K-1 AF accuracy with FA lenses

scopedude · 05-14-2016, 05:58 AM

My first post here, I guess

I just received my K-1 yesterday, so I'm still getting myself familiar with it. I also got the DFA 28-105, which is an excellent lens. I was offered to test 2 samples, one is well centered (mine), and the other one is slightly off. I also tried the DA 35/2.4 in FF mode, 50/2.8 Macro, FA 50/1.4, and FA 43/1.9.

The newer lenses, DA 35/2.4 and DFA 50/2.8 macro performed well in term of AF accuracy. The DA 35 works well in FF mode too.
The older ones, FA 50/1.4 was front focusing, but after some adjustment in AF tune I managed to get it focused well (at least at close and medium distances).
The FA 43 Ltd, however, is erratic. It was front focusing as well, but the AF tune could only be accurate at certain distance only. If I adjusted it for close range, it will be off at farther distance, and so on.

My question is, how is your experience with these older FA lenses in term of AF accuracy? I want the 43ltd so much but I'm really put off with the AF issue.

Sorry for my English

Thanks!
Winston

repaap · 05-14-2016, 07:35 AM

My FA 43 have been little hard to focus at near closest focuspoint, but other than tha it has been great and fast.

6BQ5 · 05-14-2016, 07:50 AM

I thought I read that the focusing system in all Pentax DSLR's is tuned to an ideal aperture of f/2.8. Anything faster or slower takes it away from optimal performance.

Please correct me if I am wrong.

aremmes · 05-14-2016, 08:53 AM

Originally posted by 6BQ5

I thought I read that the focusing system in all Pentax DSLR's is tuned to an ideal aperture of f/2.8. Anything faster or slower takes it away from optimal performance.

Please correct me if I am wrong.

Possibly true, but if so I'd wonder why the cameras do not stop down fast lenses when focusing, which I imagine would be a cheap and easy optimization.

stevebrot · 05-14-2016, 09:00 AM

Originally posted by 6BQ5

I thought I read that the focusing system in all Pentax DSLR's is tuned to an ideal aperture of f/2.8. Anything faster or slower takes it away from optimal performance.

Please correct me if I am wrong.

You are correct about the tuning, not correct about the models, and not quite right about the performance implications.

For PDAF, the center column focus points on current flagship models have focus sensitivity of f/2.8. The rest are f/5.6. This is fairly industry standard for enthusiast and pro model dSLR bodies. The Pentax consumer models are f/5.6 for all points. What this means with the center column is that the system has the same ability to detect an out-of-focus condition at f/2.8 maximum aperture as with a faster lens and better than with a slower lens. The fast glass penalty is that you have poorer precision for fine focus than you might with a more sensitive focus system. Notice the points of emphasis for that last sentence. Use a maximum f/2 lens and f/2.8 lens and both will have about the same number of shots lacking acceptable focus than you would using magnified live view.

Put another way, when using the center column points, the system rewards faster maximum aperture up to f/2.8 with no improvement beyond that point.

Steve

repaap · 05-14-2016, 01:41 PM

I could do this wit no trouble.

FA 43

scopedude · 05-14-2016, 04:40 PM

Pentax distributor told me they have more FA43 in stock and I could test more next week. Hopefully I could find the one that works. I tested the FA 50 more and I did find that it is more difficult to obtain consistent AF at close range.

robjmitchell · 05-14-2016, 04:43 PM

I find my 43ltd needs to be adjusted with a mid distance target (where you use it for portraiture) otherwise it will be off. Once its dialled in on a new camera I don't have a problem. I suspect the fast Aperture plus focusing at short distances throws the adjustment off compared to the rest of the range.

scopedude · 05-14-2016, 05:03 PM

Noted. Will do this when testing the new unit. They brought me silver while I asked for the black one

Inspireart · 05-14-2016, 07:05 PM

I will ordering a K-1 and a 28-105 in several weeks. How did you detect the centering issue on one of the 28-105 lenses? I am not an optical engineer and am worried about getting a poor example.

Thanks, Bruce

Class A · 05-14-2016, 07:34 PM

Originally posted by 6BQ5

I thought I read that the focusing system in all Pentax DSLR's is tuned to an ideal aperture of f/2.8. Anything faster or slower takes it away from optimal performance.

Please correct me if I am wrong.

