Z8 After One Week - What You Need To Know

[Steve]

I've read that…but then I've read that it isn't correct and I've not tried it in enough situations to make a determination on my own.

Going to DX can help - I've tested it multiple times and I thought I had it in a video out there where I showed it. If you have a distant subject, subject detection may see it in FX mode, but if you go to DX it may then go to the face or eye. I've seen it enough times to think it'1 100% true. FWIW, the a1 does not exhibit this behavior.

 

[Anjin San]

i think the reason we think this is the case is we have heard from a reputable source (sorry, now i can't remember exactly where) is that the subject detection works from the evf stream. this suggests that if it's bigger in the evf, it's bigger to the subject detection system, in the same way that if your exposure is such that you can't see a subject in the evf, the subject detection system can't see it as well either (and this part i think we've seen plenty of evidence for).

But doesn't the fact that the AF sites are the same mean that even if it is using the EVF stream and ignoring part of the pixels…it's not doing the AF calculations based on what is displayed on the EVF but on the stream of data coming off the sensor…which is the same level of detail/pixels on target/whatever we call it regardless of the fact that it's ignoring part of the stream. Not arguing with your statement…just that I had considered that already and it seems like ignoring half the pixels and AF sites doesn't make the ones that aren't being ignored work any better…the bird's eye has the same number of pixels on it in either crop. Maybe it's because none of us really understand how the AF really works at the algorithm level and it's doing calculations faster in DX…but I would think that might improve AF speed of lock on but not necessarily seeing the eye where it wasn't seen in FX.

The trouble with being a recovering engineer…is that we rarely want to hear an answer like 'just because'…we want to know why and even now 12 years post retirement and actually having not done any real engineering since I got off of submarines in 1988 I still don't like 'just because' as an answer…but I'm afraid that unless somebody from Nikon spills the beans in a manner that is allowed to be reported on we'll never know for sure whether it's true or not or why. It's the 'it's bigger in the EVF therefore bigger in the AF system' part that is missing some key detail as to why that would be the case.

 

[Anjin San]

Going to DX can help - I've tested it multiple times and I thought I had it in a video out there where I showed it. If you have a distant subject, subject detection may see it in FX mode, but if you go to DX it may then go to the face or eye. I've seen it enough times to think it'1 100% true. FWIW, the a1 does not exhibit this behavior.

OK then…it's gotta be something in the algorithm then…and I guess nobody outside of Nikon knows or maybe can tell us. But if you've seen it enough to think it's the case then I'm gonna take it as being correct even if I don't understand it.

 

[John Navitsky]

But doesn't the fact that the AF sites are the same mean that even if it is using the EVF stream and ignoring part of the pixels…it's not doing the AF calculations based on what is displayed on the EVF but on the stream of data coming off the sensor…which is the same level of detail/pixels on target/whatever we call it regardless of the fact that it's ignoring part of the stream.

my _assumption_ is that, yes, actually the AUTOFOCUS SPEED is probably actually the same, but the subject detection and AF systems work together. so, in it sees WHAT it wants to focus on more precisely and quicker.

basically, i think, it's the subject detection that works better. the af is probably the same, but the overall effect is that it is better

 

[ARClark]

I've read that…but then I've read that it isn't correct and I've not tried it in enough situations to make a determination on my own. On the face of it though…going to DX just means the outer pixels get ignored completely…so the actual number of pixels on the subject remains the same as in FX…and they're the same pixels with the same AF sites and all that stuff…so unless the fact that fewer pixels being paid attention to means that the AF calculation rate per second goes up then how it makes AF better seems strange to me. I can't say it does or does not happen…and IIRC Steve has stated that it appears to him that the Z9 AF is a little better in DX mode…but the recovering engineer in me wants to know if it is really factually better and if so why that is so. The little work I have done with it says that eye detection is a little better in DX…but that's just anecdotal since the sample size is small for me…and again with the same pixels on the eye and same AF sites why that would make it better I don't have a good grasp on. AFAIK the AF ops per second remains the same but that's the only thing I can see in the whole AF equation that might make DX mode better…and as I said the larger frame in FX makes it easier for me to keep the BIF in the frame and not clipped and I can move the DX crop anywhere in the frame later in PP.

Speaking of DX crop Steve…which button do you have it assigned to on the Z8…I wonder what you gave up to assign it there? And…if you actually said what I think you said about AF being better in DX for distant targets…do you have any insight from Nikon or anybody else or a guess as to why it's better?

