htom
New member
Is there any difference in behavior and precision between the Single Point AF area and a 1x1 size custom box (configured for C1 or C2) when subject detection is turned off for the custom AF area? Thanks in advance!
Nikon Z8 Af Area modes Single point vs Custom Area
- Thread starter htom
- Start date
If you would like to post, you'll need to register. Note that if you have a BCG store account, you'll need a new, separate account here (we keep the two sites separate for security purposes).
mathjak107
Active member
the old school single point will not use any of the new features like subject detection , if you turn it off you are left with the old single point
htom
New member
Thanks, that is my understanding as well. However, if I take this a step further, the custom area prioritizes the closest object when acquiring focus (according to Nikon), which is why I’m asking the question.
mathjak107
Active member
i think though that with the 1x1 the subject area would be the same as single point af .. the difference being the custom box with subject detection will prioritize the closest subject .
i don’t see a difference once you turn off subject detection and use the same area box size
i don’t see a difference once you turn off subject detection and use the same area box size
htom
New member
Thanks! It might be because the 1x1 size is too small, and with SD turned off, it essentially restricts the custom area focus algorithm, causing it to behave similarly to single-point AF. As we increase the custom box size, it starts to function more like the older Group AF area found on DSLRs.
mathjak107
Active member
I use the 1x1 box with subject detection and it works well as it does what it should do
It Is like single point af on steroids
It Is like single point af on steroids
htom
New member
Me too. However, there are situations where I need 100% accuracy, so switching to single-point focus gives me peace of mind. My main AF area is the 1x1 custom with Subject Detection enabled, and I use the single point programmed onto Fn1 as AF-area + AF On. Now, I'll adjust my current setup to use RSF (hold) with Subject Detection off, programmed to the joystick center press, and assign 3D to Fn1.
fcotterill
Well-known member
I have not used Single Point on the Z9 for 2+ years.... since Nikon added the custom area modes (FW 2.0, April 2022), with the bonus of the 1*1 option.
It's rare that I need to turn off Subject Detection, so don't need single point.
The C1/C2 1*1 option should, if anything, have better Closest Subject Priority; well, at least from what little is known about the innards of the AF modes outside of Nikon's R&D laboratories.
It's rare that I need to turn off Subject Detection, so don't need single point.
The C1/C2 1*1 option should, if anything, have better Closest Subject Priority; well, at least from what little is known about the innards of the AF modes outside of Nikon's R&D laboratories.
Last edited:
The biggest differences I see between a custom C1/C2 1x1 area and Single Point relate to Subject Detection but even with SD disabled Single Point really won't move beyond the confines of that area but the larger area modes including C1/C2 can stray a bit beyond the displayed box. They can stray quite a bit beyond the box outlines when SD is enabled but as they support automated area modes where the camera attempts to detect and track a subject they're not as confined as an actual Single Point area.
Like some of the others I keep Single Point as one of my AF Area options for tricky subjects that confuse subject detection as a quick way to disable SD. It's pretty rare with modern Z8/Z9 firmware but there are still some subjects like Pika and Nutria with their large round ears that can fool eye detection and in cases like that I just press a button to switch to Single Point and by doing so implicitly turn off SD in the process.
Like some of the others I keep Single Point as one of my AF Area options for tricky subjects that confuse subject detection as a quick way to disable SD. It's pretty rare with modern Z8/Z9 firmware but there are still some subjects like Pika and Nutria with their large round ears that can fool eye detection and in cases like that I just press a button to switch to Single Point and by doing so implicitly turn off SD in the process.
htom
New member
Thanks for the input Dave, much appreciated! Similarly, I've used the single point as backup for tricky situations when the SD picked something else than the subject eye.
ajrmd
Well-known member
Clearly there is some kind of logic which the camera uses to determine searching beyond the confines of the selected custom area. How it prioritizes this or how far beyond the custom area hasn't been readily characterized. I can say that when I use the custom 1x1, I don't recall the AF searching much beyond it though I haven't tested it in circumstances such as elephants, hippos, etc. where traditional wide areas fail. In my limited experience with a custom 1x1 on subjects such as Sandhill Cranes where the WA tends to want to jump to the body, I haven't witnessed that occurring with the custom 1x1 box.
