[pcarfan]

Falk,

I read somewhere that you are studying this problem and will be revealing your findings. Is this true?, if so when can we expect the results?

[jsherman999]

I do know from personal experience that if I aim my camera at the eyes, it's likely the camera won't focus on the eyes.

.

Do you find this acceptable?

.

[Christine Tham]

So, you cannot say the AF sensor focussed on feature x. It's not as simple as this. To further complicate things, a vendor can experiment with functions, e.g., replace them by ci^2 etc. Unmatched contrast edges make a correlation function small which is why they matter. But all contrast variations within an AF sensor's area contribute to the phase decision to a smaller or larger amount. It's not feature x like pupil or spoon.

I agree, and pretty much said the same thing:

I think we keep forgetting and imagine the AF sensor is seeing what we are seeing, but at the end of the day it is just registering voltage changes.

In the photo that jolepp posted - the spoon seems like the brightest object in the photo with contrasty edges hence may potentially influence the focusing.

It can be argued whether it falls within the sensor line or not.

Actually, what I've found quite interesting is that that the sensor can sometimes be influenced even by presence of high brightness/contrast well away from the sensor line.

For example, I mentioned previously my disappointment with the K10D focusing on a bright sunlit patch of grass even though my target (cyclist) covered the entire centre AF sensor area.

I had another similar experience yesterday when I was doing AF fine focus adjustment on my lenses.

I was careful not to have parts of the Lenscal exposed to direct sunlight (which in Australia is very very bright). However, towards the end, a bit of sunlight did creep in to the upper right of the frame.

Even though I was using centre AF, and the sunlight was on the top right well away from the sensor lines - I discovered it was enough to throw the focusing off by a few mm. Shifting the Lenscal away from direct sunlight once again restored the focusing.

Note - this is reflected sunlight, not sunlight hitting the lens.

[Gimbal]

I don't know if anybody is interested in this.

Normally, what a phase AF does, is this:

it takes the two 1-D images from the two sides (left/right or top/down) of an AF sensor's line sensor. It ends up with two curves ci(x) from each sensor. The curves would have large values for bright regions and low values for dark regions (normally, ci is defined such that Integral_x ci(x) dx = 0 but it doesn't matter here).

Now, it computes the so-called cross correlation function between both curves, where one curve is shifted against the other curve by a so-called phase p: Integral_x c1(x)*c2(x-p) dx

In a quick optimization, it detemines p such that this is maximized. (Actually, this isn't fundamentally different from what a contrast AF system does. Which is why I sometimes don't differentiate between phase and contrast AF, much to people's surprise normally )

This value of p is then used to determine the necessary lens shift to aquire focus.

The important point is this:

Whatever p maximizes Integral_x c1(x)*c2(x-p) dx is taken.

So, you cannot say the AF sensor focussed on feature x. It's not as simple as this. To further complicate things, a vendor can experiment with functions, e.g., replace them by ci^2 etc. Unmatched contrast edges make a correlation function small which is why they matter. But all contrast variations within an AF sensor's area contribute to the phase decision to a smaller or larger amount. It's not feature x like pupil or spoon.


Sidenote:

In the presence of noise, the computed value of p starts to be noisy too up to the point where it becomes a random value. An AF module has to decide how much light it receives (in order to take longer captures with its AF sensor) and if it can trust the value of p. Both are sources for mistakes being made by the firmware of the AF module if it is autonomous.

What I haven't understood yet is what happens next. We have a value from the AF-sensor that tells us that the lens needs to be adjusted a certain amount, and the lens has told us something about it's internal gears so that the camera can calculate how much it should turn the screw to achieve focus. (Assuming screw-drive driven lens). This is my understanding from surfing the net so it might be totally wrong.

Now the weird thing. If the lens isn't perfectly calibrated we might end up out of focus (that is reasonable). But if we press AF again the camera signals green as in perfect focus, this I don't understand.

The AF-sensor must see the phase shift but still it signals that we are in focus. If at least the lens jerked back and forth but no, it signals perfect focus immediately.

(This is with a "faulty" lens that needs adjustment and is not related to the K-5 and the FF issue. I'm just wondering how the AF works...)

I must be missing a link in this chain.

[dcmsox2004]

for falk......... please indulge my ignorance as well as my desire for a k-5........

every time i read about this i think i need to enroll in a class at MIT or cal polytech.... i find the nuts and bolts of the focus and it's related camera components most interesting, but what a brain drain.... i hope i'm not interpreting these posts as a 'how to' guide for an end user in overcoming the flaws of the k-5..... to be frank, this is why i add the k-5 to my b+h cart and then walk away from it..... scares the bloody hell out of me... either the k-5 will focus properly in low light, be it natural of man made, or it won't..... seems like all the manipulation with the settings for the k-5 still can't overcome it's inherent failings...
but... other k-5 owners state they have NO problems whatsoever... how can this be.... is it a software issue or mechanical problem only affecting a few random bodies?????
i'm even willing to have my wife kill me if i order the new silver surfer version about to be released... i'd be stunned if this 1699 beauty also exhibits ff................
just trying to assimilate this info so i can make an informed decision....
is that possible for a guy who studied fortran in college in 1974 ?????
dave m

[falconeye]

I read somewhere that you are studying this problem and will be revealing your findings. Is this true?, if so when can we expect the results?

as soon as I find a couple more hours to finish it.