I have it assigned to Fn3. I learned about it in a post, possibly at DPR. I’ll try to track it down. My understanding is that the camera will use the same view you see for AF purposes, and that by enlarging the DX crop, it can actually see the fine details better. This has resulted in the camera being able to focus on the eyes of a bird that it couldn’t focus on in FX mode. I could literally go in and out of DX and see the small focus boxes in the eyes come and go with each mode shift. I don’t have any documentation to support it, but I found it to work consistently.

 

[Anjin San]

I have it assigned to Fn3. I learned about it in a post, possibly at DPR. I’ll try to track it down. My understanding is that the camera will use the same view you see for AF purposes, and that by enlarging the DX crop, it can actually see the fine details better. This has resulted in the camera being able to focus on the eyes of a bird that it couldn’t focus on in FX mode. I could literally go in and out of DX and see the small focus boxes in the eyes come and go with each mode shift. I don’t have any documentation to support it, but I found it to work consistently.

Fn might be an option but for consistency between the Z8 and Z9 I would have to put it on Fn1 or Fn2 I guess rather than Fn3. Good thought though…

 

[Cameron T]

I have it assigned to Fn3. I learned about it in a post, possibly at DPR. I’ll try to track it down. My understanding is that the camera will use the same view you see for AF purposes, and that by enlarging the DX crop, it can actually see the fine details better. This has resulted in the camera being able to focus on the eyes of a bird that it couldn’t focus on in FX mode. I could literally go in and out of DX and see the small focus boxes in the eyes come and go with each mode shift. I don’t have any documentation to support it, but I found it to work consistently.

My guess is it's about relative size in the EVF and how sure it is that something is a (bird/face/eye/car/whatever).

The bigger it is in the evf the more sure it can be, because I think that nikon made the choice (correct or not, that's a different argument) to not spend tons of resource processing every frame if it didn't hit a specific level of confidence. Once it does, it seems to want to try to re-acquire it more again if it slips off (say, the bird turns its head away for a second).

 

[ARClark]

Fn might be an option but for consistency between the Z8 and Z9 I would have to put it on Fn1 or Fn2 I guess rather than Fn3. Good thought though…

Yes, I’ve been wondering where I’d put on a Z8 if I got one.

 

[RichF]

Thank you Steve, Wonderful information. I expect to have mine Friday. In this vid you were a fast talker, it was- Lets get this vid done, so I can play with my Z8 some more

You could always play the video at 3/4 speed or if you want faster talking, 1.25 speed

 

[RichF]

@Steve Great video. You covered all the major points. You results were very consistent with what I have heard out other reliable sources. One questions about your AF speed tests: did you run the test multiple times or only once?

 

[Ltgk20]

my _assumption_ is that, yes, actually the AUTOFOCUS SPEED is probably actually the same, but the subject detection and AF systems work together. so, in it sees WHAT it wants to focus on more precisely and quicker.

basically, i think, it's the subject detection that works better. the af is probably the same, but the overall effect is that it is better

I think this is it. The AF sensels work the same regardless of FX or DX mode, but the subject detection which tells the camera which of the sensels to use benefits from exposure changes and magnification in the video stream.

 

[James C.]

If the EVF of the Z8 vs OVF of the D500 doesn’t bother you (I’ve found the Z9 EVF very DSLR-like with added benefits), I think you’ll easily adapt to the Z8. With the Z8 in DX mode, you’ll feel right at home, and you’ll have the added benefit of the full frame when you need it.

I typically shoot in FX and crop later. But if you’re used to DX, either shoot in DX on the Z8, get closer to your subjects or use longer focal length lenses. Note that in those instances where you can fill the frame in FX, you’ll have more pixels on your subject than can be done in DX, potentially giving you a higher quality image.

I wouldn’t say “should”. You now have a choice. If DX works best, leave it in DX. However, when the subject moves in too close, you can instantly “zoom out” FX by simply pressing a programmed button on the camera.
• How might the IQ from my 500mm PF lens be different with the Z8, if at all?
No different, and your AF with the 500 PF and a TC will be faster. Plus, AF is snappy in decent light even with 1.7x and 2x TCS.

They’ll be in the 50MB range in FX, but shooting in HE* RAW in FX, which is what I use, they’ll be the same as your D500. Also, shooting in DX mode in lossless RAW, they should be of similar size your D500.

