AF microadjustment

Does any one around here really understand why this is needed?

Because if so, I would dearly love to understand a number of things.

Firstly, and most importantly, how can the way the AF system works out whether the image is in focus or not, give a different result for different lenses?

I could imagine, for instance, that if the AF sensors and the sensor are at different distances from where the light splits, you'd need to adjust it. If this was so, I do not understand why they couldn't do it at the factory. Perhaps it's something that can change over time, in which case, fair enough.

However, this does not explain how different lenses would cause the need for different amounts of adjustment. The sensor is flat, and the mirror is flat, so if the image at the sensor is in focus, why is a different amount of adjustment needed for different lenses?

I've wondered about this on and off for ages, but I'm now in the position of having a lens that back-focuses quite a lot, in the case of my first (now replaced) K-5, by MUCH more than the adjustment allows, and now with my second one, by just a little more than the adjustment allows.

I have yet to retest this with the latest firmware, but I can't see how it can be a software problem either.

So, any Pentax AF boffins out there, please explain if you can. I'm clearly missing something important, and I'd love to understand.

You just have to accept that it's true It baffles me
as well especially with manual focus lenses

Contrast AF under Live View seems very accurate and unaffected
by the AF microadjustment (K-5)

Make sure any tests that you do are under daylight.

Contrast AF works by doing more or less what you do as a human - it changes the focus by analysing the contrast of the image on the sensor. It doesn't know how far to change focus, but it keeps refining it until it is as sharp as it can get it.

Phase AF is an entirely different kettle of fish. It uses a separate sensor mechanism: Two micro-lenses capture the light rays coming from the opposite sides of the lens and divert it to the AF sensor, creating a simple rangefinder with a base within the lens's diameter. The two images are then analysed for similar light intensity patterns (peaks and valleys) and the separation error is calculated in order to find if the object is in front focus or back focus position. This instantly gives the exact direction of focusing and amount of focus ring's movement.
It is that final part which can give the error - although it can work out how much to move the lens (so it is fast) it doesn't really know whether the image is in focus - it only knows what the analysis of the split image told it.

Different lenses have different optical characteristics and some tolerance in the focussing mechanism, so telling the lens to go to focus distance X as computed from the phase AF analysis is not as good as it could be.

Hence the provision of an adjustment mechanism in the software.

Thanks for that info. It's a pity that they didn't indicate which
way to turn a MF lens in the viewfinder like they did with the
original MEF

I wonder if they ever build a camera with superfast and precise
focusing will they call it the MEF-D

Very clear. That was the point I'd been missing. Many thanks KM.

So.. going back to the lens that's back focusing more than the K-5 adjustment allows... what's best to do?

I've the same problem with some MF lenses. If you are talking about AF lenses you can use LV with Contrast AF.... it's very fast.

Well I was rather hoping for something that would be a real fix.. can it be calibrated in some way by the Pentax service dept?

Or has someone found the Debug mode yet, that might give access to more adjustment than +/- 10?

Or could it be the camera?

Sorry, I should have mentioned... it's a DA 17-70. I still haven't had a chance to retest with the new firmware, so I'll try and do that today.

Why would a manual focus lens need the AF adjusting? We turn the focusing ring until the image in the viewfinder looks sharpest and that's it.

Because the AF confirmation hexagon is just as useful with a MF lens as it is with an AF one - arguably more so in fact, since as you said yourself, many people struggle to visually identify the point of sharpest focus.

If the camera's going to tell you when the image is focus, it's nice for it to tell the truth, AF or MF.

Thanks Chris, I suspected that would be the answer.

It is true that many find manual focusing difficult. On the other hand, given that manual focusing does not need to be 100% spot on (DOField and DOFocus cover minor errors) and we used to get perfectly sharp images, AF is no better in many circumstances. But it doesn't need to be 100% accurate for the same reasons.

Macro photography is probably the exception, where ultra-critical focusing is necessary for best results.

I've always used MF with 6x6 cameras... although when it looks in
focus.. it's not really in focus from a critical POV that is

By definition anything within the DOF looks in focus and you
can't argue with that.... so if I have a subject at say 10m and
and the DOF is from say 6.5m to 20m.... anything between
6.5m to 20m is going to look in focus... I might
actually be focused at 6.6m not 10m and I lose the
DOF behind the subject, because the true DOF is
now from about 5m to 10m

This partly shows what a lot of theoretical nonsense DOF tables
are You simply can't focus accurately enough for
them to be any use.

The green hexagon is incredibly accurate, I've tested it. The
problem is transferring that accuracy to the focus ring on the
lens.

LV Contrast AF is also accurate, but no use with MF lenses.

I'm really struggling with 28mm - 35mm f/3.5 lenses because
they have a big DOF, that's why I tend to use INF + f/8 and
leave it there. If I try to focus the object I focus on is
usually OOF... possibly my screen needs shimming.

Another reason to use AF with a fully-manual lens is Catch In Focus.

The problem with DOF tables is that they assume a certain viewing size and distance of the final image, such that the out-of-focus areas are lost below the level of detail the eye can resolve at that range. If that image is blown up to 100% on the computer then the DOF will be considerably less since the image is now viewed at the pixel level. This reveals all sorts of errors in focussing.

On my GX10 I calibrated the AF until the focus agreed with the results from 100% images on the computer. However, I had to shim the focussing screen to make that agree, since otherwise MF gave results that were noticeably out (with the dioptre adjustment as correct as I could make it). Three tiny strips of Post-It note each side of the screen made it spot-on with the Samyang 85mm/1.4. That's about 0.25mm.

See this old thread.

With a 35mm f/2.8 lens I even tried using LV at 10X to find the two
points where a target a meter away went in and out of focus so that
I could take an average of the two which would have been near enough
the correct point of focus. I used a focus rack that had +/- 30mm
adjustment, but it stayed in focus all the way according to the
10X image on the screen.... so I need something with +/- 100mm
adjustment.

The green hexagon on the other hand was sensitive to about +/- 3mm

Actually this begs an interesting question regarding MF lenses.

Given that it's not the camera moving the focus to the calculated point, but the operator, would you expect the BF/FF errors not to occur?

Or at least be consistent from lens to lens on the same camera?

Another use for the green hexagon is with M42 lenses. If it works
OK you can leave the lens set at the working aperture i.e. f/5.6
or f/8 and still focus OK, because a lot of zooms are that aperture
anyway.

Try focusing a 28 to 50mm lens at f/5.6 visually and you will see
what I mean about DOF.... it looks in focus, but isn't when you
import the pic onto a PC.