pm

This really sparked a controversy, thanks James for providing the spark =)
Theories aside, i think it's really a matter of what and how you shoot, and also a matter of habits. Is there really a right or wrong method of shooting anything?
Personally, i'm not much for close ups-and typically stay no closer than 1.5 meters from my subjects with my prime 50mm. That gives about 6 cm of DOF @ f/1.8 which allows for plenty of recomposition w/o focus issues.
I would argue there is no wrong or right, there are only circumstances, artistic goals, habits and what we feel comfortable with.
Funny that this turned out to be an argue over what is wrong or right - there is no such thing as far as i can see.
/MIke
Tony September 2, 2011 05:42 pm

You have an eye for stick figures!
I like, period :)
Gareth September 2, 2011 05:19 pm

@javier - nice diagram, makes things clear and helps explain the concept of a focal plane over a sphere nicely. Perhaps James could include it in the original article.
@abhy - you sound like a nice guy. Keep smiling.
Krzysiek September 2, 2011 04:43 pm

Javier, your diagram helped me to understand why James's theory is right, and why the focus plan is more flat than sphere. In my opinion that is the key to understanging that...
Abhy September 2, 2011 04:30 pm

Absolute great post.. I just amazes me how dumb some ppl who are here opposing your view.. either they are not sure how focussing mechanism happens on a camera or using not a wide lens so that they dont notice this.. no comments for these ppl..
Dave September 2, 2011 04:15 pm

Think the whole idea behind the article suggests that with wide apertures (my guess is less than F/2), moving the camera just a smidge or if the subject moves just a smidge, the focus plane may have changed. If that is the case then you original focus point has moved. At F1.4, your area of focus is so so small that the person breathing or the photographers breathing may move the camera just enough so that the things are out of focus. I know, with focus-recompose it doesn't seem like you move the camera a lot but it moves, thus causes out of focus images. Look at the dart board, proof is in the pudding. Now at F2.8, that little movement may not do much to the focus point.
Paul Fulwood September 2, 2011 03:55 pm

Hi James
RE: Your argument combined with SPOT metering
Could I throw spot metering into the mix and ask how you can effectively centre spot meter and use a focus point other than the centre point? Essentially you have to use the centre point unless you meter and then recompose and then move the focus point and then take the shot.... dont you? Especially if its "action" photography which as you say most photography is...
Paul
Suhas Shelke September 2, 2011 03:14 pm

Its a fantastic article to learn the deceit of focus recompose :) ...I however do recompose shots for locking exposure ..hope its fine :)
RichB September 2, 2011 02:37 pm

How many photographers would be shooting "up" at that angle to the face of their subjects? I would think most would be shooting close to parallel to the horizon. The ideal position is to be halfway (in height) between the eyes and your other chosen focal point on the body, since the two radials above and below the camera would be the same length and therefore in focus.
Tom September 2, 2011 02:15 pm

Oh dear - the yobs have arrived. Why use personal abuse when all we want is a reasoned discussion. We can be right or wrong but who cares so long as we improve our photography.
I've learnt from this discussion that I don't actually know if the plane of focus is curved or flat. But I'll know by the end of the weekend when I do an experiment with a curved sheet v the same sheet kept flat. Results to come.
In the meantime Garrett, read my note above on how the camera actually rocks forward when rotated downwards and shortens the distance. And bfeldman, run away and start using your Hasselblad out in the field - but only after you've learnt to spell words like you're and Canon.
Brad September 2, 2011 01:49 pm

I have used both approaches. Recompose works well for me when the lens is closed down and moving the focus point works well for me when the lens is wide open. On average, I adjust the focus point more than recompose as I was getting more soft images with recompose than I would like. I know photogs who have much higher success than I with recompose. I feel each photog should find method(s) that work for them as I don't believe there a universal right/wrong way.
What I continue to struggle with is good in camera composition when adjusting the focus point. (just bugs me to crop afterward) I have good success with a closed lens, adjusting the focus point and doing a little recompose. However, when the lens is open I start to get more soft focus images. The latest thing I'm trying is using the back focus button and it has helped me reduce focusing errors.
BFeldman September 2, 2011 01:35 pm

Ok, so the only thing you have left to explain is why your using cannon equipment...
boss_007 September 2, 2011 01:34 pm

but if we keep the camera straight while focussing and then recompose all the time keeping the cam straight then focus and recompose works
edu September 2, 2011 01:21 pm

Focus - recompose works sometimes.
I don't know if anyone has already mentioned this but wouldn't this affect the metering? If one uses center AF then focuses, say, on the eye of the subject/model, I think one would also be metering from that spot right? Then if you move 2feet down to the waist for composition, wouldn't the metering be inaccurate also?
garrett September 2, 2011 12:42 pm

That was an awful explanation and I completely disagree with your flawed logic. If you focus on the eyes, unless you move your camera forward or backward the eyes will still be in focus no matter how you recompose.
Also, you should read about the math you are using so that you don't sound so uneducated.
Edna September 2, 2011 11:27 am

So that's precisely why i can't get that perfectly focused portrait using rule of thirds - i've abused the focus-recompose method. I promise to change my ways next time.. thanks!
Bob September 2, 2011 10:10 am

Sorry guys, but I have to agree with Don Says: September 2nd, 2011 at 12:37 am, if you don't re-focus when you shift from point A to point B then your original point of focus will still be in focus and the rest of the image will remain the same as well.
I personally use centre focus point, in combination with the back button focus custom function on my Canon 5D Mark II, which removes the chance of accidentally refocusing when trying to re-compose, while holding the shutter release half way down.
jonathan September 2, 2011 09:48 am

good article, that is all and carry on.
Sam kanter September 2, 2011 09:44 am

I have center point focus and recompose for 10 years. I have had zero focus issues. I've heard all this before, and must say I find it irrelevant in the real world.
javier September 2, 2011 08:47 am

Don, tom, and others. I am sorry if I sound harsh, but you are still not getting it. It is not the distance from the eyes to the camera but from the focus plane to the sensor plane. And the focal plane is NOT curved, that would make no sense because the sensor is flat. Here I tried to make a graphic to show what is going on:
http://imageshack.us/photo/my-images/607/focusi.png/
When one focus to the eyes by pointing upwards, the sensor plane is the red line sensor A and the focal plane is Focus A, as you rotate the camera to recompose, pointing the center towards point B, the focus distance keeps being the same as before, the new sensor plane is blue line SensorB and the new focal plane is the blue line FocusB. It is not the distance between the camera and the eyes that matters, is the distance between the planes, and once you move from the red plane to the blue one, the Eyes are no longer on the focal plane. If you want to focus and recompose YOU SHOULD NOT ROTATE THE CAMERA but move it from side to side or up up down without changing the direction to which it was pointing when you focused.
Tom September 2, 2011 08:34 am

Just measured my tripod/camera setup. The centre of the sensor is 4 inches above the centre of the tripod ball. So pointing the camera at an object 4 feet away, then lowering it to look at the replacement object 2 feet lower is achieved by rotating my camera forward and downwards, and it moves forward by a distance of exactly 2 inches (check by using similar triangles!). So there goes our focus if it's only 0.16 of a foot (= 1.92 inches).
So gotta agree with James, but for different reasons. Final comment: Peace to all mankind (just in case I don't post again before Christmas).
Andrew September 2, 2011 08:28 am

This is a good article, but I think you've missed a really important point. Your metering system takes into account which focus point you select. So if you focus re-compose your exposed will be off.
Josh September 2, 2011 08:21 am

Gak! Didn't have time to read all the comments, but this is huge for me. Based on your images, my habitual use of focus and re-frame may be the reason i struggle with nailing the focus on my portraits! I'll be back to this thread to read what others have said, but I will also definitely be playing with my focus point selection in a few shots to see if that helps with this super annoying problem!
Also, circled you. Love G+!
Catani September 2, 2011 08:15 am

It's ok if you need to stick with such a wide aperture as 1.4, wich is frequent in action photography. But setting down just one or one-and-a-half stop maybe enough to cope with the problem, I think.
Mike September 2, 2011 08:08 am

