Z9 EVF and Camera Latency: Better then R3 or A1?
Vložit
- čas přidán 27. 12. 2021
- Like my content? Support me and my reviews via Kofi!
ko-fi.com/zpeaktures
Instagram: / zpeaktures
Facebook: / zpeaktures
Sound: www.bensound.com
Contact me:
zpeaktures.squarespace.com/co...
Video Shot with
Canon C200
Canon CN-E 18-80mm / Sigma 18-35mm F1.8
Sennheiser MKH 8060 - Zábava
I want a z9
Great video. Eye opening actually. Especially the lag test.
Very good video!
great video! Thanks
Awesome video Thanks ♥
Great videos! Thank you for doing this. More testing comparing autofocus would be awesome
Trying to get a free and willing party to help me in my autofocus test haha
love ur detail review bruh
very interesting test. thank you!
Very informative video: thanks for putting in the effort! In my experience, this latency is a very real thing, especially when using strobes. I work around it by pressing the shutter slightly early to get the peak of the action (on fuji GFX50R). Faster response would make life easier.
For the Gfx system, I’ll recommend u don’t look through viewfinder. For gfx50R maybe consider opening two eyes. That’s how I use mine . Viewfinder to compose and left eye to time the shot .
Great test, thank you!
hopefully its of some use .
Nice review. Waiting for next video.
great video
Keep it up
Thank you very much ,, ur video is extra unique ,, I got to learn more ,, please do upload a video soon ,, like to learn more
Thank you for testing this. IMO, the most relevant thing is “real world” or ‘in the hand’ experience. Much like your off/on latency comparison between z9 v A1 in your last vlog. Unless you have a machine that can fire off all 3 shutter buttons simultaneously, I happily accept that everyone will have different reaction times, much like the theoretical 0 - 100 kph (0 - 60 mph) car acceleration times. However, subjective and/or a perceived difference is still what counts.
A1 can change the EVF for higher resolution at the expense of a smaller EVF and battery life. Perhaps this feature is programmable to a function button on the A1? I’m not sure I would assign a Fn button just for this - smaller EVF, higher resolution/hertz - but I still appreciate that every photographer has their own preference, needs, wants and use. Some photographers will not be bothered by latency, vertical grip, deep buffer, rugged body, weather sealing, etc. Some of us are … Each to their own and respect.
Thanks and keep up the good work!
I can say in real world it doesn’t matter too much haha. If put into car theory, it’s like 0-100 in 4 vs 3.9. Marginal and maybe your own skill matters more. Perceived is another thing, all part of the user experience. That said I think 120fps is just nice really. At least I don’t think our human vision can tell more then that.
But most importantly, we now live in the world of mirrorless where we are talking in double digit millisecond difference. In real world in the field, the ability to anticipate probably matters more and everything else is how you feel and prefernce
@@ZPProductions
I’m a bit confused by your comments and thus why undertake subjective comparisons? Is it just to gain CZcams ‘likes’ and ‘subscribers’ then?
Certainly for me, just the camera off/on latency time is crucial. My camera stays on standby after I turn it on but there are times I want to capture a shot ‘straight off the mark’ for nature and wildlife.
Heavily investing in an ecosystem - Sony, Canon, Nikon, Fujifilm - is the main reason why most prosumers don’t change their gear every few years or “latest” release. Even Nikon F-mount users (not necessarily those producing YT content) will need some convincing or perhaps have already changed and won’t want or need to change back.
Each to their own and respect.
Oh it’s purely curiosity and discussion. Think of it as purely tech geeking. YT is kinda my outlet. That’s why I comment on most post just for discussion purposes haha.
Excellent test. There is no better demonstration of the strength of the Z9 on the internet currently. I can imagine EVF latency will be the battleground of flagship cameras for generations to come, with particular attention due for the upcoming Canon R1.
exactly
Hi there! This was a great video as I'm mostly now concerned about the shooting experience when it comes to committing and moving to a mirrorless system. Since this video and the updates to the Z9 with the higher refresh rate for the EVF, have you noticed any considerable difference and an even more improved experience compared to its rivals? Thank you!