One aspect of this that has not been mentioned is the emphasis on Nearest Subject Priority for various Wide modes including C1/C2 etc. Wide modes will focus on the nearest target in the focus box - and this may be a leaf, blade of grass, or twig. Single - like Dynamic modes - does not incorporate Nearest Subject so it is a little more likely to pick up the background if focus is missed. Given the small AF box with both approaches, it won't make much difference, but in general I find the Wide modes pretty good at picking up the nearest target in a group or when there are alternatives.
As far as going beyond the borders of the AF box, subject detection prioritizes the subject, subject's head, and then the eye in that order. If the AF box is on the subject with subject detection active, the camera will normally pick up the head and eye of that subject. So you could focus on the chest of a baseball player, and it would pick up the head and eye even if outside the AF box by a small distance. I find this works best when the subject is slightly larger than the AF box, so it would not be suitable for close ups that fill the frame.
As far as going beyond the borders of the AF box, subject detection prioritizes the subject, subject's head, and then the eye in that order. If the AF box is on the subject with subject detection active, the camera will normally pick up the head and eye of that subject. So you could focus on the chest of a baseball player, and it would pick up the head and eye even if outside the AF box by a small distance. I find this works best when the subject is slightly larger than the AF box, so it would not be suitable for close ups that fill the frame.
ajrmd
Well-known member
That is the listed priority, though that's not how it works in the field. Say, I'm pointing the Wide Area S on a GBH/Sandy with the red AF area clearly over the eye. Almost 100% of the time, the SD (as indicated by the small white box) will jump to the eye, though the minute the BBF is pressed the green AF confirmation box will oscillate back and forth, frequently to the eye, often to the chest, to a highlight on the neck/bill, etc. Even when I am close to the subject, as I illustrated in another thread with a Trumpeter Swan which was large in the frame, the green AF box oscillated wildly all over the subject.
I really need an Atmos (should I start a GoFundMe?) to demonstrate this behavior on a video capture but I'll do my best to illustrate with a single frame capture. The Red box is the one recorded in the exif as the actual AF area. Again, this was shot in Wide Area S, squarely over the eye/head which is approximated by the pink colored box. Pressing BBF caused the, the green AF box to oscillate back and forth between the eye and multiple spots on the subject. As Steve and others have observed, Nikon (and other systems) struggle particularly with long necked birds. Unfortunately, this behavior occurs with other static birds where I frequently see the green AF box fixing on the chest or oscillating back and forth between the chest, a bright spot(s), and the eye, be the target a hawk or a goldfinch.
Interestingly, I encounter similar AF oscillations when shooting people too particularly in natural light, where one witnesses the green AF box dance between the eyes, forehead, nose, etc. Ironically, I never encounter AF oscillations when shooting metal birds (or canvas/lacquer) in flight with the airplane AF mode. Likewise, with well characterized animals (eg, a cat), in flat light, the AF oscillations don't occur. As the video below demonstrates the SD (gray/white boxes) followed by AF activation (green boxes) stick to the subject like glue. Now, if Nikon can develop similar recognition and stickiness for other subjects, I would be ecstatic.
In my mind, this leads me to believe Nikon has some opportunity to improve and refine its AI recognition and AF tracking abilities for birds, (some animals), and people.
You can only see EXIF info for this image if you are logged in.
This past weekend I was using Wide Small and Wide Custom with Subject Detection turned Off. Nearest subject priority worked flawlessly. I agree - the Subject detection on the Z6iii has missed in some rather obvious situations. My most recent testing suggests the problem is with Subject Detection - which simply needs a firmware update.
htom
New member
Indeed, subject detection is not as accurate as we may expect even if it’s ai trained. Nothing wrong with going back to old-school mode and nail the shot.
fcotterill
Well-known member
Since Nikon launched the Z9 launched end of 2021, they have updated its Autofocus system at least 5 times in Firmware. Two significant features have included the Custom Area AF mode added from the D6 (fw v2.0) in April 2022, and Aircraft and Bird Detection modes in Subject Detection (fw 4.10) in October 2023.
