[Cubman23]

Amateur question(s), but what will AF adjust do exactly, and how can you tell if it's necessary?

Thanks in advance!

[icywindow]

It corrects for consistent incorrect auto-focus. If you notice with a given lens or all lenses that the autofocus consistently sets the focus either in front of (front focus) or behind (back focus), but when you manually focus the lens, it gives you an in-focus shot, you can use the AF adjust to adjust the autofocus such that it focuses correctly. There are focus charts out there to calibrate for just such a purpose.

[builttospill]

There is a tolerance allowed in the production of Pentax lenses. If the lenses fall within the acceptable range they get shipped and sold to you and me. Sometimes even though a lens passes inspection, it may not work as well as it could with your camera's focusing system. This doesn't mean it won't work perfectly on another camera, as there is also a range cameras should fall into as well, I believe. So in other words if the acceptable range is -5 to +5 and your camera is -3 and mine is +3 a lens that is -3 would work great on my camera but not so great on yours.

Focus adjustment allows you to dial in the lenses to work at their best. For charts and more reading (including examples of front- and back-focusing) see here: Nikon D70 Focus Chart

[ytterbium]

May i join this topic with probably stupid question. Still it has always bothered me. Why aren't lenses made so that they require no AF adjustment?

Usually, where it is not possible to deliver absolute fixed precision it is possible to use dynamic systems that can adjust to achieve precision.
Depending on what is the most common error source (Sensor distance, AF sensor alignment, Lens design), there should have been some auto calibration procedure.

---

Last edited by ytterbium; 08-21-2011 at 01:51 AM.

[builttospill]

That's a good question and have wondered myself, but I don't know the exact production process. A long time ago I saw a YouTube video about making a lens, but it was a very old video (if I remember correctly 60s or 70s) and it was quite the process.

Also, with the making of some lenses I don't know how easy it would be to automate correction. Does it make a difference that FA Limiteds are hand-built? Has anyone sent their camera and lenses in to Pentax to be dialed in and do they know how it's done?

[Marc Sabatella]

The camera is what does most of the work in the focusing, not the lens. It's the camera that "looks" at the image to decide if it's in focus or not, and it keeps turning the lens until it is. So really, it should never be the case that the lens could be at fault. But it's very easy for the camera to be - all that is required is for the AF sensor to be a slightly different distance from the mirror than the main sensor is.

There are certain special corner case situations where something in the lens itself could cause a focus problem, but these would be special unusual inconsistent focus problems, not the sort of consistent ones that are always the camera's fault.

For instance, if the focus gearing is off, so the camera turns the screw a certain amount to yield what it thinks will be proper focus, but the camera turns too far or not far enough in response, then the camera,s first attempt at focusing might be off. It should keep trying until it gets it right, but it doesn't necessarily. So it might stop slightly off. If ths is the case, then you'd see different results depending on whether the focus started out in front of or behind the subject. And you'd also see the focus get better with repeated trials. Hardly anyone ever reports this.

The other lens issue that could cause focus problems would be if the focus shifts as you stop down the lens. This can happen, but it would more likely be an unfortunate part of the design than any manufacturing defect. And the symptom would be, perfect focus wide open, with slight misfocus as you stop down. Of course, DOF increaes as you stop down, so you'd probably never notice this effect unless it was severe. And again, hardly anyone ever reports these exact symptoms.

EDIT: I keep waiting for someone to offer some other technical explanation for how a lens can possibly contribute to focus problems. I'm not saying there isn't some other technical explanation out there. But it's been like five years since focus problems first started to get attention, and as yet, I haven't seen any reason to believe the vast majority of focus problems are anything but misaligned focus sensors in the camera. The fact that people have sent lenses in for service or had them exchanged and reported better results is, to me, proof only that doing accurate testing is very difficult, and that the placebo effect can be very strong.

[Just1MoreDave]

EDIT: I keep waiting for someone to offer some other technical explanation for how a lens can possibly contribute to focus problems. I'm not saying there isn't some other technical explanation out there. But it's been like five years since focus problems first started to get attention, and as yet, I haven't seen any reason to believe the vast majority of focus problems are anything but misaligned focus sensors in the camera. The fact that people have sent lenses in for service or had them exchanged and reported better results is, to me, proof only that doing accurate testing is very difficult, and that the placebo effect can be very strong.

The newer cameras that can do Contrast Detect AF in Live View might prove your point. Isn't that using data from the main sensor itself? If CDAF works and phase detect has consistent errors, the lens is not the problem.

[ytterbium]

Focusing procedure must be know quite precisely to tell anything more detailed.

I must agree with mark, that for most part the AF depends on camera. But that just makes my question as to why cant they be made to auto-adjust even more solid. Clearly the final result is based on some sort of assumption by camera, that can be variable.

