December's CHALLENGE

The topic to inspire your creative juices this month is BOXES Please don't forget to vote on November's LEAVES challenge and please re-vote if you already did but before the recent forum upgrade.

See more
See less

Zuiko 300 F/2.8 AF calibration

  • Filter
  • Time
  • Show
Clear All
new posts

  • Zuiko 300 F/2.8 AF calibration

    Hi all,

    Sharing my experiences of lens calibration with the community.

    I have so far not been able to achieve satisfactory results with my 300 F/2.8 lens when used with M4/3 cameras, in particular when using it with the teleconverters. Usually photos have shown sign of back focus, quite often pronounced. For the sake of completeness there is nothing wrong with the lens. That lens has delivered superb results when used with the E-5, but has been some of a disappointment since I jumped on the m43 train compared to what the 300 F/4 delivers.

    The other day I came home from a photo session using the 300 F/2.8 but got almost no good photos, a big disappointment. Having some some spare time last Sunday I spent the mid-day using a recently acquired calibration tool to come to clarity what is going on.

    The tool
    If you are familiar with how the tool should be used you can directly jump to the method section.

    Basically the tool works like this: Set up the camera and calibration tool so that they are aligned with the target area in same plane as the camera sensor. I used two tripods for this
    The camera should be placed at recommended distance from target. I have read different recommendations and decided to go for the one that says 25 times focal length (meaning 7,5 meters in this case). This would reflect the closest distance at which you will shoot when conditions are favorable. Closer distance would mean it becomes easier to record front/back focus since the scale fills more of the frame, placing the camera at around 4 meters I would have filled the frame however I wanted to calibrate the setup for a realistic distance that would be useful in practice.
    When shooting:
    Shoot wide open to have as shallow depth of field as possible (F/2.8 in this case)
    Turn IS off and use electronic shutter to eliminate any blurring by IS/shutter shock
    Shoot in good light conditions (ideal is mid day, light clouds, calm weather) to eliminate any shadows that might bias the assessment and make sure shutter speed short enough and that the wind does not result in shake blur.

    The method is the following:
    I found it was easiest to evaluate the focus when underexposing -2/3 EV. The contrast in the ruler was about right at this setting; 0EV would blow the highlights, -1 EV would reinforce the blacks, making the details in the ruler harder to read.

    Aim at the centre of the target, after focus is confirmed take a picture.

    De-focus randomly front/back between shots and repeat the procedure to get a population of samples in order to mitigate potential focus variations between shots. I decided arbitrarily to take 10 shots for each sequence. Looking at the results (see below) I think that was sufficient.

    After 10 shots are taken, in the camera menu "AF Calibration": Change the setting towards + or -. The scale goes from -20 to +20. There is no direct information about how much each step transforms in terms of compensation/distance, the answer was given after reviewing the samples and having them displayed in a graph using linear regression. I decided arbitrarily and based on reading articles related to this subject to change in steps of 5 at the time, which turned out to be about right. By adjusting in steps of 5 you will end up with up to 9 measurement points; one for -20, -15... +15, +20.

    When done with the shooting: easiest is to review the pictures using the computer. You could of course do it on the LCD however it is more difficult. And you would need using a spreadsheet (Excel) anyway to get the results displayed in a graph which makes it much easier to determine the calibration needed to apply. See below.

    For each shot:
    inspect the scale and assess where focus is. The scale goes from -6 to +6, negative values are towards camera, positive are away from camera. This task can be a tad challenging since you will that a number of lines look all equally sharp. I found that a method that worked fine is to match the lines/digits that appear equally unsharp, this way you narrow down to a final pair of lines that appear equally sharp, the focus should halfway in between. To validate this method I re-visited the results and made a second assessment after having rested my brain and eyes, and found the uncertainty using this method is around 0,5 when comparing the two assessments. I used the average of the two assessments when analyzing the results.

    (Edit): A method that I found worked even better and saved time was to increase the contrast and brightness enough for the lines out of focus to vanish. Then the focus offset is easy to assess, example below where focus is about right within a few mm:

    For each series of 10 shots:
    Inspect each photo using the method above and make a note where lines/digits appear as most sharp on the scale. Negative values are for front focus (towards the camera), positive for back focus (away from camera). Enter the values in an Excel table.

    As a next step have Excel calculate the average for the 10 samples and if you like also the standard deviation to get a better feeling how consistent the population of samples is.
    Repeat this for each series of 10 shots. Note that you may get series where focus goes out of bounds (sinc the scale on the ruler goes from -6 to +6), these series are useless and no need to explore further. In theory you could end up with as many as 9 series within scale. In practice it should be enough with 5 series to proceed with next step.

    Have Excel create a chart with calibration value on X-axis (scale -20 to +20) and focus offset on Y-scale (scale -6 to +6). The average values should appear on something looking like a sloping line, by have Excel draw a linear regression line it makes the next task easier. Thanks to this line you also get more assurance that the samples show consistency and you will calibrate your camera based on more samples than just shooting at one arbitrary compensation.