I briefly tried, I did not gain any significant better response time with it. It still does put z9 slightly ahead to me and in the field , my experience still puts z9 the most responsive from view to click. But this is not picking and any of the 3 brand flagship in ES mode are really close :)
Крутой тест, а если сравнить z9 с зеркалкой Никон допустим d780 какая разница в задержке?
In theory in order to remove the human element, you could to set up a second camera to film (hi speed) or photograph both the EVF/back LCD image and the monitor at once, and then you could measure the processing delay by subtracting the number visible in the viewfinder from the one on the monitor.
Of course that wouldn't account for potential differences in delays from pressing the shutter button until the image is acquired.
I did think about that solution but the issue is the evf is really small and pretty much very close. I don't think there's any known system that achieve the DOF to see both in relatively clear terms. Using two cameras even of the same brand will induce other synchronisation issues.
The other issue is the delay of the capturing, may need some rig with motorised actuator to actually be fully accurate.
But that said maybe something to ponder on but since only high end flagships have such low latency, I don't quite think its really worth to create a whole rig to do it haha.
Ok probably best to reverse this logic here. If your reaction time is 230 vs 250 ms that means you have to predict about 10% better to keep the subject in frame while doing a burst. But if you have it in frame the different fps means given that delay you have in first shot will be followed by a shot either every 50ms or 33ms. So in the end it means to me that the 'old' delay we had on evf's has become so low its easier to adapt to stay ahead and the z9 you need to train yourself a little less (about 10%). Also would be nice to see the delays tested with the different quality and speed (evf) settings of these cameras. Great work just not 100% sure how to process it for real use. I guess best way would be to see it as a complete system and do end to end testing on real subjects and see how well you get the first and following shots.
Pretty much agree with you. There are not many scenarios I feel will truly benefit from this greatly between the 3. It does give that perceived advantage I guess. The only scene I can imagine at least for portraiture is studio in motion shots. Or anywhere you can't burst and very opportunistic in nature like street. That said as you say less than 10 % difference in real world may not make any difference at all. With probably a max difference of 20ms, that's quite fleeting and about 1 frame in 50fps.
good vid mate. how does the latency time compares to best DSLRs you think, like d6-d5 etc
The ovf has no latency but the shutter and mirror flipping does have it. Based on a comparison I did with my r5 running on mechanical shutter, it takes a considerable time.
My preliminary thoughts is u have no optical latency (ovf) but the mechanical action and firing latency will probably be greater. As such I believe that dslr on firing the first shot maybe slower, but when it comes to tracking it may be as good or even better then this guys but with a lower fps.
@@ZPProductions oh good i'll check that vid.
Is there a way to capture both the actual stopwatch and the evf in the same frame? The time difference between what the stopwatch displays and what the evf displays will indicate the evf lag.
That's something I responded in another thread. Mainly is evf is very small and close, so no camera will have the dof to do it unless I have a stop watch to the size of of a say a TV then just capture it fuzzy to get a rough outline for the numbers. Maybe something to explore.
The other thing is the capturing process, Nikon Dual stream isn't just about lowering EVF latency but also the camera capturing latency. Now the capturing part of things is probably harder to test as you will need to hit a mechanical shutter button and probably can't be done without extensive rigging to remove the human element. Since the responsiveness of the camera is pretty much a combination of this two, just testing evf latency without the capturing one may not be a complete one. Pretty much why the title is EVF and Camera latency.
