I benchmarked the WiFi throughput of the Z7 on various networks and connections method, using the 'Connect to PC' functionality and Nikon's Wireless Transmitter utility.
Based on a hunch I tested the transfers in two modes: 1) While it was in the 'Connect to PC' setup menu and 2) While it was in the shooting mode (Z7 supports background transfers). As expected, the background transfer mode is noticeably slower, likely due to the transfer competing for internal CPU time with the shooting mode
Setup
Sony 64GB 440 MB/s XQD
Infrastructure mode (camera connects to existing AP network): Netgear Nighthawk XRS R7800 (AC2600, 800 Mbits/sec theoretical). Camera 8 feet from router. WPA2-AES security.
Direct mode (camera creates AP network): TP-Link TL-WN722N USB WiFi Adapter (150 Mbits/sec theoretical). Camera 4 feet from network adapter
5Ghz Infrastructure:
In 'Send To PC' menu - Transfer time: 2:18 (2.34 MB/s)
In shooting mode - Transfer time: 2:55 (1.85 MB/s)
2.4Ghz Infrastructure:
In 'Send To PC' menu - Transfer time: 2:21 (2.29 MB/s)
In shooting mode - Transfer time: 2:53 (1.87 MB/s)
5Ghz Direct:
In 'Send To PC' menu - Transfer time: 2:22 (2.28 MB/s)
In shooting mode - Transfer time: 2:53 (1.87 MB/s)
USB attached to camera (Windows Explorer drag 'n drop):
In Setup menu - Transfer time: 0:15 (21.59 MB/s)
In shooting mode - Transfer time: 0:19 (17.04 MB/s)
XQD Card Reader MRW-E90 (Windows Explorer drag 'n drop):
Transfer time: 0:01 (> 300 MB/s)
For comparison, my D7200 gets around 2.40 MB/s in 2.4Ghz Direct mode (D7200 doesn't support Infrastructure mode).
Summary
The Z7 offers no improvement in WiFi speeds vs previous generation bodies that have built-in WiFi, even though the Z7 supports 5Ghz connectivity vs 2.4Ghz in previous bodies. This is likely due to performance limitations of the cameras' embedded processor / OS.
snapsy wrote:
I benchmarked the WiFi throughput of the Z7 on various networks and connections method, using the 'Connect to PC' functionality and Nikon's Wireless Transmitter utility.
Based on a hunch I tested the transfers in two modes: 1) While it was in the 'Connect to PC' setup menu and 2) While it was in the shooting mode (Z7 supports background transfers). As expected, the background transfer mode is noticeably slower, likely due to the transfer competing for internal CPU time with the shooting mode
Setup
Sony 64GB 440 MB/s XQD
Infrastructure mode (camera connects to existing AP network): Netgear Nighthawk XRS R7800 (AC2600, 800 Mbits/sec theoretical). Camera 8 feet from router. WPA2-AES security.
Direct mode (camera creates AP network): TP-Link TL-WN722N USB WiFi Adapter (150 Mbits/sec theoretical). Camera 4 feet from network adapter
5Ghz Infrastructure:
In 'Send To PC' menu - Transfer time: 2:18 (2.34 MB/s)
In shooting mode - Transfer time: 2:55 (1.85 MB/s)
2.4Ghz Infrastructure:
In 'Send To PC' menu - Transfer time: 2:21 (2.29 MB/s)
In shooting mode - Transfer time: 2:53 (1.87 MB/s)
5Ghz Direct:
In 'Send To PC' menu - Transfer time: 2:22 (2.28 MB/s)
In shooting mode - Transfer time: 2:53 (1.87 MB/s)
USB attached to camera (Windows Explorer drag 'n drop):
In Setup menu - Transfer time: 0:15 (21.59 MB/s)
In shooting mode - Transfer time: 0:19 (17.04 MB/s)
XQD Card Reader MRW-E90 (Windows Explorer drag 'n drop):
Transfer time: 0:01 (> 300 MB/s)
For comparison, my D7200 gets around 2.40 MB/s in 2.4Ghz Direct mode (D7200 doesn't support Infrastructure mode).
Summary
The Z7 offers no improvement in WiFi speeds vs previous generation bodies that have built-in WiFi, even though the Z7 supports 5Ghz connectivity vs 2.4Ghz in previous bodies. This is likely due to performance limitations of the cameras' embedded processor / OS....Show more →
BSPhotog wrote:
Here's a silly question, does Snapbridge work?
Yep, tried it today. Here are the benchmarks (Bluetooth). These are with the camera located 2 feet from my iPhone.
Time to begin transferring new photo: Approx 9 seconds
Time to complete transfer 2 Megapixel 1.0MB image: Approx 12.5 seconds (81.92 KB/s)
So the lag to begin transferring is a bit long at 9 seconds, but if you shoot multiple photos the camera transfers them all back-to-back (ie, you don't get multiple 9-second start lags).
I also tried measuring Snapbridge over WiFi but found it too buggy to use.
