This is a rewrite of my post after detecting errors in my argument. I shouldn't do "back of the envelope" calculations at 2 AM when I am supposed to sleep

Originally posted by PentaxPoke My guess is that because it is a theoretical issue that does not show up in practice.

You are probably right in normal cases.

I am as much concerned about non-planar focal "planes" as I am about recompose DoF errors.

E.g., when I checked my DA* 16-50 for corner sharpness at 16mm fully open, I found MUCH better figures than any test, e.g., Photozone.de. But only if I manually focussed the corners using LiveView. Focussing with a corner AF point could help a bit here, too.

As for the recompose DoF error: you can [easily

] compute the effect. If you map the focal plane into subject space you see that recomposition moves the center of the focal plane on the surface of a sphere. E.g., 50mm on APS.C has 27° FoV. If you recompose by 2/3 of the distance from center to edge, this is a turn by 9°. This moves the subject away from focal plane by

x = d (1 / cos (9°) - 1) = ~0.012 d = ~ d/100 (or ~1% of d).

where d is the distance to the subject.

Now, this may induce a focus error if your DoF is smaller than 1% of the distance.

Let's check for a portrait situation:

d = 1.0m (50cm APS.C FoV) with 50mm and f/1.4: DoF extends 1.2cm in front and behind subject. The recompose error moves it by just 1.2cm, so DoF now extends from 0cm in front and 2.4cm behind subject.

So, you roughly entirely lost the DoF area in front of the subject in this rather extreme situation.

In a less extreme situation, e.g., d=3.2m or f/2.8, you would half the DoF area in front of the subject rather than entirely loose it.

In an even more extreme situation, e.g., d=0.7m or f/1.2, the subject would indeed move out of the DoF area and become unsharp even for non pixel peepers.

I guess you won't see it in a majority of your photography and personally, I happily recompose. But since you asked, I thought I'll do the math for you.

If this guy ever needs to provide a sample photo, it would have to be portrait shot from the side and focussing on the rear eye

Originally posted by Original, erroneous version: Originally posted by PentaxPoke My guess is that because it is a theoretical issue that does not show up in practice.

You are probably right.

I am as much concerned about non-planar focal "planes" as I am about recompose DoF errors.

E.g., when I checked my DA* 16-50 for corner sharpness at 16mm fully open, I found MUCH better figures than any test, e.g., Photozone.de. But only if I manually focussed the corners using LiveView. Focussing with a corner AF point could help a bit here, too.

As for the recompose DoF error: you can easily compute the effect. If you map the focal plane into subject space you see that recomposition moves the center of the focal plane on the surface of a sphere. E.g., 50mm on APS.C has 27° FoV. If you recompose by 2/3 of the distance from center to edge, this is a turn by 9°. This moves the subject away from focal plane by

x = 2 d sin^2(9°/2) = ~0.012 d = ~ d/100 (or ~1% of d).

where d is the distance to the subject.

Now, this may induce a focus error if your DoF is smaller than 1% of the distance.

Let's check for a portrait situation:

d = 3.2m (50cm APS.C FoV) with 50mm and f/1.4: DoF extends 10cm in front and 11cm behind subject. The recompose error moves it by just 4cm, so DoF now extends from 6cm in front and 15cm behind subject.

So, you roughly halved the DoF area in front of the subject in this rather extreme situation.

I guess you won't ever see it in your photography and personally, I happily recompose. But since you asked, I thought I'll do the math for you.

If this guy ever needs to provide a sample photo, it would have to be portrait shot from the side and focussing on the rear eye