Reading mode: Light Dark

Canon EOS 90D Initial Findings at PhotonsToPhotos

Go to last post
B

bclaff

Veteran Member
Messages
14,415
Solutions
24
Reaction score
13,408
Location
Metro-West Boston, MA, US
I performed a Read Noise in DN analysis and the results agree with those by cgarcia.

7e9db192390c4af692bdc044db864782.jpg.png

My estimates of DxOMark Landscape Dynamic Range (8MP normalized Engineering Dynamic Range (EDR)) also agree.

The implications for Photographic Dynamic Range (PDR) are not so dramatic.
PDR takes all noise sources into account, not just read noise.

I have a mathematical model to estimate PDR from Read Noise in DN and Full Well Capacity (FWC).
Here's a chart of my PDR estimate based in the above Read Noise in DN and an educated guess at FWC:

36fd55796f96401f812a08ddd8848946.jpg.png

Please remember this is a very early estimate.
However, I don't see evidence of any dramatic improvement.

FWIW, the sensor in 90D does seem to be identical to that in the M6 Mark II

--
Bill ( Your trusted source for independent sensor data at PhotonsToPhotos )
 
Last edited:
bclaff said:
FWIW, the sensor in 90D does seem to be identical to that in the M6 Mark II
Click to expand...
Are there any striking differences in performance (one being better than the other?)

Chris
 
Chris Mak said:
bclaff said:
FWIW, the sensor in 90D does seem to be identical to that in the M6 Mark II
Click to expand...
Are there any striking differences in performance (one being better than the other?)
Click to expand...
At this point they look identical.
Although sometimes cameras with identical sensors perform differently for other reasons. For example a smaller body might not dissipate heat as well and therefore might be slightly noisier.

We'll know better whenever I get more complete and appropriate test files.
 
So when Canon said they got more resolution without degradation they were being very precisely truthful, leaving no room for optimism of improvements beyond stated goals.
Still, it’s a step in the right direction toward the 40MP APS-C sensor that I really want.

It would also be neat to see them make a 12MP video and low light focused sensor at the same time though.
 
Last edited:
bclaff said:
Please remember this is a very early estimate.
However, I don't see evidence of any dramatic improvement.
Click to expand...
Are you aware that there are ISO 100 samples from the 90D that give both ~2.7DN and ~3.7DN in the masked areas (both at higher f-numbers, so it isn't a microlens push" for low f-numbers). Either the specific cameras are different, or different electronic readout methods are used.
 
John Sheehy said:
bclaff said:
Please remember this is a very early estimate.
However, I don't see evidence of any dramatic improvement.
Click to expand...
Are you aware that there are ISO 100 samples from the 90D that give both ~2.7DN and ~3.7DN in the masked areas (both at higher f-numbers, so it isn't a microlens push" for low f-numbers). Either the specific cameras are different, or different electronic readout methods are used.
Click to expand...
Can cite the ~2.7DN example for closer inspection.
 
bclaff said:
Chris Mak said:
bclaff said:
FWIW, the sensor in 90D does seem to be identical to that in the M6 Mark II
Click to expand...
Are there any striking differences in performance (one being better than the other?)
Click to expand...
At this point they look identical.
Although sometimes cameras with identical sensors perform differently for other reasons. For example a smaller body might not dissipate heat as well and therefore might be slightly noisier.

We'll know better whenever I get more complete and appropriate test files.
Click to expand...
How can it me the "same" sensor? The 90D is 28 MP on a crop and the 6D2 is 26 MP on FF. They may perform similarly, but just the pixel density alone would suggest the 90D is a new sensor.
 
Silvex said:
bclaff said:
Chris Mak said:
bclaff said:
FWIW, the sensor in 90D does seem to be identical to that in the M6 Mark II
Click to expand...
Are there any striking differences in performance (one being better than the other?)
Click to expand...
At this point they look identical.
Although sometimes cameras with identical sensors perform differently for other reasons. For example a smaller body might not dissipate heat as well and therefore might be slightly noisier.

We'll know better whenever I get more complete and appropriate test files.
Click to expand...
How can it me the "same" sensor? The 90D is 28 MP on a crop and the 6D2 is 26 MP on FF. They may perform similarly, but just the pixel density alone would suggest the 90D is a new sensor.
Click to expand...
Looks like you mis-read; M6 Mark II not 6D Mark II :-)
 
bclaff said:
Silvex said:
bclaff said:
Chris Mak said:
bclaff said:
FWIW, the sensor in 90D does seem to be identical to that in the M6 Mark II
Click to expand...
Are there any striking differences in performance (one being better than the other?)
Click to expand...
At this point they look identical.
Although sometimes cameras with identical sensors perform differently for other reasons. For example a smaller body might not dissipate heat as well and therefore might be slightly noisier.

