Fujifilm GFX100S: Focus BKT Feature Shows 90% Failure Rate with Fujifilm GF 20-35mm f/4 and 35-70mm too
Related: botched design, focus stacking, Fujifilm, Fujifilm GF 20-35mm f/4 R WR, Fujifilm GFX100S, Fujifilm Lenses, Fujifilm medium format, Fujifilm mirrorless, imcompetence crisis, medium format, please fix this, please fix this camera, please fix this Fujifilm
If you think you’re getting a sharp INF-focus image with Fujifilm Focus BKT, you probably aren’t.
Issue: Fujifilm Focus BKT often ends prematurely, failing to capture at frame at infinity (INF) focus.
Defocus varies from just noticeable to obviously blurred to ocassionally OK. As if some junior engineer did integer-math rounding and rounded down, taking N-1 instead of N images.
Unless you manually focus at INF (warning: don’t depend on autofocus) and manually capture a sharp image for comparison, you have no basis for assuming that Focus BKT has delivered a sharp image at INF for the last frame. Your buggy 100MP camera instead captures a 14MP or 24MP or 32MP blurred final image for the focus stack.
Partly into my trip, I got suspicious of the final frame—too late to save my Tenaya Canyon and Dana Canyon focus stacks, the large majority of which were ruined by incompetent Fujifilm firmware. Too late did I discover the failure, both with the 20-35mm and the 35-70mm. Had I not discovered the problem, I would have had dozens more focus stacks with usharp INF frames in later days. But as it stands, Fujifilm ruined half of my work.
Fujifilm GFX Focus BKT was/is a demonstration of Fujifilm engineering sloppiness, with a 90% failure rate on my recent trip. I can say that with confidence by having (once I discovered the behavior) shooting a final INF frame manually and seeing an obvious improvement in sharpness over and over.
Do not count on Fujifilm GFX100S/100 Focus BKT to reach INF.
Do not count on Sony A7R V Focus Bracket to reach INF.
Your results will vary by lens, but I now state unequivocally that all focal lengths are subject to this problem.
Bugs happen, but a failure to test/verify is unforgivable
Especially after years have passed and users report the issue.
Companies like Fujifilm seemingly have no formal mechanism (that I am aware of) to report gross bugs like this. Insular engineering. Where is the rapid product improvement program? Does not exist.
Why don’t these companies test and fix their broken shit before shipping it to customers?!!!
Will it ever get fixed? Sony Focus Bracket has been broken since release with the Sony A7R V. And I reported on problems with years ago. And here we are.
Working around it
Your only recourse, whih must be done for every focus stack, is to take an infinity-focus frame. Be sure to use manual focus, because over and over I observe autofocus variability. You cannot count on Fujfilm autofocus even in 100% magnified Live View because roughly 1 in 4 times it will produce a slightly defocused image—easy to see by repeatedly invoking AF-ON.
But taking that frame requires tediously switching to single-shot mode from , then back again*. Making things take far longer than if things worked right. And that intervening time delay can ruin a focus stack because the lighting is changing—happens a lot. A ridiculous situation of design by jackasses.
Nikon is the ONLY company that does it right—Sony focus bracketing is a dumpster fire, and so is Fujifilm at all focal lengths.
* A make-work anti-design. Fujifilm and Sony your implementation is idiotica and pedantic, please just copy Nikon and make it practical to use in the field.
Example
This is a typical failure, but results vary from much worse to subtly less sharp. Most cases are simlar to what is shown below.
Below, frames 3/4/5 of a 5-frame focus stack at f/9. The GFX100S was set to take up to 29 frames, step size 4. This is typical of what the GFX100S delivered over and over.
Until I caught it, it ruined 25 or so focus stacks—bugs are understandable, but a failure to test and verify is unforgiveable. Damn you Fujifilm—you ruined a lot of my work. Work that can literally never again be repeated (August in a year that had all-time record snow likely to never be of this scale again).
Even at reduced size it is obvious that the last frame is out of focus! Luckily for this particular scene I had shot an aperture series just prior which shows what INF focus should look like ("INF").