Sony user here (a7iv) I love this competition because is what really push manufacturer to do better . Can you compare the Lcd screens?. I only do photos and I hate flipping screen of a7iv, would love a z9 kind:)
The z9 implementation is the best for photography. But the R3 screen is the nicest hahaha. Most pixel and sharpest
that is the good thing about the z9, you can get 20 fps with any lense and not just special lense like sony but if you want the best buffer you need the fastest cf express card only if you shoot in raw losless compress. If you shoot in lossy raw, like many people do for action then the buffer is not a problem. Ricci talk got 1600 photos in the highest lossy 14 bit raw format(he*) with the z9
Actually when I was trying out the buffer is rather Low when choosing certain cards . SanDisk 128 gb only gives a buffer of 48 shots for lossless and 92 for he*. The 512 version gave significantly more. That said unless you intending to smash the shutter forever, the buffer clear speed is the fastest, in the 2-3s range
@@ZPProductions You do need to use a fast cf express type b card. Maximum write speed they state on the card mean nothing. They can say 1600mb/s maximum write speed but they will only have 800mb/s sustain write speed. You need the fastest sustain write speed to get the best buffer, the best cards right now are the delkin black with the 128gb version being the fastest at 1710mb/s sustain write speed . Prograde cobalt have really good sustain write speed, all the other are slower and you will get a smaller buffer. Ricci talk used the delkin black when he got 1600 photos in he*
@@2906rick yup. The z9 is very picky with the cards. I probably won’t be purchasing specific cards since all my other cameras I use mainly SD unless I’m recording video and I don’t quite need to burst more then 300 shots. Though the concept of abandoning deep internal buffer in exchange of very fast write mechanism is an interesting thought from Nikon. This is quite similar to cinema cameras and their camera raws but writing many small files are harder then a continuous write in a video file due to processing overheads.
@@ZPProductions Wonder if this is how the Z9 can keep the long video recording times (2 hours+) without overheating. They simply eliminate writing to an internal buffer (additional process which can generate heat) but write directly to the CF Express card.
@@victorlim5077 interesting thought process. I’m not too sure really. It can just be Nikon using less buffer and more heatsink considering the body weight
which setting did you use on the A-1 EVF?, btw the standard used in calculating human response time is .75 seconds. I used to be an fatal accident investigator and all mathematical formulas to calculate the time it takes for a human being to perceive AND react were based on that number
I was on High+ / 240fps
I based the human reaction time on recent papers and also open sources sites that gather reaction time to button clicks. The closest reference was one experiment on seeing a colour change and mouse click with no accuracy required since it’s the shortest motion to activation (just finger motion) , very similar to how we click a shutter.
Fatal accident is it based on vehicular based accidents ? I’m actually really interested in human reaction time since it makes for better software designs.
@@ZPProductions Reacting to a color is not sufficient, If you were reacting to a scene and deciding whether to hit the shutter its a complex calculation, is the light right, is the composition right much different. Its actually more like accident avoidance than a single stimulus. Also are you still using the 3rd party lens? That will further skew the results as the viewfinder is waiting on feedback from the lens, also do you have focus set to release? if not it will slightly delay the signal to confirm focus. I would like to see the test with focus set to release and a newer linear motor lens mounted. The Northwestern University developed developed most of the Accident Investigation formulas, you may find more information on reaction times on their site.
@@longrider9551 scene based analysis and reaction time is based on complexity of decision and if there’s some sort of preamble to it. Givens complex situation like say sports I’ll agree with you since a choice needs to be made. However since the focus of this test was latency times, I don’t think we need to take it in consideration as long as it’s consistent in all test.
I rerunned the test for the A1 and released priority with the 50mm f1.2 GM on multiple runs (each run 6 reading) . Based on today condition I did worst haha. And just for confirmation I took the Nikon out to compare, I did worst but slightly better then the Sony still. Also on release priority and using a Z native lens. Also did some with Canon and its quite similar. But surprisingly I deteriorated lesser there.
Maybe I’ll test on some other permutations when I have more time. If you have any other settings that may change the result do say.
@@ZPProductions that sounds terrific. Really nice work, I would only say if the delta of the latency times is less than the reaction time coefficient it may be a non issue, some folks will be faster than the coefficient some slower so the impact might end up be user controlled. Very interesting topic, you could also do a mass in motion test to see how the weight of the camera and lens effect the photographers ability to redirect on a moving subject. Human fatigue also plays a role in reaction times, its a fascinating subject and this series of videos is so interesting. ✌
Great work. How about the latency of the eye detect? On the Z7II the latency of the eye detect is too slow (IMO) for serious sports shooting, or any time you need to move the camera off the eye for a spit second and back. Would love to know how the Z9 performs in this respect. Thank you.
