Z9 AF Operation Observations by Thom Hogan

[Chris K]

OK, I just did my own test. I went into a small walk-in closet with a 60W-equivalent LED and pointed the camera at some low-contrast carpet and at a colored-striped notebook that I'd say is an "easy" target for an AF sensor. Exposure was 1/500 f/5.6 ISO1600. Cameras were in Single AF, Continuous AF.

First I shot my Z6+85/1.8S:

• Live view exposure simulation on (giving a dark EVF): The camera hunted and usually found both targets but I could not hear the motor tracking. If I moved the framing, the camera would re-hunt.
• Live view exposure simulation off (giving a "good" EVF exposure): The camera immediately found both subjects and I could hear the AF motor "jitter" as it tracked. If I moved the framing, the camera adjusted as you'd expect
• Live view exposure simulation on, but the ISO cranked to ISO6400: The viewfinder looked the same as when "exposure simulation" was turned off, and the results were also the same

I conclude from this test that the Z6's autofocus is affected by its live view. Which we all already knew.

Next I shot my Z9+85/1.8S:

• Live view exposure simulation on (giving a dark EVF): The camera immediately found both subjects and I could hear the AF motor "jitter" as it tracked. If I moved the framing, the camera adjusted as you'd expect. (As an aside, I noticed an occasional brightness "blip" when first pressing the AF-ON button, as if exposure simulation was temporarily turned off.)
• Live view exposure simulation off (giving a "good" EVF exposure): The camera immediately found both subjects and I could hear the AF motor "jitter" as it tracked. If I moved the framing, the camera adjusted as you'd expect
• Live view exposure simulation on, but the ISO cranked to ISO6400: The viewfinder looked the same as when "exposure simulation" was turned off, and the results were also the same

I conclude from this test that the Z9's autofocus is unaffected by its live view. Which is contrary to Thom Hogan's report.

I obviously can't say Thom is incorrect; I didn't see his test. But I'm seeing different results. Anybody else want to test?

 

[John Navitsky]

i think you want to add ISO adjustments to your test. ie, instead of upping the ISO to 6400, drop it down to 64 or something and see if it impacts the Z9's ability to AF. i suspect this is really what Thom is getting at

 

[Chris K]

In this room ISO1600 gave an EVF image that was so dark that I could barely make out an image. I could see the white target, but the carpet was very nearly "black". Upping the exposure (simulation) to ISO6400 gave a pretty nicely-exposed (simulated) viewfinder image, which incidentally looked very similar to the viewfinder image I got when exposure simulation was turned off.

Doesn't that cover the issue?

If not I can try again in daylight, but I figured a low (and artificial) light test would be a better, more extreme test.

 

[John Navitsky]

was the camera able to AF with the ISO set to 1600?

 

[Chris K]

Yes. ISO didn't matter for either camera. The viewfinder brightness affected the Z6's performance, but the Z9 performed the same no matter the ISO and no matter the brightness of the viewfinder.

I suspect based on this test, that if you adjust the "viewfinder brightness" setting that you might increase/decrease AF performance on the Z6, as well.

 

[John Navitsky]

interesting

 

[Chris K]

Yeah, I can't explain Thom's opinion, but I'm obviously not going to say he's wrong, because I don't know what his test was, nor if my methodology is sound.

I do disagree with the idea that live view is affected by exposure, however. EVF brightness, yes. But not photographic exposure.

On my Z6, I'm very familiar with the effect of a dark viewfinder on AF. Shooting macro with flash, I turn off exposure simulation in the EVF because the camera is great at macro AF, but only if the EVF is bright enough. All you have to do is make the EVF bright by turning off EVF exposure simulation. The photographic exposure is irrelevant, in my experience.

 

[John Woodworth]

Some thoughts:

@John Woodworth, your point is an interesting one: if the EVF images affects focus, Picture Control should also affect focus. So, does the camera focus better or worse if you were to set the camera to B&W low contrast versus color high-contrast? Seems like this would be easy to verify using your hypothesis!

Perhaps print out a rectangular card divided vertically into two colored rectangles of the same tone. Then focus on the card using Picture Control set to "Vivid" and again using "Monochrome".

I wouldn't think it would focus in Monochrome if the focus system was using the same inputs as the EVF. In Monochrome, the two colored sides of the card would just show up as the same shade of gray so the card would have no feature to focus on.