There is no "tuning" to f/2.8.

Some of the AF areas are capable of resolving at an AF base of f/2.8.
That means they provide higher precision than the standard f/5.6 areas.

Lenses faster than f/2.8 are a challenge for the AF system but to the best of my knowledge there is no "brickwall" characteristic.

Faster lenses often have a number of aberrations and small manufacturing faults have exaggerated effects. In understand AF systems struggle with that more than the fact that these lenses are faster than f/2.8.

With respect to lenses slower than f/2.8: As long as a lens does not exhibit a so-called "focus-shift" phenomenon, an AF system will not be affected, i.e., there is no AF performance penalty for shooting at apertures higher than f/2.8. Again, to the best of my knowledge, modern lenses contain calibration data regarding their focus-shift behaviour so the camera is able to compensate for known focus-shifts.

If a lens does not quite exhibit standard performance, all regular computations and compensation mechanisms are off and one can get AF problems. Such non-standard behaviour is more pronounced on very fast lenses, hence they represent a bigger challenge.

Originally posted by scopedude

I tested the FA 50 more and I did find that it is more difficult to obtain consistent AF at close range.

I don't think you should expect too much AF consistency with this lens at wide apertures and close range (both factors contributing to very shallow DOF). My K100D can struggle very hard to lock focus with it. Haven't used it on my K-5 II in such situations yet.

scopedude · 05-14-2016, 09:45 PM

Originally posted by Inspireart

I will ordering a K-1 and a 28-105 in several weeks. How did you detect the centering issue on one of the 28-105 lenses? I am not an optical engineer and am worried about getting a poor example.

Thanks, Bruce

Hi Bruce,

I always do at two distances, first at infinity wide open, and minimum distance. To make it easier at close distance I take photo of long 24" ruler. It will be easy to detect if the focus plane shifts front and back from left to right.

These tests are also useful to see if the lens is optimized at short or long distance. For example, canon 24-70 L ii tends to be flatter at short distance but has curvature at infinity that doesn't get better even at f8.

In case of DFA 28-105, the two samples behaved differently on the ruler test

Winston

Class A · 05-14-2016, 09:58 PM

Originally posted by scopedude

To make it easier at close distance I take photo of long 24" ruler. It will be easy to detect if the focus plane shifts front and back from left to right.

I don't think it is "easy" to ensure that the sensor is perfectly aligned to the ruler. Commercial tools use dedicated alignment constructions for this purpose.

A better decentering test is to chose an object at a fair distance and put it in all four image corners in consecutive shots (using manual exposure & manual focus fixed for all shots, and a tripod).

stevebrot · 05-14-2016, 10:31 PM

Originally posted by Class A

i.e., there is no AF performance penalty for shooting at apertures higher than f/2.8.

You are correct, there is no brick wall hard limit in actual practice for most subjects. The penalty is if you need critical focus when shooting at apertures wider than f/2.8. The precision (probability of having the focus plane coincident with the focal plane) will be no better than if you were doing the AF at f/2.8. The same range of focus is being detected by PDAF as being "not out of focus" as with a slower lens. This is a matter of physics and how the sensor works and is not related to the quality of the lens. The system is quite literally "blind" to the focus lapse, though in a fuzzy way.*

The easiest way to demonstrate is, unfortunately, not available on the K-1; using the PDAF focus confirmation in real-time comparison to a well-calibrated split-image rangefinder. Moving slowly through the focus point with an f/2 or faster lens at 50mm or longer focal length, the PDAF system will attain focus before the rangefinder images mate and will continue to indicate focus for a fair distance the other side after they diverge. Equivalent comparison is possible without the split image, though more clumsily, by making a best effort using PDAF focus confirmation and do the comparison by switching to magnified live view. Missed focus should be readily apparent. Do a trial of 10-20 attempts to get an idea of the real world impact.

Steve

* Probably the best resource I have found that illustrates in fairly simple terms how split-image, microprism, and PDAF systems work is the PDF article by Doug Kerr. Yes, all three systems work on the same principles and all three have similar limitations. The first parts are very interesting, but for those who lack patience, the PDAF stuff starts on page 12 and that dealing with aperture dependency (balancing black-out against precision) on page 17.

http://dougkerr.net/Pumpkin/articles/Split_Prism.pdf

scopedude · 05-14-2016, 10:58 PM

Originally posted by Class A

I don't think it is "easy" to ensure that the sensor is perfectly aligned to the ruler. Commercial tools use dedicated alignment constructions for this purpose.