The z9 is the slowest among the 3. Still faster then z7ii when it comes to drawing The Eye box
Good to know thanks!!
Interesting test. What I would like to find out is tracking on birds in flight. Which camera acquires the bird first and which is the most sticky? I have heard from another CZcams reviewer Jared Polin although by and large the Z9's AF has caught up to both Sony (A1) and Canon (R3 & R5), it is nevertheless not quite as fast nor sticky especially when adapting to F-mount lenses. I would like to know whether that is indeed the case.
Fro is a problematic reviewer. Also look at the comments. He did not use the Z9 correctly. The settings where not set to maximum “stickiness”, if that’s what counts. Also why be so hung up on only AF? What about everything else which is equally or more important?
That’s probably really hard to test for me. I don’t have access to super teles and my birder friends all quit recently. Unless The manufactures them self loan me a copy it’s probably hard to gain access.
JP is a Sony shill, not a reviewer :)
@@tor2919 he knows very well, what is he doing - this way he can piss on Nikon :)
Does lower resolution on z9 give it this advantage? I never noticed the lag on A1 on a conscious level so I will look for it next time. How is evf resolution on z9 vs A1 especially in fine tuning manual focus? The best manual focus viewfinders we're on Old Nikon F bodies like FM2 and I missed being able to easily manual focus on later Nikon digital bodies. A1 makes it a lot easier to shoot without using focus assist etc
Probably need to try. Actually my test here is really to find the marginal difference, that tiny bit of lag/frame jump is something I only notice after some observation and it has to be panning above a certain speed.
Now testing for mf, probably visit it one day.
I am not sure because battery life ratings are way off. but one thing i noticed is with EVF Sony and Canon had less shot per battery then LCD. is it due to higher resolution and higher frame rate.
Nikon shows the EVF live view at screen resolution whereas Sony and I believe Canon show live view at a resolution lower than the resolution of the EVF. From experience with different displays (normal displays for gaming/movies and EVF for cameras) I can say that native resolution is always sharper and therefore mf should be easier with Nikon EVFs.
Nikon also has my preferred style of focus zooming while focusing manually. On Sony it zooms out when even tapping the shutter button but on Nikon it zooms out only after taking the picture or toggling the mode yourself
@@TheDeltaMoo Someone who was familiar with workings of Sony A1 evf (9m dots) and did some research said it does dip in resolution down to 5million dots during a 30fps burst. I upgrade from A1 from A9 and A7riv. A7riv was a nice upgrade from the almost pixellated 1.2 megapixel(3 million dots) a9 viewfinder but a7riv af was unusable for me especially in lower light. Haven't tried Z9 but even with resolution drop a1 evf has higher pixel count. not sure i can tell the difference from a1 if I try Z9 viewfinder as they all look good to my tired eyes.
I mean, if you've ever used a 1440P 240Hz or 120Hz monitor, you'll realise there's not a huge difference, however difference between 60Hz & 120Hz is alot.
I want to see 3 camera for the portrait as same setting! Many many thx
Probably that will take sometime. I don't have the equivalent lens from all three system haha. Sony and canon was easy since I can use adapted EF lens, but Nikon with a distinctly different mount is going to be hard. Unless Im sponsored or get a get a free loan of 50mm F1.2 from each of the system haha.
Actually ill be doing a video for this today. Watch out for it around GMT +8 12am
I wonder if the z9 results can be even better now cause of the 120fps evf update
I will retry once it’s out. But this will be a long process again haha
@@ZPProductions Oh yeah, I forgot that it is not available yet. Looking foward to it lol
For lag increments, Use an A7s3 240fps to shoot evf, aim camera at a fan and count how many frames the image froze/blackout. A1's lag is better at 240hz for sure
For better results shoot a slow ceiling fan, this way if "burst start" skips frames it's easier to see. If aiming up is difficult, Find a mirror.