Interesting discussion. I think the most compelling argument for using focal point selection rather than "focus and recompose" is the imperfection of lens optics. Most generally used lenses are neither perfectly curved or perfectly flat - but the question is always, for a particular lens, how curved vs. how flat? So the easiest way to handle the unknown (lens curvedness vs. flatness) is to use direct focal point selection, particularly when working with large apertures and close subjects.
Having said that, however, the right technique is the one that works most reliably for the individual photographer, their chosen subjects and shooting style.
Tom September 2, 2011 08:02 am

I agree with Don. The distance from the camera to the eyes is unchanged after recomposing. So the eyes should still be in focus. The focal plane is actually a curved surface - defined by being at a fixed distance from the camera sensor. The reason that the 20 on the dart board was out of focus is most likely that the camera was rocked forward as you recomposed onto the centre.
The major disadvantage of focus/recompose is that the exposure will be set on the face, which is likely to be different from the general background with different peoples' skin tones ranging from very light to very dark. Then your shot will be over or underexposed but the eyes should still be in focus assuming a reasonable depth of field.
Kind regards
Tom
Ronald Thain September 2, 2011 07:50 am

I'm surprised no one has mentioned Hasselblad's innovation, True Focus and Absolute Position Lock.
As their cameras have only a centre focusing sensor, they have included a motion sensor in their latest cameras to compensate for the error when focusing and recomposing. The article, with clear diagrams, in the link below, explains the problem and should clarify the situation for everyone.
http://press.hasselblad.com/media-library/background-articles/true-focus.aspx
John Appleyard September 2, 2011 07:43 am

Overly simplistic. I use the single center point focus and recompose for most stuff and multi-point continuous focus for fast moving action [like cycling and boxing], basically because I don't have the time to pick the point. The point you pick is the one that will give you the parts of the images that are in focus and control the DOF. If you really understand DOF you can do this.
Terje Bergesen September 2, 2011 07:28 am

The author is correct in his conclusion, but completely and utterly wrong in his explanation. As has been pointed out (within a fairly negligeable amount) the distance from the camera to the eyes of the subject remains the same when you tilt. The adjustment of the camera doesn't change this enough for it to make a difference. The drawing also shows this exactly, when the camera is tilted, the distance to the subject center changes, but the distance to the eyes do not.
HOWEVER, the author is correct that what was in focus will (if the DOF is shallow enough) get noticeably out of focus. This is NOT because the distance to the focus point has changed, it hasn't, but because (as someone else points out) the plane of focus changes.
Forget about the drawing of the line from the camera to the body, imagine a line 90 degrees on the line from the camera to the eyes. Imagine that as two pieces of wood, exactly 90 degrees. Now move the red line down while keeping the line 90 degrees, you will end up with a line that is no longer focused on the eyes, not because the distance to the subject changed, it didn't in any meaningful way, but because the plane of focus changed.
Gareth September 2, 2011 07:27 am

What a great discussion. Thanks to those who explained the plane vs sphere stuff. I can now see the point the article is making, even if it only applies at the extremes of certain parameters.
Anthony September 2, 2011 07:21 am

Unless I am shooting a model who is completely still, I am shooting a moving subject. So most of the time, I am using the continuous servo auto focus on my D300s. I have just starting using the separate focus points and it is getting easier. Question: Can I shoot in continuous servo auto focus and still shoot and recompose? If so, how?
Scottc September 2, 2011 06:31 am

Wow, quite some response to this article.
DoF is a factor, why would a photog not consider it? Seems like common sense to me, represented by some complicated math here. If it doesn't work the first time.......
There is a place for all the focus methods, hint: that's why they've been on DSLRs for such a long time. The most important part of the article was the ref to the iphone DoF app, now that I will be taking a look at.
Focused and recomposed, at f/11: http://www.flickr.com/photos/lendog64/5828675939/
Cheers
Carson September 2, 2011 06:30 am

This is overwhelming! With so many opinions and arguments for and against this issue it's hard to know what to do. I guess the best thing to do is practice for ourselves and see what the results are! :)
Tronam September 2, 2011 06:12 am

Thanks John B for your thorough real world test. It helps balance out the largely hyperbolic conclusions of this article.
Trapper September 2, 2011 05:57 am

The second sample is not only out of focus it is also darker. Why?
Wayne September 2, 2011 05:43 am

Forget about my previous post. Conventional wisdom would tell me the focus lies in an arc from the camera where as previously stated it is a focal plane, inferring that the focus is parallel to the camera. Certainly makes a case for mutliple focus points.
dok September 2, 2011 05:40 am

Thanks John, I was about to write kinda the same explanation. With small angles, we can usually take the approximation : cos(x) = x.
Nonetheless, very good idea to have opened this subject, thank you James.
PHugger September 2, 2011 05:31 am

What James should have done was show an image that was focused on the dartboard 20 using one of the upper focusing points (no re-composition). This would eliminate any lens effects. The plane of sharp focus will vary across the plane unless your lens is perfect. Your point should be - make sure that everything you want in focus in in the same plane and parallel to the focal plane of the camera. If you change the plane of sharp focus too drastically when focus/composing you would expect the deviations you are seeing.
Wayne September 2, 2011 05:29 am

I still can't follow this - You're saying that if I was taking say, a half body portrait shot, I should use an off center focus point to focus on the eyes. Wouldn't this still result in the body being slightly out of focus? After all, the focus plane will still follow an arc down from the eyes with the plane of focus being slightly behind the body. This would be the same result I would get when using the focus and recompose method.
John Belmont September 2, 2011 05:11 am

I've been using center-focus and recompose for years, and if it has damaged my work, I have not noticed. However, in view of all the comments, I decided to do a test in the most practical way I could think of. The camera is a 50D with the EFS 17-55mm set at 55mm. Tripod, IS was ON, and with a 2-sec delay. Center-focused on the glass bulb of my household thermometer at a distance of 1 meter from the sensor. All functions manual, except for AF. ISO 400 f2.8 @ 1/30". Checked the image on the camera at highest magnification. Lovely, lovely sharp image of the bulb and the numbers and hash marks above it. I wanted to lick the blub but decided instead to go on with the experiment.. Now here's the crucial test. Switched the lens to MF and without changing the focus took 8 photos representing recomposition of the bulb to the center of each of the eight "thirds" going from upper left to lower right. With this model of center-focus and recompose I was disappoint to see that the image was noticably degraded in all 8 peripheral positions. Changed to f3.5 @ 1/50 and repeated the experiment. All 8 thirds provided just about equally but infinitesimally degraded images compared with the original center-focused center-shot image. Finished the experiment at f4.5 @ 1/30. All eight thirds were in very nice focus and not noticably different from the center-focused, center-shot image. I conclude that at practical settings for hand-held work (which is what center-focus and compose is all about), the approach is perfectly satisfactory. OH, and a side benefit of having made these tests: It looks as though that beautiful EFS 17-55 produces comparable images all over the periphery. Nice piece of glass !
Nick September 2, 2011 05:10 am

I totally agree with the theory explained here, but there is a question of practicality / speed to be considered.
I started using the focus and re-shoot method many years ago because I was fed up with cameras choosing a point away from the main subject. Yes, you can select a point, but when you're shooting wedding groups, the guests have to be captured quickly and there just isn't the time to start messing around selecting the appropriate point.
Referring to the diagram, as Krzysiek has mentioned, the focus will still be on the head, wherever you move the lens. If you shoot at 1.4 - 1.8 the tummy (the nearest point) will be out of focus, whether you re-compose or not. The same would apply with the camera held at eye level, where the distance to the feet would be far greater and they would be out of focus.
Therefore, my view is that the method used should suit the user and the situation. If you want all of your subject to be in focus at short distance, then shoot at 5.6 - 11. For specialist low depth of field shots, the key is to maintain relative distance.
gazza September 2, 2011 04:48 am

you have explained it bang on and for all of the doubters out there let them carry on taking non sharp images, i change my foucus points to get the best image and when you do a 100% crop you will see the differance.
gazza
Cliff Lewis September 2, 2011 04:46 am

