R5 vs EOS R shadow recovery samples

[Kit Lens Jockey]

Trying to bring everybody the content that I would want to see if I didn't have my hands on an R5 yet. I did some comparisons of the R5 vs the EOS R for how well they can recover shadows in dark photos at various ISOs. All photos taken with an RF 50mm 1.2. (From here on I'm just going to refer to the EOS R as R.)

All photos were resized to the same dimensions to negate the different sensor megapixels. No noise reduction used anywhere.

First, an underexposed photo at low ISO. I'm only going to include an original from the R5 here because the R didn't look too different.

R5, 1/60sec, f4, ISO 200


And brightened up version, R vs R5

R5, 1/60sec, f4, ISO 200, brightened 2.75 and raised shadows


R, 1/60sec, f4, ISO 200, brightened 2.75 and raised shadows


R5 detail, 1/60sec, f4, ISO 200, brightened 2.75 and raised shadows


R detail, 1/60sec, f4, ISO 200, brightened 2.75 and raised shadows


Next, same photo at 10000 ISO. But I did not intentionally underexpose it, so I'm not going to post the original. I just pushed the shadows up on both the R5 and the R.

R5, 1/800sec, f4, ISO 10000, pushed shadows by 100


R, 1/800sec, f4, ISO 10000, pushed shadows by 100


R5 detail, 1/800sec, f4, ISO 10000, pushed shadows by 100


R detail, 1/800sec, f4, ISO 10000, pushed shadows by 100

 

[Kit Lens Jockey]

Same thing at 20000 ISO and then 40000 ISO

R5, 1/1600sec, f4, ISO 20000, pushed shadows by 60


R, 1/1600sec, f4, ISO 20000, pushed shadows by 60


R5 detail, 1/1600sec, f4, ISO 20000, pushed shadows by 60


R detail, 1/1600sec, f4, ISO 20000, pushed shadows by 60


And finally 40000 ISO.

R5, 1/3200sec, f4, ISO 40000, pushed shadows by 60


R, 1/3200sec, f4, ISO 40000, pushed shadows by 60


R5 detail, 1/3200sec, f4, ISO 40000, pushed shadows by 60


R detail, 1/3200sec, f4, ISO 40000, pushed shadows by 60

 

[Kit Lens Jockey]

Overall I think the R5 is a little better. It's not the earth shattering improvement I would have hoped for, being about a four year newer sensor. But to pack 50% more pixels into the sensor and still have the noise be as good or slightly better, I'll take it.

 

[digigal]

Thanks for posting that.

 

[Joules]

Great, thanks. High ISO performance is strongly influenced by shot noise, so getting any improvement is already a good thing.

Anyway, all that's left is the obligatory base ISO 5 stop push (preferably with the shadows also raised 100) to see if any banding remains (I don't think there will be, after all it is already so much better on the 90D, M6 II and 1DX III sensor).

I think they managed to eliminate or reduce a good bit of the banding in the R via firmware at some point. But for people coming from any other Canon camera, that's one area where actual improvements can be expected.

 

[Chris.Chapterten]

Thanks for posting this. But why not test iso 100? The differences should be most pronounced at base iso when comparing dynamic range.

 

[Kit Lens Jockey]

Thanks for posting this. But why not test iso 100? The differences should be most pronounced at base iso when comparing dynamic range.

Ask and ye shall receive. I just figured I would do ISO 200 because outside of landscapes on a tripod, most people don't have the luxury of staying right at base ISO under most conditions. Anyway, here are a few more shots. I decided to use something with a little more detail in the shadows. Please ignore the dusty glass!

All shots taken with the R are at firmware 1.4.0

R5 , 1/13sec, f9, ISO 100 (Again not going to include the R to compare the original since they're pretty similar.)


R5 , 1/13sec, f9, ISO 100, pushed 5 stops and shadows raised by 100


R , 1/13sec, f9, ISO 100, pushed 5 stops and shadows raised by 100


R5 detail , 1/13sec, f9, ISO 100, pushed 5 stops and shadows raised by 100


R detail , 1/13sec, f9, ISO 100, pushed 5 stops and shadows raised by 100


Actually the R5 looks like it has much less nasty color noise when you push shadows to the extremes!

 

[Kit Lens Jockey]

Another detail crop of a darker area. The R5 definitely holds together better when recovering very dark shadows. There's a ton less color noise.