 

[eaj101]

It seems that we don't know *at what point* in the EVF data stream AF occurs. Many of these suppositions could be true (e.g. AF occurs before picture control). I think what's important is less the technical implementation and more the functional impact. At its simplest, I think Thom is saying that the camera's ability to *create* an acceptable exposure, through ISO or gain or something, affects AF.

 

[Chris K]

Agreed. In the end, knowing how it work isn't as important as knowing that it does work, and how to make it perform.

Personally, I'm satisfied that I don't need to turn off exposure simulation to get good AF on the Z9, and that's why I prefer it for use with flash over my Z6.

 

[EricBowles]

I found the observation that the Z9 af is better if the image displayed in the VF is properly exposed interesting: https://www.zsystemuser.com/nikon-z-system-news-and/reader-questions-answered.html

I read his comments a little differently.

The sensor of the Z9 has a dual readout. The camera has a fast processor. The smaller amount of data directed for the EVF represents a partial readout of the sensor. It's adequate for an EVF and also adequate for AF, but it's a fraction of the data captured by the sensor for creating the RAW file. This allows both the EVF and AF refresh rates to be much faster - because the camera is processing a lot less data. In addition, the AF Area mode selected can direct the camera to the specific portion of the frame being used for focus. If a smaller portion of the frame is used for focus, the data is processed more quickly than using the entire frame, but it is still a limited readout of the dual readout for that specific area.

The way the data is processed requires contrast and in some situations also requires color. In low light, there is less contrast. As Thom suggests, focusing on the bright areas or bright elements of a dim scene provides more contrast than dim, low contrast areas.

I'm not sure about the impact of ISO. Thom suggests raising ISO is a strategy to improve AF in dim conditions - which could be possible. It's just a matter of applying gain to the readout used for the EVF and focus. But gain or any type of signal amplification is applied globally, so there is little reason why it can't be applied to AF automatically and outside the EVF - but it might slow processing.

I don't think the Picture Control setting is going to make any difference in AF speed. It's a post processing choice. It does affect the EVF and uses the same readout. But if increasing contrast or similar techniques are useful, it would be better for Nikon ot simply implement that in the software design for AF than to have various users coming up with their own approaches which are likely to be more wrong than right. I suspect any test that is positive is driven by some other factor such as contrast or brightness in the focus target.

 

[Ken Miracle]

First what @EricBowles said.

One of my favorite features in my Z9 having a live histogram in the EVF. I can quickly adjust for the neutral gray metering of the camera as needed with quick Exposure Compensation (EC) changes that I can see happening in the histogram. I use the lens control ring for EC on my Z100-400 and Z800PF. and the EC button on top for other lenses.

I have noticed that when I have moved my histogram to the right to a more "proper exposure for me" hence usually more light to the sensor that the Z9 AF does seem to be better in the field but that is just anecdotal observations after using the Z9 and the Z800 pf and Z100-400 in varying light from very bright to snowing and low light of overcast morning and evenings.

 

[Calson]

Autofocus relies on contrast detection and so more light reaching the AF sensors is going to help the autofocus system. EV gain with slower shutter speeds or a high level of signal amplification is not going to help the sensors, only adding more light with a larger aperture.

With my D1x there were situations where I needed a f/1.4 lens to have the autofocus work at all. IR AF assist from a speedlight was needed for slower lenses. It was not until the arrival of the D3 that I could get away with using only f/2.8 zoom lenses.

Autofocus still depends on contrast and works better with subjects that provide this as with a leopard as compared to a white Charolais cow or even a GBH.

 

[Ken Miracle]

Autofocus relies on contrast detection and so more light reaching the AF sensors is going to help the autofocus system. EV gain with slower shutter speeds or a high level of signal amplification is not going to help the sensors, only adding more light with a larger aperture.

With my D1x there were situations where I needed a f/1.4 lens to have the autofocus work at all. IR AF assist from a speedlight was needed for slower lenses. It was not until the arrival of the D3 that I could get away with using only f/2.8 zoom lenses.

Autofocus still depends on contrast and works better with subjects that provide this as with a leopard as compared to a white Charolais cow or even a GBH.

