How can front(back)-focusing be a problem of the lens?

[wuyongzheng]

I think front(back)-focusing is purely a problem of the camera body, because the body controls and evaluates the focusing. The lens doesn't know whether it's focused or not, it is the body which does the quality control.

However, I see lots of complains on the lens having front(back)-focusing problem. Some even say that one lens has focusing problem but other lenses are correct on the same camera body. I don't understand this. It's like saying "This lens only shoots 5 megapixel image."

Of course, if the AF screw or in-lens motor fail, the lens cannot focus at all. But that's different from front(back)-focusing where the body confirms the focusing to be succeed.

Another possible related problem is the lag of the AF mechanism in the lens. When the body says "stop", the lens continues moving for a small amount of time. This can be fixed by half-pressing the shutter a few times. However, some people complains their lenses to be consistently front-focusing, and this method doesn't help at all.

Could someone enlighten me on this? What are the possible lens characteristics which cause front(back)-focusing?

 

[dRebelXT]

Take a ruler, place it flat on the table, put the camera in the same direction as the ruler and set to F/2.8 (Must have very narrow DOF), manual focus to 10cm line, take a steady picture.
Inspect the picture, if it shows 9cm is in focus instead of 10cm, it's front focus.. if it shows 11cm, then backfocus.
It's due to the mechanical part in the lens which slides too far or less.

 

[rendition]

Testing it like dRebelXT mentioned in a controled environment is one way... another long term way is... if you start seeing all your pictures OOF even though you die-die know that you've focussed on the subject that's supposed to be sharp. You can tell when a lens has back/front focusing, it's not the usual 'not-so-sharp' look but just purely off focus, and again, it should apply to all the photos taken on that same particular lens. If it's just one or two then it's prolly human error...

 

[scorpioh]

Hmm, think his question is not really answered.
From what I think is a logical expalanation:
You must first understand how the camera works. There is a mirror in the body which allows you to see through the lens from your view finder. This mirror is a translucent one as at the same time, it allows an alternate light path to pass to the bottom of the body where the AF unit is. When you snap your picture, the mirror slaps up and allows the image to fall on the sensor.
a) focus issues with AF
If the distance from the lens to the AF unit is not equal to the distance to the sensor plane, you'll get your focus issues.
b) focus issues with MF
Similarly, if the distance from the lens to the focus screen is not equal to the distance to the sensor plane, you'll get focus issues as well.
Such focus issues arises even with very small distance disparity and are hard to control during the manufacturing process. The best way to correct them is to use bodies which offers focus offset corrections.

 

[nstclicks]

I've always wondered about this myself too. scorpioh's explanation about distances between focusing screen/autofocus unit and sensor are all in the camera body. Which means it's the body's fault Maybe it has more to do with the angle at which light strikes the sensor, rather than the distances, and that may be more sensitive to each individual lens.

I suppose the best way of looking at it is that is the system's (body + lens) fault. Manufacturing tolerances mean that sometimes some combinations of body and lens will be a bit off. Possibly tolerances on lenses are less stringent than on bodies so it's conventional to set the offset on the lens rather than on the body. I'm sure some poor sods have lenses that have to be recalibrated for every body.

The interesting question would be whether it's possible to have good focus in live view but bad focus when taking the photo.

 

[scorpioh]

I've always wondered about this myself too. scorpioh's explanation about distances between focusing screen/autofocus unit and sensor are all in the camera body. Which means it's the body's fault Maybe it has more to do with the angle at which light strikes the sensor, rather than the distances, and that may be more sensitive to each individual lens.

I suppose the best way of looking at it is that is the system's (body + lens) fault. Manufacturing tolerances mean that sometimes some combinations of body and lens will be a bit off. Possibly tolerances on lenses are less stringent than on bodies so it's conventional to set the offset on the lens rather than on the body. I'm sure some poor sods have lenses that have to be recalibrated for every body.

The interesting question would be whether it's possible to have good focus in live view but bad focus when taking the photo.

Erm, not really. The distance is measured from the lens itself. So both the body and lens could be at fault. If live view relies on imaging on the sensor itself, then I don't think a photo could be off-focus once focued on live-view.

 

[wuyongzheng]

Erm, not really. The distance is measured from the lens itself. So both the body and lens could be at fault.

Could you explain a bit? I have the same doubt as nstclicks. Let's take the image here (http://www.leongoodman.com/d70focus.html, the first image titled "Nikon DSLR focusing system") as an example. If the length of the yellow ray and the purple ray are different, no matter how you move the lens, they are different. If they are the same, they are always the same regardless of the lens. So this should be purely problem of the body.

 

[scorpioh]

Could you explain a bit? I have the same doubt as nstclicks. Let's take the image here (http://www.leongoodman.com/d70focus.html, the first image titled "Nikon DSLR focusing system") as an example. If the length of the yellow ray and the purple ray are different, no matter how you move the lens, they are different. If they are the same, they are always the same regardless of the lens. So this should be purely problem of the body.

Hm, u make sense, I am loss now. Ha.

 

[calebk]

The problem lies in the communication between the lens and body. The lens first transmits optical information to the body through the lens, and the body determines the focus distance for the subject to be in focus. The body then transmits this data to the lens, but if the lens is miscalibrated, when it focuses to what it thinks is the correct focus distance, it will, in fact, err from its intended correct focus distance.

For instance, if the body determines that should the lens acquire a focus distance of 3.5cm, it will render the subject in focus, but when the data is transmitted, the lens thinks that 3.5cm is 3.3cm on its distance scale, and goes to 3.3cm - you have a front focusing lens.

What calibration does is ensure that the lens and the body are on the same page. When the body says correct focus distance is at 3.5cm, the lens does not go to 3.3cm and think that it is the correct distance.

 

[wuyongzheng]

For instance, if the body determines that should the lens acquire a focus distance of 3.5cm, it will render the subject in focus, but when the data is transmitted, the lens thinks that 3.5cm is 3.3cm on its distance scale, and goes to 3.3cm - you have a front focusing lens.

I doubt SLR's AF works this way. Are you talking about the active AF system which is mostly applied on point-and-shoot? How does the body know that the subject is at 3.5cm in the first place?

 

[travelfotografer]

AFAIK, that is how phase-detect AF (the primary AF system on SLR) works. The phase-detection sensor knows how much and in what direction out of focus is the area under the AF point. The instructions of how much to change the focus and in what direction is sent to the focus motor. So if lenses are not calibrated correctly, and when given an instruction to move so much, but ended up moving a little more or less because of bad calibration, FF/BF can come from the lens.

How does the body know that the subject is at 3.5cm in the first place?

 

[kaikibbler]

It does. The misconception is in thinking that the camera AFs using the quality of the resulting image: That's only true for contrast-detect in liveview, and it's unacceptably slow for example for sports, birds, etc.

In a sense, the body has the functional equivalent of a rangefinder, like one you might use for golf.

I doubt SLR's AF works this way. Are you talking about the active AF system which is mostly applied on point-and-shoot? How does the body know that the subject is at 3.5cm in the first place?