A better decentering test is to chose an object at a fair distance and put it in all four image corners in consecutive shots (using manual exposure & manual focus fixed for all shots, and a tripod).

It's meant to be a practical way to test the lens. There is a grid in K1's viewfinder that would really help. I usually redo the test at home with a tripod like you described. That said, using the ruler with the VF grid is not difficult at all - because when you find a decentered lens you'll be surprised at how far the focus plane shifts. In case of the 28-105, the left frame was in focus in front of the ruler, and the right frame was way behind the ruler. Very noticeable. When I did the long distance test, the left and right frame showed different sharpness too. Pentax people came to my office with 2 K1s and 2 28-105s, and some other lenses so I had to decide which one to take home

I've been doing it for years after getting less than ideal samples (many of them) from Canon and Zeiss ZE/ZF. I'm doing mostly landscape, architecture and some astrophotography that's why it's important for me.

05-14-2016, 05:58 AM	#1
scopedude Junior Member Join Date: Apr 2013 Posts: 43	K-1 AF accuracy with FA lenses My first post here, I guess I just received my K-1 yesterday, so I'm still getting myself familiar with it. I also got the DFA 28-105, which is an excellent lens. I was offered to test 2 samples, one is well centered (mine), and the other one is slightly off. I also tried the DA 35/2.4 in FF mode, 50/2.8 Macro, FA 50/1.4, and FA 43/1.9. The newer lenses, DA 35/2.4 and DFA 50/2.8 macro performed well in term of AF accuracy. The DA 35 works well in FF mode too. The older ones, FA 50/1.4 was front focusing, but after some adjustment in AF tune I managed to get it focused well (at least at close and medium distances). The FA 43 Ltd, however, is erratic. It was front focusing as well, but the AF tune could only be accurate at certain distance only. If I adjusted it for close range, it will be off at farther distance, and so on. My question is, how is your experience with these older FA lenses in term of AF accuracy? I want the 43ltd so much but I'm really put off with the AF issue. Sorry for my English Thanks! Winston

05-14-2016, 09:00 AM	#5
stevebrot Otis Memorial Pentaxian Join Date: Mar 2007 Location: Vancouver (USA) Photos: Gallery \| Albums Posts: 42,007	Originally posted by 6BQ5 I thought I read that the focusing system in all Pentax DSLR's is tuned to an ideal aperture of f/2.8. Anything faster or slower takes it away from optimal performance. Please correct me if I am wrong. You are correct about the tuning, not correct about the models, and not quite right about the performance implications. For PDAF, the center column focus points on current flagship models have focus sensitivity of f/2.8. The rest are f/5.6. This is fairly industry standard for enthusiast and pro model dSLR bodies. The Pentax consumer models are f/5.6 for all points. What this means with the center column is that the system has the same ability to detect an out-of-focus condition at f/2.8 maximum aperture as with a faster lens and better than with a slower lens. The fast glass penalty is that you have poorer precision for fine focus than you might with a more sensitive focus system. Notice the points of emphasis for that last sentence. Use a maximum f/2 lens and f/2.8 lens and both will have about the same number of shots lacking acceptable focus than you would using magnified live view. Put another way, when using the center column points, the system rewards faster maximum aperture up to f/2.8 with no improvement beyond that point. Steve Last edited by stevebrot; 05-14-2016 at 10:32 AM.