I agree with what you are saying. The trouble with multiple focus points is that the camera can choose the wrong one, particularly if something is closer to the camera than the subject is. But that happens mostly when the subject is not particularly close to the camera, where recomposing does not make as much of a difference. So I would say that focus-recompose is best when the subject is more than a few feet away and other things are closer. Multiple focus points are best for a close subject where recomposing can affect the focus.
My film camera, a Canon Elan IIe, had an eye-tracking feature. It had three focus points, and I could choose my preferred point just by looking at it. The sales person when I bought my DSLR tried to convince me that most people who owned that model never understood how to use that feature. But I do miss it.
Roman September 2, 2011 04:35 am

I can't fully agree with your article.
Your maths are ok, but i want the eyes sharp after recompose not the center! The head, in your example, is more fare away than the center to the sensor.
Your example pictures show only the limiting resolution of the lens wide open and outside the center.
Other risks by choosing not the center AF are problematic resolutions outside the center of some lenses. If the lens is to soft wide open outside the center the AF will not be accurate.
javier September 2, 2011 04:28 am

One does *never* focus the lens on a point, one focuses the lens at a *distance*. Just go look at an old lens with manual focus ring. The focus "points" on the sensors are just locators that will be used to make an estimate of the distance. Once you fix and lock your focus at say 1.5 meters, it will keep focused everything that is at a plane 1.5 meters away from the camera sensor/film, plus/minus the depth of field. If by moving the camera you change the distance to whatever you were focusing at, then your focus is gone. Again, if you are shooting at f/11 at 20m distance it wont make a difference if your point of interest is at 20.5 meters or at 19.80, but if your subject is close to the camera and the aperture is wide, a shift of a couple of centimeters will ruin your photo. But as many people say, go and test this yourself, it is the only way you will learn.
If you want to see the math, here we go: suppose you focus at a distance a, and you have a depth of field d, so everything between distance a-d and a+d will be on focus (we assume we are focusing closer than the hyperfocal distance). If now you tilt your camera an angle x, then the old "point" of focus is now at distance (of the focus plane) a*cos x, we can assume x is small enough so that cos x is positive, so to keep our old point on focus, we need
a - d < a*cos x) < a + d
second inequality is trivial because cos x is always smaller than 1, so we can ignore it, so we need
a - d < a* cos x)
dividing by a we get
1 - (d/a) < cos x
Here, if d is big compared to a, then d/a will be a reasonably big number, so 1-(d/a) will be way smaller than 1 and a reasonably big variation of the angle x will still keep things on focus. BUT if d is small compared to a, then even a small variation of x will put our point of interest out of focus. In James' drawing, a is small, but d is much smaller and x is very big.
If we focus beyond the hyperfocal distance, then d becomes very large, and thus everything stays on focus.
John Douglas September 2, 2011 04:25 am

Point of minor terminology. The "focal plane" in normal use is not the plane "out there" that the camera is focusing on. It is the plane within the camera at which the image is in focus. As I understand it, the normal terminology for the plane "out there" that is brought into focus on the focal plane is the "plane of sharp focus." Of course, this usually only matters to view camera and Lensbaby users - not to me.
Jason St. Petersburg Photographer September 2, 2011 04:23 am

You have convinced me to start moving my focus points off center when using my 50mm lens at large apertures and while close to my subject. I am still not convinced though that for a shot like this...
http://jasoncollinphotography.com/blog/2011/9/1/dslr-photography-lesson-with-terri-and-her-nikon-d80-in-st-p.html
...where I am shooting at f/11 32mm about 5 feet away from the subject, that setting the focus on the subject's face and then recomposing will produce a less sharp shot than if I had moved the focus point off-center to the subject.
Stacy Gardner September 2, 2011 04:13 am

I REALLY appreciate this article. I shoot wide open often and I have not contemplated how a difference in distance from this "shift" affect the focus. Now I need to contemplate a new way of shooting. Thanks for bringing this up and being so clear. I would love to hear your thoughts on better alternatives to focus and recompose.
cpando September 2, 2011 04:12 am

thanks for the post, the example and to everybody for the discussion. I also used to think as the focal plane as a sphere, DUH!!! can't you see the name? focal PLANE! (well, forgiveme. I'm not a native english speaker) it can't be a plane and a sphere at the same time (ok maybe it actually was incorreclty named).
But yes, if we shoot a wall it should all be in focus (well if it is a flat wall and not a spherical one).
I guess the lens designers had solved this problem for us, and the lens optics take care of converting the "logical" focus sphere into a more practical focus plane. I think it has to do with the lens being actually a semy sphere by itlesf.
From now on, I'll limit my focus-recompose.
Thanks for your post, and I hope you get some extra points for getting so much comments.
John Douglas September 2, 2011 04:03 am

James,
Simply dissing everyone else's understanding of math doesn't contribute to the discussion. Your pythagorean theorem calc was correct (though, as some have pointed out, the example person you use would be too tall for the NBA). The problem is that at every step of your post you exagerate the effects of "focus and recompose." You can check my previous post for the math showing that the real world implications of using focus and recompose, which summarize to "not much, except in fairly extreme situations."
As for your "real world" test, you put your camera very close to the target, set a REALLY wide open aperature and then swung the camera through a pretty extreme arc. It's no surprise the result is the recomposed "20" is out of focus. I'll do the math for you on your example shots, as best I can, given the lack of data provided by you. I'm guessing you were about six feet from the dartboard, based on the way it looks through your normal lens. I am also assuming you did not crop, so the shots were taken in portrait orientation. Your 50 mm lens has a 39.6 degree angle of view (top to bottom in portrait orientation). Based on that, I did a quick take off from the screen images and estimated you swung your camera through 16 degrees in recomposing. Given that, here's the math:
cos of 16 degrees = 0.9612616959
The distance from the camera to the plane of sharp focus at the location of the "20" after recomposing = 6 feet divided by cos 16 degrees = 6.2418 feet
Distance from the "20" on the dartboard to the plane of sharp focus after recomposing =
.2418 feet = 2.9 inches
Your DOF for your 50 mm lens at f/1.4 at 6 feet = 4.32 inches
(provided by the DOFMaster app).
Since you were "perfectly focused," we'll assume you were focused dead in the middle of the DOF, so you had 1/2 the DOF to play with when the plane of sharp focus was moved, or 2.16 inches.
Ergo, your final plane of sharp focus was (2.9 - 2.16) or .74 inches outside the DOF at the location of the "20" after recomposing. Not surprising the "20" is fuzzy in the recomposed shot.
Of course, the problem is (1) you were very close to the subject, (2) using a really wide aperature and (3) you swung the camera through a pretty wide arc to recompose the shot. Adjust just one of those elements and the final image would have been fine.
Of course you can potentially do better on final image sharpness of your ponit of interest if you have a focus point close to the spot where you want the key element to be when the scene is properly composed and you use that focus point to AF. Perhaps not, though, since the center focus point on many cameras is more sensitive and is consequently more likely to get focus right. "Doing better" also assumes that you have the time to pick your focus point, set it over the target spot and shoot. If you let the camera pick the focus points for you, well, good luck. In quick shooting situations, focus and recompose is by far the best choice. IMHO, of course.
John Douglas
Dana Lane September 2, 2011 03:54 am

Wow, this has sure generated a lot of responses. :)
The big thing to remember is that it isn't purely distance to the focal plane, you also have to remember that the focal plane is a straight plane parallel to the sensor.
The reason the OP's comment is correct, even though you may maintain the same exact distance from the eye's to the sensor is that the eye's will no longer be on the focal plane that will now reside 4.5 feet from the sensor, but going PARALLEL to the sensor, thus actually .5 feet BEHIND the eyes now. Does that make more sense?
Photogoofer September 2, 2011 03:45 am

Alex has the right answer. Unless you're at an EXTREMELY shallow depth of field, the difference is minuscule. I just suggest, in that case, do it the old fashioned way: manual focus!
Simon SC September 2, 2011 03:43 am

This is absolutely correct. The only exception I would add is that this technique actually works wonderfully well for compact cameras with small sensors. They generally have such huge depth of field that it doesn't really matter if focus is slightly off
char September 2, 2011 03:36 am