We'll know better whenever I get more complete and appropriate test files.
Click to expand...
How can it me the "same" sensor? The 90D is 28 MP on a crop and the 6D2 is 26 MP on FF. They may perform similarly, but just the pixel density alone would suggest the 90D is a new sensor.
Click to expand...
Looks like you mis-read; M6 Mark II not 6D Mark II :-)
Click to expand...
Yes I did. Now it makes total sense. It wouldn't make sense for these two to have different sensors.
 
bclaff said:
John Sheehy said:
bclaff said:
Please remember this is a very early estimate.
However, I don't see evidence of any dramatic improvement.
Click to expand...
Are you aware that there are ISO 100 samples from the 90D that give both ~2.7DN and ~3.7DN in the masked areas (both at higher f-numbers, so it isn't a microlens push" for low f-numbers). Either the specific cameras are different, or different electronic readout methods are used.
Click to expand...
Can cite the ~2.7DN example for closer inspection.
Click to expand...
I am not sure what that sentence is asking or saying, but if you mean where are the ~2.7 DN samples, they are in DPReview's sample images. Originally, I saw "FroKnows"' samples on his website, and the 4 ISO 100s were all ~3.7 DN, suggesting a defective camera or maybe a different mode of readout if ~2.7 is possible.
 
John Sheehy said:
bclaff said:
John Sheehy said:
bclaff said:
Please remember this is a very early estimate.
However, I don't see evidence of any dramatic improvement.
Click to expand...
Are you aware that there are ISO 100 samples from the 90D that give both ~2.7DN and ~3.7DN in the masked areas (both at higher f-numbers, so it isn't a microlens push" for low f-numbers). Either the specific cameras are different, or different electronic readout methods are used.
Click to expand...
Can cite the ~2.7DN example for closer inspection.
Click to expand...
I am not sure what that sentence is asking or saying, but if you mean where are the ~2.7 DN samples, they are in DPReview's sample images. Originally, I saw "FroKnows"' samples on his website, and the 4 ISO 100s were all ~3.7 DN, suggesting a defective camera or maybe a different mode of readout if ~2.7 is possible.
Click to expand...
I have not downloaded them all. Cite a specific image with ~2.7DN please.
 
bclaff said:
I have not downloaded them all. Cite a specific image with ~2.7DN please.
Click to expand...
IMG_-146.cr3 is one.

I think I see what is going on - perhaps you're looking at the average, which would be closer to ~3 DN with this file. Measuring noise is getting problematic, because of all the doctoring of "RAW" levels by manufacturers.

The fact is, the three color channels have different read noise in DN as given in the RAWs, but those pushed reds and blues are irrelevant, unless headroom is lost in those channels do to clipping, where those channels have less DR. In daylight, for example, those higher DNs for red and green may have no practical meaning. For incandescent, then the blue channel can lose a bit of DR

.--
John
 
John Sheehy said:
bclaff said:
I have not downloaded them all. Cite a specific image with ~2.7DN please.
Click to expand...
IMG_-146.cr3 is one.

I think I see what is going on - perhaps you're looking at the average, which would be closer to ~3 DN with this file. Measuring noise is getting problematic, because of all the doctoring of "RAW" levels by manufacturers.

The fact is, the three color channels have different read noise in DN as given in the RAWs, but those pushed reds and blues are irrelevant, unless headroom is lost in those channels do to clipping, where those channels have less DR. In daylight, for example, those higher DNs for red and green may have no practical meaning. For incandescent, then the blue channel can lose a bit of DR
Click to expand...
Your supposition is correct and although I understand your reasoning I always report values based on all the channels.

Here's the details:

3513bcf79a494928aabbd9d33a4055ae.jpg.png

Here's what's published:

6386b13aea0543d6b67280718f2385d7.jpg.png

Note I combine 4 channels not 3 as I treat the two green channels separately.

--
Bill ( Your trusted source for independent sensor data at PhotonsToPhotos )
 
I only use the two green channels combined.

The first Canon camera I remember doing this was the 5D4:

DR_5D4_RGB.png


So the red channel had a negative scaling (I think I didn't measured the maximum DN it could reach, likely I used the same 16383 for all).

The 1DX2 already had on-sensor ADC but still didn't messed the channels:

DR_1DX2_RGB.png


This is perhaps some kind of white balance normalization, due to software lazyness. If we don't properly select the unscaled channels, we are just introducing a new kind of noise source (let's call it software noise) in the measures :-D
 
Last edited:
If scaling were negative noise would be lower not higher.

There are no gaps in the histograms nor any beat pattern.

d872efeaf0a1468795600ad71b798863.jpg.png

So it's a bit of a mystery I think as to what's going on.
It does look a bit like white balance pre-scaling but perhaps in the analog domain.

--
Bill ( Your trusted source for independent sensor data at PhotonsToPhotos )
 
In simple speak, what is the bottom line of the 90 d in comparison to the 80 d and 7d2 regarding DR and high ISO?
 
bclaff said:
MAC said:
In simple speak, what is the bottom line of the 90 d in comparison to the 80 d and 7d2 regarding DR and high ISO?
Click to expand...
Initial indications are that the 90D will be a lot like the 80D in PDR and Low Light ISO but naturally with more resolution.
Click to expand...
ok, thanks

since my 7d2 is a bit better at iso6400 than my T7i (same sensor as 80d), I was hoping the 90d was a tad better at iso6400 than the 80d which trails a bit in low light iso
 
You must log in or register to reply here.

Keyboard shortcuts

f
Forum
m
My threads
Mobile site
About
Editorial content
Cameras & Lenses
Community
All content, design, and layout are Copyright © 1998–2025 Digital Photography Review All Rights Reserved.
Reproduction in whole or part in any form or medium without specific written permission is prohibited.
When you use DPReview links to buy products, the site may earn a commission.
©GPS Media - Guides, Products, Services.
Back
Top