Canon releases an official statement about the EOS R5 and EOS R6 heat concerns

Quarkcharmed

EOS R5
Feb 14, 2018
1,219
1,076
Australia
www.michaelborisenko.com
Within a given technological generation same format sensors are essentially equal in high ISO noise performance.

That's true, but the conversation somehow slipped into the high-ISO performance topic. Original post doesn't say anything about high ISO, as far as I can see. It says about postprocessing away differences between different sensors in general.

If you're expecting the R5 to be leaps and bounds better than the 5Ds/sR at high ISO, or the R6 to be better than the R5, you've set yourself up for disappointment.

I'm hoping for a 1-stop increase at the base ISO, and I believe Canon promised that to PDR guys. Also it was repeated at B&H broadcast. I don't remember Canon has ever promised concrete figures before, so this is something. But that most likely applies to the base ISO performance.
High ISO increase would be nice but highly unlikely to expect from the R5, if its performance is close to 1DxIII.


significant improvement can be achieved in 16-bit sensors as we can see in the chart above.

UPD: I've re-read the conversation and apparently the high ISO is mentioned there. So partially I missed the point. Still I stand you can't 'postprocess away' the difference, if there's any. And optimal slider settings don't compensate the difference either.
 
Last edited:

dtaylor

Canon 5Ds
Jul 26, 2011
1,630
1,186
I'm hoping for a 1-stop increase at the base ISO, and I believe Canon promised that to PDR guys.

R6 vs R5? Or both vs 5D4? I don't know if you'll get that but you might. Base ISO DR is a different beast than high ISO noise and if they improved the circuitry the R5/R6 might be better, but the 5D4 and 1DX mark III are already quite good. I suspect the R6 sensor is basically the 1DX mark III sensor, so if there's a DR improvement it may only appear in the R5.

significant improvement can be achieved in 16-bit sensors as we can see in the chart above.

I would guess that's more about the larger MF sensor than 14-vs-16 bit ADCs.

Still I stand you can't 'postprocess away' the difference, if there's any. And optimal slider settings don't compensate the difference either.

I don't know why you would cling to that position when you can literally download RAWs from dpreview or Imaging Resource and quickly 'post process away' small noise differences between an older, higher resolution sensor and a newer, lower resolution sensor.
 

Quarkcharmed

EOS R5
Feb 14, 2018
1,219
1,076
Australia
www.michaelborisenko.com
R6 vs R5? Or both vs 5D4? I don't know if you'll get that but you might.

Canon said we could expect an about 1-stop improvement in the R5 against the R/5DIV. Which is more than I hoped for. tbh it was a major factor why I preordered it. Otherwise I'd have waited for the reviews. Very rarely we get concrete figures on DR from Canon, so I trust Canon's reputation in this case. Hopefully it's all true, but I think they're scratching the very bottom already and further improvements will only come with 16-bit FF sensors.

I would guess that's more about the larger MF sensor than 14-vs-16 bit ADCs.

Perhaps it's also a big factor, but higher bitness means the noise/fluctuations in the lowest bit are less significant, that's why 16 bit sensors have more room for noise reduction. You can see GFX 100 cropped to FF is still better than A7rIV


I don't know why you would cling to that position when you can literally download RAWs from dpreview or Imaging Resource and quickly 'post process away' small noise differences between an older, higher resolution sensor and a newer, lower resolution sensor.

Downsamling of the higher-res image is not a valid postprocessing - as I argued earlier in this thread, suppose you have a 20mp image and I have a 45mp image - it's your task to manipulate *your* image, not mine. :)
And if you apply say some noise reduction to your image, I can also apply noise reduction to my image and I'm ahead again.
 

Starting out EOS R

EOS R5 - RF24-105mm F4L, RF70-200mm f2.8L
Feb 13, 2020
293
314
Canon said we could expect an about 1-stop improvement in the R5 against the R/5DIV. Which is more than I hoped for. tbh it was a major factor why I preordered it. Otherwise I'd have waited for the reviews. Very rarely we get concrete figures on DR from Canon, so I trust Canon's reputation in this case. Hopefully it's all true, but I think they're scratching the very bottom already and further improvements will only come with 16-bit FF sensors.



