Oly MFT lens for shallow DOF

The DoF you get is defined by math. So it seems we cannot disagree???

The problem is in the definition of the circle of confusion.

Look at the image below. (This won the teddybear challenge on dpreview.)

The bear is about the size of a human head. This shot was taken at 34mm f/4. You would argue that the bear must be completely within the DoF and everything must be sharp. DOFMaster says that at 34mm f/4 and a camera to subject distance of 1 meter the total depth of field is 10cm. The reality is that DOFMaster computes using a circle of confusion that is something like 3 pixels large. And indeed, if you look at the full size image you will see that most of the bear isn't blurred by more than 3 pixels. But I don't call that pin-sharp.

The nose and eyes of the bear are pin sharp but if you look not 5 but just 2 to 3 cm behind the eyes there is already serious blur. My experience with DOFMaster is that when it tells me the DoF is X then in reality the DoF is less than half of that. And as a result the calculation of hyperfocal distance is also way off because it accepts a blur at infinity of 3 pixels.

Now replace the bear by a human head and let's increase the distance to 1.5 meters and slap on the 45mm at f/2. You say the head will be sharp. DOFMaster says that everything in a range of 6.5cm will be acceptably sharp. That is probably about right: if you focus on an eye then everything within 3cm in front or behind that eye will not be blurred more than 2 to 3 pixels. But I don't call that sharp. The reality is that the tip of the nose and certainly the ears will be very clearly unsharp. What a DoF calculation calls "acceptable focus" is not acceptable in my book. If you accept up to 3 pixels blur as being sharp then you are right that f/2 is good enough for a head shot. My experience is that for a head shot I am not satisfied even with 45mm f/2.8.