Archive 2024 · Canon R5 (and R5M2) vs Nikon Z8 (low light AF performance)
AmbientMike Offline • • • • • • [X]
p.2 #1 · Canon R5 (and R5M2) vs Nikon Z8 (low light AF performance)
bernardl wrote:
True, although I have been using Nikon cameras since 1986 I find the AF compatibility matrix of pre-AF-S lenses released before 1998 to be completely irrelevant as I haven’t used any in more than 15 years except my 135mm f2.0 DC that I use in MF or on my D5, but it’s so far behind the 135mm f1.8 Plena in performance that I only use it when I am after a vintage look.
I can use a single adapter (the FTZ) to mount any F mount Nikon released in the past 70 years in MF mode on my Nikon mirrorless bodies and find that impressive.
So again the FTZ is amazingly good for actual working photographers trying to use their AF-S F mount Nikkors released since 1998. Lens collectors can still use their very old Nikkors in MF mode and are mostly fine with it as well.
Does the Canon adapter enable AF on pre USM lenses? Why set the start of history on the year when Canon decided it was ok to abandon earlier lenses without any solution to preserve AF? How is that an objective comparison of both systems?
Is it coherent to both chase the latest improvements between an R5 and hypothetical R5II, that will require the latest RF lenses anyways, and to stress heavily the importance of lenses older than 25 years on Nikon side?
The reality is that both Nikon and Canon adapters do an equally great job at ensuring continued usage of lenses that practically matter to photographers.
On the weight topic I find it a bit ironic to both highlight the weight difference of the R5 vs Z8 (which is real) and to boast about the 2 heaviest normal zooms in the history of photography, the 28-70mm f2.0 and 24-105mm f2.8. If weight is a major concern then Sony is obviously the way to go.
Glad they finally came out with an adapter that you can use pre-AI lenses on? My FM-2n and EM annoyingly couldn't do that, or many (any?) DSLR's other than the Df. Canon had a couple big issues, Nikon a lot of seemingly never ending smaller ones, like less expensive DSLR bodies not metering using MF lenses, even though Canon does
I find the 24-105/2.8 & 28-70/2 much more impressive than permanently attaching a relatively old object to a tele permanently.
Feb 27, 2024 at 06:00 PM
AmbientMike Offline • • • • • • [X]
p.2 #2 · Canon R5 (and R5M2) vs Nikon Z8 (low light AF performance)
CanadaMark wrote:
Look again - Bill Claff's data shows that the R6II, R5 and R3 are all cooking their RAWs starting right at base ISO (the downward facing triangle indicates baked in NR). This wouldn't be an issue really if the option was given to the user to disable this in-camera, but unfortunately you have no choice.
Sony and Nikon are not baking NR into their RAW files. The only real exception being that Sony does bake some in on some of their bodies beyond ISO 51,200, but not before.
How could you know no NR is applied? I don't believe that, older bodies have more noise, if you go back far enough, but it may just largely be they put NR in camera as opposed to any better hardware. Banding isn't as common vs the 5D and rebel xt, for instance, but did they actually improve the sensor and other components, or just put anti banding software in the camera, probably easy to do
p.2 #4 · Canon R5 (and R5M2) vs Nikon Z8 (low light AF performance)
snapsy wrote:
I'm not sure what method Bill is using to detect NR but it's missed at least a few cameras. For example the Sony A7III:
In that particular case, it wasn't missed, rather he deemed it photographically irrelevant - these are his words:
I judge (by eye) 2D FTs of black frames and assign "noise reduction" where is appears to be "obvious"; those points warrant a down-triangle on my charts.
(I also try to ferret out digital scaling when used in place of analog gain; not always obvious.)
So, FWIW; in the case of the ILCE-7M3 I see small amounts of signal processing but didn't deem it photographically relevant, so I let it pass.
------------
So, if you see a down triangle on one of Bill's charts, the baked-in NR is "obvious".