    To calibrate the camera/lens combination: Look at the diagram and find out where the linear regression line intersects the Y axis at 0 offset, then read the calibration value on the X axis. That is the value which you should enter in your camera menu.

    Repeat above for each camera/lens combination.

    Case study
    I calibrated the following gear:
    • Cameras: E-M1, E-M1 Mark 2
    • Lens: Zuiko 300 F/2.8
    • With and without EC 14

    This means 4 combinations in total.

    Result: all combinations were back-focusing, as I suspected. The EC-14 amplified the back focus quite a lot.

    Looking into the chart and rounding to closest integer I found the following calibration should be applied
    • Zuiko 300 F2.8: -2
    • Zuiko 300 F/2.8 + EC14: -8

    E-M1 Mark 2
    • Zuiko 300 F2.8: -5
    • Zuiko 300 F/2.8 + EC14: -9

    Finally I proceeded with the Zuiko 300 F/2.8 with EC20 however the method failed as the back focus was so pronounced that most measurements fell off scale, only when setting -15 or -20 I could get something that looked reasonably sharp. I will have to prolong the ruler for useful results. In the meantime I will calibrate this lens at -20.

    Discussion: how bad was the back-focus before calibrating it?

    My Gallery on 500px

  • #2
    Re: Zuiko 300 F/2.8 AF calibration

    Tord, I made a similar device myself a few years ago.

    I didn't use such a precise method of analysing the results, rather I just took a series of photos, with varying + and - in-camera offset and looked to see which was in closest focus.

    In the end I made a second "ruler" and rather than having numbers and lines, I used a strip of newsprint. I specifically used a newspaper, as the grain of the actual paper is more obvious and it eases the decision on which line of print is in most accurate focus.

    The reason for the two paxolin sheets making up the jig, is that the rear one houses a laser pointer, shining through a hole in the middle of the centre target on the front sheet.
    This enables easy setting up (with the lens cap on for sensor safety). A small bubble level in the base completes the device, which has a tripod mounting hole tapped into the bottom.

    I originally made the gizmo, as I was having problems obtaining accurate focus on the E-3 when using a 112mm filter on the front.
    The lens eventually went back to Japan, where my findings were confirmed, and it was explained that my particular lens couldn't be calibrated for use with/without a filter, and with/without a teleconverter.
    Obviously with later cameras, it's possible to come up with my own in-camera calibration data.
    Best Regards

    The nearest I have to a home page.
    They're Watching You!


    • #3
      Re: Zuiko 300 F/2.8 AF calibration

      I always thought it strange why a camera with dual contrast and phase AF can not do this all by itself by repeatedly adjusting the PDAF adjustment value and switching between PDAF and CDAF until the CDAF confirms that accurate focus was obtained by PDAF. It can then do this for all focal lengths for all PDAF sensors!
      Cameras: E-M5, E-PM2, OM40, OM4Ti
      Lenses (M.Zuiko Digital): 7-14mm/F2.8, 12-40mm/F2.8, 40-150mm/F2.8+TC1.4x, 12-50mm/F3.5-6.3, 14-42mm/F3.5-5.6 EZ, M.ZD 40-150 F4-5.6 R, 75-300mm/F4.8-6.7 Mk1, 12mm/F2, 17mm/F1.8
      Lenses (OM Zuiko): 50mm/F1.2, 24mm/F2, 35mm/F2.8 shift
      Lenses (OM Fit): Vivitar Series II 28-105mm/F2.8-3.8, Sigma 21-35mm/F3.4-4.2, Sigma 35-70mm/F2.8-4, Sigma 75-200mm/F2.8-3.5, Vivitar Series II 100-500mm/F5.6-8.0, Centon 500mm/F8 Mirror
      Learn something new every day


      • #4
        Re: Zuiko 300 F/2.8 AF calibration

        This is very thorough guys well done. Can it be made into a sticky so it's easy to find when needed.


        • #5
          Re: Zuiko 300 F/2.8 AF calibration

          Originally posted by OM USer View Post
          I always thought it strange why a camera with dual contrast and phase AF can not do this all by itself by repeatedly adjusting the PDAF adjustment value and switching between PDAF and CDAF until the CDAF confirms that accurate focus was obtained by PDAF. It can then do this for all focal lengths for all PDAF sensors!
          That "self-calibration" would have been a cool feature, which technically should be feasible to implement. The reason why it is not available could be because Olympus want people to buy m4/3 lenses rather than given the 4/3 lenses "extended life"?

          My Gallery on 500px


          • #6
            Re: Zuiko 300 F/2.8 AF calibration

            After prolonging the ruler (actually, taping a longer ruler on top of the existing) I was able to calibrate the Zuiko 300 F/2.8 with the EC20 attached. Now that the method is developed and proofed it did not take that much time, just had to translate the longer rule scale and marks to what would have been read if the original had been long enough

            Here is the summary.

            As you can see the back-focus was quite pronounced (bad). In a matter of fact with the EC20 we reach the limit for what the +/- 20 compensation menu allows to correct.

            What also makes me wonder is if attaching a TC amplifies any existing front/back focus issues? Anyone here with insight or knowledge that would like to comment or elaborate on this?


            My Gallery on 500px