1. Welcome to Mu-43.com—a friendly Micro 4/3 camera & photography discussion forum!

    If you are thinking of buying a camera or need help with your photos, you will find our forum members full of advice! Click here to join for free!

Lens auto focus accuracy

Discussion in 'Native Lenses' started by Yves Giot, Jun 12, 2018.

  1. Yves Giot

    Yves Giot Mu-43 Rookie

    13
    Apr 7, 2018
    Hello,

    I thought that you didn't have to check autofocus with a EM1 markII as it is using directly the sensor. After seeing more pictures out of focus than with my Nikon D700 (might be user error too), I decided to check autofocus on my lenses with the data color calibration tool. I noticed that my
    40-150 F2.8 Zuiko Pro: is perfectly centered or close
    45mm f1.2 Zuiko Pro : Front focus slightly but noticeable (about 1/3 back, 2/3 in the front see photo)
    12-40 F2.8 pro : Very heavy front focus (all focus is in front - almost no depth of field in the back). This could may be explain why I had so many pictures blurry using this lens.

    I tried to use focus adjust to correct but looks like it doesn't do anything with S-AF (I read threads saying so, I tried but no difference between -20 and +20 so I guess it doesn't do anything with S-AF.

    I tried the same tests with another camera EM1 (mark1) and got very similar results.

    The 12-40 I obviously have an issue and probably need to send it to Olympus but what about the new 45mm F1.2. Should it be closer to 50/50 or not (see picture) ? Any other experiences ?

    Thanks
     

    Attached Files:

  2. Clint

    Clint Mu-43 Hall of Famer Subscribing Member

    Apr 22, 2013
    San Diego area, CA
    Clint
    There has been some debate about focus tuning for m4/3 whether using S-AF or C-AF. Focus testing is not just about the accuracy of a single focus, also consider the consistency of focus precision.

    I've found it very helpful to know how 5-10 shots at -20, -15, -10, -5, 0, 5, 10, 15, and 20 shows up in a pattern. And then significantly reduce the testing criteria to 5 points on both sides of what seems the most accurate focus. And then try several days later on the same test to see if the results are repeatable.

    When looking at many photos and determining the best focused lines seemed to play games with my vision. I became a quick believer in FocusTune software.

    Consider the depth of field which can complicate trying to figure out exactly where the camera focused. e.g.
    - a 45mm at f/1.2, 4 ft from target should have near 1 inch (23mm) of DOF.
    - 40mm at f/2.8, 4 ft from target should have around 2 inches (50mm) DOF.

    For several m4/3 lenses my initial testing in S-AF leads me to believe the OM-D E-M1s have outstanding focus consistency and precision.

    If you have not you might want to read Phocals procedures for testing.
     
  3. panamike

    panamike Mu-43 Hall of Famer Subscribing Member

    Jul 5, 2016
    Lincolnshire UK
    Just to prove S-AF does not need adjust, i adjusted my EM1MK11 for my 100-400 @400 it needed +10, any way i sold the lens and bought a 40-150, in C-AF it was way out but in S-af its spot on, i had set the focus adjustment in the wrong section and it was working on all lenses but only in C-AF.
     
  4. I can't say I've ever had a problem. I can't even understand how you need or can adjust anything for S-AF. If I put dioptres or an extension tube on it completely screws with AF distances but AF still works just fine...
     
  5. Ross the fiddler

    Ross the fiddler Mu-43 Hall of Famer

    I would have to ask what AF target size is being used with those results. The latest FW for the E-M1MkII helps there with small AF target.
     
  6. Growltiger

    Growltiger Mu-43 All-Pro

    Mar 26, 2014
    UK
    I think the problems are all down to experimental design and are not to do with the lenses.
    Were the results you quote using S-AF, it isn't clear?

    Note that the problems you show are greater the smaller the focal length, I don't think this is coincidence. You need to put the camera nearer or use a small focus point and make completely certain that only the flat area is being detected in the focus point.
     