Feb 27, 2024 at 06:11 PM
AmbientMike Offline • • • • • • [X]
p.2 #5 · Canon R5 (and R5M2) vs Nikon Z8 (low light AF performance)
sirimiri wrote:
It's all crap! Go pinhole and you'll never worry about focus acquisition again!
This man has the answer!! Modern sensors high iso make f/120 much more usable
p.2 #6 · Canon R5 (and R5M2) vs Nikon Z8 (low light AF performance)
CanadaMark wrote:
So, if you see a down triangle on one of Bill's charts, the baked-in NR is "obvious".
Obvious in the context of technical measurebating or in actual real world images? I guess I'll go on being blissfully ignorant of the low ISO NR forced on me by Canon in an attempt to cover up their inferior sensor technology.
p.2 #7 · Canon R5 (and R5M2) vs Nikon Z8 (low light AF performance)
AmbientMike wrote:
How could you know no NR is applied? I don't believe that, older bodies have more noise, if you go back far enough, but it may just largely be they put NR in camera as opposed to any better hardware. Banding isn't as common vs the 5D and rebel xt, for instance, but did they actually improve the sensor and other components, or just put anti banding software in the camera, probably easy to do
The nice thing about facts is that whether or not you believe them, they're still true
Bill has lots of in-depth articles on his website if you care to read them, and he also makes his email available if you have your own objective data that you would like to share, or if you have questions about his processes.
Some cameras bake NR into their RAW files, others do not. Some are worse offenders than others. All of Canon's current full-frame RF bodies (R5, R6II, R8 and R3) are cooking their RAWs to a degree that is obvious when examining the data. Whether or not you care is entirely up to you.
Older bodies have more noise for lots of reasons, most of them hardware - if you want an example, have a look at how much better Canon's sensors got once they finally moved the ADC on-chip around ~2015.
p.2 #8 · Canon R5 (and R5M2) vs Nikon Z8 (low light AF performance)
rscheffler wrote:
Obvious in the context of technical measurebating or in actual real world images? I guess I'll go on being blissfully ignorant of the low ISO NR forced on me by Canon in an attempt to cover up their inferior sensor technology.
Call it whatever you want, but some cameras are baking material amounts of NR into RAW files that is not user-controllable. It is impossible to apply NR without a loss of detail, and without an uncooked RAW from the same camera, it's not something you can easily quantify. If it's not something that bothers you for your usage, then you don't need to worry about it. Personally, I prefer having as much control as possible over my RAW files for the same reasons I don't want lens corrections, sharpening, or vignetting correction baked into my RAWs. The baked-in NR is also why, for example, the R5 is not ISO invariant until after the ISO values that Canon is adding NR.
Some people will not care if their RAWs have baked-in NR, but it's hard to argue that it wouldn't be better to have a menu option and let the user decide. If there is baked-in NR, it also makes it that much more difficult to objectively compare to other bodies when making a purchase decision or similar. Bill also estimates that the NR in the R5 is making it look about 2/3 stop better than it is. Also, baked-in NR tends to be something that camera manufacturers actively try to hide, it is not viewed as a 'feature' - that should tell you something
p.2 #9 · Canon R5 (and R5M2) vs Nikon Z8 (low light AF performance)
CanadaMark wrote:
In that particular case, it wasn't missed, rather he deemed it photographically irrelevant - these are his words:
I judge (by eye) 2D FTs of black frames and assign "noise reduction" where is appears to be "obvious"; those points warrant a down-triangle on my charts.
(I also try to ferret out digital scaling when used in place of analog gain; not always obvious.)
So, FWIW; in the case of the ILCE-7M3 I see small amounts of signal processing but didn't deem it photographically relevant, so I let it pass.
------------
So, if you see a down triangle on one of Bill's charts, the baked-in NR is "obvious"....Show more →
Thanks, didn't see Bill's comment on that, which reading now I disagree with. It's clear from Jim's measurements (which I verified as well) the camera is performing noise reduction across the frequency range. The fact it doesn't show up on a spatial representation of an FFT is not the measure I would use to qualify it as photographically significant. In fact you can do a certain measure of gaussian blur and not have it show up clearly there. Jim is using an FFT but graphing the result, which is much more sensitive to NR than the spatial representation that Bill is relying on by eye.
p.2 #12 · Canon R5 (and R5M2) vs Nikon Z8 (low light AF performance)
I have no real world experience with either. But the 3rd party lens support canon offers is why I didn't even consider them. I was between Nikon and Sony and since Nikon "never" was going to release the Z8, I went sony just so many choices.