Perhaps it's also a big factor, but higher bitness means the noise/fluctuations in the lowest bit are less significant, that's why 16 bit sensors have more room for noise reduction. You can see GFX 100 cropped to FF is still better than A7rIV




Downsamling of the higher-res image is not a valid postprocessing - as I argued earlier in this thread, suppose you have a 20mp image and I have a 45mp image - it's your task to manipulate *your* image, not mine. :)
And if you apply say some noise reduction to your image, I can also apply noise reduction to my image and I'm ahead again.
You guys are amazing. This is possibly the longest thread I've seen and whilst you don't agree on one aspect and probably never will lol, the conversation has generally been well meaning which is welcome. I'll admit some of the stuff you've been discussing is way above my skills set but keep going it's always interesting. :)
 

dtaylor

Canon 5Ds
Jul 26, 2011
1,630
1,186
Downsamling of the higher-res image is not a valid postprocessing

You don't need to explicitly down sample anything. Just compare sensors at the same view size. It's pixel peeping that's not valid because it inherently magnifies one image more than another.

as I argued earlier in this thread, suppose you have a 20mp image and I have a 45mp image - it's your task to manipulate *your* image, not mine. :)
And if you apply say some noise reduction to your image, I can also apply noise reduction to my image and I'm ahead again.

So if you agree that a high MP image can take more NR than a low MP image, why are we debating this point? A 5 year old 5Ds/sR has more noise at ISO 12,800 than a 1DX mark III (by about 0.5ev). You can see that at the same view size. But it also retains more detail. So you can NR the 5Ds/sR to have the same noise levels as the 1DX mark III, yet still have at least as much if not more detail. And you can't just apply more NR to the 20mp file to "stay ahead" because it will just make it even softer compared to the 50mp file.

And this brings us all the way back to the original point: you lose nothing with higher pixel density. An R6 will not be a "low light monster" compared to an R5. They are likely to have similar high ISO performance to within 1/3ev or less. Edit: but even if they were different, you've got 25 extra MP to trade off in NR.

Base ISO DR is a different beast and does rely on circuit design and pixel size, so it's hard to say which will have better DR until tests come out.
 
Last edited:

privatebydesign

EOS-1D X Mark III
CR Pro
Jan 29, 2011
9,473
3,855
We already have the core results of the R6 high iso performance as it shares the sensor to a major degree with the 1DX MkIII, and that showed no high iso improvement over the 5D MkIV.

1595164030010.png
 
  • Like
Reactions: dtaylor

Quarkcharmed

EOS R5
Feb 14, 2018
1,219
1,076
Australia
www.michaelborisenko.com
So if you agree that a high MP image can take more NR than a low MP image, why are we debating this point?

If you mean "downsampling = NR" it's not really a valid 'postprocessing'.
It can, but it doesn't eliminate the difference in noise - after downsampling it's not the same original image. My point was simply that you can't postprocess away the difference.

A 5 year old 5Ds/sR has more noise at ISO 12,800 than a 1DX mark III (by about 0.5ev). You can see that at the same view size. But it also retains more detail. So you can NR the 5Ds/sR to have the same noise levels as the 1DX mark III, yet still have at least as much if not more detail.

According to PTP, they have about 0.4 stop difference at ISO 800, and just 0.2 stops at ISO 12800. But that's not just any view size, it's the size used in PTP metrics. The resulting difference may change if we change the target size.
 
Last edited:

Quarkcharmed

EOS R5
Feb 14, 2018
1,219
1,076
Australia
www.michaelborisenko.com
We already have the core results of the R6 high iso performance as it shares the sensor to a major degree with the 1DX MkIII, and that showed no high iso improvement over the 5D MkIV.

Exactly. My hope (based on Canon's remarks) was the R5 would show the DR improvement at the base ISO. I don't expect any significant improvement at high ISOs. My hope is, at high ISOs it won't have heavy banding in the shadows like 5DIV. But it wasn't promised by Canon though.
 

privatebydesign

EOS-1D X Mark III
CR Pro
Jan 29, 2011
9,473
3,855
Exactly. My hope (based on Canon's remarks) was the R5 would show the DR improvement at the base ISO. I don't expect any significant improvement at high ISOs. My hope is, at high ISOs it won't have heavy banding in the shadows like 5DIV. But it wasn't promised by Canon though.
You have turned your argument about high iso noise in so many knots through the various threads you now agree with me! Pathetic. Oh and thanks for the images that proved the point you were arguing days ago when I realized it was a complete waste of time.

