How do autofocus points work in relation to DOF?

[jwbryson1]

Let's say your DSLR has 20+ autofocus points and you are shooting wide open, say 2.8 or larger. In that case, it's possible that your DOF will be shallower than the positions that you autofocus points are tied to in the viewfinder. So, which wins?

I guess the DOF is the DOF no matter what, and just because the autofocus point(s) appear to be focusing on a near object and also on a far object in the view finder, that means nothing. DOF wins.

Is that correct?

 

[amolitor]

There is a plane of focus, no matter what. The lens is focused to some distance, and that's the stuff that's in focus, period. Then there's DoF, which describes how quickly things fall out of focus, as they are farther off the plane of stuff that's in focus.

The AF points just tell you what points the software is using to guess what a good plane is to be focused at. The focusing system should be able to determine how far out of focus the lens is at every AF point. That is, it can tell for every AF point, how far away from the plane of focus the object precisely at that AF point is. This allows the system to, in fact, measure how far away from the lens the objects are at each AF point. This gives it a very very crude 3D model of the world out there in front of the lens.

The software will then select some collection of the AF points to use, based on I do not know what, but surely depending on AF mode, and will place the plane of focus on presumably some sort of average distance based on how far each of those "active" AF points is.

You can imagine the software saying 'ok, points 1, 2, 7 and 12 are just all over the stupid place, but 3, 4, 5, 6, 8, 9, 10, and 11 all seem to be about the same distance and grouped together, so I am gonna bet that's the subject, so let's average up the distances.. and bias a little toward the ones more in the middle of the group.. and call it good!' It's gonna light up AF points 3, 4, 5, 6, 8, 9, 10, and 11 to give you a hint what it's thinking about.

 

[jake337]

Huh? Your DOF plane will run parallel with your image sensor unless you use a tilt/shift lens or movements in a bellows.

After reading your question again I think I understand what you are asking and I am unsure how to answer.

So lets say you are taking a pretty close up portrait, shooting wide open(f1.4), while using all your focus points(auto/dynamic/etc) and while looking through your viewfinder multiple points will "pop" up but in your head you know your DOF is so thin your focus is either on one or the other points, correct?

Could it be focusing in between the highlighted points?

 

[jwbryson1]

So lets say you are taking a pretty close up portrait, shooting wide open(f1.4), while using all your focus points(auto/dynamic/etc) and while looking through your viewfinder multiple points will "pop" up but in your head you know your DOF is so thin your focus is either on one or the other points, correct?

Pretty close to my question... Using your facts above, (portrait at f/1.4), the DOF will be very shallow (let's say 2 or 3 feet), but in the viewfinder the autofocus points may be highlighted on (1) the subject's eyes (within the DOF) but also on (2) a tree or shrub 10 feet behind the subject and outside of the DOF.

The viewfinder appears to show that both the eyes and the tree/shrub are being "focused on" by the camera, but we know that the DOF will stop far in front of the tree/shrub. So, I guess what happens is that the DOF starts where it will start and will stop where it stops, and the fact that the tree/shrub appears to be something that the camera is focusing on means nothing because the DOF ends many feet in front of the tree.

 

[bhop]

DOF and focus points are unrelated..

DOF is controlled by the aperture blades.. and the focal length too to some extent I guess.

 

[amolitor]

If the camera indicates that it's using AF points on the subject's eyes, as well as one on a shrub in the distance, I am pretty sure it's really telling you 'I am about to focus in completely the wrong place!'

 

[kundalini]

I only use one focal point for a portrait. That would be on the near camera eye and that's what be considered the important bit. The rest of the image can fade away according to your DOF.

I'm guessing you're asking about a protrait/person shot.

 

[Big Mike]

but in the viewfinder the autofocus points may be highlighted on (1) the subject's eyes (within the DOF) but also on (2) a tree or shrub 10 feet behind the subject and outside of the DOF.

That doesn't happen. If it does, then either the subject and the shrub are very far away from the camera, or your camera is malfunctioning.

As mentioned, the lens can only physically be focused to one point (distance). So when you see more than one AF point light up, it (should) mean that those points are the same distance away.

Personally, I almost never use the 'Auto point selection' option. A always use a single point so that I'm the one making the decision on where to focus, not the camera.

 

[KmH]

DOF and focus points are unrelated..

DOF is controlled by the aperture blades.. and the focal length too to some extent I guess.