What you get in your solution is just fps and what are the frames in between. It does not represent what’s happening in real world. lag means how much time has pass in real world vs time the frames shows on evf. What you are describing is time between frames and no doubt A1 will win because it’s more frame but doesn’t mean the frame is timely.
As explained to some below, the only real way is showing a view with both the evf and the real world together taken with one frame. This is a common test used for measuring monitor lag. The issue is depth of field as the evf is very close and the timer/ fan is much further. Additionally this does not answer the second lag Which is time taken for the shot. That requires a mechanical actuator on the shutter button with some computer vision to solve it.
Thanjs for replying. Actually I was referring to "burst start increment lag". Normal lag in Sony is really low like 10ms, but much higher when burst starts (like 50ms).
For absolute lag calculation best to have 240hz pc monitor, and a lens like RF5.2mm F2.8 L Dual Fisheye Lens
@@gordonyz4 oh u mean that burst lag, then it’s a rather good idea !
I don’t think that special lens can focus serperately. But good idea to investigate since I do have a R5
@@ZPProductions won't focus separately, but can adjust viewfinder dipoter to shift the focus distance. I definitely believe we can adjust to a similar distance as monitor
Interesting. But getting the shot depends on the photographers. Anticipation and experience really counts. Although your test is interesting, I am not smart enough to make sense of it.
Anticipation is key. The difference is small and some training to speed up reaction time and prediction will probably get you further then changing body haha.
All my test show is given a situation where you can't burst and highly unpredictable, the z9 gives a small edge. I don't think it will break any record but it will help a little more haha
Can you test with a DSLR and OVF?
I don't have access to one now so its hard
Optical viewfinders have virtually no lag as we are dealing with Light speed (300000 km/s). On a human scale it is instantaneous.
@@V2KPhotography there’s no lag but there’s the latency still of flipping up the mirror and opening/closing of the shutter. It’s a interesting thought though maybe I should see if I can borrow a 1dx m3 to try or d6
I wonder if you could have gotten your baseline response time by using a DSLR, which would presumably have no latency?
There’s still mechanical latency. Mirror and shutter all. OVF just means no lag in the view but there may be still latency from the shutter press to the capturing of the shot
I just want you to know my car shifts a gear in less than 100ms, so these cameras need to step it up.
233 milliseconds = 5 frames @ 24fps
How is the LCD latency?
Didn’t try that out yet. Maybe before my next review
Why am I reminded of Spiderman : No Way Home? 🤔
Just watched it recently before I complete my review. Great movie haha
just a few observations, if you set the A1 to 240fps the frame skip hardly appears and there is no blackout either. furthermore, the latency calculation is useless because what arrives at the viewfinder or on hdmi is not what arrives at the viewfinder, in fact if the display has a 60fps refresh you will always have at least 170ms between one frame and another.
And this if you set it to ntsc, because otherwise the camera works at 50hz, that obviously can produce only 50fps…that is 200ms
At 240fps the A1 still has that frame skip. That jerk is always there regardless how many fps I mess with it. Been staring at it the whole morning. It’s fast but there. This is more for perceived smoothness rather then any form of real world effect.
60fps is close 16.6ms. And lastly you are mixing evf refresh with hdmi and system refresh. It’s constant 60fps unless shutter speed do not permit. It’s like the A1 constantly at 240fps regardless of type.
That said 16ms is the space Between each frame so it’s also possible my results are within margin of error
1000ms/60fps is NOT 170ms - try harder! :)))
The "frame skip" is visible on your video as well - and it is daylight:
czcams.com/video/FKIjwgdJ7eY/video.html
@@ZPProductions yes I know that when the shot is released it presents a jump, but this also depends on the shooting, light and optical type settings.
In general, the viewfinder of the A1 at 240fps offers the best possible vision for viewing in the sports field, without changes in frequency or resolution and with a fluidity unknown to the others (in panning).
In the video I see the nikon produce jumps but in a homogeneous way, probably due to a lower frequency, in any case it is more homogeneous. I doubt it will be suitable for fast panning though.
As for the refresh rate of the display you are right they are 16ms I miscalculated, in any case they do not correspond to what is visible in the viewfinder, never, because they follow another type of output.