05-14-2016, 10:31 PM	#14
stevebrot Otis Memorial Pentaxian Join Date: Mar 2007 Location: Vancouver (USA) Photos: Gallery \| Albums Posts: 42,007	Originally posted by Class A i.e., there is no AF performance penalty for shooting at apertures higher than f/2.8. You are correct, there is no brick wall hard limit in actual practice for most subjects. The penalty is if you need critical focus when shooting at apertures wider than f/2.8. The precision (probability of having the focus plane coincident with the focal plane) will be no better than if you were doing the AF at f/2.8. The same range of focus is being detected by PDAF as being "not out of focus" as with a slower lens. This is a matter of physics and how the sensor works and is not related to the quality of the lens. The system is quite literally "blind" to the focus lapse, though in a fuzzy way.* The easiest way to demonstrate is, unfortunately, not available on the K-1; using the PDAF focus confirmation in real-time comparison to a well-calibrated split-image rangefinder. Moving slowly through the focus point with an f/2 or faster lens at 50mm or longer focal length, the PDAF system will attain focus before the rangefinder images mate and will continue to indicate focus for a fair distance the other side after they diverge. Equivalent comparison is possible without the split image, though more clumsily, by making a best effort using PDAF focus confirmation and do the comparison by switching to magnified live view. Missed focus should be readily apparent. Do a trial of 10-20 attempts to get an idea of the real world impact. Steve * Probably the best resource I have found that illustrates in fairly simple terms how split-image, microprism, and PDAF systems work is the PDF article by Doug Kerr. Yes, all three systems work on the same principles and all three have similar limitations. The first parts are very interesting, but for those who lack patience, the PDAF stuff starts on page 12 and that dealing with aperture dependency (balancing black-out against precision) on page 17. http://dougkerr.net/Pumpkin/articles/Split_Prism.pdf Last edited by stevebrot; 05-14-2016 at 10:37 PM.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
AF accuracy with DA Limited lenses	biz-engineer	Troubleshooting and Beginner Help	15	02-26-2015 06:58 PM
Pentax K-5 AF accuracy	Christine Tham	Pentax K-5 & K-5 II	181	02-06-2012 11:00 AM
For Sale - Sold: M42 / K / KA / FA Lenses Grab bag of Vintage to Current lenses. 50mm FA f/1.4	MikeDubU	Sold Items	10	02-09-2009 12:45 PM
AF Accuracy's Dependency on Lenses and Yellow Light	RiceHigh	Pentax DSLR Discussion	10	02-19-2007 06:51 AM

05-14-2016, 07:35 AM	#2
repaap Pentaxian Join Date: Oct 2010 Photos: Gallery \| Albums Posts: 6,242	My FA 43 have been little hard to focus at near closest focuspoint, but other than tha it has been great and fast.

05-14-2016, 07:50 AM	#3
6BQ5 Veteran Member Join Date: Jun 2013 Location: Nevada, USA Photos: Albums Posts: 3,348	I thought I read that the focusing system in all Pentax DSLR's is tuned to an ideal aperture of f/2.8. Anything faster or slower takes it away from optimal performance. Please correct me if I am wrong.

05-14-2016, 08:53 AM	#4
aremmes Senior Member Join Date: Oct 2013 Location: Philadelphia Photos: Albums Posts: 217	Originally posted by 6BQ5 I thought I read that the focusing system in all Pentax DSLR's is tuned to an ideal aperture of f/2.8. Anything faster or slower takes it away from optimal performance. Please correct me if I am wrong. Possibly true, but if so I'd wonder why the cameras do not stop down fast lenses when focusing, which I imagine would be a cheap and easy optimization.

05-14-2016, 01:41 PM	#6
repaap Pentaxian Join Date: Oct 2010 Photos: Gallery \| Albums Posts: 6,242	I could do this wit no trouble. FA 43

05-14-2016, 04:40 PM	#7
scopedude Junior Member Join Date: Apr 2013 Posts: 43 Original Poster	Pentax distributor told me they have more FA43 in stock and I could test more next week. Hopefully I could find the one that works. I tested the FA 50 more and I did find that it is more difficult to obtain consistent AF at close range.

05-14-2016, 04:43 PM	#8
robjmitchell Veteran Member Join Date: Oct 2007 Location: Melbourne Aus Posts: 1,776	I find my 43ltd needs to be adjusted with a mid distance target (where you use it for portraiture) otherwise it will be off. Once its dialled in on a new camera I don't have a problem. I suspect the fast Aperture plus focusing at short distances throws the adjustment off compared to the rest of the range.

