The sensor in the upcoming Canon EOS R3 is designed and manufactured by Canon

Michael Clark

Now we see through a glass, darkly...
Apr 5, 2016
3,657
2,142
There are noise comparison tools and if you for example compare the 1D X and the 1DX Mark III, there are eight years between them and you see some noise improvements, but not a lot. It is definitely much less than one stop in eight years. I would have expected at least two stops of improvement in that time frame. In the eight years before that 1D X, we easily had over two stops of improvement.

Here is a noise comparison tool to play around with:


Everyone had more improvement, noise wise, between 2004 and 2012 than they have between 2012 and 2020.

We've been approaching the theoretical limits of the random nature of photon distribution in a light field of given intensity over time (Poisson distribution noise, which is a property of light, not of cameras and sensors) for several years.
 

AlanF

Stay at home
CR Pro
Aug 16, 2012
8,289
10,367
There are noise comparison tools and if you for example compare the 1D X and the 1DX Mark III, there are eight years between them and you see some noise improvements, but not a lot. It is definitely much less than one stop in eight years. I would have expected at least two stops of improvement in that time frame. In the eight years before that 1D X, we easily had over two stops of improvement.

Here is a noise comparison tool to play around with:

You would need to break the laws of physics to have a two-stop improvement in noise at the iso12800 in that comparison tool setting! At high isos, the limiting factor in noise is mainly statistical fluctuations in the low number of photons captured (photon shot noise rather than electronic noise) and the signal to noise is given by sqrt(number of photons captured). Modern sensors have been greater than 80% efficient at capturing photons for several years and so there is room for only a tiny improvement, a fraction of a stop, and which is why you have seen only some improvement.

Edit: Michael, your reply was posted while I was writing mine and so sorry to repeat what you said.
 
Aug 7, 2018
228
180
What about the dark noise at ISO 100? For a long time Canon at a problem with underexposed areas that are pushed a few stops up. Unfortunately that can't be seen in those noise comparison tools. It was really shocking to me how noisy the dark areas get with a 18 megapixel full frame sensor from Canon. I hope at least that has improved somehow.

Also the pink light that was visible in long exposed high ISO shots looked really bad. My idea was to basically use my camera as a night vision tool and to a very long exposure in almost complete darkness. That was no problem at all with analogue cameras, but even with the 1D X the results were not usable.

I more and more come to the conclusion though that I might never need a new camera at all as long as my old one does not break. Noise did not improve much, I really don't need more than 3 fps and for architecture even the worst autofocus works fine. The only thing I like about the new cameras is the IBIS, but for that price I might be able to buy a very good gimbal that even beats IBIS.
 

AlanF

Stay at home
CR Pro
Aug 16, 2012
8,289
10,367
What about the dark noise at ISO 100? For a long time Canon at a problem with underexposed areas that are pushed a few stops up. Unfortunately that can't be seen in those noise comparison tools. It was really shocking to me how noisy the dark areas get with a 18 megapixel full frame sensor from Canon. I hope at least that has improved somehow.

Also the pink light that was visible in long exposed high ISO shots looked really bad. My idea was to basically use my camera as a night vision tool and to a very long exposure in almost complete darkness. That was no problem at all with analogue cameras, but even with the 1D X the results were not usable.

I more and more come to the conclusion though that I might never need a new camera at all as long as my old one does not break. Noise did not improve much, I really don't need more than 3 fps and for architecture even the worst autofocus works fine. The only thing I like about the new cameras is the IBIS, but for that price I might be able to buy a very good gimbal that even beats IBIS.
You complained about the lack of improvemet in noise at high iso, and that's what Michael and I explained to you. You wrote "for example compare the 1D X and the 1DX Mark III, there are eight years between them and you see some noise improvements, but not a lot." In fact, there is a huge improvement at low iso - you can see that in DR vs iso plots (and you can see the improvements using the DPR noise comparison tools if you know how to use them). At iso 100, the 1DXIII is over 2 stops improved and the R5 nearly 3 stops. Improvements at low iso reflect the improvement in circuit noise. The fact that the DR vs iso plots are nearly linear throughout shows there is now little improvement at low iso for the R5 etc.

 
  • Like
Reactions: Chris.Chapterten

AlanF

Stay at home
CR Pro
Aug 16, 2012
8,289
10,367
I wonder why the R6 has a lower dynamic range at low ISO than the R5. Shouldn't larger pixels help when it comes to dynamic range?
There are some artefacts with the R5 due to a change in amplifier circuitry between isoi320 and iso400. Larger pixels do have greater full well depth but they also have larger dark current, but that's not the problem.
 

Joules

doom
CR Pro
Jul 16, 2017
1,655
1,996
Hamburg, Germany
Shouldn't larger pixels help when it comes to dynamic range?
No, not if you care about an image rather than an individual pixel.

Also, the R6 uses the 1DX III sensor. While that is an excellent sensor, Canon uses some even more improved technologies in the R5. And the R3 will obviously push that even further, being both stacked and back side illuminated.
 

Kit.

EOS 5D Mark IV
Apr 25, 2011
2,125
1,461
Not sure what to make of this. A Thai site is showing 134,900 Baht (about 4300 USD) for R3.

https://ipricethailand.com/ราคา/canon-eos-r3/#
I would expect it to be just a misreported price for R5.

