[devorama]

There's been a lot of speculation about the size of the AF points on the K-3. One of my biggest complaints with the K-5 I own is that the focus points are so large that I'll often focus on an unintended object behind my subject due to the size of the focus points. I said my initial testing seemed to indicate the K-3 points were smaller, but the official review here said they were about the same. So I thought I'd set up a more controlled test.

I taped a piece of wrapping paper to some curtains hanging in my house. This was to provide a higher contrast background than the brown curtains. Then I used a lightstand and a boom to hold a high contrast focusing target about 3 feet from the curtain. My goal was to create a high contrast focus target, with a reasonably high contrast background. I would pivot my K-5 and my K-3 from the same tripod location with the same lens and try to get a rough idea of how far from center the target would have to be before the camera (set to center spot) would focus on the background instead of the projecting target. My test lens was a Sigma 24mm F/1.8 lens set to F/2.0.

I took hundreds of frames, moving the camera slightly up, down, left, and right to try to figure out the boarders of the AF zone. The results confuse me.

For the K-5, I estimated the center focus point to be approximately 200x200 pixels. This is of a frame height of 3264 pixels. That's about 6% of the image height.
For the K-3, I was astounded to measure an approximate center focus point of 520 pixels! For a frame height of 4000 pixels, this would be about 13% of the image height!

I checked my K-3 to make sure I didn't do anything crazy, let set the expanded AF zones or anything weird. But I can't find any setting out of the ordinary. Has anyone else tried such a test? Maybe I need a test where the background is of equally high contrast? Maybe the expanded AF abilities of the K-3 change how it transitions from one focus point to another? I'm really at a loss to explain this.

Attached Images

View Picture EXIF

File Date

 

[Adam]

When I did a similar test, I noticed that the center point was larger than all the other points. Have you tried the others in SEL mode?

[devorama]

Nope, I've only tested the center point. I didn't consider that they may be different sizes. I guess it might make sense since those center 3 points are F/2.8 sensitive ones. I guess for the purposes of portraits, the outer points are more important anyways. My subject's dominant eye I focus on is usually closer to the corner point. Back to the drawing board!

[rawr]

devorama, your test reminds me of a simple but effective AF spot focus test UnknownVT did on his K-x a few years ago.

[unkipunki]

My goal was to create a high contrast focus target, with a reasonably high contrast background....... Has anyone else tried such a test? Maybe I need a test where the background is of equally high contrast?

Not a test as such but maybe a little illuminating on this subject

I have been using LensAlign MKII to tune my lenses which involves aligning a specifically designed high contrast target . The alignment is achieved by aligning three small holes on a back plate (60mm behind) through three small holes on the font plate. These holes are ~2.5mm in diameter. This is using AFS and centre point with camera typically 30 to 40 x lens focal length form the target. Going through this alignment process with my K-5 and K-5iis I never got focus lock on the back plate through the small holes, but with the K-3 this happens about 20% of the time.

The final set up mounts a card without holes in it over the alignment plate so as to prevent accidental lock on the back plate during calibration Photography. But as I said I didn't ever lock on the back plate with either of the K-5 models. So I am not sure is this absolutely proves that the centre focus points are smaller on the K-3 but if not what other explanation is there. Seriously, I'm interested to know if something else could be at play with this.

On another note I have noticed the centre cross of the LiveView grid is not always the same spot through the view finder. It is often a little to the left, right, below , or above. This usually off by about the width of the focus square. I'm guessing that this is very slight variation in individual lenses, but again there may be some other explanation.

[twitch]

I think the test needs to be done with a high contrast small target and a zero contrast background (like a smooth plain coloured wall). This removes the variable of the camera making a choice of which target it wants to focus on if it has the choice of two targets. By having only one viable focus target you can see exactly where the outer reaches of the AF point is.

[skyer]

I read a review of K-3 and K-5 IIs owner. He wrote that he had made several tests and now it's obvios to him that K-5 IIs is able to autofocus in dimmer light than K-3.
Has anyone had the same experience?

[rawr]

K-5 IIs is able to autofocus in dimmer light than K-3.

I'd be interested to hear how he did his tests. The AF sensitivity of both is equal, at -3 to 18 EV, according to the specs.

However the K-3 can also meter exposure down to -3 EV, compared to the K-5II's exposure range of EV 0-22, which should (in theory) give the K-3 an advantage over the K-5II, since K-3 exposure info is actively fed into the AF system.

[skyer]

I'd be interested to hear how he did his tests. The AF sensitivity of both is equal, at -3 to 18 EV, according to the specs.

However the K-3 can also meter exposure down to -3 EV, compared to the K-5II's exposure range of EV 0-22, which should (in theory) give the K-3 an advantage over the K-5II, since K-3 exposure info is actively fed into the AF system.

Yes, manuals of K-3 and K-5 IIs say that the AF sensitivity of K-3 and K-5 IIs is equal. However, that owner of K-3 and K-5 IIs wrote that he pointed the cameras at the same objects that were not lit well (in almost darkness) with the same lens. He found out that in such circumstances the AF of K-5 IIs is much more responsive.
He also tested the center AF point of K-3 and came to the same conclusion as was written above - unfortunately, the center point of K-3 is even bigger than the K-5 IIs'. I asked him to test other points as well but he haven't tested it yet.

[jbinpg]

I really don't care about the size of the center AF spot on my K-3. All I know is that the focus is faster and more accurate on my K-3 than on my K-5 IIs, especially so in AFC.

Jack

[skyer]

I really don't care about the size of the center AF spot on my K-3. All I know is that the focus is faster and more accurate on my K-3 than on my K-5 IIs, especially so in AFC.
Jack

Would you, please, make a quick test regarding the AF sensitivity of K-5 IIs vs K-3?

[jbinpg]

If I had two of every lens that would be possible. But tests using different lenses on two different bodies intended to test the same operational mode make no sense to me. Actually some sun around today so I am heading out the door. Will post anything relevant when I return.

Jack