I will warn you that the Z9/Z8 AF is very different than your D850. Some things are better and some are worse. But that’s a whole other subject.

Thanks for the feedback.

 

[John_DE]

It does and I don't think there is a way to disable it. (Well, gaff tape )

Yep, I have a small piece of black gaffers tape over my Z9’s light. Never use the focus light anyway and certainly don’t want a blinking light when taking a night sky photo when out with other photographers….

 

[bleirer]

I've read that…but then I've read that it isn't correct and I've not tried it in enough situations to make a determination on my own. On the face of it though…going to DX just means the outer pixels get ignored completely…so the actual number of pixels on the subject remains the same as in FX…and they're the same pixels with the same AF sites and all that stuff…so unless the fact that fewer pixels being paid attention to means that the AF calculation rate per second goes up then how it makes AF better seems strange to me. I can't say it does or does not happen…and IIRC Steve has stated that it appears to him that the Z9 AF is a little better in DX mode…but the recovering engineer in me wants to know if it is really factually better and if so why that is so. The little work I have done with it says that eye detection is a little better in DX…but that's just anecdotal since the sample size is small for me…and again with the same pixels on the eye and same AF sites why that would make it better I don't have a good grasp on. AFAIK the AF ops per second remains the same but that's the only thing I can see in the whole AF equation that might make DX mode better…and as I said the larger frame in FX makes it easier for me to keep the BIF in the frame and not clipped and I can move the DX crop anywhere in the frame later in PP.

Speaking of DX crop Steve…which button do you have it assigned to on the Z8…I wonder what you gave up to assign it there? And…if you actually said what I think you said about AF being better in DX for distant targets…do you have any insight from Nikon or anybody else or a guess as to why it's better?

After Steve mentioned it in a video I tested it in my R5. In full frame I kept stepping back and using people eye AF until it failed. I then switched to crop and could then continue to step back and have success. I forget my exact number, but I think it was about another 30-40% improvement.

 

[padrepaul]

Thanks Steve, it answered my questions and like you I'll be pairing it with my Z9. Ordered with NPS last week, hopefully good news in coming days.

 

[NorthernFocus]

OK then…it's gotta be something in the algorithm then…and I guess nobody outside of Nikon knows or maybe can tell us. But if you've seen it enough to think it's the case then I'm gonna take it as being correct even if I don't understand it.

The only thing that makes sense to me is that the AF algorithm is only analyzing the data relevant to the DX cropped image which is roughly 40 percent of the data. By processing less data it simply works faster/better.

 

[Len Shepherd]

I wasn't really aware of the charging option you mentioned, I'm intrigued.

There plenty of these devices around with a leading name in the UK Anker.
Size varies with Storage ability.
As an approximate indication they are about the same size as a smart phone and a distinctly heavier than a single spare EN-EL-15 battery.

They can also top-up mobile phones, laptops etc.

 

[Steve Pumilia]

Thanks Steve that answers pretty much all of my questions. Seems like it fits the bill for my needs -- a Z9 backup/lighter travel camera.

 

[Geoff_G]

Fn might be an option but for consistency between the Z8 and Z9 I would have to put it on Fn1 or Fn2 I guess rather than Fn3. Good thought though…

I have the LFn2 button programmed for FX-DX switch but it may end up in the i menu

 

[Geoff_G]

The only thing that makes sense to me is that the AF algorithm is only analyzing the data relevant to the DX cropped image which is roughly 40 percent of the data. By processing less data it simply works faster/better.

Bingo, that was my thought as well.

 

[fcotterill]

But doesn't the fact that the AF sites are the same mean that even if it is using the EVF stream and ignoring part of the pixels…it's not doing the AF calculations based on what is displayed on the EVF but on the stream of data coming off the sensor…which is the same level of detail/pixels on target/whatever we call it regardless of the fact that it's ignoring part of the stream. Not arguing with your statement…just that I had considered that already and it seems like ignoring half the pixels and AF sites doesn't make the ones that aren't being ignored work any better…the bird's eye has the same number of pixels on it in either crop. Maybe it's because none of us really understand how the AF really works at the algorithm level and it's doing calculations faster in DX…but I would think that might improve AF speed of lock on but not necessarily seeing the eye where it wasn't seen in FX.

