This is going to be completely off topic. Please skip ...
Originally posted by Class A Class A, thank you very much for pointing me to this.
I stand corrected!
I thought I considered what appears to be called Brenizer method (I wasn't aware of the name but already had the idea myself). Point is that I had a flaw in my thinking.
As a matter of fact, if you stitch e.g. 2x2 photos each taken at 50mm f/1.4 (with a combined FoV of a 25mm lens) then the CoC (which remains constant at say 10 pixels) is reduced to half the size relative to the stitched image height (because it doubled). I.e., DoF appears to be
less shallow than from a single 50/1.4. I.e., stitching
increases DoF.
From that, I drew the
false conclusion that panorama stitching can't be used to produce
more shallow depth of field.
My error is that I would have had to compare with a single 25mm lens image though. So, what aperture would create the panorama's CoC relative to image height (say 5 pixels using our example)? Although the CoC is smaller, it turns out that the aperture must be wider, f/0.7 to be precise. I.e., the effect only depends on the physical aperture diameter in mm. And of course, a 25mm f/0.7 would be quite a lens ...
It actually is a special case of lens equivalence, considering panorama stitching creates a sensor with increased size (and a lens with increased image circle).
Thanks for pointing my error out. Panorama stitching is indeed good to create shallow depth of filed.