Thanks for the testing. I have a super basic question.. some people have complained that the image in the EVF does not show the actual taking exposure, but just a viewing image. So their captures are not what they expected for exposure.
Can the camera be set up so, for example, a half press of the shutter button shows the actual shooting exposure in the EVF? I can do that with my Olympus OMD and even Leica M with the EVF attachment.
That way I can adjust exposure compensation before taking a shot.
I'm assuming that all mirrorless cameras allow this, but...
Desmolicious wrote:
Thanks for the testing. I have a super basic question.. some people have complained that the image in the EVF does not show the actual taking exposure, but just a viewing image. So their captures are not what they expected for exposure.
Can the camera be set up so, for example, a half press of the shutter button shows the actual shooting exposure in the EVF? I can do that with my Olympus OMD and even Leica M with the EVF attachment.
That way I can adjust exposure compensation before taking a shot.
I'm assuming that all mirrorless cameras allow this, but...
Thanks!
Yes, there's a toggle that lets you switch between exposure preview and a constant EVF/LCD brightness, the latter of which is useful for flash photography. Nikon calls it "Apply settings to live view" on the Z7.
snapsy wrote:
Yep, tried it today. Here are the benchmarks (Bluetooth). These are with the camera located 2 feet from my iPhone.
Time to begin transferring new photo: Approx 9 seconds
Time to complete transfer 2 Megapixel 1.0MB image: Approx 12.5 seconds (81.92 KB/s)
So the lag to begin transferring is a bit long at 9 seconds, but if you shoot multiple photos the camera transfers them all back-to-back (ie, you don't get multiple 9-second start lags).
I also tried measuring Snapbridge over WiFi but found it too buggy to use.
Interesting, I was thinking about remote triggering via SnapBridge specifically. It is garbage on the D500, but hoping for an improvement. I believe that hinges on WiFi connectivity which, as you said, is prohibitively buggy.
Jim Kasson wrote some excellent manual focusing tips here. Here are some additional tips I offered:
The digital rangefinder in MF mode is based on what's under the active AF point. As such, be sure to enable the focus point display in MF mode, that way you'll always know what you're MF'ing to (a10 Focus Point Options -> Manual focus mode -> ON). If this setting is OFF then the AF point in MF mode is only displayed momentarily when you move the active AF point.
Another benefit of having the active AF point always displayed in MF mode is that it's color coded to match the rangefinder's focus indication. A green box indicates in-focus, a flashing red/green box means at/near focus, and a red box indicates OOF. That way you don't always have to take your eyes off the subject area you're focusing too, rather than looking down to the rangefinder.
Unfortunately the active AF point is only displayed at the zoomed-out magnification level. As soon as you start zooming in it disappears. This is important because if you pan around the screen (using the multi-selector control) then you're also moving the active AF point, and thus what the rangefinder is determining focus indication from, even though the focus point is not displayed. Although the active AF point is not displayed, it'll always be at the center of the screen in the magnified views since the Z7 keeps the screen centered around the active AF point you're moving as you pan. I suggest turning on the 3x3 grid (A9 Framing grid display), that way you can use the center cross-hair (intersection of horizontal/vertical grid lines) to know what the rangefinder is using for focus determination in the magnified views.
Irrespective of the AF area mode in AF mode (pinpoint/single-point/wide-area), the camera forces the area mode back to single-point in MF mode. It would be nice if this weren't the case, ie the rangefinder worked over a larger AF area.
A known technique on Sony FF MILC cameras is to use the exposure zebra feature as a quick and easy visual method for ETTR. The Z7 supports exposure zebras too but they're only available in movie mode (ie, when the stills/movie knob is set to movie). This is a hassle because the movie mode only supports JPG stills, so if you want to use this in your workflow you have to meter in movie mode but then switch to stills mode to shoot raws. More hassle still is that video/stills mode save their own exposure settings, so once you dial in your exposure in movie mode based on the zebras you have to re-dial them in after you switch to stills mode. Anyway, here's the process and results...
First, to enable zebras, to go Custom Setting Menu -> g Movie -> Highlight display. Set the threshold to 255.
With the camera pointed to a blue sky with a thin-layer of high clouds, I adjusted the shutter speed to just above where zebras can be seen (ie, no zebras), then adjusted the shutter in 1/3EV steps brighter, taking photos at each shutter speed.
Loading the files into RawDigger, the zebras pretty much matched the raw clipping, so like the Sony it's an accurate method. Loading the files into ACR provided a pretty close match as well. Here's a GIF animation showing the 1/3EV photos I took, starting at 1/100 (no zebras), 1/80 (zebras appearing near bottom of FOV), 1/60 (more zebras), and 1/50 (almost all zebras). The highlight display in ACR pretty much matches the zebras I saw on the camera.
As has been documented, the Z7 will focus at your shooting aperture, up to f/5.6, after which it'll keep focusing at f/5.6 and only stop-down to a smaller aperture (higher f/number) for the actual exposure.
