EOS R5: NR always applied in raw files under ISO 640..

dilbert

I'm New Here
Aug 12, 2010
20
9
One of the first things I looked for after the R5 release was results on photonstophotos - e.g. https://www.photonstophotos.net/Charts/PDR.htm#Canon EOS R5

What does that mean? Bill Claff has posted his findings on dpreview:

Summary: With NR turned off (in any user accessible settings), the camera still applies it if ISO is less than 640. From the dpreview thread, Bill estimates the benefit is about 2/3 of a stop.

I can see why Canon might do this from a marketing perspective but if you are a stills shooter, is Canon being honest here? Do stills shooters care about this? Not if your ISO is over 640 but...
 

AlanF

Stay alert, control the camera, save photos
CR Pro
Aug 16, 2012
7,089
6,463
its a bit of a both: a noise reduction under ISO 640 and( in addition) a intermediate gain boost at ISO 400.. :)

https://www.photonstophotos.net/Charts/RN_ADU.htm#Canon EOS 1D X Mark III_14,Canon EOS R5_14

on unrelated note I am really liking how 1dx3 (R6) ISO 6400 noise level is as low as R5 at ISO 3200... DR be damned :D
Claff does write in the notes: "These raw values are not appropriate for comparing camera models because they are not adjusted for gain or area." Do you know what he means by that?
 

SecureGSM

2 x 5D IV
Feb 26, 2017
2,368
1,240
Claff does write in the notes: "These raw values are not appropriate for comparing camera models because they are not adjusted for gain or area." Do you know what he means by that?
as per Sharlin's post above: .. In particular they're not normalized for pixel size .. meaning that R5 45MP image normalised to 20MP resolution of R5 will likely deliver a similar result.
 

Joules

EOS R
CR Pro
Jul 16, 2017
1,035
1,155
Hamburg, Germany
as per Sharlin's post above: .. In particular they're not normalized for pixel size .. meaning that R5 45MP image normalised to 20MP resolution of R5 will likely deliver a similar result.
I don't think that is what is meant.

The PDR at PTP does take sensor size and resolution into account, based on my understanding. That's what is meant by the statement about the 8 X 10 inch print held at arm's length being used as the base for comparison, as far as I know.

The difference in gain mentioned is relevant, as ISO 400 for example does not have to mean the same thing on different cameras, and therefore you can't say 'this camera is better at this ISO' without knowing that the ISOs you are referring to are actually comparable.
 

AlanF

Stay alert, control the camera, save photos
CR Pro
Aug 16, 2012
7,089
6,463
Bill Claff of photonstophotos is obscure to non-specialists at the best of times. I am an experienced scientist and I don't know what he is getting at for much of the time.
 
  • Like
Reactions: BeenThere

SecureGSM

2 x 5D IV
Feb 26, 2017
2,368
1,240
I don't think that is what is meant.

The PDR at PTP does take sensor size and resolution into account, based on my understanding. That's what is meant by the statement about the 8 X 10 inch print held at arm's length being used as the base for comparison, as far as I know.

The difference in gain mentioned is relevant, as ISO 400 for example does not have to mean the same thing on different cameras, and therefore you can't say 'this camera is better at this ISO' without knowing that the ISOs you are referring to are actually comparable.
Hi Joules, I value your opinion. And yes, you are correct. There are multiple factors at play here. I tend to agree with what Privatebydesign was able to demonstrate: it appears that R6 and R5 high ISO performance is equaly strong when normalised for sensor resolution. it does make sense. We have seen R5, R6, 1Dx3 DPR scene comparison side by side. And difference is negligible if not indistinguishable completely.
we found that R5 produced much stronger artefacts (false colours) in the top left corner of the image Than R6. But that's expect-able.
 
Last edited:

Sharlin

EOS R
CR Pro
Dec 26, 2015
1,324
1,093
Turku, Finland
The PDR at PTP does take sensor size and resolution into account, based on my understanding. That's what is meant by the statement about the 8 X 10 inch print held at arm's length being used as the base for comparison, as far as I know.
Yes, PDR is of course normalized, which is why we don't see a stop of difference between the R6 and R5's ISO performance. But what SecureGSM linked was specifically raw non-normalized read noise measurements in (log) DN aka Digital Numbers, which is a post-ADC representation of the analog readout from a photosite. Smaller photosites means more gain needed means more noise per photosite.