[stevebrot]

My experience is that refocussing once often improves focus accuracy.

Conventional wisdom is that the Pentax AF algorithms do a double check before signaling "least OOF" detected, though that does not preclude subsequent attempts once the mechanicals come to full rest.

It may well descend into a tight random well.

My thoughts as well, with the randomness being contributed from multiple sources.


Steve

[beholder3]

The stock focus screen on most dSLRs has a sensitivity of about f/4.0. The split image on aftermarket screens is sensitive to about f/1.4 at best

This is not correct, or at least I have never heard of anything close. "split image" prism focussing screens work exactly like computerized PDAF to the extent that I would call split prism focussing PDMF. The split prism is a single centered "line sensor".
And just like with electronic PDAF the F-something given is not "sensitivity" but describes a geometrical parameter: how far on the outside edges of the image circle do the two focus sensors look through the lens? It also means that any lens slower than this F-something inevitable will hard block light rays from hitting the focus sensor.
This then would automatically mean you can not use a "F1.4" split image screen with any lens slower than F1.4, since you would only see black split prisms.

BTW for the small split prism usage, very high "F-numbers" would be bad, since while it would allow very good precision, you'd be unable to see corresponding image parts unless you are already very close to correct focus.

[Lev]

What I suggested is the known element of imprecision as far as detection is concerned. Do the comparison again with the D FA 50 set to f/2.8 and the D FA 100 wide open to allow similar DOF to affect the final results. Remember to adjust the distance for the D FA 100 to provide equivalent magnification. The D FA 100 will be at a slight disadvantage, but I would expect AF precision to be similar.


Steve

Ok Steve thanks a lot I'll do the test. Thanks everybody!

[stevebrot]

This is not correct, or at least I have never heard of anything close.

There have been several threads over the years regarding the stock screen.

"split image" prism focussing screens work exactly like computerized PDAF to the extent that I would call split prism focussing PDMF. The split prism is a single centered "line sensor"

This is true. The difference is in the tuning of the prism design. The better split-image screens (e.g. Nikon K3, Katz Eye, and others) have the prism cut at a variable pitch to allow sensitivity at fast and moderated apertures while resisting black-out at narrower apertures. PDAF detectors address this issue by using two sets of prisms. I may have been off-base at stating an f/1.4 figure (not having an f/1.2 lens for comparison), but seem to remember claims to that effect at one point or another when doing research early-on regarding PDAF and 3rd-party screens.


Steve

[stevebrot]

And just like with electronic PDAF the F-something given is not "sensitivity" but describes a geometrical parameter: how far on the outside edges of the image circle do the two focus sensors look through the lens? It also means that any lens slower than this F-something inevitable will hard block light rays from hitting the focus sensor.

Yep, that is how it works. The f/2.8 detectors on Pentax cameras probably "black out" at an aperture probably not much narrower than f/5.6. The system deals with that by pairing the f/2.8 detectors with an f/5.6 unit trained to the AF point, extending functionality to that of the slower Pentax AF lenses.



Steve

[Lev]

What I suggested is the known element of imprecision as far as detection is concerned. Do the comparison again with the D FA 50 set to f/2.8 and the D FA 100 wide open to allow similar DOF to affect the final results. Remember to adjust the distance for the D FA 100 to provide equivalent magnification. The D FA 100 will be at a slight disadvantage, but I would expect AF precision to be similar.


Steve

Steve, you was right, done testing. Very similar results (DFA 100mm vs DFA* 50mm). Both at 2.8, slight deviation within focusing area for some amount of sequential shots and then boom out of focus. Then I made final tests with 50mm - 12 in focus, 3 out of focus.

[gm4life]

I think this thread explains what I witness once and a while with this lens. I thought it was that I was a crummy photographer or moving to much while open to F1.4... I still love the lens no doubt but I too get maybe 10 good and one bad... This AF thing could explain it and I double checked for front or back focus also... No adjustments were needed.

[pschlute]

I think this thread explains what I witness once and a while with this lens. I thought it was that I was a crummy photographer or moving to much while open to F1.4... I still love the lens no doubt but I too get maybe 10 good and one bad... This AF thing could explain it and I double checked for front or back focus also... No adjustments were needed.

The point is that while we may expect PDAF to be 100% accurate every time we focus, the truth is it is simply not.

Do a test with the camera on a tripod with centre point AF only, and an unambiguous high contrast target. Between each attempt, make sure the lens is starting from a closest focus or infinity focus position (hand in front of the lens works well for this or use quick-shift). No need to look at the pictures, each time the camera achieves focus look at the barrel position distance scale, or take an iphone picture each time. They will all be very close but you will be lucky to achieve 10/10 identical ones.

This is why it is so important when doing a Autofocus Fine Adjustment to do 10 tests and adjust your AF/FA setting to the majority vote, and not rely on a single focus acquisition.

PDAF is nearly 40 year-old technology. Add in the variables of sensor versus AF-sensor misalignment (hence AF/FA menu), and something as simple as a speck of dust on the AF sensor and you have even more scope for inaccuracy.

[Digitalis]

For a wider perspective on PDAF accuracy, even my Canon and Nikon cameras get things wrong. 2 or 3 frames OOF* is a common miss rate for PDAF. My lenses and camera bodies are painstakingly calibrated so that AF accuracy with PDAF is at least in the 90th percentile.



*even with a static subject.

[Lev]

Thanks guys you make my life a lot easier.

Well, I must say that this might be real advantage of mirrorless system, not the fact that you see what you get, not thousands of focus points across the frame, but accuracy.

[clackers]

Thanks guys you make my life a lot easier.

Well, I must say that this might be real advantage of mirrorless system, not the fact that you see what you get, not thousands of focus points across the frame, but accuracy.

Their PDAF is less accurate AFAIK, because it lacks large dedicated AF modules, they re-purpose selected pixels from the sensor so they obviously can't be big or there'd be blank bits in every picture taken.

The CDAF side of mirrorless has poor performance you can see by putting your DSLR into LiveView.

[pschlute]

Thanks guys you make my life a lot easier.

Well, I must say that this might be real advantage of mirrorless system, not the fact that you see what you get, not thousands of focus points across the frame, but accuracy.

Well it is not quite that simple. For sure there is no back/front focus issues with CDAF and overall the success rate should be higher than PDAF. But, CDAF is slower. Probably not a problem with static objects but for sports/wildlife it can be.

You also need to take charge with CDAF just as much as you do with PDAF. If your focal area is not large and in the centre of the frame you still need to adjust the cameras AF point/s to get accurate focus.

[Paul the Sunman]

I don't think anyone has mentioned the recent revelation that central point AF on the K1 is NOT the same as Spot AF. The latter is supposed to be more accurate (or was it sensitive?). Has anyone checked to see if the consistency of AF differs between the two?

[beholder3]

I don't think anyone has mentioned the recent revelation that central point AF on the K1 is NOT the same as Spot AF. The latter is supposed to be more accurate (or was it sensitive?). Has anyone checked to see if the consistency of AF differs between the two?

The recently found Pentax statements are on size of the central focus point cross, not on accuracy or sensitivity.
And I have to say that my personal impression does not match the official statement.

[Lev]

guys, just in case, has anyone tested this lens on focusing chart?