The disconnection between micro adjust AF settings and the nature of lens design...

This is a Sigma 50mm f1:1.4 lens with a Sony A mount. It's a great lens but my camera can't reliably focus it and neither can yours. Even if you use a focus align jig and take great pains to calibrate the hell out of it. Is there something wrong with the lens? Nope, I get the same behavior from the Zeiss 50mm f1.4 for the Canon and also the manual focus Carl Zeiss 85mm 1.4 in any of the major brand mounts. So what's the deal?

It's pretty simple really. All these fast lenses have a common attribute called focus shift and the simplest explanation is that the point of correct focus shifts as you stop the lens down. The micro-adjust AF controls in all of cameras are amazingly simple and stupid. They are all made to do one calculation per lens. But if you calibrate for the wide open setting (f1.4) which is the stop you paid all that hard earned money for the lens system's focus point will shift as you stop down. The setting at f2.8 when used on my Sony a99 is three or four points different than the f1.4 stop. In theory you could test and divine a calibration setting for each f-stop but there's no way to load more than one setting into the camera for each lens.

If you were amazingly compulsive you could calibrate all the critical stops and third stops and make a chart. Then when you grab your camera to shoot you can check your f-stop, consult your data and set the correct number of every f-stop. You'd probably only need to do the stops from wide open to about f4 because at that setting depth of field masks the errors. Mirrorless cameras set focus at the shooting aperture for their contrast detection AF so they tend to be much, much more accurate. I wonder if there is a downside to the inclusion of phase detection AF points on a sensor as relates to focusing point accuracy.

Why do I bring this up? Because I spend a few quality hours yesterday recalibrating all of my Alpha lenses for proper AF performance and the lenses with gaussian designs and modified Planar designs all exhibited this sort of obvious focus shift. My strategy for the 50mm 1.4 Sigma lens is to set the highest focusing accuracy at the most commonly used f-stop, which for me is f2..5. I actually fudged and set my for 2.8 because I tend to shoot a lot of stuff at f4. Now that I've re-calibrated the lens is much sharper at f4 than my previous experiences with it.

The nice thing about the Sony a99 camera is the ability to do magnified live view focus (with focus peaking) with manual focus lenses. With a quick magnification to 8x or higher it's easy to nail focus at any of the wider f-stops with a high degree of accuracy.

The focus shift is not a fault of one lens company or another, it's inherent in the type of optical design typically used in high speed lenses.

Just a head's up for people who are interested in micro-AF adjusting. You might start out by considering which of the fast apertures on your lens you use the most and start there. By the time you get to f4 you really are mostly stymied by the DOF of the lens. For my money, unless you consistently shoot wide open you might think of optimizing for the ultimate sharpness at two stops down.

You may find that your lenses are a lot sharper than they get credit for. Panasonic and Olympus shooters? This is one thing you just don't have to worry about.

Studio Portrait Lighting


Peter F. said...

Good... I'm glad you told me I don't have to do anything with my Olympus 45 F1.8 and 75 F1.8 on an EM1 and EM5 body. Right? This is because of the contrast detection?

Peter F.

Ananda Sim said...

Bear with me. I understand that the focus discrepancy is caused by the focus sensor being one thing (phase detect but that's beside the point) and the shooting sensor being another thing. And the focus sensor is reading the focussing screen? Or is the light going direct to the focus sensor?

With mirrorless cameras, whether the focus sensor is Contrast Detect or Phase Detect, the focus sensor and the shooting sensor is on the same plane and is the on same piece of silicon. There should be one less impediment then.

Kirk Tuck said...

Yep Peter. Not necessary with the CD-AF cameras. Might be another reason to start shooting the faster glass on the mirror less cameras instead of the dino-cams.

jlemile salvignol said...

Kirk, OK for Pana and Oly. Ditto for the V1 and the K-01.
Using my Pentax KA 85mm f/1.4 becomes a real pleasure after 15 years of disappointment and misunderstanding. Glory to the K-01 with a carry speed VF-3 of course!

Brad Calkins said...

Interestingly, the E-M1 offers pdaf adjust on EVERY focus point, for both wide and tele setting on the lens. With 37 af points you can configure 74 adjustments! I dridn't go anywhere close to that far, but I did setup the three central points so that left of center is back focused a bit, and right with a bit of front focus. Then I can bias it a bit ahead or behind for moving subjects. Floating elements seem to mostly eliminate what you are talking about, and the last lens I owned without one was the Zeiss 50mm Planar (if memory serves). The AF issue you mention was annoying to work around!

Gato said...

Interesting thought about on-sensor PDAF.

Actually, playing with my NEX 6 it appears to focus wide open, then stop down. So unless it checks again after stopping down there could still be a problem.

I remember back in my view camera days I learned to always stop down as much as possible before focusing to guard against shift.

Craig Yuill said...

I have a feeling I need to do some micro adjustments for one or two of my DSLR lenses. I was hoping you or someone else could mention a product that works well. Doing the micro adjustments two stops down from maximum aperture sounds like good advice.

Gato said...

Update to my earlier comment -- with Setting Effects on the NEX 6 views and focuses stopped down, so should focus shift should not be an issue.

Only with Setting Effect Off does it focus wide open -- something to think about when I'm using flash in the studio.

atmtx said...

The beauty of contrast detect focusing. Another disadvantage of the old DSLR world.

Michael Watkins said...
This comment has been removed by the author.