Feb 27, 2024 at 10:08 PM
AmbientMike Offline • • • • • • [X]
p.2 #13 · Canon R5 (and R5M2) vs Nikon Z8 (low light AF performance)
CanadaMark wrote:
The nice thing about facts is that whether or not you believe them, they're still true
Bill has lots of in-depth articles on his website if you care to read them, and he also makes his email available if you have your own objective data that you would like to share, or if you have questions about his processes.
Some cameras bake NR into their RAW files, others do not. Some are worse offenders than others. All of Canon's current full-frame RF bodies (R5, R6II, R8 and R3) are cooking their RAWs to a degree that is obvious when examining the data. Whether or not you care is entirely up to you.
Older bodies have more noise for lots of reasons, most of them hardware - if you want an example, have a look at how much better Canon's sensors got once they finally moved the ADC on-chip around ~2015....Show more →
Glad you know it all, if you could point me to the Bill Claff article that states these non Canon cameras don't have any Noise Reduction. Not that he's 100%.
Your interpretation of a graph, does not a fact make
Canon had low noise on aps by the 2010 60D, if not the 7D. FF certainly by 2012 5D3/1Dx/6D
p.2 #14 · Canon R5 (and R5M2) vs Nikon Z8 (low light AF performance)
CanadaMark wrote:
Call it whatever you want, but some cameras are baking material amounts of NR into RAW files that is not user-controllable. It is impossible to apply NR without a loss of detail, and without an uncooked RAW from the same camera, it's not something you can easily quantify. If it's not something that bothers you for your usage, then you don't need to worry about it. Personally, I prefer having as much control as possible over my RAW files for the same reasons I don't want lens corrections, sharpening, or vignetting correction baked into my RAWs. The baked-in NR is also why, for example, the R5 is not ISO invariant until after the ISO values that Canon is adding NR.
Some people will not care if their RAWs have baked-in NR, but it's hard to argue that it wouldn't be better to have a menu option and let the user decide. If there is baked-in NR, it also makes it that much more difficult to objectively compare to other bodies when making a purchase decision or similar. Bill also estimates that the NR in the R5 is making it look about 2/3 stop better than it is. Also, baked-in NR tends to be something that camera manufacturers actively try to hide, it is not viewed as a 'feature' - that should tell you something ...Show more →
Irrelevant. What matters is results. If a camera maintains a high level of accutance with lower noise, it is smart design.
In case you haven't noticed, computational imaging is making meaningful improvements. This is just another example.
p.2 #15 · Canon R5 (and R5M2) vs Nikon Z8 (low light AF performance)
AmbientMike wrote:
Glad they finally came out with an adapter that you can use pre-AI lenses on? My FM-2n and EM annoyingly couldn't do that, or many (any?) DSLR's other than the Df. Canon had a couple big issues, Nikon a lot of seemingly never ending smaller ones, like less expensive DSLR bodies not metering using MF lenses, even though Canon does
I find the 24-105/2.8 & 28-70/2 much more impressive than permanently attaching a relatively old object to a tele permanently.
I don't believe that pre-AI lenses can be used with the FTZ, apologies if my comment was misleading. On the other hand my Nikon mirrorless bodies all meter with the MF lenses I own (from Nikon, Leica, Voigtlander and Zeiss) and on the Zf the camera looks for the eyes and allows you to zoom to 100% on the eye, which is nothing short of amazing for finely focused portrait with manual glass.
If you find those 2 zooms impressive then Canon was the right system for you. Do you own them or is it a theoretical statement?
