I did a lot of handheld shooting while on my recent trip where assessing AF accuracy is just an impression (eg with portraits of my daughter). In general, AF accuracy seemed pretty good, but how do you gauge that in truth, when shooting at f/8 or f/11 most of the time? You cannot. Stopping down masks any errors pretty well, so that most of the time things work out OK. And we just throw away the bad ones, shrugging-off the loss. Hence the cognitive commitment to ‘“excellent AF” is maintained without creating cognitive dissonance as to one’s adored camera.
On the tripod, I sometimes saw grossly-off focus and/or smaller soft-focus errors that led me to stick to my manual focus plan whenever working on a tripod. With both the Fujifilm GF 35-70mm f/3.5-4.5 and the faster Fujifilm GF 45mm f/2.8. For landscape at distance, that is a big deal, where there is no margin for error.
At closer range, such as with dim but not dark dusk, self timer, picture of myself and my daughter, the GFX100S misfocused twice in a row so badly that f/11 could not save things—focus landed 30 feet distant instead of at ~10 feet. Not even sure how you could design it that badly, given the solid surface that I gave it to focus on. Only because I checked and tried a 3rd time and coddled it as it focused and verified focus did I not lose that image, which meant something personal to me. You cannot go back and replay life events. Is a professional camera professional when you cannot rely on focus, when it claims focus has been acquired (beep, light) but is actually ruinously off?
Why can’t I turn off PDAF and get the camera to use CDAF? Better yet, a slower but high precision and accuracy focusing mode, instead of a fail-fast-and-pretend algorithm? And analyze the damned image post-capture, automatically, at the point of focus, why not? Let the hapless user know the camera screwed the pooch!
Do NOT rely on the GFX100S to focus accurately for landscape.
This becomes obvious particularly at distance. Where after AF, things looked a little soft in magnified Live View (even if focused magnified). When manual focus obviously improves upon AF, you know you cannot rely on the camera. So even if the camera delivers acceptable ≠ optimal results 80% of the time, that’s about 10X too poor—it’s a science fair project.
Dr S writes:
I agree! I discovered very early the AF failed at distance and used manual all the time. Funny, I just accept it because with a 100MP camera one should use a tripod.
Did you try eye AF with your daughter? I'm sure you did. Results? We'll see.
DIGLLOYD: Eye AF problematic on Fujifilm (unlike with Sony GM lenses where I can use the button on the lens—too much hassle to go back and forth while hiking.