P.S. And no I'm not interested in rehashing it or relitigating your argumentative nonsense.
 
Last edited:

PureClassA

Canon since age 5. The A1
CR Pro
Aug 15, 2014
2,116
813
Mandeville, LA
Shields-Photography.com
Magic Lantern on a 5D Mk3 is able to display the sensor temperature.
One can easily monitor how the sensor temperature increses pretty fast.
It stays at a somewhat high temperature after some time, so yes, you
can run it all day in preview mode.

But you will have an already elevated temperature on your hands when
you finally start recording. R5/R6 are no different.

I had the camera running and recording externally for 1.5 Hours STRAIGHT. Plus another 10-20 mins before hand doing ore-roll checks. EOS R never even got warm. Not remotely. And at 1080p, it is using the whole sensor. Power consumption and heat is ALL about the recording/encoding which in this case done by Atomos. And If you’re doing THAT serious of work where you are spending 20mins on pre-roll doing shot alignment etc... You are either pro enough to be shooting with a Cinema Camera or you are rolling externally.

and If you are shooting in 8k, you’re filling the card Up in 20 mins. Drone? The Pro grade Octocopter has a flight time of 20 mins. So you cant exceed that anyway. And again, who in the hell is shooting at 4k120 for 20-30mins straight??? And if that’s a routine thing for you, then you should be shooting REDs (if they dont break) or ARRI.



in
FF6F4637-3B64-4BA7-8FE5-077F0240D265.jpeg
 

dtaylor

Canon 5Ds
Jul 26, 2011
1,630
1,186
If you mean "downsampling = NR" it's not really a valid 'postprocessing'.

I don't know what your hang up is on downsampling or why you keep brining it up. If downsampling a higher MP file to the same view size of the lower MP file makes the noise visually equal, then it was literally equal to begin with at the sensor. It's not downsampling which is 'invalid', it's pixel peeping. Viewing a 45mp file at the same size as a 20mp file is no different than viewing Adox CMS 20 at the same print size as Portra 400. You can't insist on printing Adox CMS 20 twice as large...which it can handle...and then claim Adox CMS 20 has more grain.

Of course people buy high resolution systems specifically so that they can enlarge more. But nobody is expecting to make large prints from very high ISOs.

after downsampling it's not the same original image.

A permanent resize (as opposed to a resampling to a specific device's underlying pixel/dot matrix) does change the image, but keeping the "original image" is not a goal anyone cares about.

My point was simply that you can't postprocess away the difference.

And you are observably wrong. You absolutely can trade resolution (sampling rate) for SNR, and photography is hardly the only field where that is done.

According to PTP, they have about 0.4 stop difference at ISO 800, and just 0.2 stops at ISO 12800. But that's not just any view size, it's the size used in PTP metrics. The resulting difference may change if we change the target size.

Only if you force the sizes to be different.
 

Quarkcharmed

EOS R5
Feb 14, 2018
1,219
1,076
Australia
www.michaelborisenko.com
Of course people buy high resolution systems specifically so that they can enlarge more. But nobody is expecting to make large prints from very high ISOs.

I think we're talking about the same thing but using different semantics.
What I meant was, if you have technical differences between sensors such as noise, you can't equalise them. As in my example, if sensor A is noisier than sensor B, you can't just apply noise reduction to A and tell 'now the sensors are the same'. They're not the same because postprocessing makes comparison meaningless: you can apply the same noise reduction to B and make it better again, or apply even more NR to A and make it even 'better' than B.

The same applies to downsampling (no matter from high ISO or low ISO images). Trading off resolution for NR doesn't actually make sensors 'equal'.

But in the narrow practical sense, when you need files/prints of the same size - absolutely. Yes, within some limits and by using digital manipulation, you can produce images of the same resolution and IQ from different sensors.
But equalisation will always come at a cost of information loss (from one sensor or both). If you consider information loss, it's evident that digital manipulation can't really make one poor sensor on par with a better sensor.
 
<-- start Taboola -->