[falconeye]

is that possible for a guy who studied fortran in college in 1974 ?????
dave m

Dave, my post above probably wasn't meant for you. It wasn't K-5 related at all. I tried to explain the inner workings of a single AF point sensor for the curious mind. I assumed prior knowledge of Correlation function (statistical mechanics) - Wikipedia, the free encyclopedia which I guess was a bit over the edge

[areidjr]

as soon as I find a couple more hours to finish it.

Well, stay out of here and finish it.

[falconeye]

Well, stay out of here and finish it.

Well, currently, I am in office and hunt people electronically to do their job (and it's 7:40 pm around here but deadlines are mercyless ...). To have a look in here from time to time between phone calls and mails helps me to keep some sanity To finish my report needs a quieter time

[areidjr]

Well, currently, I am in office and hunt people electronically to do their job (and it's 7:40 pm around here but deadlines are mercyless ...). To have a look in here from time to time between phone calls and mails helps me to keep some sanity To finish my report needs a quieter time

Of course I was only joking.

I pop in from time to time when I need a short break to clear my mind. Sometimes reading the posts here do more harm than good

I am anxious to read your report, BTW.

[Class A]

The AF-sensor must see the phase shift but still it signals that we are in focus. If at least the lens jerked back and forth but no, it signals perfect focus immediately.

This is because the AF-sensor doesn't observe a phase shift anymore. The subject is in focus, as far as the AF-sensor is concerned.

In a closed feedback loop the lens focus will be adjusted until focus has been achieved (excluding exit conditions to be used when the latter is impossible). So "worn gears" or other similar explanations that are sometimes used don't apply for explaining significant amounts of FF/BF.

The problem is that if the AF-sensor "sees a sharp image" this doesn't mean that the camera sensor does as well. Potential reasons for this include a) misalignment in the camera (will be consistent for all colour temperatures) and b) the AF-sensor is most sensitive to green while the camera sensor records the full spectrum. Lenses have chromatic abberations and (sometimes depending on the f-ratio) don't always focus all frequencies into the same focus plane. If an image is predominantly "red" then the lens doesn't focus the majority of the image into the same plane as the one that the "AF sensor picked" to be sharp.

[Rupert]

I could be wrong......but I can't seem to find any K5 shooter here that has sent his camera to Pentax Service with a focus issue and received it back repaired/not repaired? There may indeed be a repair for this, I see that on the Nikon D7K Forum there are several reports of focus issues like the K5 that were repaired by Nikon, and the shooters were well satisfied.

Surely someone is going to try this instead of just going for a new body every time they get a bad focusing camera? Before I give up on the K5, I would like to hear from some that have actually given Pentax a shot at making a repair. We are assuming firmware, but it could be that all that is needed is some hardware fix or some tweaking at a Certified Service Center?
Regards

[Smeggypants]

Apparently over two weeks ago there was a firmware fix with Pentax Germany that was going to be, after some beta testing, released to the public in two weeks.

IOW it should have been with us now.

I'm fed up with tackling FF deniers and doing more tests. I just want a fix. It's about time Pentax opened it's mouth. Release a firmware fix or tell us what you are going to do about this fault.

I've been holding off sending my K-5 back for replacement/Repair becuase of the alleged FW fix. I can't hold off forever.

[creampuff]

Apparently over two weeks ago there was a firmware fix with Pentax Germany that was going to be, after some beta testing, released to the public in two weeks.

IOW it should have been with us now.

I'm fed up with tackling FF deniers and doing more tests. I just want a fix. It's about time Pentax opened it's mouth. Release a firmware fix or tell us what you are going to do about this fault.

I've been holding off sending my K-5 back for replacement/Repair becuase of the alleged FW fix. I can't hold off forever.

Seriously I wonder why you don't take positive steps and have a refund or an exchange. In another post you indicated that you discovered the focus issue on your camera body within a few hours of getting it. So what is stopping you from getting your presumably defective camera swopped for another one or sending it in for service? It seems the most logical step would be to replace your dud copy rather than hope for a supposedly rumored firmware fix.

[mtngal]

Falk, while I don't understand much of what you wrote, I found it very interesting. Thanks for posting it, I'm going to have to think more about what you wrote and see if I either need to approach my camera's possible AF problem differently. Also the discussion about cameras focusing on something unintended because of light source changes and so on give me a lot more to think about also.