    • Agree Agree x 1
  7. Mikehit

    Mikehit Mu-43 Regular

    162
    Jan 26, 2018
    With purely contrast detect based on the sensor (as most mirrorless has been) then focus should not be a problem. As I understand it, when they started putting phase detect pixels on the sensor, they introduced the same source of errors seen in DSLR which is why the EMII also has autofocus micro adjustment just like DSLRs have.

    The more things change, the more they stay the same...
     
    • Like Like x 1
  8. Yves Giot

    Yves Giot Mu-43 Rookie

    13
    Apr 7, 2018
    Hello Thanks for feedback,

    I agree that it shouldn't be a problem with contrast detect, that is why I didn't even look at it first (I should have because now lens is out of warranty)
    I realize that my photo resolution has been reduced and cannot see what I see in Lightroom.

    To answer a few questions
    1) I believe I used the small focal points but in any case I made sure the whole green box is well within the target area and is across lines to facilitate auto focus
    2) I tested all lenses at around 40mm (40-150 at 40, 45 at 45, 12-40 at 40) so focal length is basically very similar in all tries
    3) I tested multiple times, every time taking 3 pictures. Probably did it 4 or 5 times with 2 cameras and got same results each time, for that it was very consistent. (probably 20 pictures for each lens). refocusing each time.

    Again the 45mm 1.2 might be within tolerance with 1/3 in the back and 2/3 in the front (I am not sure what those would be) but my 12-40 is clearly off, it is barely in focus with all the depth of field up front (nothing in the back). I would post a picture but looks like resolution doesn't help anyway.

    May be I didn't test right. Hopefully that is the case. But the reason I started this whole process in the first place is because I noticed that I had a lot more photos out of focus than with my D700 which puzzled me. I would have thought the opposite.

    Focus adjust doesn't seem to do anything for S-AF. It should for CAF based on what I read but I didn't test that.

    Thanks
     
  9. johnvanatta

    johnvanatta Mu-43 Veteran Subscribing Member

    237
    Aug 5, 2014
    Oakland, CA
    There's a setting buried in the menus somewhere that allows you to take a picture before autofocus has finished. Some like it on for AF-C, but I think you can turn it on for AF-S too (no idea what that'd be useful for).

    The CDAF on the EM-1 line is highly accurate, moreso than any PDAF DSLR out there.

    What about manual focusing?
     
  10. Ross the fiddler

    Ross the fiddler Mu-43 Hall of Famer

    You would be referring to Priority Release then in Menu A1.
     
  11. panamike

    panamike Mu-43 Hall of Famer Subscribing Member

    Jul 5, 2016
    Lincolnshire UK
    Going to risk getting shot down in flames for this but for my type of subjects i dont take a lot of notice of test charts, i go by what it does in the field, the grass in this picture shows possibly slight error but near enough for me.
    At a different distance i may get a slightly different result so i look for an average that works.

    42724753162_51b3c97a38_o.
    E-M1MarkII    M.40-150mm F2.8 + MC-14    180mm    f/4.5    1/2000s    ISO 320
    P6130277
    by electric.mike, on Flickr
     
    • Like Like x 2
    • Agree Agree x 1
  12. Ross the fiddler

    Ross the fiddler Mu-43 Hall of Famer

    That looks pretty good to me. Is the focus on the eye or the body?
     
  13. panamike

    panamike Mu-43 Hall of Famer Subscribing Member

    Jul 5, 2016
    Lincolnshire UK
    Cant say as i was waiting for it to fly so most likley had the center 9 af boxes on, i would guess body.
     
    • Like Like x 1
  14. Growltiger

    Growltiger Mu-43 All-Pro

    Mar 26, 2014
    UK
    You have still not said if your tests were done using S-AF or C-AF?
     