The trouble with being a recovering engineer…is that we rarely want to hear an answer like 'just because'…we want to know why and even now 12 years post retirement and actually having not done any real engineering since I got off of submarines in 1988 I still don't like 'just because' as an answer…but I'm afraid that unless somebody from Nikon spills the beans in a manner that is allowed to be reported on we'll never know for sure whether it's true or not or why. It's the 'it's bigger in the EVF therefore bigger in the AF system' part that is missing some key detail as to why that would be the case.

This was discussed last year here, according to feedback at that time by Thom H, " the exposure at the focus position is the one that will determine how well the camera focuses. If the subject under the focus cursor is well exposed, the focus system works fine. If your subject is backlit and underexposed—common in birds against sky situations—focus is going to be more problematic."
As I understand the processing - inasmuch as Nikon has revealed - this points to the AF engine using the image stream post-exposure to calculate focus position, using contrast patterns etc. It follows that all these processes happen within the EXPEED7 CPU. In other words the AF engine is mining the image stream fed on to the EVF

 

[arbitrage]

Of all the top end MILCs I've tried, some do recognize the subject/eye better in DX mode. This was most obvious on R5, R3, somewhat obvious on Z9.
On the A1 it almost works in reverse, it often will recognize the bird in FX but not in DX which is just mind boggling to me.
On the A7RV I noticed a different behaviour than the A1. It was now seeming to recognize in DX more than in FX just like the Canon and Nikon cameras. Although it wasn't as consistent as the Canon and Nikon with this behavior.

 

[MotoPixel]

Yes, I’ve been wondering where I’d put on a Z8 if I got one.

I have my Z9 set to switch between FX and DX with the "Record" button next to the ISO button. Still works for record when in Video mode. It's right where you need it when you want it.

 

[ElenaH]

But doesn't the fact that the AF sites are the same mean that even if it is using the EVF stream and ignoring part of the pixels…it's not doing the AF calculations based on what is displayed on the EVF but on the stream of data coming off the sensor…which is the same level of detail/pixels on target/whatever we call it regardless of the fact that it's ignoring part of the stream. Not arguing with your statement…just that I had considered that already and it seems like ignoring half the pixels and AF sites doesn't make the ones that aren't being ignored work any better…the bird's eye has the same number of pixels on it in either crop. Maybe it's because none of us really understand how the AF really works at the algorithm level and it's doing calculations faster in DX…but I would think that might improve AF speed of lock on but not necessarily seeing the eye where it wasn't seen in FX.

I noticed that soemtimes the second Expeed 7 stream (which coming from the sensor) lives its own life. Sometimes it gives me a AF confirmation on subject but by checking images I realise that it focused on background. Yes, it is true if the focus-box is a normal small in AF-C mode and not a pin-box in AF-S. and half-a-box hit the background or what so ever. But sometimes I am sure I focused on the subject ;-)

Lets assume, that the second stream got slower just for milliseconds becasue of writing speed on the card or buffer or some other reasons (environmental temperature, etc). Theoretically semething can shange during those milliseconds and the focus can be different than confirmed in EVF. Just a thought.

When there are two streams I, as a user, can think that the second stream is doing what the first is showing or what I can see. But there can be seomthing what I cannot see. The camera reaction is must faster that the reaction of my brain (perception and interpretation).

 

[John Navitsky]

I noticed that soemtimes the second Expeed 7 stream (which coming from the sensor) lives its own life. Sometimes it gives me a AF confirmation on subject but by checking images I realise that it focused on background. Yes, it is true if the focus-box is a normal small in AF-C mode and not a pin-box in AF-S. and half-a-box hit the background or what so ever. But sometimes I am sure I focused on the subject ;-)

Lets assume, that the second stream got slower just for milliseconds becasue of writing speed on the card or buffer or some other reasons (environmental temperature, etc). Theoretically semething can shange during those milliseconds and the focus can be different than confirmed in EVF. Just a thought.

When there are two streams I, as a user, can think that the second stream is doing what the first is showing or what I can see. But there can be seomthing what I cannot see. The camera reaction is must faster that the reaction of my brain (perception and interpretation).

my suspicion is that the informational overlays in the evf aren't always up to date.

so it think there are actually three things. the two evf streams, and the "information overlay" "stream". and it's pretty clear that latter one isn't super precise or locked to the content it's being overlayed on.

from what they've been saying, it seems like the evf stream is never delayed, which might be why the overlays sometimes don't match if they DO require processing that takes some amount of time.

ymmv