 

RayValdez360

Soon to be the greatest.
Jun 6, 2012
646
406
39
Philadelphia
You would need to break the laws of physics to have a two-stop improvement in noise at the iso12800 in that comparison tool setting! At high isos, the limiting factor in noise is mainly statistical fluctuations in the low number of photons captured (photon shot noise rather than electronic noise) and the signal to noise is given by sqrt(number of photons captured). Modern sensors have been greater than 80% efficient at capturing photons for several years and so there is room for only a tiny improvement, a fraction of a stop, and which is why you have seen only some improvement.

Edit: Michael, your reply was posted while I was writing mine and so sorry to repeat what you said.
Couldnt they use software and AI to help with noise and detail or at least have an advanced correction mode in case purists dont like it.
 

AlanF

Stay at home
CR Pro
Aug 16, 2012
8,289
10,367
Couldnt they use software and AI to help with noise and detail or at least have an advanced correction mode in case purists dont like it.
That's what we do now in post-processing or the camera does to various degrees in RAW to jpeg processing, and has done for generations and is ever improving. But, it doesn't alter the signal/noise of the sensor or its dynamic range. We already have the choice of using jpegs from Canon's in-built software or as "purists" we process the RAW files ourselves. As processors increase in power, not doubt Topaz AI Denoise or its equivalent might be able to work at 30 fps to output jpegs.
 
  • Like
Reactions: EOS 4 Life

briangus

EOS 90D
CR Pro
Apr 6, 2017
104
173
Bangkok
I would expect it to be just a misreported price for R5.

Thats the price of the R5 in Thailand
The shop it links to has a coming soon page with no price.
Coming soon here means at least 3 months after it is actually released.
 
Aug 7, 2018
228
180
There are some artefacts with the R5 due to a change in amplifier circuitry between isoi320 and iso400. Larger pixels do have greater full well depth but they also have larger dark current, but that's not the problem.
I noticed that the same cameras show less dynamic range in that diagram if you crop them to APS-C. So it seems to take into consideration the pixel density of an image that helps to recover more detail. I wonder if there is another tool that gives me the dynamic range without considering the pixel density. It feels wrong to say that an image loses dynamic range by cropping it. It still is a smaller part of a larger image.
 
Aug 7, 2018
228
180
Couldnt they use software and AI to help with noise and detail or at least have an advanced correction mode in case purists dont like it.
Unfortunatlely camera manufacturers already apply some noise reduction algorithms on RAW images that you can't disable. I really hate that. RAW for me means really raw pixels. That secret noise reduction is a form of cheating. They just want to make their cameras look good on paper when it comes to "RAW noise with noise reduction disabled".
 

neuroanatomist

I post too Much on Here!!
CR Pro
Jul 21, 2010
25,050
3,007
I noticed that the same cameras show less dynamic range in that diagram if you crop them to APS-C. So it seems to take into consideration the pixel density of an image that helps to recover more detail. I wonder if there is another tool that gives me the dynamic range without considering the pixel density. It feels wrong to say that an image loses dynamic range by cropping it. It still is a smaller part of a larger image.
It may feel wrong to you, but it's just math. Image height is a component of the DR measurement, a cropped image from a FF sensor has less height, therefore the DR is lower (in a non-linear way). See this.
 
  • Like
Reactions: AlanF

AlanF

Stay at home
CR Pro
Aug 16, 2012
8,289
10,367
I noticed that the same cameras show less dynamic range in that diagram if you crop them to APS-C. So it seems to take into consideration the pixel density of an image that helps to recover more detail. I wonder if there is another tool that gives me the dynamic range without considering the pixel density. It feels wrong to say that an image loses dynamic range by cropping it. It still is a smaller part of a larger image.
The reason why they appear to have lower DR when cropped to APS-C mode isn't because of pixel density, it is because of the way DR is measured. They define DR as the dynamic range you see when you enlarge the image to 8"x10" and hold it at a distance equivalent to arms length. A crop image has to be enlarged 1.6x the height and 1.6x the width of a FF image, and this lowers the measured DR.
 

EOS 4 Life

EOS R
Sep 20, 2020
856
692
If one looks at the thermal imaging test Roger Cicala did with an R5, it's most definitely the CFe card that is the hottest place in the camera when shooting 8K video and saving it internally.
Kolari Vision did a teardown of the internals and they now have a heatsink mod for the R5 for $400.
I am not sure that it gained enough overheat time for the mod to be worth it but Kolari Vision is an interesting company.
 

Franklyok

EOS 90D
Oct 24, 2018
123
44
Will we have 15 stops of DR on R3 or should we wait for more, R1 , or other 20 stops DR rumored cameras? I have patience.
 

David - Sydney

EOS R
CR Pro
Dec 7, 2014
836
696
www.flickr.com
Kolari Vision did a teardown of the internals and they now have a heatsink mod for the R5 for $400.
I am not sure that it gained enough overheat time for the mod to be worth it but Kolari Vision is an interesting company.
Interesting solution! Similar to the hack with copper heatsink that someone on youtube did but with Koalari's warranty.
What is doesn't mention is how to reduce the temperature on the back panel. It would be foolish to leave the LCD turned in and be damaged when video is being taken. Low temperature burns would also be an issue to be aware of.
Tilta advertises a cage/peltier cooler for the R5 with the cage being a massive heatsink but it doesn't appear that it can be ordered for some reason. It would reduce the back panel temperatures effectively but needs a separate power supply.

Has anyone measured the 8k raw lite record time when firmware 1.3 was released?