This is just some speculation. I don't know whats the exact math behind this, but i suspect that lenses with decentering defects, lenses whose optical path doesn't align with center of the mount (and focus/imaging sensors), have poor edge performance or complex light path might distort phase information in very odd ways.
But knowing that there are several AF sensors besides center one makes me believe that optical path alignment isnt that critical. Still not putting any bet on this, there might as well be compensation applied that relies on lens being centered relative to the mount/sensor.

Imagine if the optical axis is offset up or down, how the registered phase shifts are offset. What would happen if the optical axis isnt perfectly perpendicular to the AF sensor? This all can happen in lens.

Well thats just one lens related property i could think of. Not even sure if it can affect AF, even if it does, i don't see how some software calibration could cure it. Unless the offset is fixed, id expect it to require lens repair or replacement.

P.S. I like this discussion. I've had a lot of AF trouble in past. I hope it keeps alive and qualitative.

[builttospill]

Marc, I'm not disagreeing with you--please don't think I am, especially since your post is completely logical and I think you are right, to a point. I don't have the technical details you might be looking for, but my next question would be why does Pentax make a camera that stores correction information for multiple lenses if it's just the camera?

If I'm correct the K10D stores focus adjustments for up to five Pentax lenses, and I think they upped it to 20 lenses with the K20D. I don't know the numbers for the K-7 and K-5 as I didn't follow them as much. But logic here suggests it may be a combination of the camera and lens contributing to a focus error, right?

[ytterbium]

P.S. Just thought of another possible defect i didn't mention. Situation where lens elements are centered on the same optical axis, but optical axis of each group isn't parallel to main axis.
I'm just not sure if it's not the same as centering defect. Here i have drew some of the cases i'm thinking of:




Another question is, if any of those defects can produce significant AF error while retaining small impact on overall picture quality. Shifts certainly can remain unnoticed.

[icywindow]

Just a thought, but I wonder if any lens aberrations could account for some of this? Some could be more visible but would not add much to focus error - coma for example would just produce halos - while others might be less detectable except for focus (astigmatism due to a mismachined element? misalignment?).

[Marc Sabatella]

Marc, I'm not disagreeing with you--please don't think I am, especially since your post is completely logical and I think you are right, to a point. I don't have the technical details you might be looking for, but my next question would be why does Pentax make a camera that stores correction information for multiple lenses if it's just the camera?

A misaligned focus sensor would probably cause the magnitude of the error in practice to differ from lens to lens, or at least, according to focal length. Also, if the spherical aberration thing is going on (the issue that causes focus to be off when stopped down), then might be something you'd only need to correct on certain lenses.

Dave has a very interesting thought - are there people with lenses that consistently show misfocus when using live view? I can't recall ever seeing any posted tests that demonstrate this.

[icejam]

My K-r is suffering from FF problem only with one lens - Sigma 100-300mm. I will print some more focus charts tonight and see if the problem persists with LV focus.
But thew fact that problem itself only exist with this one lens points to lens being at fault, not the camera.
ytterbium may be just right with his explanation.

[Marc Sabatella]

I don't really understand how those pictures lesn defects would cause AF to be off if the sensors are aligned correctly. After all, whatever defect the focus sensor sees is the same defect the imaging sensor sees. If the image is correctly in focus at the AF sensor, how could it not be at the image sensor, since they are the same distance away from the focal point?

Here's a hypothesis - the chromatic aberration of the lens is such that different wavelengths of light reach focus at different points. So the red might be in focus at one position of the focus ring, but blue might be in focus a fraction of a turn away. Then, tf the AF sensors happen to be more sensitive to red but the scene itself is more blue, then we'd perceive the scene as being out of focus if focused acording to red.

If this were the case, then the perception of misfocus would depend on the color of the scene. That would be an interesting test for someone to perform. We already know this does happen with respect to strongly yellow light like tungsten, which is why some cameras (eg, K-5) correct for this specifically with an additional color sensor in the AF system.

But whether this could happen to any noticeable extent for colors exposed by daylight is an interesting thought to consider. It also seems that any such problem would make manual focus equally problematic in scenes that contained a mix of colors - you'd notice your red objects in focus but your blue ones not in the same scene, or vice versa.

Anyhow, again, I'm nto saying that I absolutely believe no lens has ever been the cause of a focus problem. But I think you get the sense of why I'm so skeptical. That plus the fact that I know from experience how hard it is to consruct a properly controlled focus test, and we've all seen some of the more obvious mistakes posed in these forums over and over again (eg, ambiguous focus target, test chart skewed from shooting axis, shooting in tungsten light, not using a tripod, not accounting for field curvature, etc).

As for the fact that you only see the problem with one lens, that still doesn't prove anything in itself. A small error in the AF sensor alignment would only show up with very shallow DOF, which you will tend to see with only your fastest and/or your longest lenses. Also, such errors would likely be focus distance dependent. Meaning you might not ever be able to see the problem when focusing on a chart at half a meter, but you might see it at three meters. Since you'll tend to shoot tests at different distances depending on focal length, that can also come into play.

[debmalya]

Here is a link that demonstrates how passive phase detection auto focus system works: Autofocus: phase detection