WilbaW
Forum Pro
Until recently I had very little experience with doing autofocus microadjustment, but I had followed the discussions about various method for years in anticipation of having a body with the function. So the fun begins with a 7D II, and DotTune on the 400/5.6, 70-300L, 100L Macro, 40/2.8 STM, and 15-85, no worries.
I found the long lenses tended to have a "long tail" towards infinity, where confirmation gradually deteriorated over three or four MA numbers, rather than cutting off suddenly like it usually does towards MFD. I wondered if that might bias the MA settings towards infinity, so I checked AF accuracy using the EOS Utility and ended up revising a few numbers towards MFD as predicted. I also found a better way to assess focus accuracy with the EOS Utility. I can't recall reading about the method I came up with, so it might be new, might be ancient, dunno.
To assess focus accuracy using images from the camera you need some kind of reference, like a slanted scale, or a series of shots using different MA values, or how many clicks in the EOS Utility it takes to match the focus of a CD AF shot. In many of those methods you're looking for very subtle differences between data points clustered at the bottom of a curve, so I wondered if it was possible to compare gross differences that more readily reveal discrepancies. I borrowed an idea from working with slanted scales - it's easier to judge a focus error by comparing how out of focus are two marks on the scale equidistant from 0.
Here's the method I used with the EOS Utility, after DotTuning the lens to get it pretty close to spot on. I have the EU set to save images on the PC.
Comparing images 2 and 4 tells me whether the current MA setting is front or back-focussing. If 2 is more out of focus than 4, the MA setting is causing front-focus, and if 4 is more out of focus than 2, it's causing back-focus. Then adjust by 1 click towards infinity or MFD as appropriate, and retest from step 3.
It's much easier to compare the two out of focus images and judge which is more so, than it is to judge the difference between two well-focussed images. It can also be interesting to compare similar images over a number of sets (over a range of MA settings), for instance all the 1s, all the 2s, or all the 2s and 4s as pairs.
I often found that one setting, e.g. +2, gave apparent front-focus, and the next setting, e.g. +3, gave a similar level of back-focus, so perfection would be in between, somewhere around +2.5. At this point we really are splitting hairs, and in the field I expect other real-world factors would have much more of an influence on focus accuracy.
I'm interested to hear your comments and experiences with this way of assessing focus accuracy with the EOS Utility.
Teleconverters... I had no problem with the Canon 1.4 III on the 400/5.6, but my Kenko TC will lock up the camera if I use a per lens adjustment. I can get around that by using Adjust All by Same Amount, but it's a complete PITA because I have to know what setting I need for each lens, and I have to change it manually before I swap lenses. I also have to remember to do that before I switch to a C mode that is set to use Adjust by Lens. It can be done, but for me it's simply too much hassle to use the Kenko TC with 7D II (it's fine with the 60D).
I found a similar lock-up problem with old manual lenses on chipped adapters. If I were going out to shoot all day with one, I'd probably change to Adjust All by Same Amount to make use of the confirmation in MF mode.
I found the long lenses tended to have a "long tail" towards infinity, where confirmation gradually deteriorated over three or four MA numbers, rather than cutting off suddenly like it usually does towards MFD. I wondered if that might bias the MA settings towards infinity, so I checked AF accuracy using the EOS Utility and ended up revising a few numbers towards MFD as predicted. I also found a better way to assess focus accuracy with the EOS Utility. I can't recall reading about the method I came up with, so it might be new, might be ancient, dunno.
To assess focus accuracy using images from the camera you need some kind of reference, like a slanted scale, or a series of shots using different MA values, or how many clicks in the EOS Utility it takes to match the focus of a CD AF shot. In many of those methods you're looking for very subtle differences between data points clustered at the bottom of a curve, so I wondered if it was possible to compare gross differences that more readily reveal discrepancies. I borrowed an idea from working with slanted scales - it's easier to judge a focus error by comparing how out of focus are two marks on the scale equidistant from 0.
Here's the method I used with the EOS Utility, after DotTuning the lens to get it pretty close to spot on. I have the EU set to save images on the PC.
- Defocus the lens and switch to AF.
- Start Remote Shooting but don't open the Live View window yet.
- Hover the mouse over the shutter button in the Remote Shooting main window to trigger PD AF.
- Open the Live View window. (Wait for the LV window to open fully before you move the mouse for the next step.)
- Click the shutter button in the RS window to take image 1. (You could click the shutter button to take image 1 in step 3, but since the metering is different in viewfinder and Live View modes, the subsequent image might have different brightness, which makes it harder to compare focus. Taking the image after the Live View window is fully open makes the brightness consistent, or you could use manual settings with unchanging light and click in step 3.)
- Click the "<" button in the LV window a number of times, or the "<<" button once, to slightly but clearly defocus the lens towards MFD. (Click carefully, watching for the button to flash yellow with each click.)
- Click the shutter button in the RS window to take image 2.
- Click the "<<<" button in the LV window to grossly defocus the lens for the next test.
- Close the LV window to go back to PD AF.
- Hover the mouse over the shutter button in the Remote Shooting main window to trigger PD AF again.
- Open the Live View window. (Wait...)
- Click the shutter button in the RS window to take image 3.
- Click the ">" button in the LV window the exact same number of times as in step 6, or the ">>" button once, to shift the focus towards infinity. (Click carefully...)
- Click the shutter button in the RS window to take image 4.
- Click the ">>>" button in the LV window to grossly defocus the lens for the next test.
- Close the LV window.
- autofocussed,
- defocussed towards MFD,
- autofocussed,
- defocussed the same amount towards infinity.
Comparing images 2 and 4 tells me whether the current MA setting is front or back-focussing. If 2 is more out of focus than 4, the MA setting is causing front-focus, and if 4 is more out of focus than 2, it's causing back-focus. Then adjust by 1 click towards infinity or MFD as appropriate, and retest from step 3.
It's much easier to compare the two out of focus images and judge which is more so, than it is to judge the difference between two well-focussed images. It can also be interesting to compare similar images over a number of sets (over a range of MA settings), for instance all the 1s, all the 2s, or all the 2s and 4s as pairs.
I often found that one setting, e.g. +2, gave apparent front-focus, and the next setting, e.g. +3, gave a similar level of back-focus, so perfection would be in between, somewhere around +2.5. At this point we really are splitting hairs, and in the field I expect other real-world factors would have much more of an influence on focus accuracy.
I'm interested to hear your comments and experiences with this way of assessing focus accuracy with the EOS Utility.
Teleconverters... I had no problem with the Canon 1.4 III on the 400/5.6, but my Kenko TC will lock up the camera if I use a per lens adjustment. I can get around that by using Adjust All by Same Amount, but it's a complete PITA because I have to know what setting I need for each lens, and I have to change it manually before I swap lenses. I also have to remember to do that before I switch to a C mode that is set to use Adjust by Lens. It can be done, but for me it's simply too much hassle to use the Kenko TC with 7D II (it's fine with the 60D).
I found a similar lock-up problem with old manual lenses on chipped adapters. If I were going out to shoot all day with one, I'd probably change to Adjust All by Same Amount to make use of the confirmation in MF mode.
Last edited: