How to tell which AFMA is sharpest?

I'm conducting a lens calibration test for my Canon 180mm macro. I've narrowed the best auto focus micro adjustment (AFMA) down to a ballpark somewhere between -5 and -1 points. I'm having a hell of a time in PS trying to tease out the sharper image. Is there a way to enhance the images in such a way to make the winner make itself obvious? I tried adding Smart Sharpen, but it's still too close to call. If you're going to say I should split the difference and choose -3 points, I hear ya. But I want to be perfect.

What did you use for your focusing chart? A monitor, or zoomed in at 400%? Either option is poor for determining precise settings IMO.

I used the LensAlign target. The image I posted is a screenshot in PS of the lettering on the target zoomed in to 3200x. The Focus Tune software recommends an AFMA somewhere around -7 points, but my manual eyeball testing method convinces me it should be lower than that. Software is good for getting you close, but I believe doing it by eye will get you the closest. The test images were shot in RAW and converted to ProPhotoRGB 16bit for maximum IQ.

If they both look the same, split the difference and settle on -2

But if you really want better data, shoot a resolution chart like the ISO 12233 chart. Then look at which setting resolves the lowest.

left over right, but neither one looks particularly sharp to my eye...I would use a print chart rather than an electronic display; or output actual crops and post them for review. Doing a screengrab from your monitor is introducing additional artifacting and such that make it nearly impossible to gauge here.

I was hoping there was a PS filter technique. There are visible differences between the left and right snapshots I posted. For example, the left one has a greater degree of color variation amongst the edgemost pixels. Perhaps that is a telltale sign of greater sharpness. I suspect that the more blurred a group of pixels get, the more uniform in color those pixels become. Am I right, or is the opposite true?

Looking at pixels at 3200x isn't going to tell you much. You need the line pairs to be able to quantify at what point does the optics/sensor pair fail to resolve any more detail. If you look at my example, the sample on the left starts failing at about 8 lines pairs. The lines get muddy and are barely resolvable but on the right at 8 they are still easily discernible and even at 10, you can still see the pairs.

gjl711 wrote in post #18576246
If you look at my example, the sample on the left starts failing at about 8 lines pairs. The lines get muddy and are barely resolvable but on the right at 8 they are still easily discernible and even at 10, you can still see the pairs.

I'm sure it's easy to see the difference between 0 and +10 AFMA on the ISO12233 chart. But have you tried comparing +9 vs +10 points? Not so easy I bet.

texshooter wrote in post #18576256
I'm sure it's easy to see the difference between 0 and +10 AFMA on the ISO12233 chart. But have you tried comparing +9 vs +10 points? Not so easy I bet.

The changes are miniscule at that point. Every 8 is the equivalent of the depth of focus at the sensor level. A +/-1 change isn't going to be very noticeable at all, even if you were shooting with an f/1.2 lens.

Shoot the camera and lens combo as you would in real life and then adjust. It is that simple. The easiest way that I do this, developed from the days of the 1D3, is to put a high contrast small object in short grass or on a driveway, and go back to a point that the object barely covers the center point. Take shots and adjust from there. Works every time, and I can do this at a game or marching band contest if lights are different each time, in no time at all.

My god, zooming in 3200x and asking about more Photoshop filters to determine lens AFMA? At 3200x, you're just seeing individual pixels and aliasing patterns instead of actual contrast characteristics. Likewise about using filters: which will induce some form of artifacting. Personally, I find the ruler to be the key judge in a focusing aid. You should be able to see at what mark the camera is in focus, and if that mark is not your intended focus. I can see whether it's front or back focused, and then I dial in a couple points, shoot, and see if it's dead on or not. I'm pretty certain my AFMA values are +1 or - 1 if not dead on.

Just last week I tried to help a local guy with this problem. He had cycled through various MFA settings and could not see any difference in test images.

After 1/2 hour reviewing the basics, I finally discovered he had made all his shots in live view mode!

If you can't tell the difference at 100%, it doesn't matter. Looking deeper is pointless, it isn't real.

texshooter wrote in post #18576191
I'm conducting a lens calibration test for my Canon 180mm macro. I've narrowed the best auto focus micro adjustment (AFMA) down to a ballpark somewhere between -5 and -1 points. I'm having a hell of a time in PS trying to tease out the sharper image. Is there a way to enhance the images in such a way to make the winner make itself obvious? I tried adding Smart Sharpen, but it's still too close to call. If you're going to say I should split the difference and choose -3 points, I hear ya. But I want to be perfect.

Hosted photo: posted by texshooter in
./showthread.php?p=185​76191&i=i256399412
forum: RAW, Post Processing & Printing

I think the one on the right looks sharper. If you want to take all the guesswork away you may consider this.

http://www.reikan.co.u​k/focalweb/

Snydremark wrote in post #18576210
left over right, but neither one looks particularly sharp to my eye...I would use a print chart rather than an electronic display; or output actual crops and post them for review. Doing a screengrab from your monitor is introducing additional artifacting and such that make it nearly impossible to gauge here.

Right one looks sharper to me.