Hmmm... doesn't make sense. Focus on eyes, recenter camera to forehead to allow more headroom in the frame, for example) and the focal point on those eyes (which is what you want) has NOT changed unless you have inadvertently let go of that pre-focus button. The M5MkII is notorious for back-focusing though so that might be tripping you up?
Ben Strasser September 2, 2011 03:25 am

James, I understand what you are saying but I have a problem with your initial premise. I assume your example is a portrait situation being that your subject is 4 feet away then again you would not use a 50mm f1.4 lens at 4 feet for portraiture. Given your scenario, I would have my camera at eye level of the subject and focus on the eye from there. There will always be errors in focus which is why depth of field should be considered.
I find the “focus-recompose” or better yet the “zoom-in, focus and recompose” method to be very accurate. (provided your lens is parafocal) I also find that the outside focus points of the camera to be less accurate than the center point due to the fact the outside sensors are single dimensional while the center sensor is a cross-type sensor with two dimensional contrast detection.
Aaron September 2, 2011 03:20 am

Of course this will be an issue wide open at close range..... this method is perfectly valid for doing something like run-and-gun wedding shots, or portraits using a lens in the telephoto range. Shooting a 50mm on a full frame camera at 1.4 from the short distances.... the focus should always be nailed down AFTER composing. Focus and then recompose is perfectly valid for most any other type of photography.... even studio shooting at normal apertures (8-11).
Don September 2, 2011 03:18 am

I have to agree with the author esp when shooting fast lenses.
Sam September 2, 2011 03:16 am

THANK YOU for writing this. This is an excellent and useful article for specific DOF situations.
Simon September 2, 2011 03:15 am

Don (the "but the distance to the face is still the same guy) is wrong for this reason: The plane of focus is intended to be flat, not spherical. Lens designers work hard to try to achieve this.
In fact, of course, as this is something created by man, it's imperfect. It turns out that the focal planes of almost all lenses are neither perfectly flat, nor spherical, they tend to be approximately flat, but with wobbles in them. This is actually yet another reason why focus and recompose should be avoided when possible. The "correct" focus point will give you accurate focus in the point you care about (the eyes for portraitists) regardless of the perfection, or lack thereof, of the lens.
The reason focus and recompose is so common, of course, (this might have been said already) is that in the old days, we only had one accurate focus point, that was the split image/microprism dot in the center of the screen. All else was just ground glass. More recently, there are plenty of cameras that have only a few focus points, and even my D700 doesn't have focus points neatly spaced on the rule of thirds, they're mostly bunched in at the middle, but it's completely true that if you're wide open, you'd better focus with the correct pont, for at least three reasons, all stated clearly in this comment stream now.
Danferno September 2, 2011 03:10 am

I'm not convinced. Your distance from sensor to eyes stays pretty much the same no matter how you tilt your camera (unless you turn it away from the eyes, then it becomes infinity). The reason your example image is out of focus is because you don't rotate your camera with the sensor as pivot. The center will usually be roughly the middle of your head if you're shooting through the viewfinder.
Why do you think the bolt of your tripod goes in where your sensor is located (at least mine does)? Exactly, so any horizontal shift doesn't change the sensor-to-subject distance (+stability). Unfortunately vertical shifts will still get you a displacement of a couple millimeters.
tl;dr If you're a ninja you can shift your camera in such a way focus-recompose without issues, otherwise you better decrease the aperture or move further away (but not because of the reasons stated in the article).
Desiree September 2, 2011 03:04 am

This explains a lot. I actually thought that there was something wrong with my camera when it didn't produce sharp results. Thanks.
Sylfest September 2, 2011 03:04 am

I think there are some errors in this article.
Don's point is surely correct - it's irrelevant whether the centre of the image is in focus or not after you have recomposed (you weren't interested in that point anyway). If we consider the important focus point, then rotating the camera to recompose the image should have hardly any effect on its distance from the sensor.
(So what's causing the effect shown in his 'proof'?)
Also, if the depth of field is 0.16 ft, that doesn't mean it extends equally 0.8 ft in front and behind the focus. It extends less to the near limit and more to the far limit (this is the principle behind the hyperfocal distance method for landscape photography).
Aside from this, it seems like the author's using a rather extreme case to make his point, so I remain sceptical.
FrancoisM September 2, 2011 03:02 am

Also, for those stating that the focal plane is in the shape of a sphere – it is not.
Then it would been have called focal sphere !
James Brandon September 2, 2011 02:49 am

Also, for those stating that the focal plane is in the shape of a sphere - it is not.
Again, just grab your camera and go test this out, I really encourage you to try it for yourself so you can see it first hand...
Grab your camera and walk up to a wall and photograph it and most shallow depth of field you can attain. If you shot the wall straight on, you'll see that the entire wall is in focus, no matter how shallow the depth of field.
If the focal plane were spherical, the center of the wall would be in focus and the wall would gradually become blurry towards the edges (much like a lens baby)
konspir September 2, 2011 02:43 am

Don is right.
The focus remains in the A distance.
James Brandon September 2, 2011 02:42 am

Thanks for the feedback everyone. I see a lot of people starting to get why this method is flawed. For those that are arguing and saying my math and diagrams are all wrong, have at it. I assure you the math and the explanations are sound.
Like @Paul stated, instead of arguing, just go try it for yourself! Find a subject that approximately fits the dimensions in the diagram and replicate what I did. You may stop trying to defend focus-recompose :-)
Scott Lewis September 2, 2011 02:35 am

You're argument is as strong for "never shoot anything without the focal plane being perfectly parallel to the subject" as it is for not focus then recompose. If you're shooting more than a headshot of a person at 4' at f1.4 you're going to have eyes in focus but nose not and other similar issues regardless of how you focus and compose unless you're parallel. In the dartboard photo you see this in that the 20 is less focused in the right hand image but the bullseye is more focused. Maybe the rule should be "frame your shot, find the camera position that keeps focal plane subject to important items to keep in focus, select the focal point that preserves the focus you want for this composition." Or maybe selecting an aperture with more DOF is appropriate for shots where a few inches matters....
Your point is well made. It raises a series of other issues worth considering. Did you also try multi-point focus on the dartboard to see if the camera would choose the right point for you? On the headshot it would likely try to choose the whole face which, with that shallow DOF, might still get the eyes wrong.
Scott
Efrain Bojorquez September 2, 2011 02:33 am

@Don:
The problem here is that you're thinking of the sensor as a single point in the center of your camera... or in the center of the actual sensor, to be more precise.
If you think about the sensor as a table you can get a better idea of the whole thing. If you put an axis that runs through the center of and parallel to the top of the table (it doesn't matter in which direction), and the table is flat on the floor, you won't be able to rotate the table around this axis without moving it away from the floor.
For any given leg length (assuming they are all equal), there if only one angle at which all four legs will be touching the floor.
To complete the analogy, the top of the table would be your camera sensor; the floor, your focal plane. If the distance to the eyes of the subject to the center of the sensor is the length of the legs, then there would only be one angle at which the eyes are in focus.
Brad September 2, 2011 02:32 am

@Don etc. The issue is that the camera moves through an arc, but the focal plane is not spherical, and therefore it is always tangential to the axis of rotation. This, I believe you will find, places your focal point behind the point of your original focus as Mr. Brandon stated (and drew, sort of. hehe) I have experienced this with my 50mm manual focus lens, and have had to learn how to compensate for it. Not a bad skill to learn actually. When the depth of field is shallow on my wide lenses I also notice this same issue, and have therefore made the trek back to the focus points group.
PHugger September 2, 2011 02:31 am

You say -
"That means that if you focus on your subjects eye, move the camera down to their chest to recompose, then your focal plane is now half a foot behind your subject!"
What is your subject, the entire body? As close as you are, parts of your subject will be at the correct focal length (A) and parts are not (B). This will hold true for any flat subject that is sufficiently large or close.
What is the sharpest part of any image? Probably the center. The lack of Sharpness and Fringing says more about your lens than about your theory. Your point is well taken, but your theory and math are not well thought out. What you have proven is that an in focus image is sharper near the center than it is near the edges using your lens.
Mako September 2, 2011 02:28 am

