[mrohank]

Hello fellow people!

I have a strange problem with my K-50 which I bought just recently: the focus assistance with my Helios 44M works strangely. Generally it is front focusing, sometimes back-focusing, and sometimes nailing the focus perfectly. So far I have tried the following after setting the camera up to maximum ISO of 400 (to avoid ISO-based blur), f/2 (as it is the minimum) and tried to focus as close as possible in many trials:

- I went to live view and nailed the focus there. Then, without moving the camera, I switched to the viewfinder. It showed the green dot (focus confirmation) sometimes, and did not in other times (I, of course, focused at the center of the frame).

- Checked all the steps with AF fine adjustment on both sides. When I went negative, there were more positive returns (I mean there is a front-focusing problem) but it was not consistent - sometimes -5 was what was the appropriate one, sometimes -7, sometimes -10 was not enough.

- Checked with different distances between me and the subject. The farther I went, the better the focus became - yet clearly was not good enough. But -5 / -7, as long as the subject was far enough from me, was sharp enough.

- Changed f/2 to 2.8 and 4. Of course with more DOF there was more accuracy, but the strangeness did not disappear.

Besides these, I went through the following thread, yet no solution still:

AF Fine adjustment - PentaxForums.com

So, any suggestions on what to do? What seems to be the problem right now is that at f/2, AF system goes kinda mad and cannot nail most of the shots, and as I move down, it gets kinda better (yet not reliable 10/10 times). I am used to using manual focus lenses already, but am also used to using AF assistance, and especially with close-ups and portraits I use AF assistance pretty much.

Thanks in advance for any recommendations, suggestions, ideas, or comments!

Ioannis

[victormeldrew]

As you have so neatly tested...the focus confirmation in the viewfinder is rather imprecise.

[jatrax]

If you are using the AF 'beep' as focus confirmation be aware that it is not particularly accurate. The 'beep' indicates a range that approximates the DOF of perhaps f/5.6 or f/8 so if you are using f/8 it will be reasonably accurate. At f/2 not so much.

Also this uses only the center AF point and that will pick up whatever is within it's view that has the most contrast. Depending on the subject and your distance from it this can vary considerably. The AF 'point' is not a 'point' at all but an area that can be larger than many expect.

As you already determined, Live view focusing will be the most accurate.

[Adam]

Yes, there is certainly a bit of tolerance when it comes to viewfinder AF confirmation. One technique you could use is to try to find the middle of the confirmation range. DoF and of course live view are probably more reliable, though, as others have pointed out.

Adam
PentaxForums.com Webmaster (Site Usage Guide | Site Help | My Photography)



PentaxForums.com server and development costs are user-supported. You can help cover these costs by donating or purchasing one of our Pentax eBooks. Or, buy your photo gear from our affiliates, Adorama, B&H Photo, KEH, or Topaz Labs, and get FREE Marketplace access - click here to see how! Trusted Pentax retailers:

[jatrax]

One other thought is that at f/2 the DOF is very thin and just the motion of pressing the shutter button can move the camera enough to ruin the shot. Proper stance and grip can minimize this but after having some narrow DOF shots not come out well I studied my own stance and noticed that when I pressed the shutter button I moved my head ever so slightly down and forward. Enough at f/2 to make a difference.

When you say: "without moving the camera" were you on a tripod or hand held?

[stevebrot]

As you have so neatly tested...the focus confirmation in the viewfinder is rather imprecise.

Yes, imprecise is the appropriate term. As hinted above, the precision reflects an ability to detect an out-of-focus condition the same at f/2 as at f/5.6. It is fairly likely that out of 20 focus attempts half will fall on one side of the intended mark and half on the other with the degree of variance being about the same on both side. In other words, what you (the OP) observed is what is expected and affects the PDAF auto-focus to a similar degree (same sensor...same problem). One can attempt global fine focus adjustment (the only kind available with manual focus lenses), but that will not improve precision.

That is cold comfort when one would prefer consistent plane of focus at the intended subject. One can improve their odds a little by using the optical view finder. Its ability to detect OOF is about f/4. Yes, the optical viewfinder is rather imprecise too.

Conventional wisdom is to use magnified live view where critical focus is required. Another, more involved, option is an aftermarket focus screen with a split-image focus aid. Depending on design, such a screen will reliably and consistently detect OOF to at least f/1.4 or even f/1.2. That precision comes at the price of the screen, the process of installation, and the possibility that the calibration might require adjustment with different thickness shims.