05-14-2016, 05:03 PM	#9
scopedude Junior Member Join Date: Apr 2013 Posts: 43 Original Poster	Noted. Will do this when testing the new unit. They brought me silver while I asked for the black one

05-14-2016, 07:05 PM	#10
Inspireart New Member Join Date: Mar 2016 Posts: 4	I will ordering a K-1 and a 28-105 in several weeks. How did you detect the centering issue on one of the 28-105 lenses? I am not an optical engineer and am worried about getting a poor example. Thanks, Bruce

05-14-2016, 07:34 PM	#11
Class A Pentaxian Join Date: Aug 2008 Location: Wellington, New Zealand Posts: 11,251	Originally posted by 6BQ5 I thought I read that the focusing system in all Pentax DSLR's is tuned to an ideal aperture of f/2.8. Anything faster or slower takes it away from optimal performance. Please correct me if I am wrong. There is no "tuning" to f/2.8. Some of the AF areas are capable of resolving at an AF base of f/2.8. That means they provide higher precision than the standard f/5.6 areas. Lenses faster than f/2.8 are a challenge for the AF system but to the best of my knowledge there is no "brickwall" characteristic. Faster lenses often have a number of aberrations and small manufacturing faults have exaggerated effects. In understand AF systems struggle with that more than the fact that these lenses are faster than f/2.8. With respect to lenses slower than f/2.8: As long as a lens does not exhibit a so-called "focus-shift" phenomenon, an AF system will not be affected, i.e., there is no AF performance penalty for shooting at apertures higher than f/2.8. Again, to the best of my knowledge, modern lenses contain calibration data regarding their focus-shift behaviour so the camera is able to compensate for known focus-shifts. If a lens does not quite exhibit standard performance, all regular computations and compensation mechanisms are off and one can get AF problems. Such non-standard behaviour is more pronounced on very fast lenses, hence they represent a bigger challenge. Originally posted by scopedude I tested the FA 50 more and I did find that it is more difficult to obtain consistent AF at close range. I don't think you should expect too much AF consistency with this lens at wide apertures and close range (both factors contributing to very shallow DOF). My K100D can struggle very hard to lock focus with it. Haven't used it on my K-5 II in such situations yet.

05-14-2016, 09:45 PM	#12
scopedude Junior Member Join Date: Apr 2013 Posts: 43 Original Poster	Originally posted by Inspireart I will ordering a K-1 and a 28-105 in several weeks. How did you detect the centering issue on one of the 28-105 lenses? I am not an optical engineer and am worried about getting a poor example. Thanks, Bruce Hi Bruce, I always do at two distances, first at infinity wide open, and minimum distance. To make it easier at close distance I take photo of long 24" ruler. It will be easy to detect if the focus plane shifts front and back from left to right. These tests are also useful to see if the lens is optimized at short or long distance. For example, canon 24-70 L ii tends to be flatter at short distance but has curvature at infinity that doesn't get better even at f8. In case of DFA 28-105, the two samples behaved differently on the ruler test Winston

05-14-2016, 09:58 PM	#13
Class A Pentaxian Join Date: Aug 2008 Location: Wellington, New Zealand Posts: 11,251	Originally posted by scopedude To make it easier at close distance I take photo of long 24" ruler. It will be easy to detect if the focus plane shifts front and back from left to right. I don't think it is "easy" to ensure that the sensor is perfectly aligned to the ruler. Commercial tools use dedicated alignment constructions for this purpose. A better decentering test is to chose an object at a fair distance and put it in all four image corners in consecutive shots (using manual exposure & manual focus fixed for all shots, and a tripod).

05-14-2016, 10:58 PM	#15
scopedude Junior Member Join Date: Apr 2013 Posts: 43 Original Poster	Originally posted by Class A I don't think it is "easy" to ensure that the sensor is perfectly aligned to the ruler. Commercial tools use dedicated alignment constructions for this purpose. A better decentering test is to chose an object at a fair distance and put it in all four image corners in consecutive shots (using manual exposure & manual focus fixed for all shots, and a tripod). It's meant to be a practical way to test the lens. There is a grid in K1's viewfinder that would really help. I usually redo the test at home with a tripod like you described. That said, using the ruler with the VF grid is not difficult at all - because when you find a decentered lens you'll be surprised at how far the focus plane shifts. In case of the 28-105, the left frame was in focus in front of the ruler, and the right frame was way behind the ruler. Very noticeable. When I did the long distance test, the left and right frame showed different sharpness too. Pentax people came to my office with 2 K1s and 2 28-105s, and some other lenses so I had to decide which one to take home I've been doing it for years after getting less than ideal samples (many of them) from Canon and Zeiss ZE/ZF. I'm doing mostly landscape, architecture and some astrophotography that's why it's important for me.