R5 detail , 1/13sec, f9, ISO 100, pushed 5 stops and shadows raised by 100


R detail , 1/13sec, f9, ISO 100, pushed 5 stops and shadows raised by 100

 

[Chris.Chapterten]

Ask and ye shall receive. I just figured I would do ISO 200 because outside of landscapes on a tripod, most people don't have the luxury of staying right at base ISO under most conditions. Anyway, here are a few more shots. I decided to use something with a little more detail in the shadows. Please ignore the dusty glass!

All shots taken with the R are at firmware 1.4.0

R5 , 1/13sec, f9, ISO 100 (Again not going to include the R to compare the original since they're pretty similar.)
View attachment 191703

R5 , 1/13sec, f9, ISO 100, pushed 5 stops and shadows raised by 100
View attachment 191704

R , 1/13sec, f9, ISO 100, pushed 5 stops and shadows raised by 100
View attachment 191705

R5 detail , 1/13sec, f9, ISO 100, pushed 5 stops and shadows raised by 100
View attachment 191706

R detail , 1/13sec, f9, ISO 100, pushed 5 stops and shadows raised by 100
View attachment 191707

Actually the R5 looks like it has much less nasty color noise when you push shadows to the extremes!

Wow! thanks for taking the time to do that. It definitely looks like the R5 has improved at base iso. I think it would be quite rare to need to push an image that much but great to see that it holds up so well.

Can I ask what picture profile you are using in the raw conversion? I've noticed in a few sample images from the R5 that it seems to have a slight magenta cast when compared to the R.

 

[Kit Lens Jockey]

Wow! thanks for taking the time to do that. It definitely looks like the R5 has improved at base iso. I think it would be quite rare to need to push an image that much but great to see that it holds up so well.

Can I ask what picture profile you are using in the raw conversion? I've noticed in a few sample images from the R5 that it seems to have a slight magenta cast when compared to the R.

I didn't even take note of that when I was doing it. Looks like I was using the "camera standard" profile. But Photoshop also gives me a warning "Profile not installed, rendering is incorrect." So yeah I don't think I have things quite dialed in yet. I wouldn't worry about the magenta cast.

 

[Chris.Chapterten]

I didn't even take note of that when I was doing it. Looks like I was using the "camera standard" profile. But Photoshop also gives me a warning "Profile not installed, rendering is incorrect." So yeah I don't think I have things quite dialed in yet. I wouldn't worry about the magenta cast.

Good to know, thanks!

 

[Kit Lens Jockey]

In light of the reports that the R5 applies noise reduction even on its raw files at low ISOs, I decided to do a comparison of the previous files, but I added noise reduction to the files from the R. I also corrected the white balance on both files so the "M" in the Mobil logo is set to be the neutral gray.

I think the R still has a lot more nasty looking green color noise that is not present in the R5 file, even after noise reduction is run on the R file.

R5 detail , 1/13sec, f9, ISO 100, pushed 5 stops and shadows raised by 100, white balance corrected


R detail , 1/13sec, f9, ISO 100, pushed 5 stops and shadows raised by 100, white balance corrected, noise reduction added

 

[Chris.Chapterten]

In light of the reports that the R5 applies noise reduction even on its raw files at low ISOs, I decided to do a comparison of the previous files, but I added noise reduction to the files from the R. I also corrected the white balance on both files so the "M" in the Mobil logo is set to be the neutral gray.

I think the R still has a lot more nasty looking green color noise that is not present in the R5 file, even after noise reduction is run on the R file.

R5 detail , 1/13sec, f9, ISO 100, pushed 5 stops and shadows raised by 100, white balance corrected
View attachment 191826

R detail , 1/13sec, f9, ISO 100, pushed 5 stops and shadows raised by 100, white balance corrected, noise reduction added
View attachment 191827

thanks for doing this. That has been my conclusion too. The R5 still retains more detail and colour information in the shadows despite any supposed noise reduction. It's honestly hard to see any ill affects of the noise reduction even if it is being applied. R5 images look very crisp/sharp to my eyes

 

[Kit Lens Jockey]

thanks for doing this. That has been my conclusion too. The R5 still retains more detail and colour information in the shadows despite any supposed noise reduction. It's honestly hard to see any ill affects of the noise reduction even if it is being applied. R5 images look very crisp/sharp to my eyes

I agree. I mean maybe the R5 raws at low ISO do have noise reduction baked in. But even if they do, I can't seem to replicate their clarity or color consistency in post with EOS R raw files.