Focus Recompose has so far worked well in all of my portrait shots. Focus points are annoying most of the time because there are incidents that there are no available focus point in the area where I want my main subject to be in the composition. So I need to change my composition because of it. Not unlike with Focus Recompose, the moment that you have locked the focus point by using the AF Lock, you can choose any composition you like with your subject always in focus. If mathematics would prove me wrong, well my photo outputs, in the other hand, tells me I'm doing the right thing. Just my opinion. Nice article anyway.
Hansa September 2, 2011 02:26 am

Would it matter if the camera was set to continuous focus mode instead of single focus mode?
Nina September 2, 2011 02:24 am

Totally agree!!
John Douglas September 2, 2011 02:14 am

I think you need to work on both your Watson drawing skills and your math. ;>) There is a potential problem, but it is not one that would occur in most reasonable scenarios.
Yes, the plane of sharp focus is perpendicular to the axis of the lens (except with tiltable lenses). And, yes, focusing and then recomposing by rotating the camera will move that plane of sharp focus behind the original point of focus. However, I think you seriously exagerate the effect.
The reason line B is significantly behind the stick figure in your drawing is because you drew it significantly longer than line A. Recomposing through the angle of swing shown in your sketch would have nowhere near that amount of shift in the plane of sharp focus. Indeed, in most situations, the shift is well within the depth of field.
As a quick example, by my calculations, if you use a 50 mm lens at f/4 and focus on your subject at 20 feet, then swing the camera 10 degrees to recompose, the plane of sharp focus will be 3.6 inches behind your original point of focus. However, you would have a 7.6 FOOT depth of field to play with, so the effect would not be noticeable. Indeed, good luck on getting your camera's autufocus to be any more accurate than that.
Moving in closer, to, say, 6 feet from the subject will get you a little closer to a problem, but still in the safe zone. At 6 feet from the subject, using a 50 mm lens set to f/4, when you swing 10 degrees you move the plane of sharp focus to 1.1 inches behind the point of original focus, but you would still have a 7.8 inch depth of field and, again, the effect should not be noticeable.
Wider aperatures, longer lenses and larger swings than 10 degrees could create a situation where you could run into a problem, but it's not likely. Longer lenses reduce your angle of view, so it becomes less likely you will recompose through a larger angle. Using much wider aperatures is the most likely way to run into a problem, but it's difficult. For example, at 10 feet from the subject with a 50 mm lens set to f/1.8, if you swing the camera 20 degrees, your plane of sharp focus would be 7.68 inches behind your original point of focus, while you only had a 5.4 inch depth of field to work with. However, after swinging 20 degrees, your original point of focus would no longer be in the viewfinder, so focus would be the least of your problems. ;>)
I'd be happy for someone to check my math (I did the math in a hurry), or to provide comments.
Paul September 2, 2011 02:11 am

One other point that needs to be made through this all, is that depth of field varies according to the size of the chip in the camera. So the 5D used in this example is a full frame chip, while those of us with 7Ds and under have fairly significantly smaller chips, making for wider depth of field. Having said that, I've been really working on getting my focus razor sharp, and frequently run into problems. This may explain why. 90% of the time I set the focus point and use the closest to the right one (although as with the comment above, 7 focus points makes that tricky) but there are times I won't shift through to the right one if there's a shot I don't want to miss. Even shooting a portrait, sometimes there's a fleeting glance that is just perfect, no time to refocus!
Josh September 2, 2011 02:10 am

It's always confused me why people think that focus/recompose is such a good way to go. I mean, if you're using a 35mm @ F8, go for it, but at that point, you could focus anywhere on your subject and get the right focus and the argument "the center is more accurate" doesn't really matter. Thanks for providing such a clear explanation of it.
Anyway, I think it would be super cool to get a G+ invite, and it says your invites are full... Oh well.
Brad Hayami September 2, 2011 02:08 am

Wow, really agree with Carla Costa. If you're taking candid portraits, where the moment is there and gone in a second, focus-recompose really seems like the only option to me. When I am taking pictures of still life, landscapes, buildings, etc ... this is a different story. Totally different photography experience, where you can slowly dial in your settings and create the perfect shot. As far as candid portraiture goes, I can't imagine all of the shots I would have lost if I took the time to select and dial in the best auto focus point for each picture.
That said, of course this article is right, and the cost of the focus-recompose method is that often the proper focus will be lost, especially when shooting wide open (as I often do). I guess my solution to that is to take as many pictures as possible using a continuous shutter, knowing that many will be out of focus ... I usually end up with a few that are in focus. Admittedly sometimes they are all out of focus, but as Carla said, I think that is a small price to pay for the ability to quickly capture a fleeting moment. That's not lazy, just a recognition that for certain types of photography, perfect timing is more important than perfect technique.
Alex Gac September 2, 2011 02:07 am

I hate to say this... but both the author and Don are right about the geometry... but they're just making different assumptions.
Author James is assuming that the axis of rotation when recomposing is NOT the sensor surface. This is a reasonable assumption, because when most of us recompose we do so by moving our heads and camera together. The axis of rotation is a half foot or more behind the sensor. In that case, the distance between the sensor and the desired point of focus will change. Another assumption is that the focal plane is flat. It is actually curved.
Don's assumption is that the sensor is on the axis of rotation. This would hold true if the camera were mounted that way on a gimbal style tripod head. (Gimbals are commonly used for the balanced action they afford for large lenses, but they are equally well suited to control rotational and optical alignment.) Another assumption is that the focal plane is spherical. It is curved, yes, but not quite spherical.
Having said all that... any change in the distance or angle between the subject and sensor will result in a change of focus, however minor.
When I'm shooing an event, I focus and recompose all the time. If it's a sporting event, I typically just keep subject in the center of frame and crop for composition later (I'd rather loose a few megapixels in post than accidentally cut off the ball-hand in camera!!!). In the studio, I have all AF points active but always rely on manual focus on a properly composed shot with camera on tripod.
Sam September 2, 2011 02:06 am

Great follow up article! This certainly clears things up -pardon the pun!
Ken Hurst September 2, 2011 02:06 am

Although I agree with, and am not surprising by your tests, I think a better title for this would have been "Critical Focus at Extremely Limited Depth-Of-Field". Anytime you use an aperture close to f1.4 any kind of change in position or movement of the camera whether you focus using the autofocus system or manually there is potential for the focus to not end up where you want it. Typically I wouldn't myself move the camera that much when recomposing and at that distance the small change probably wouldn't be critical. (I may have to conduct my own experiment to prove that point!) But - when shooting up close for macro or near macro shots recomposing after using center focus point could make a huge difference although in almost all cases you're going to want to use manual focus for macro or even near macro shots anyway. Check out my blog where I test my 100mm macro to make sure it's functioning properly and that live view is accurate when focusing manually : http://kenhurstphotography.blogspot.com/2011/07/canon-100mm-f28-macro-lens-test.html
Efrain Bojorquez September 2, 2011 02:06 am

I believe there are certain photos/situations where you can use focus-recompose with no problem.
In the example you show above there's an obvious fail to the method. you are shooting at a short distance which will make the relative differences between A and B greater as the distance to subject decreases... if you're using a big aperture you have a problem with this method.
However, if you're shooting a landscape, then the difference between A and B will be meaningless: for one, the shift in the angle will be minimal giving a very small difference (relative to subject's distance)... and, on top of that, no matter what aperture you're using, you are probably focusing beyond hyperfocal distance.
Personally, I use the focus-recompose approach when shooting wide and at not to so big apertures (landscapes, generally)... and never when I shoot a portrait, with no regard to aperture or subject distance. In the end, you almost always have a composition in your head before you actually take the shot. Why not set the camera up for that composition before hand?
kimberly September 2, 2011 01:53 am

Love the article...but isn't Don's reply above correct regarding recomposing? If you are not re-focusing on B..then A should still be in focus, right? Or am I missing something?
Lisa September 2, 2011 01:53 am