In fact, if you change the frequency of the viewfinder to 240fps that of the display and hdmi remains the same, the latency is calculated on the viewfinder not on the display (unless you want to calculate that for some reason)
@@FedericoGallinari " I doubt it will be suitable for fast panning though." Maybe read some reviews, maybe on DPR:
"Simply put, the Z9 offers one of the best viewfinder experiences of any camera on the market. In some ways, it's the most DSLR-like experience in the mirrorless camera universe, while also providing an experience a cut above even the likes of the Sony a1 with its nominally higher-resolution 9.44M-dot electronic finder."
That´s it.
@@FedericoGallinari there’s no jumps while shooting for the Nikon. It’s more like the limits of 60fps.it May feel alittle rough when fast panning. That’s when 240fps maybe better. This is similar to gamers feeling the need of a faster refresh monitor as whip pans have a difference. But for the purpose of tracking that will be not much difference.
That said I’m no sports , motor sport or birding photographer, this few are the genres that will pan fast enough.
I can't make sense of this. Let's say the EVF is updating at 60 FPS (a modest rate, not the 120 FPS or 240 FPS these cameras may be capable of) If the image being presented is 0.250 sec behind the actual scene, that means the EVF is about 15 frames behind, and consequently there are 15 frames in the video pipeline. So you are claiming the electronics has 15 (or more) frames in the video pipeline?
Maybe you can give a more complete description of what you are measuring and how you are measuring it.
To be exact, there’s the human reaction time added to the numbers as explained in the video, what we perceive till when we react is anywhere from 150-300ms. So when I press the shutter, we are around 15 frames behind what’s seen. (Or more if Nikon/Sony) but it doesn’t change when the number first appears in the viewfinder.
So what we are measuring is the human reaction time + the camera latency upon receive the shutter command. And depending on the implementation by the manufacturer this number can vary. Example is I did a previous video on the R5 and the mechanical shutter incurs another 20-30ms of time to react to the shot. In the case for this camera with no mechanical shutter, it’s about how fast they dump the frame you are trying to capture out.
Well, for a "modest" EVF, this is a pretty darn good review! So again, not everything is about numbers! This is the same as saying that "my BMW is 3 times better than Formula 1 because it has 3 times the engine capacity." :)))
"But when I used the Sony a1 after the Nikon Z9 for actual shooting, I was underwhelmed.
That's because the Sony drops in resolution while you're focusing, regardless of what your display quality settings are. And when you've stopped focusing, the resolution rises again and the sensor stabilizer seems to hiccup and do a bit of a reset."
@@ZPProductions
And the latency you are measuring is dominated by the human response time. The camera EVF latency is a small fraction of the human response time and the variation in human response time. That means in order to measure the EVF latency, you need to measure the combination of human response time and camera latency to a small fraction of the variation in human response time. You need to find a better way to make your measurement. This data has no credibility for establishing any difference in EVF latency between these cameras.
@@dclark92064 As far as test goes without extensive equipment, the only way is establishing a high sample rates to minimise potential variance with swapping between system to take in account of human condition. I did carried on testing for another 30 takes after making this video, resulting in 60 samples per camera and the resulting gap is around 15ms. If you have a better way do suggest. Most other variables are consistent, which means iso, lighting conditions, the software used to generate the timings.
As you said, bulk of the numbers are human response, this is quite unlike say the R5 in mechanical shutter mode which has a significant difference and average timing of 300ms. I believe that the numbers I measured are in the realm of potential variance and may have no significant impact to most use cases.
R3 is not a flagship
@@prokremelskidezolati1426 because it's not it's a camera between the r5 and 1dxiii go read canons own website. Stop calling it a flagship it's not.
Seriously mate, talk at a normal speed that people can understand. I have to watch all your videos on 0.75 playback speed to even have a chance to comprehend what you are trying to say or enable subtitles. Speak clearly, it's not about setting the World speed talking record ok, if your audience can't understand you.
This is not how you measure latency.
you talk too fast, need improvement
You need improvement if you can not follow him :))
But wants better lag on evf :))