A behavior that hasn't been documented is how the Z7 always momentarily cycles back to what appears to be f/5.6 after taking an exposure. For example on my 24-70 f/4Z, if my shooting aperture is f/4 the camera will focus @ f/4, shoot @ f/4, then after the exposure is over the aperture will stop down to what appears to be f/5.6 for a fraction of a second before returning back to f/4. It does this irrespective of 'Apply settings to LV setting' and also independent of the focusing mode (even does it in MF).
It might just be a firmware oddity. Or there may be a specific reason Nikon wants to do this. I can't think of a reason off hand.
One side effect of this behavior is that the camera is noisier than it otherwise needs to be, esp in silent shooting mode. It also puts more wear on the aperture mechanism than necessary.
snapsy wrote:
As has been documented, the Z7 will focus at your shooting aperture, up to f/5.6, after which it'll keep focusing at f/5.6 and only stop-down to a smaller aperture (higher f/number) for the actual exposure.
A behavior that hasn't been documented is how the Z7 always momentarily cycles back to what appears to be f/5.6 after taking an exposure. For example on my 24-70 f/4Z, if my shooting aperture is f/4 the camera will focus @ f/4, shoot @ f/4, then after the exposure is over the aperture will stop down to what appears to be f/5.6 for a fraction of a second before returning back to f/4. It does this irrespective of 'Apply settings to LV setting' and also independent of the focusing mode (even does it in MF).
It might just be a firmware oddity. Or there may be a specific reason Nikon wants to do this. I can't think of a reason off hand.
One side effect of this behavior is that the camera is noisier than it otherwise needs to be, esp in silent shooting mode. It also puts more wear on the aperture mechanism than necessary....Show more →
Actually, if you put camera in real low light, it will no longer focus at shooting aperture. I tested it with 28E under real low light at f2, the camera will try to open the aperture to do focus.
A method to do automated focus testing on the Z7 was established and demonstrated by me here. Short version - Nikon bodies provide a mechanism that allows both its Autofocus (AF) and Manual Focus (MF) to be precisely controlled under programmatic control over USB (and WiFi).
Regarding the MF control, there's a command you can issue to the camera that directs it to move the lens's focus position closer or further away by a specified amount relative to the lens's current focus position. The range of movement positions supported by this command are +/- 32,768, but the actual range of positions supported for a given lens is much smaller. For example on my 24-70Z @ 70mm, the range between infinity and MFD is about 4,400.
This gave me an idea - I wondered if this command could be used to measure how precise a lens motor's positioning mechanism is. By precise I mean how much the focus varies if I repeatedly direct the lens to focus from infinity to a fixed position.
The motivation behind this idea is to see how much shot-to-shot AF variation on a system is the result of lens motor variation vs PDAF sensing variation.
Before I continue there are a lot of provisos and pitfalls here. First, I can't be certain that the MF lens movements directed by this command would be the same movements the camera itself would command during an AF operation. For instance, whether or not the MF lens movement involves a different command/control path that might be more accurate than an AF movement. And of course an AF movement might involve multiple lens movements of varying distances, which itself might introduce movement variation not seen in a single movement. Base on some initial observation between my MF script and AF operation the movement appears the same to my eye, but that's just a cursory, non-scientific observation. Just to be clear - I'm far from certain that this experiment demonstrates what I'm hoping it demonstrates.
So here's the experiment. I set up my Z7 and 24-70Z to focus on a $100 bill near MFD. I tilted the camera relative to the bill to create thin DOF, to allow any differences in shot-to-shot lens positioning to be visible to the eye. A better approach would be to measure MTF like Jim does, but I don't have the seutp/scripts to do that yet.
My script repeats the following 25 times:
1. Focuses to infinity
2. Focuses to a fixed value that brings the eye of Benjamin Franklin into focus [this value was determined through trial and error]
3. Capture and download a basic JPG
I then loaded all 25 JPGs into PS and created an animated GIF out of it, using a 100% crop that includes OOF areas both nearer and further from the eyes, and then viewed at 200% of that crop. The differences are hardly noticeable, outside of some micro-movements between frames from me siting near the camera. Each frame in this animation is displayed for 1/3 of a second - to help distinguish frames I toggle an asterisk every other frame. Here's the animation:
t's clear the 50G has noticeably more focus motor variation than the 24-70Z.
Based on these results one might expect the AF variation would match the single-drive MF variation exhibited above, producing slightly different AF results shot-to-shot. In fact this doesn't happen. Here are the AF results for 10 images, again under scripted control, AF from infinity each time.
This demonstrates the self-correcting, partially-closed loop mechanism employed by an AF system. After command the lens to move to a specific focus position based on PDAF sensing (same as my MF drive test), the camera is obviously re-sensing PDAF and determining the focus element arrived at the wrong position, which then leads it to send another movement command to the lens. It repeats the process until focus is correct, or at least within a margin of error close enough to correct.
For the D800 I can actually hear this iterative process because it's relatively slow. On the Z7 the process is much faster.