I read this and hadn't tried selecting a focus point on my camera yet (please understand that I have only had it a few months and have been learning all of its other features; further, my previous cameras were manual SLRs from 80s). What was fascinating - the manual is WRONG. Good thing, though - it is actually faster and fewer steps than the manual said to select my own focus point vs. allowing the camera to select automatically. It is only one button to access the function and then either a few button presses or a few wheel clicks to select which point is active. When the point you want is lit up you return to shooting! Simple. Getting back is a matter of highlighting the center point again and hitting set. Easy.
Now I really have to wonder if I ever will use Focus-Recompose again...
Kent West September 2, 2011 01:51 am

Agreed, especially when you are using a fast lens (2.8 or less). Try using the center point, focus on the brides eyelashes from above, then recompose. Her eyelashes will be out of focus every time.
javier September 2, 2011 01:50 am

@gareth: that is not correct, when you rotate the camera the whole FOCAL PLANE rotates with it, effectively placing the focal plane behind your subject (in the example of pointing at the eyes and then rotating back to the center of the body). The only case in which focus-and-recompose works is when you SHIFT THE CAMERA, without rotating it, keeping it inside the plane containing the sensor when you focus. Unless you are using a tilt-shift lens, It is a matter of distances between planes, so rotating the camera is ALWAYS going to destroy your focus.
Joe September 2, 2011 01:49 am

I can see the idea of being against focus-recompose in this instance. It makes sense that at extremely shallow depths of field, that this is important.
My propensity for using it depends upon other factors, however. Pro-sumer level SLRs and below only have a cross-type focus point in the center, which I've found does indeed perform better than the point type ones that typically surround the center one. Higher-cost cameras don't have this issue as they have more cross-type focus points and as a result, eliminate an advantage of using the center point for focus-recompose.
the way I see it, the focus-recompose method is a compromise that sometimes needs to be made. This is more so on less expensive cameras, but can be just as critical in the right situation regardless of camera as well.
Just my $0.02. Great article, BTW!
Tim Krenzke September 2, 2011 01:49 am

What you have presented is true...in specific cases. If the camera-to subject distance is increased, or the aperture is closed down, the effect you present is mitigated or eliminated. In general, I agree with you and I seldom use the "focus and recompose method." But, in some circumstances with understanding of the conditions, it will work!
Ming-Tzu September 2, 2011 01:48 am

I am new to photography but my understanding was that by focusing and recomposing, you're not actually changing the focus from the eye to the body. That by pressing the shutter button halfway to focus on the eye, and then recomposing while still having the shutter button depressed, you're not re-focusing on a totally different plane (the body) but just positioning the area in focus to a different section of the composition.
Or maybe I don't know what I'm talking about lol.
Would be interested to know the answer to this though. I use the focus-recompose method exclusively now.
J September 2, 2011 01:48 am

@ Don:
I agree with you.
@ James:
You write: "If the length from your camera to your subjects [sic] chest is 4 feet and the length from your subjects [sic] chest to their eye is 2 feet, then the length from your subjects [sic] eye to your camera is 4.5 feet."
I agree with your sentence above. I disagree with your conclusion. The important distance is the distance from the camera to the subject's eyes. Once you lock focus on the eyes, all other objects 4.5 feet from the camera will be in focus. Thus, the distance from the camera to the subject's chest is irrelevant (assuming you want the subject's eyes -- and the subject's chest -- to be in focus). As others have said, your picture does not take the curvature of the field of focus into account.
@ Anyone:
Please correct me if I am wrong. It wouldn't be the first time :)
Ryan O. Hicks September 2, 2011 01:45 am

While some of this could be valid I suggest you take a read from Bob Atkins article on this subject.
"So when won't it be OK? Well, the greater the angle though which the camera is turned, the greater the focus difference, so the wider the angle of view of the lens, the greater the possible focus difference will be. The faster the lens and the closer the focus, the smaller the depth of field will be. So the worst case would be a close focused fast wideangle lens, shot wide open."
http://www.bobatkins.com/photography/technical/focus_recompose.html
Mark September 2, 2011 01:44 am

For some types of photography this is very important. Portraits, fashion, studio, and others that use wide apertures it is critical. For other types of photography it is the wrong approach to use other focus points.
I shoot wildlife with attention to action shots. For this I use center point only, AI Focus (on Canon 7D) and it is all I can do to keep the center point on the animal at the moment of movement. Thus I compose my images by cropping. Yes I lose effective megapixels, but I have some to spare with the 7D. The AI Focus means that if I try to recompose, the focus changes to the new spot in milliseconds. I am also usually at 400mm lens length (640mm equivalent) and at f/7.1 or f/8.0 so my depth of field leaves a little room for error although not much. However, if I fussed with changing focus points, I would miss the shot altogether since I am tracking the eagle from its perch to the moment I grabs a fish from the river and less than 100 yards away sometimes.
Changing focal points or using full manual or any of a number of suggestions just won't work for this type of photography and I suspect that there are others like sports photography that are also not applicable to this type of technique.
I also suspect that many of the comments to the original article were from users that had point and shoot cameras or DSLR's with kit lenses. I you are incapable of wide apertures, this is not an issue at all.
My Camera World September 2, 2011 01:44 am

Both groups are right. First when you need to get a shot always use what you know best. If that centre focus then great. For many non-action shots this works great with focus and recompose and if you have to quickly go to an action shot, no time to change.
But if you are after the best quality then perfect focus is paramount. With fast moving action then proper framing is required with focus on target. With experience and practice you will know which side to have focus point, to be ready for the shot.
Its good to get out of your comfort zone and try new approaches. Learn how this works for you and adapt to your style.
Niels Henriksen
Carlo Parducho September 2, 2011 01:40 am

Thank goodness geometry was the only math class I was good at ^_^
Shariq September 2, 2011 01:39 am

Very interesting, thanks. I am fascinated by the theory behind photography as well and this threw up some interesting food for thought. However, intuitively, I only use focus-recompose when shooting objects a fair distance (say, at least 30 feet away from me) simply because the focal planes allow a bit more lee way. I don't do it for close-ups because the focal planes are too narrow. If one sticks with this rule of thumb, doesn't all the math become a moot point?
Thanks for mentioning the DoF app though! I use an Android phone and found a replacement app for it as well!
Evan September 2, 2011 01:37 am

Hagen said: "Recomposing forces the camera to now expose on the centre point (yes even if weighted, average or other). So if you are of the recompose school: AE lock, recompose, AF lock recompose. Getting kind of complicated."
This is actually the method I use, and once you have it down it's pretty easy:
1. Point at the area you want to meter. Hold AE lock (I use my custom thumb button for this).
2. Point at the area you want to focus. Hold the shutter halfway
3. Recompose and *click*
The whole process takes about 2 seconds, and works like a charm at small apertures--for landscapes, or anytime you want a large depth of field. Of course, now that I understand why it doesn't work at f1.8, I need to learn a new technique for portraits. Thanks James!
SS Manak September 2, 2011 01:37 am

In the example above the camera is at chest level. After recomposing there is backfocus.
Now if we assume that the camera is at eye level, as is the case most times and redraw the diagram, it will show front focus.
Now in case we draw a focus plane, ie, line at 90 degree to line of sight, the story becomes quite complex.
Moral of story, should not recompose while under some conditions (not DOF) you may get good focus.
Gareth September 2, 2011 01:36 am

@phil:
Yes, good point. On the dart board shot he used a tripod, so the point he rotated the camera about is significantly below the sensor, hence the distance from sensor to focal point changed resulting in an out of focus shot.
So in summary, focus/recompose is a perfectly valid thing to do - providing you rotate the camera about the sensor (i.e. the sensor position doesn't change). In reality that's probably quite hard to achieve.
Peely22 September 2, 2011 01:35 am

I'm afraid you may be wrong on the Pythagoras Theorum explanation. Whilst the distance is indeed different between the Camera and Point A to the Camera and Point B, you are still holding focus on the Point A, and the distance hasn't changed.
There may be slight variance due to lens thickness towards the edges? I don't know; but, provided you haven't moved your camera away, then the distance between the camera lens and the eyes you focused on is still teh same.
The difference between the two dartboard pics is down to some setting changing, most probably your metering. look at teh difference in the clarity and the blacks.
Sorry, it just doesn't stack up.
I would however agree, that given the time, choosing the right focus point must be the best method.
David Bowes September 2, 2011 01:32 am

