High-speed Lens Focus Shift Explained
I really love shooting with high-speed lenses, like my Nikkor 85 mm f/1.4 AF-S. In some ways, these lenses are like finicky race horses; they aren’t always as well-behaved as you’d like.
The Nikkor 85 mm f/1.4 is legendary for its beautiful out-of-focus rendering (bokeh), combined with being sharp at the focus plane. This beautiful bokeh is achieved by a lens design that avoids the use of any aspherical lens elements.
The price paid for this bokeh is an effect called “focus shift”, caused by spherical aberration. The outer portions of the lens will focus the rays of light a little differently from the inner portions. If you stop the lens aperture down, those outer light rays are cut off, and don’t contribute to the image.
The spherical aberration effect results in the best-focus plane to be located at what’s called the “circle of least confusion”. As you stop down the lens, the “circle of least confusion” shifts, until it stops shifting at typically f/4 or so.
If Nikon engineers had used aspherical lens elements in their design, they could have virtually eliminated any spherical aberration. This would have given the lens even higher resolution at large lens apertures (with virtually no improvements at smaller apertures). All designs involve trade-offs, though.
Lenses with aspherical lens elements translate into worse bokeh, all else being equal. You start to notice that out-of-focus blobs look like sliced onions, with concentric rings of light-dark patterns. Lights that are visible in the background at night really emphasize this effect. The outer edges of light blobs should gradually melt into the background; they shouldn’t show a ring of light around the edge of the blob.
This aspherical tendency is only a generalization, however. As computer modeling gets better, lens bokeh is getting better with lenses having aspherical elements. Sigma, for instance, has a single aspherical element in their 85 mm f/1.4 Art lens; its bokeh can’t compete with the Nikkor, in my opinion, but I can’t say it has ugly bokeh, either.
Circle of least confusion with spherical aberration
The diagram above lets you visualize what happens as you change the lens aperture. The plane of best focus is located at the “circle of least confusion”, where the light rays get focused into the narrowest bundle. This light bundle always has a non-zero diameter, but gets best at around f/4.0 on the Nikkor 85 mm f/1.4 AF-S.
I got the diagram above (I added the labels and arrows) from this site. Many thanks to this organization for making a great graphic depiction of the “circle of least confusion”.
The circle of least confusion travels from left-to-right in the diagram as you stop the lens down. With a small aperture, the light rays near the outer portions of the lens get cut off, and the remaining rays (which are consistently focused at a point) now predominate. At a small-enough aperture, focus shift stops.
If you keep stopping the lens down, then diffraction starts to take over. The light ray bundle starts to expand again, although it no longer shifts. Resolution starts to degrade in proportion to the expansion of the light bundle.
Note that spherical aberration is a result of lens design, and doesn’t correspond to manufacturing variation. You won’t find a lens copy that eliminates spherical aberration, so you can stop looking for one.
Nikkor 85mm f/1.4 AF-S lens elements
The picture above is from the official Nikon web site, showing the lens elements, which are pure “spherical” shapes. Spherical lens elements have a constant radius on each surface, which makes them much easier to grind than an aspherical surface. The constant radius translates into smooth out-of-focus backgrounds.
When I do focus fine-tuning on my camera, I have to note the aperture that corresponds to each fine-tune value (from wide-open until about f/4.0). Unless you shoot using contrast-detect (live view), you’ll need to change the fine-tune value to match the aperture, or else your pictures will be slightly out of focus. Phase-detect auto-focus uses the lens at its widest aperture, which is why you get the focus error. Contrast-detect uses the shooting aperture, which is why you don’t get any focus error in that mode.
Some sample calibrations for my cameras look like this:
D7000 f/1.4 = tune +1, f/4.0 = tune -4
D7100 f/1.4 = tune +12, f/4.0 = tune +8
D500 f/1.4 = tune +3, f/4.0 = tune 0
D610 f/1.4 = tune +7, f/4.0 = tune +2
The focus shift consistently moves away from the camera as the aperture closes, so the “+” focus-tune needs to decrease to compensate as the aperture closes. As a result, the fine-tune value needs to be decreased as the lens is stopped down. After f/4.0, the focus shift is no longer noticeable.
Focus calibration chart, f/1.4, left side rotated further away
Focus chart, f/4.0 showing focus shifts further to the left (away from camera)
You can see in the focus charts above how the plane of focus shifted to the left (away from the camera) when stopping down.
To fix this, the focus fine-tune would need to change from [+7 at f/1.4] to [+2 at f/4.0] for this camera (to shift focus toward the camera). The focus chart was rotated to 45 degrees, with its left side further away from the camera.
I used the MTF Mapper program to analyze the photos of the focus chart. It makes it really easy to locate where the focus plane is. It also lets me know how much sharper the lens is when I stop down the aperture!
In case you thought of this, I tried to press the “depth of field preview” (Pv) button while I focused. The theory here is that the lens would be stopped down for the phase detect, to eliminate focus error. Unfortunately, the camera refuses to focus while the “Pv” button is pressed. Oh, well.
I try to pay as much attention to the backgrounds as I give to the main subject in photographs. Bokeh is really, really important to me. That’s why I love my 85 mm lens so much that I’m willing to put up with its annoying focus shift. If only this lens had vibration reduction…