Some great points ... and I know in some cases I have moved my exposure to the right in the histogram by opening my aperture if I wanted to cut back background noise or lower ISO and AF appeared to be a bit better in the couple of cases that stuck in my memory this week. However for most of the week I was wide open with aperture unless I was accommodating multiple birds or something very close where I needed more depth of field.

Why would higher ISO signal amplification not have a possible impact since it is impacting the sensors sensitivity to light ... after all the new Starlight View is said to make AF possible at night but again I have not done it ... I do know Nikon says it increases the light going to live view but the rest is a mystery to me ... you, @Steve and others understand that stuff better than I do

 

[bleirer]

Some great points ... and I know in some cases I have moved my exposure to the right in the histogram by opening my aperture if I wanted to cut back background noise or lower ISO and AF appeared to be a bit better in the couple of cases that stuck in my memory this week. However for most of the week I was wide open with aperture unless I was accommodating multiple birds or something very close where I needed more depth of field.

Why would higher ISO signal amplification not have a possible impact since it is impacting the sensors sensitivity to light ... after all the new Starlight View is said to make AF possible at night but again I have not done it ... I do know Nikon says it increases the light going to live view but the rest is a mystery to me ... you, @Steve and others understand that stuff better than I do

The thing is ISO does not make a sensor more sensitive to light. There is only one sensitivity, the base ISO level and everything over that is gain applied after exposure, but if the z9 AF is using part of the feed from the sensor to the evf it seems to be plausable that exposure settings impact AF.

 

[Alistair]

The thing is ISO does not make a sensor more sensitive to light. There is only one sensitivity, the base ISO level and everything over that is gain applied after exposure, but if the z9 AF is using part of the feed from the sensor to the evf it seems to be plausable that exposure settings impact AF.

This.
ISO does not affect exposure and Thom emphasises exposure as being the relevant variable. His is an interesting theory and one worth further examination.

 

[John Navitsky]

to be fair, that’s kind of a pedantic perspective. what most people mean when they point to iso, they typically mean with all other variables the same. sure that might be better termed exposure, but both are trying to say the same thing

 

[Alistair]

Call me pedantic if you will, but the distinction is rather important. Exposure refers to the amount of light a sensing plane is exposed to. There are only two ways to affect exposure. There is no point to then discuss and conduct experiments on a third variable not affecting exposure.

 

[John Navitsky]

i’m not calling you pediatric, just pointing out people who said iso generally meant exposure

 

[bleirer]

The original post though refers to focus being better when the viewfinder is bright enough, no matter how it got bright, including iso. I think we all agree iso is not exposure. Here is a direct quote: "...Some of you may remember that one way to deal with focus in low light was to either switch to a higher ISO or to move to a manual exposure that was intentionally set too high..." which shows he thinks iso as well as exposure works, if it brightens the evf.

 

[dwillar]

shutter speed, aperture, ISO = ?

DOH "The Exposure Triangle" or "Camera 101 Basics"

Is there anything in the Z9 manual that gives definitive answers as to how AF works or is it locked up in a safe with other proprietary information?

I asked Dr. Spock. His answer: Illogical!

How can one have a "proper" exposure if one does not have a subject in focus?

Me thinks this entire thread is more than a wee bit out of focus.

Intertaining, yes, but none the less grosely OOF!

 

[Chris K]

Yeah, Thom is clearly talking about viewfinder brightness, and the things you might do to increase viewfinder brightness. I think the rest of us are as well?

I'll agree with John and say that ISO is a good stand-in for exposure in this particular situation. Shutter speed and aperture change the mount of light hitting the sensor, and thus create more variables that confound an AF test. ISO is simply the "volume dial" for the sensor, and much better to use to see how the camera reacts to viewfinder brightness.

I think my earlier test turning live view exposure simulation is even better confirmation, if I may say so. One setting makes the viewfinder either bright or dark, but has NO other effect on the photo or the amount of light hitting the sensor. The Z6's AF performance changes when you flip this setting, and the Z9 stays the same. I think that's all the evidence you need that the Z6 is AF'ing off its EVF image, while the Z9 is not.

In any case, if Thom were speaking about exposure, then I think we would all just say "duh" and walk away. More light = better AF. That's universally true of all cameras in all circumstances. (Well, up to the point you overwhelm the camera with too much light for its sensor.)