If anybody needed a G+ invite still help yourself
https://plus.google.com/i/StoyWt65Rdc:Qo5JnydCnr0
Thom September 2, 2011 01:31 am

You present a compelling argument here. I've been noticing in shoots that I'm been using my 50mm prime lens I've been having issues getting the focus on the eyes tack sharp like the prime is supposed to be able to produce, while taking shots of centered items have been near perfect. I've never thought of the angles problem here but it makes perfect sense to me. I have a wedding shoot coming up this weekend that I'm going to try and not use the focus / recompile method and go back to using the other focus points for my low-light wide aperture shots to see if that helps any. If nothing else I can do what I've been doing with slightly out of focus shots and produce them for 4x6 pics.
Phil September 2, 2011 01:26 am

@gareth:
That makes sense. It also makes sense why most "focus-recompose" don't look right. If I'm doing a f/r (my new shorthand for it) I'm on a tripod, so when I angle the camera my axis of rotation, the ball joint, is 3 to 4 inches below the sensor. If I'm on a monopod, it's even worse since the easiest way to change the angle is to tile the entire monopod. The axis then is several *feet* below the sensor!
Alan September 2, 2011 01:24 am

Hmmmm...I've always been a focus-recompose guy but this is definitely something to make me reconsider that. Well explained, cheers.
Lou McLaughlin September 2, 2011 01:24 am

I do agree with you on the idea of not using just the center focus point. I think there are better reasons not to focus and recompose. Exposure differences to me are the main reason. In your example, you said you were using a tripod. Depending on the head, your camera could have actually been moving forward quite a bit when tilting it down. In your drawing example, I have another feeling about that. If the focal plane of the camera remains in the same place in regards to distance from the ground and the camera is just pivoting off of that point, and you don't move either forward or backwards of course; the distance to the eye is still going to be greater than the distance to the chest. Since you're supposed to focus on the eye, that's the distance you would want. I can' t draw, so I'll try to explain. Without moving the camera, you could draw a line from the center focus point to a point on the same plane as the eye, then draw a line from the upper focus point to the eye You are still creating the same basic triangle, but the longer line is the line from the focal plane to the eye, which is correct in most cases. I hope I have made myself clear. One other thought, It is totally absurd to focus with the center point and recompose on a tripod! You know your camera is moving quite a lot when you move it on a tripod, especially when tilting up and down! Happy shooting!
Steve M September 2, 2011 01:23 am

I am inclined to agree with Ovidiu and Don. If it is 4.5 feet to the eyes when pointing the center of the image up to the eyes, it is still 4.5 feet to the eyes when pointing the center of the image straight at the chest. I'm not sure what impact there might be in changing the angle of the focal plane to the eyes, but the distance is certainly static. Yet another factor might be focusing on the eyes using the sweet spot in the center of the lens and then moving to a spot closer to the edge of the lens, but still the distance from the plane of the sensor to the eyes should not be changed.
Stephanie September 2, 2011 01:23 am

So then what is the best method for getting the sharpest focal point in an action shot, when your desired point of focus is not in the center of the frame?
Paul September 2, 2011 01:22 am

Matters not about challenging the geometry. Just try it - shoot wide open using the recompose method and you`ll be forever moaning that the lens you are using is soft. In fact it`s for the very reason James highlights here. Sometimes you hit lucky most times not.
Gareth September 2, 2011 01:20 am

I think I agree with Don. (2nd Sep @ 12.37).
Assuming you just rotate the camera about an axis that runs through the sensor, the distance from the sensor to the focal point (the eye) remains the same, so it will still be in focus.
The out-of-focus shot of the dart board probably came about because the camera was not rotated around the sensor, so the distance from sensor to focal point has changed. Obviously it's hard/impossible to rotate exactly around an axis through the sensor, so using the other focal points is still a good idea.
Tim September 2, 2011 01:19 am

The proof is also that Hasselblad has some kind of technology to correct for the focus problem (haven't used it, just read about it once).
However, film cameras dictate that you use focus and recompose by placing the focusing method of choice in the middle of the screen. So maybe the whole focus and recompose debate needs to be on the list of "well, it was good enough for film, and look at all the great pictures that were produced."
Jon DeJong September 2, 2011 01:17 am

There is a third method that I often use. I'm not saying it's a good idea, just that I use it, and would love to hear thoughts. With DSLR's having such a high pixel count these days, I find myself focusing on the subject and composing (cropping) afterwords. Obviously, this has it's downsides, and they're potentially significant. But I take a lot of action shots. This allows me to use the center focus point, but still end up with some decently composed shots.
Rich O September 2, 2011 01:15 am

Fantastic article. I have been a big (ab)user of Focus/Recompose for awhile (too long, lol). WOW, what an eye opener. I am not a professional, but do consider myself to be in the advance amateur category, and I often teach others; helping them to take better photos....well, I'VE just been schooled! lol. Thank you so much for the tip.
There were plenty of times when I couldn't achieve critical focus and always wondered: Was it my lens? My Camera...nope it was me and now I know. But, as BENOXI pointed out, it seems that "focus drift" is more critical when using wider apertures and perhaps not so much when using smaller apertures. If I may, I would like to add to that: if I am understanding this right.....Any time there is a shallowing of DOF; be it wider apertures or long focal lengths, one should be aware of the "Curvature of view" and be mindful of "focus drift." Does that seem like a fair assumption?
thanks again.
Lou September 2, 2011 01:14 am

Kinda new to DSLR. From the get go I didn't really comprehend how focus recompose would work. After all....I have focused on the eye now that point is at the body. I use different focal points all the time now. With that said my newbie xs has a whooping 7 points......hence my looking to upgrade to the 7D. Focal points may be the biggest single reason I'm choosing that model camera.
I wish they had points towards each of the corners somehow.
Matt Owens September 2, 2011 01:10 am

James, I agree with you on this debate. If I'm shooting at F11 or so for landscapes, I will still use the focus, recompose method just because it's quicker and won't affect the focus that much. But for shallow portrait work, I always change my focus point to make sure my images are tack sharp. It doesn't take much movement at F1.4 to ruin an image.
Krzysiek September 2, 2011 01:08 am

The picture wasn't working for me for the first time - but now I see why it isn't correct. Why is the vertical line straight not curved? Shouldn't it instead of looking like this: ----| look: ------). If not - why?
Diane September 2, 2011 01:08 am

Wow, amazing article!!! I 'll be using focus points from now on!!
ErikKerstenbeck September 2, 2011 01:07 am

Hi
I have been anticipating this article when you mentioned it on Twitter - I totally agree with your point of view and I like the Stick Figures! ;-)
I would have failed with this shot had I used Focus Recompose:
http://www.stumbleupon.com/su/2s6RfA/kerstenbeckphotoart.wordpress.com/2011/08/28/the-corvette-river/
Carolyn September 2, 2011 01:06 am

Completely agree. I shoot food at close range and wide apertures, and focus-recompose would be completely useless in that situation. I use single point AF with a non-center focus point almost always.
javier September 2, 2011 01:06 am

PS: In case James ran out of G+ invites, here you have some more:
https://plus.google.com/i/ehgC9QCxEpo:AAVPggjkxi8
Chris Humphrey September 2, 2011 01:05 am

So the other moral of the story (i.e., apart from using another focus point than the center one) is to make sure your depth of field will be sufficient that focusing and recomposing will not throw your center of interest out of focus.
Very good demonstration of your point.
Eden September 2, 2011 01:03 am

I'm keen to hear the feedback
Alan Hendry September 2, 2011 01:03 am

Great write up I have used the recompose method for years, buy forgot about the math!
Cool stuff thanks
Alan
javier September 2, 2011 01:03 am

Agreed. The only situation in which it really doesn't matter is when using wide-angle lenses focusing far away, i.e typical landscape shots. For close portraits or macro shooting, "focus and recompose" will bring a lot of pain. Oh, FWIW, I am actually a mathematician, and I vouch that James got this one right ;-)
Alex September 2, 2011 01:02 am