  15. Reflector

    Reflector Mu-43 All-Pro

    Aug 31, 2013
    With adapted lenses: (in my experience)
    When in S-AF, the E-M1II will throw focus rapidly like a DSLR's PDAF motions but once it gets close it will do a CDAF shuffle for the final adjustment.
    In C-AF, the E-M1II generally throws focus all in one go. It may or may not make an additional adjustment afterwards.
     
  16. Stanga

    Stanga Mu-43 Top Veteran

    540
    Oct 16, 2016
    I noticed a similar issue on the Panasonic GX8 and G80. In S-AF you can get a bit of extra sharpness in the focusing if you use the PIP display enlargement. You can actually see the camera adjusting for a sharper focus compared to the one that is achieved without the PIP enlarged display.
    I also noticed that when you use C-AF, press the shutter button lightly so that the camera does a refocus, and then take your finger off the shutter button, the tracked subject inside the focus box on the screen is far sharper and stays that way. I can read all the letters on the lowest line on the reading test chart at my optician's, so I don't think that I am imagining the difference in sharpness...
     
  17. wimg

    wimg Mu-43 Top Veteran

    515
    Dec 10, 2016
    Netherlands
    There generally is a substantial difference between area in focus in front of the focus point and area in focus beyond, depending on the focusing distance. Generally speaking when focusing up close the larger part of the in focus area is in front of the focus point, say 2/3 of the in-focus area in front, while that reverses when focusing at larger distances, generally speaking that is.

    With AF it depends a lot on the focusing algorithm, however, as focusing is not continuous, but a bunch of discrete incremental steps due to the way focusing motors work. In addition, both lenses and camera bodies are calibrated towards a standard with minimum and maximum distance deviation, independently of each other, and if one is at the minimum threshold, the other at the maximum, it will provide different results to both being somewhere in the middle. This is also why modern cameras allow calibration by the user.

    IOW, you may be doing everything correctly, and still get differences between the two systems, unless your lens are calibrated to the body used. This is exactly what I had done out with my FF lenses, although so far it was not necessary for my MFT setup.

    Kind regards, Wim
     
  18. PakkyT

    PakkyT Mu-43 Hall of Famer

    Jun 20, 2015
    New England
    Why would this be needed? The lens doesn't control AF. The camera will keep moving the lens until it achieves what it thinks is proper focus, right?

    Calibration of the lens to the body used to be a thing that was sometimes needed for mirrored cameras where the image to the focus sensor and the image to the imaging sensor were not the same thing, which is what I am assuming you are talking about when you refer to your FF stuff.

    To the original poster, did you try this test outside in good sunlight so the contrast on the calibration tool? From the photo you attached the "whites" looked a little muddy to me if that was the lighting you used for your testing.
     
  19. Mikehit

    Mikehit Mu-43 Regular

    162
    Jan 26, 2018
    The lens and the camera body communicate regards the amount of movement of the focussing mechanism which is why different lenses will AF quicker or slower. The camera AF algorithms also permit a pre-defined number of iterations so if after the defined number of adjustments the AF is deemed by the algorithm to be within the DOF it will permit the shot to be taken. If it is still outside that range it will go down as 'hunting'.
    So yes, the lens does affect AF.
     
  20. wimg

    wimg Mu-43 Top Veteran

    515
    Dec 10, 2016
    Netherlands
    This is needed because AF motors are stepper motors. The lens may not control AF, but the electronics do control the stepper motor, and these have to be calibrated. The camera AF system that sends the signals to the lens for the stepper motor to be used, has to be calibrated as well. Calibration in both cases is done for a specific range, which may therefore limit AF accuracy.

    The camera will keep moving the lens until it achieves focus, especially in CAF-mode, but even so, we are dealing with a stepper motor, which means it is rarely exactly 100% correct. And in PDAF mode it is less correct, generally speaking. The only fortunate thing is that FLs are shorter than with FF, and therefore MFT is more forgiving due to larger DoF.

    Kind regards, Wim
     
    • Informative Informative x 1
  1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.