I am sure you know this, but the TC equipped Nikon Z super tele are far more than existing lenses to which TC were attached. The Nikon 400mm f2.8 TC and 600mm f4 TC are new designs optimized to include TC elements in the middle of their optical groups. And the value is both the optical quality and the possibility to instantly switch the TC. Have you tried using one? I own the 400mm f2.8 TC and it is nothing short of fabulous. At that price point you wouldn't expect anything less of course.
p.2 #16 · Canon R5 (and R5M2) vs Nikon Z8 (low light AF performance)
garyvot wrote:
Irrelevant. What matters is results. If a camera maintains a high level of accutance with lower noise, it is smart design.
In case you haven't noticed, computational imaging is making meaningful improvements. This is just another example.
Which is precisely the point. It is possible to get better results with the highly computational intensive denoising algorithms implemented in raw conversion software.
It's really a matter of good enough. Some people are happy with in camera raw NR, others prefer to tap fully into the potential of the sensor. My personal preference is for autonomy and I prefer to do things myself but I fully understand that others prefer to trust their camera to do it for them.
I don't consider this to be a huge problem and I am glad to see that Canon finally found a way to provide their users with decent dynamic range since the R6/R5/R3 generation. The 5D mkIV was starting to show encouraging signs already. If this were the only gap I wouldn't hesitate to consider Canon.
If you consider 10.63 to be equal to 11.63 then yes...
Two Sony sensors from the same generation, it's perfectly normal that the APS-C sensor is roughly one stop behind the FF one.
Cheers,
Bernard
Feb 28, 2024 at 01:00 AM
AmbientMike Offline • • • • • • [X]
p.2 #19 · Canon R5 (and R5M2) vs Nikon Z8 (low light AF performance)
bernardl wrote:
If you consider 10.63 to be equal to 11.63 then yes...
Two Sony sensors from the same generation, it's perfectly normal that the APS-C sensor is roughly one stop behind the FF one.
Cheers,
Bernard
Doesn't look like the link copied the models selected for the graph on the top. You might have to do it manually.
Apparently Fuji mis-states their iso. You kind of have to be careful jumping to too many conclusions on these graphs
Glad you like your 1.4, since you're stuck using it. Have to see if the rumored 1x-2x tc from Canon comes out. One potential reason they didn't build one in
p.2 #20 · Canon R5 (and R5M2) vs Nikon Z8 (low light AF performance)
CanadaMark wrote:
Call it whatever you want, but some cameras are baking material amounts of NR into RAW files that is not user-controllable. It is impossible to apply NR without a loss of detail, and without an uncooked RAW from the same camera, it's not something you can easily quantify. If it's not something that bothers you for your usage, then you don't need to worry about it. Personally, I prefer having as much control as possible over my RAW files for the same reasons I don't want lens corrections, sharpening, or vignetting correction baked into my RAWs. The baked-in NR is also why, for example, the R5 is not ISO invariant until after the ISO values that Canon is adding NR.
Some people will not care if their RAWs have baked-in NR, but it's hard to argue that it wouldn't be better to have a menu option and let the user decide. If there is baked-in NR, it also makes it that much more difficult to objectively compare to other bodies when making a purchase decision or similar. Bill also estimates that the NR in the R5 is making it look about 2/3 stop better than it is. Also, baked-in NR tends to be something that camera manufacturers actively try to hide, it is not viewed as a 'feature' - that should tell you something ...Show more →
For the casual reader, the type of noise reduction baked into the raw files is only detectable at the smallest signal levels, that you don't see unless you push the exposure or lift the shadows several stops at low ISO. So to start with, it is not relevant at all for normal photography, and especially irrelevant for low light photography. There you have complete control to set the noise reduction to zero.
Further, the effect of the baked in NR, when you see it, is not disturbing. In the old days, with fixed pattern noise, that was disturbing and could ruin your image in a very few cases. I have yet to se ONE real image from R5, R8 etc, where the effect of the baked in NR had ANY significant impact on the photographic result.
Of course it is theoretically better no not have the NR baked in, but making this a criterion for staying away from a brand, is not going to be productive.