Sorry, I am such a ray of sunshine. So...What to do? Manual focus using the stock focus screen will yield quite acceptable results most of the time for hand-held shooting at moderate apertures. (The "sway factor" is real.) You will likely do better than the same subjects using an AF lens. When you want perfect focus, use a tripod and live view.


Steve

[Jeff Lopez]

you should select another WB setting, as I found this affects the green dot and AF. RiceHigh confirmed that, and I did some tests with a K30 and it was mostly true. There is a thread around that I brought from DP from a guy who describe the issue about certain light wavelenghts that screw phase detection focusing. It seems that the beep derives from what AF chooses as focused, instead the other way around..

[mrohank]

First of all, thank you all for the replies. I have kept those in mind and after receiving my Sigma 24mm 2.8, I tried to find out what the problem actually is. Although these might not be some new stuff for you, I wanted to share my findings here:

1- AF assistance works properly between f/2.8 and f/5.6 (or 8 the most). At 2 and 2.5, it is not reliable at all. This was the reason of back-focusing at times.

2- As the subject gets closer, it is more prone to front-focus. At around 2 meters and beyond, it is more consistent.

3- I expected the AF sensors to be smaller than the square on the viewfinder, but they are actually bigger. Somehow, strangely, right-upper side of my center AF point likes to work more and/or faster than the other areas. Also, even more strangely, that is the part which likes to be cheated and give false positives more.

4- Not exactly WB, but the light type has effects on it. In shadow, at home, however indirect sunlight it is, AF was more prone to fail than outside.

5- I have never used the beep, so I do not know how it reacts. But in the comparison of the green dot and the AF confirmation red square, I found red superior to green (well, saying this way, it sounded like Nikon is superior to Pentax ). The green dot has a much larger tolerance. Red square, though, is sometimes (or most of the times?) working only as a notifier of the green dot.

6- I realized that my eye got better as I tried to nail focus myself, and tested the system accordingly. More than half of the time I could find the correct focus when I was shooting close objects. Yet with the farther ones, my success level failed.

7- Lastly, AF is more precise depending on the distance between me and the subject and the side I am approaching the subject. If I am close, approaching from infinity and vice versa makes AF more precise and reliable. It is not scientific and just personal experience, though.

So thank you all for the answers again, they helped me!

[stevebrot]

you should select another WB setting, as I found this affects the green dot and AF. RiceHigh confirmed that, and I did some tests with a K30 and it was mostly true.

I am not in a position to test this, but would be very surprised if it were true or even mostly true, particularly if auto-WB is used. Auto-WB is assessed in real time during image processing (long after the AF has done its work). If using fixed WB, a high degree of diligence would be required to make sure that the setting accurately reflects the subject illuminance (hand held meter with spectral information?). FWIW, I have found RiceHigh to only be mostly right even when he is "right".

Out of curiosity, what WB setting would you suggest?

There is a thread around that I brought from DP from a guy who describe the issue about certain light wavelenghts that screw phase detection focusing.

This is a well-known characteristic of PDAF systems, but one that was dealt with some years back by most of the industry including Pentax (2010?).

It seems that the beep derives from what AF chooses as focused

Yes...When doing manual focus, the red indicator, the beep and the green hexagon all occur when the PDAF detects other than the out-of-focus condition (not the same as in-focus or focus attained).


Steve

[stevebrot]

But in the comparison of the green dot and the AF confirmation red square, I found red superior to green (well, saying this way, it sounded like Nikon is superior to Pentax ). The green dot has a much larger tolerance. Red square, though, is sometimes (or most of the times?) working only as a notifier of the green dot.

The beep and green hexagon happen when best focus has been detected. The red marker indicates the active focus point and is often lit at the same time as the beep and green hexagon (always when doing manual focus).

BTW...Welcome to the Pentax Forums!

Steve

[stevebrot]

7- Lastly, AF is more precise depending on the distance between me and the subject and the side I am approaching the subject. If I am close, approaching from infinity and vice versa makes AF more precise and reliable. It is not scientific and just personal experience, though.

It is best practice to do a mix of from-MFD* and from-infinity attempts when doing focus assessment.


Steve

* MFD = Minimum Focus Distance

[frizzz]

I would first propose to switch back TWO funktions of Reset.
You have to search them- i think on C3 or 4 and one other under a different function.
Then you will see more.....