It's assumed each new software incorporated updated Deep-Learning algorithms from Nikon's latest iteration in training Subject Detection. It's far less clear whether/how Nikon changed behaviour of other AF modes, especially Closest Subject Priority in the Group family (C1, C2, Small, Wide) also Auto AF and 3D.
It's probably reasonable to assume Nikon's flagship Mirrorless camera has the most advanced AF system compared to the 3rd generation siblings of the Z9. Several corollaries follow....
An obvious fact is early reviews of these cameras are obsolescent, at best, unless the author has updated the results with the respective firmware.
Do reviewers succeed at comparing AF systems objectively? The ideal methodology would use advanced robotic camera operator(s) and targets, which test how AF systems differ between cameras, conditions etc. The individual photographer's skill level must be a critical factor, especially with muscle memory and experience at tracking moving subjects (wildlife, sports). This factor applies particularly to behaviour of the subject detection system: subject variations, besides tracking, lock on etc.
Although they use essentially the same AF system, the Z9 siblings - Z8, Zf, Z6 III - have received comparatively less attention in Firmware updates. The Z6 III and Z8 appear to be wanting, considering reports of behaviour of their subject detection subsystems. These could reflect software differences besides their respective hardware features.
It's assumed each new software incorporated updated Deep-Learning algorithms from Nikon's latest iteration in training Subject Detection. It's far less clear whether/how Nikon changed behaviour of other AF modes, especially Closest Subject Priority in the Group family (C1, C2, Small, Wide) also Auto AF and 3D.
It's probably reasonable to assume Nikon's flagship Mirrorless camera has the most advanced AF system compared to the 3rd generation siblings of the Z9. Several corollaries follow....
An obvious fact is early reviews of these cameras are obsolescent, at best, unless the author has updated the results with the respective firmware.
Do reviewers succeed at comparing AF systems objectively? The ideal methodology would use advanced robotic camera operator(s) and targets, which test how AF systems differ between cameras, conditions etc. The individual photographer's skill level must be a critical factor, especially with muscle memory and experience at tracking moving subjects (wildlife, sports). This factor applies particularly to behaviour of the subject detection system: subject variations, besides tracking, lock on etc.
Although they use essentially the same AF system, the Z9 siblings - Z8, Zf, Z6 III - have received comparatively less attention in Firmware updates. The Z6 III and Z8 appear to be wanting, considering reports of behaviour of their subject detection subsystems. These could reflect software differences besides their respective hardware features.
Last edited:
macwalter
Well-known member
I use c1 custom set to single point and a button programmed to turn off and on subject detection. One of Steve’s ideas. It works great.
ajrmd
Well-known member
One shouldn't have to, and that was almost unheard of with my Canon or Sony gear.
macwalter
Well-known member
I find subject detection to work pretty well in many situations. Not so well with small subjects in cluttered settings.
You do what is necessary to get the shot. If I can turn off subject detection and have a 95% success rate, I might do that. I've even turned it off when subject detection had a 98% success rate - one miss out of 50 previous images - because I had an upcoming shot at a trophy presentation that had one chance and could not be repeated. Some of the old school methods work very well.
Sometimes AF success is unexpectedly good. With the R5 a friend found they had nearly 100% success photographing turtles, but at the same session had 0% success with photographing snakes. Apparently the turtles were close enough to birds through evolution that the camera achieved focus on the eye, but snakes were different and did not work the same way. I made a test photographing brindle patterned dogs and found subject detection did not work at all - but it was 100% accurate for every other dog with different coloring.
macwalter
Well-known member
I have a hard time with snakes. Being invariably on the ground and often in unappealing positions. Also trying to get the whole animal in focus where even f16 doesn’t seem to be enough. in doing warblers if I’m close enough I will use cus single point and watch where the green box goes. If it goes to the eye I will try first with subject detection on then if it drifts to nearby objects turn it off. I take short bursts and refocus constantly. Maybe there is a better way but that seems to works best for me. I seem to remember the r5 was better than both Sony and Nikon in certain situations