This is one in a series of posts on the Fujifilm GFX 100. You should be able to find all the posts about that camera in the Category List on the right sidebar, below the Articles widget. There’s a drop-down menu there that you can use to get to all the posts in this series; just look for “GFX 100”.
I’ve discussed automatic focus shifting features in cameras many times. In the D850 and the Z7, Nikon calls the feature “focus shift shooting”. Fuji calls the implementation in the GFX 50S, GFX 50R, and GFX 100 “focus bracketing.” I think Nikon’s name is more descriptive of what the feature is there for than Fuji’s, but I’ll call it focus bracketing in this post. At the very bottom I’ll tell you why I don’t think it is apt.
In this post, I gave you a visual look at the focus bracketing step size, and I’ll be mor quantitative here. But I’m going to derive my results publicly and in detail, because I think the Fuji focus bracketing feature is fundamentally misunderstood by many. As an example, just the other day I got this question from a well-known web photographer:
What’s the right step size in the GFX 100 for f/8?
If that looks like a perfectly reasonable question to you, read on, and find out why it’s not.
In order to quantify the focus bracketing step size, I used the following setup:
The camera has a 110 mm f/2 lens on it and is 1.65 meters from the center of the ramp, which has a checkerboard on it that I use for computer analysis of the plane of focus. The ramp is lit from above and slightly to the front by a Westcott LED panel behind a diffuser.
I set the step size to one, and made focus shifted series at f/2, f/2.8, f/4, f/5.6, and f/8. I used a Matlab program that I had previously written to analyze the plane of focus shift in the object field (along the ramp):
The curves are approximately linear; which means that each step is the same distance. The distance changes faster for the narrow f-stops than the wide ones.
I then converted the focus shift from the object plane to the image plane (at the sensor), and changed the units from millimeters (mm) to micrometers (um);
Then I computed the diameter of the circle of confusion (CoC) implied by each shift in the focal plane:
The f/2 curve is noisier, because the focal shift in um is divided by the smallest number to get to CoC diameter. There’s a shift in the f/2 curve, too. I think the focus shift of the 110/2 near wide open is the main culprit. In order to calibrate that out and get a better estimate of the slope of the five curves, I fitted a line to each minimizing the least-squares error:
Regardless of the f-stop, the step size is that which produces about a 1 um diameter CoC. The camera is compensating for the f-stop to make your life easier.
Now you see why the question “What’s the right step size in the GFX 100 for f/8?” is poorly framed. The right step size doesn’t depend on the f-stop.
The step size that you chose in the GFX 100 gives you the implied CoC diameter in um.
A 1 um CoC diameter per step is extremely conservative, and you may never want to use steps that fine. A useful way to think about the CoC diameter for precision work is to compare it to the pixel aperture, which in the GFX 100 is about 4 pixels on a side. When the CoC is 4 um, about half the resultant blur comes from the pixel aperture, and half from defocus.
The way that I define focus bracketing, it’s taking several images on both sides of the estimated proper plane of focus, with the intention of picking one such image for use. Focus stacking is taking several — or hundreds — of images with the intention of blending them all in postproduction. Why is “focus bracketing” a misnomer as the name for Fuji’s feature? It’s because the Fuji user interface is lousy for that. To do focus bracketing right, you’d focus, and tell the camera how many steps of what size, and it would start the focal plane on the near side of the place you focused, have the center shot of the series on the plane where you focused, and put the last half of the shots further way. The way the feature actually works is you pick the near plane, and it moves the focal plane further and further away with each shot. This is great for stacking, but not for bracketing, since you have no way to precisely find the near plane in that case.