Or at least the lens f-stop setting - and the point of focus distance, and the image sensor size.
If subject scale is maintained in the image frame, focal length has very little effect on total DoF, but has more of an effect of the DoF distribution.

Point of focus distance has a more immediate effect on total DoF than the lens aperture does.

Auto focus points have no relation to DoF other than they are used to select the point of focus. From there, DoF results from a combination of other variables.

There can only be 1 point of focus in the scene. In other words, if the camera I set to a mode that displays more than 1 focus point, the camera can only use 1 of them for final focus.

http://cpn.canon-europe.com/content/education/infobank/depth_of_field/depth_of_field.do
Understanding Depth of Field in Photography
Understanding Camera Autofocus
http://en.wikipedia.org/wiki/Autofocus
http://imaging.nikon.com/lineup/dslr/d800/features03.htm

 

[TCampbell]

If you let the camera decide which AF point to use, then technically it'll only use one of them to lock focus. By coincidence, it's possible that more than one AF point may be on a subject at the same distance (especially if you're focusing on something which is "flat" and parallel to the sensor plane.)

Some cameras have a mode that asks the camera to maximize focus of a subject (A-Dep -- automatic depth of field) and it can try to make both the near and far part of a subject focused. Sometimes it's a pseudo a-dep in that it doesn't do anything to alter depth of field... it simply tries to maximize the number of AF points that can lock focus.

Ignoring A-DEP and just using standard focus modes, when the objects in the viewfinder are at assorted distances and you allow the camera to auto-pick the AF point, then a Canon camera will always select the AF point which can lock focus at the CLOSEST focusing distance (that's how they programmed their algorithm.) I don't know how other brands do it.

Suppose your subject is a person at the far end of a table. In the middle of the table are some flowers. If there are focus points covering the flowers and other focus points covering the intended subject, then the camera will focus on the flowers and NOT on the subject because the flowers are closer to the camera. This is where you need to take control of the focus system and tell the camera which focus points you want to use for focus rather than let it auto-pick.

The camera doesn't need to test focus at each point to find the focus point with the closest subject. Due to the way phase-detect focus works, the phase split is shifted when an out of focus subject is either too close or to far. But the shift is in one direction if the focus is too far and it shifts in the opposite direction if the focus is to close. That means the computer can easily tell that one focus point has locked focus, and all other focus points are out of focus but the shift is all in the same direction and they're all farther away (based on the direction of the shift.)

This only applies to phase detection AF and does not work with contrast detection AF.

 

[KmH]

FWIW, the actual focus point is smaller than the indicator that is displayed.

The actual focus point is not likely to be centered in the indicatoir either.

 

[jwbryson1]

Good answers, thanks. I follow Big Mike's rule and use a single focus point most of the time. The question popped into my head, however, so I thought I'd post it here.

 

[ScottMac]

There is no relationship whatsoever between depth-of-field in an image, and the focal point used to acquire focus.

DOF is purely a function of the aperture setting on the lens at the time an image is taken, while the point of focus in an image is a function of the algorithm of the camera, relative to the focus settings within the camera, together with the how the user moves between capturing focus and capturing the image. Try not to over-complicate the issues, or for that matter, connect them. They are totally independent of each other.

 

[KmH]

DoF is entirely dependent on, and permanently anchored to, the point of focus.

DoF is distributed in front of and behind the point of focus.

 

[ScottMac]

[FONT=arial, sans-serif]Depth-of-Field = "The distance between the nearest and the furthest objects in an image that are judged to be in focus." and this distance is a relationship between the aperture of the lens, the focal length the image is shot at, and the distance of the "point (plane) of focus" from the camera. It bares little relationship to any little focus point highlighted in the viewfinder of the camera at the time of image capture.

To calculate the DoF of your camera and lens at a particular aperture setting, at a specific focal length, together with a "focus distance from the camera, have a look at this site. Depth of Field Table

".....[/FONT]DoF is entirely dependent on, and permanently anchored to, the point of focus......" This statement will be confusing to many, and fudge the issue when discussing DEPTH in the term Depth-of-Field. If ".....DoF is ENTIRELY DEPENDENT ON ... POINT OF FOCUS" are you saying that DoF is NOT dependent on aperture? I think not..... While it may be "permanently anchored to the point of focus", it is the DEPTH (short or long, and how long?) that should be under discussion, and not where in the image this in-focus portion actually occurs. [FONT=arial, sans-serif]

[/FONT]