Consider the following grid:
What you’re looking at is point spread functions from my lens simulator on the lens axis. Defocus occurs across the rows, with the red square marking the nominally-focussed points. Spherical aberration (SA) changes from the top to the bottom, with the value of the SA for each below that row. The top four rows are overcorrected SA, which, somewhat confusingly but following optical convention, is indicated by negative numbers. The top four rows are undercorrected SA, which is indicated by positive numbers.
A few things to notice:
- The shape of the point spread function (PSF) is not symmetric with defocus, except in the row where the SA in zero.
- Changing the sign of the SA reverses the way the PSFs change.
- Actual best focus is not nominal best focus except where SA is zero. SA introduces focus errors.
With a Siemens star target, you can see the effect upon images:
Unsurprisingly, the same asymmetry is visible.
Now you know why undercorrected spherical aberration is prized by some portraitists. It has the ability to make the background deliciously creamy. Overcorrected SA makes the foreground look that way, which is not in general as valuable an attribute for photographers. Nikon has made lenses with the suffix DC that allows control of the SA, and they are valued by some photographers.
There is a measure of focus accuracy called root-mean-square (RMS) point spread function (PSF) size. here’s how it varies with defocus:
Here is an illustration of how defocus and SA interact:
Another way to look at it:
Leave a Reply