NICE tips! Thanks! I've started noticing that the focus-recompose method isn't perfect and wasn't 100% sure why! Thanks for the explanation and the tip on fixing it!
Krzysiek September 2, 2011 01:01 am

Correct me if I am wrong...
If you are focusing on a person's eye, focus, and then recompose and shoot, then the distance between the eye while focusing and while recomposing is still the same - therefore the eye is in focus - maybe the chest is out of focus - but that's why you are recomposing - to have the focus on the eye...
Carla Costa September 2, 2011 01:00 am

Well, I think anyone can agree focus is more accurate using the proper focus point, but the debate here is more about cost x benefit. I use focus-recompose and I often get bothered by how it's not perfect, but I really can't afford the time that it would take me to choose the right spot while I'm shooting - I do candid photography. For other types of photography, that might be worth it, but in cases like mine I think it becomes impossible.
The truth is none of those methods are ideal. They should just make cameras with buttons that make you take less than a second to select the spot, I think I'd like it to have a tiny "joystick" (does that make sense?).
norbe September 2, 2011 01:00 am

hmm seems dear writer you forget about AI Servo Focus...
Hagen September 2, 2011 12:59 am

Against recompose: On one side, I agree that focus-recompose is problematic for narrow DoF, I also believe the vast majority of people will not have this problem: they are either too far away, or they are out near f4 or more.
Center points are better at different cross-type focusing, so IF you have problems, switch to it and recompose.
FOR using other focus points: make the time to learn it and you'll see how fast you can be. After all, it won't take any longer than focus and recompose. It is a tool that is specific to its intended purpose and for that, it is far superior to focus and recompose. As james says: any subject coming towards to (people walking etc) need a different focus point.
Recomposing forces the camera to now expose on the centre point (yes even if weighted, average or other). So if you are of the recompose school: AE lock, recompose, AF lock recompose. Getting kind of complicated.
Summary: the other focus points are there for a reason - learning to use them will only improve your photography.
Bud Branch September 2, 2011 12:59 am

Try back-button focus - it decouples the focus trigger from the shutter. Custom-function on Canon.
FrancoisM September 2, 2011 12:59 am

Hi James,
very clear explanation ... noe I know why some of my pictures are a little bit out of focus sometimes.
I so miss the eye lead focusing points from my EOS 5. It was a charm to select one of the 5 focusing points just by looking at them.
On a side note: what are you doing with a dart board in your office ? Aren,t you supposed to work there ? LOL
Or maybe you stick your boss picture on it to thrown the darts ? ;)
Francois.
Stuart September 2, 2011 12:57 am

"the length from your subjects chest to their eye is 2 feet" - this would make the people you're photographing about 12 feet tall!
Most people wouldn't be shooting at f/1.4, how about some examples based on f/8 or f/11 where the DoF is greater and there is more chance of the "20" being in focus?
david crundwell September 2, 2011 12:57 am

Hi James,
I'm not sure I totally agree with your logic. If I use single servo autofocus and focus (tilted upwards as you suggest) on a subject's eyes and then move the angle of the camera downwards to compose the shot better, the distance from the lens to the subject's eyes is still the same (with a small margin for error during tilting) and therefore in all likelihood the eyes will still be in focus. Yes, the body is closer to me, thus possibly in a different focal plane and may well be slightly out of focus. But the eyes are in focus still and isn't that the point of the exercise?
Phil September 2, 2011 12:57 am

James, thanks for breaking down and explaining *why* this won't work. I discovered this myself the hard way when I first started experimenting with my camera, and I noticed it especially when I used a particularly shallow depth of field. I just guessed I was doing it wrong and moved on. It's nice to know that it wasn't me, per se, but the act of trying to do it was doing it wrong :)
Incidentally, your invite link for Google+ isn't working for me. It takes me to the Google+ main page. If you have any spare invites, I'm at phil.frederick@gmail.com
Ovidiu September 2, 2011 12:54 am

@DON:
For what its worth, I do think you are absolutely right. I'm curios if there will be any follow-up statement from the article's author on this matter :-)
Mario K. September 2, 2011 12:52 am

@ James
I agree that the drawing is a bit misleading. You also move your sensor plane when you recompose e.g. from the body to the eye. So the two focus planes are not parallel to each other.
Jeremy Randall September 2, 2011 12:51 am

Nice post. Should "then our 0.16 ft plane of focus is actually 4.5 feet behind our subject" really be "then our 0.16 ft plane of focus is actually 4.5 feet from the camera" (ie the plane of focus moved .5 feet not 4.5 feet)? And, of course, shooting at an aperture with a larger depth of field can make the recompose stay in focus -- f1.4 is pretty extreme! :)
benoxi September 2, 2011 12:47 am

i agree with the author up to a degree. however in my case, mostly i would be shooting at apertures around f2.2 - f2.8. As little the difference may be, at such apertures the focus drift is less of a problem , provided the center of your recomposed shot is not too far from center of your initial focus. Focus-Recompose is a good method, it's very fast and effective when you've learnt about it. The way i digest this article , the author highlighted a very crucial potential for error within the method(we should always be aware of the ''curvature of view'' we make when we recompose)
happyspace September 2, 2011 12:45 am

Excellent article! Thank you! You answered each of my "but what about...." that came up as I read through. Articles like this one really make me appreciate DPS.
Ron September 2, 2011 12:40 am

Great article. I ALWAYS use the focus recompose method, but rarely at such large apertures; usually 5.6 or 8.0 when shooting weddings. For those paper-thin depths of field with a 1.8 or less, I often tripod and manually focus. Thanks for the great information. Sure would be great to get on google+ sometime.................
Mimi September 2, 2011 12:39 am

Agree!! Great post!
Don September 2, 2011 12:37 am

Hi James,
I'm not sure (and you photo proof seems to suggest otherwise) your geometry theory is really accurate. Yes, Line A is longer than Line B... But you aren't refocusing when the camera is in the B position. So, what is in the center position at point B is out of focus and the eye (point A) should still be in focus because the distance from A to your camera is still the same regardless of where the camera is pointed. And since you are focused for distance A, anything that is A distance away from your camera should be in focus...
Now, if you are re-focusing at point B after tilting it down, then yes, point A is out of focus. But, I don't think you are so line length B is meaningless. It is just used to see if what you aim at is in the dof for your aperture...
I think of it more like a sphere... If you focus at point A (i.e. 4 feet), anything that is in the sphere around you at a distance of 4 feet (and in the field of view of the lens) is in focus...regardless of where the camera is pointed...you just get a different section of the sphere in focus...
I think...
Michael Fang September 2, 2011 12:34 am

This is exactly the reason I started to use other focusing points to directly compose the image 2 years ago.
However, if you are a manual focus savvy, you might be able to get away by manual offset. I personally cant do it, but I believe that there are people who can.
Anyway, very nice post James!
Dana Lane September 2, 2011 12:33 am

I shoot a lot of portraits at F1.4 and I never use the focus and recompose method. I always select the focus point that rests where I want the eyes to be in my frame and use that point to focus the eyes and shoot without moving the camera. I shoot with a Sony A850 and I have the little joystick on the back of the camera linked to my focus points, so I can change it on the fly with my thumb, while the camera is up to my eye. It really is the only way to make sure images are tack sharp where you want them to be.
Great use of images to prove your point, nicely written up. :)
Raymond September 2, 2011 12:31 am

Agreed wholeheartedly. I have always tried to avoid using just the focus recompose method as much as I can. It is vital for me to get the sharpest possible shots at all times and I think focus recompose is just plain laziness from the photographer's part. Just my 2cents.
andybonn September 2, 2011 12:29 am

i agree with focus recompose. it's quick, and it's self-defeating. thanks for the article and the invite to google+(although it has run out from your link ) /hint. hhehehe

DPS offers a free weekly newsletter with: 1. new photography tutorials and tips 2. latest photography assignments 3. photo competitions and prizes Enter your email below to subscribe.
Email:
Get DAILY free tips, news and reviews via our RSS feed.