Canon R6 - Excellent AF but disappointing dynamic range???

[sid_19911991]

Hi everyone,

I recently tested the Canon R6 for birding paired with Canon 100-400 USM ii.

I found the AF tracking & initial target acquisition to be the best that I have ever used. It was much faster than the D500/500 pf combo (mine) & A9i/200-600 combo (friend's).

I was able to track swallows in-flight (small birds) with great ease unlike ever before. The AI in it is extraordinary I guess.

However, I was extremely disappointed with its dynamic range performance in low light. It threw up a lot of noise even at ISO 800-1600 in low light.
The stationary sunbird that I shot had a lot color noise in its feathers.

A quick glance at DX0 rating told me R6's dynamic range is rated at 13.4 ev at ISO 100 & that it is not a BSI sensor.

1) Your thoughts?
2) How does the non BSI sensor Canon R5 fare?
3) Will the BSI sensor type in Canon's R3 up its dynamic range?

 

[FB101]

Canon has a long history of having files that are more noisy at intermediate ISO levels, that’s not unique to the R6 and why I eventually left canon for Nikon. Their claim has always been that the trade off was ok because

1. They actually kept noise increase under control at the highest isos. For sport shooters indoors it was a good trade off but i can’t say that in recent years canon’s claim has been very valid against the newest Nikon’s and Sony’s. But when the 1Dx came out, the claim was true.
2. Better colours in Adobe. Canon has long prided themselves in their relationship with Adobe and for the longest time it was true - you could drop a canon raw file in LR and colours were spot on. When I switched to Nikon I had to adjust seriously because colours were off on import. Over the last decade others have built better relationships with Adobe and the profiles have improved, while canon drifted away to the point that the R5/6 profiles in LR are largely regarded as poor. To see R6 colours you need to use a 3rd party profile. Jan Wegener has a video on that and how much of a difference it made in his workflow. Or you need to use the canon editing program...
3. finally, canon always claimed their files print well regardless of noise and I have found that to be true at least for models after the 1Dmk4. With the 1Dx they finally got rid of noise banding which was really bad up to that point.

Candidly, for the last 20 years canon has been more about ergonomics, optics, workflow than they have been about ultimate sensor science. I don’t know that it will change going forward. Other brands have seen between 1/2 and 2/3rd noise improvement with BSI, maybe canon will see a full stop - we’ll know soon now that the R3 is out.

i tested the R5 for a week - it’s a huge step forward compared to the 5Dmk4 but it still lags a little behind the D850 on image quality but you really need to pixel peep to notice or care.

 

[sid_19911991]

Canon has a long history of having files that are more noisy at intermediate ISO levels, that’s not unique to the R6 and why I eventually left canon for Nikon. Their claim has always been that the trade off was ok because

1. They actually kept noise increase under control at the highest isos. For sport shooters indoors it was a good trade off but i can’t say that in recent years canon’s claim has been very valid against the newest Nikon’s and Sony’s. But when the 1Dx came out, the claim was true.
2. Better colours in Adobe. Canon has long prided themselves in their relationship with Adobe and for the longest time it was true - you could drop a canon raw file in LR and colours were spot on. When I switched to Nikon I had to adjust seriously because colours were off on import. Over the last decade others have built better relationships with Adobe and the profiles have improved, while canon drifted away to the point that the R5/6 profiles in LR are largely regarded as poor. To see R6 colours you need to use a 3rd party profile. Jan Wegener has a video on that and how much of a difference it made in his workflow. Or you need to use the canon editing program...
3. finally, canon always claimed their files print well regardless of noise and I have found that to be true at least for models after the 1Dmk4. With the 1Dx they finally got rid of noise banding which was really bad up to that point.

Candidly, for the last 20 years canon has been more about ergonomics, optics, workflow than they have been about ultimate sensor science. I don’t know that it will change going forward. Other brands have seen between 1/2 and 2/3rd noise improvement with BSI, maybe canon will see a full stop - we’ll know soon now that the R3 is out.

i tested the R5 for a week - it’s a huge step forward compared to the 5Dmk4 but it still lags a little behind the D850 on image quality but you really need to pixel peep to notice or care.

Nice insight.

1) I did use Canon DPP & still wasn't happy.
2) I guess if Canon can bring out BSI sensors in its mid-range bodies or use Sony sensors, I would switch to Canon in a jiffy.
3) It will be interesting how R3 performs with a BSI sensor.

 

[bleirer]

Photons to photos tests and ranks low light performance objectively in EV units. The low light performance for your d500 is 9.68, the a9 is 11.05. The R6 is 10.66 and the r5 is 10.76. The Nikon d850 comes in at 10.36 and the d5 at 11.17.

So by objective testing the r6 is slightly better than the d500 you tested against, which makes me wonder if something else might be going on in your testing to give you such different results from photons to photos. Cropping for example can show more noise. I guess you'd have to shoot the same scene with both cameras and process both in a similar way. Or do what we all do, pick the one that looks best overall to our subjective eye even if testing says otherwise.

 

[FB101]

Photons to photos tests and ranks low light performance objectively in EV units. The low light performance for your d500 is 9.68, the a9 is 11.05. The R6 is 10.66 and the r5 is 10.76. The Nikon d850 comes in at 10.36 and the d5 at 11.17.

So by objective testing the r6 is slightly better than the d500 you tested against, which makes me wonder if something else might be going on in your testing to give you such different results from photons to photos. Cropping for example can show more noise. I guess you'd have to shoot the same scene with both cameras and process both in a similar way. Or do what we all do, pick the one that looks best overall to our subjective eye even if testing says otherwise.

Do those measurements differentiate between colour noise and luminosity noise? I find the former far more objectionable than the latter (maybe my film days) and canon to historically have noise that has more colour to it and is less randomly distributed, two factors that I feel impact how we react to noise in a picture, regardless (within reason) of how much noise there actually is.

 

[sid_19911991]

I find some of the ratings of DXO mark not in line with the performance in real world usage. So I always use it as a frame of reference and not as gospel.

Well, that's the best my brain can make sense of...

As per my casual testing, the luminosity noise level performance of the R6 was comparable to Sony/Nikon cameras/sensors, but the color/chromatic noise to be incredibly bad.

In my case, there was plenty of colour noise on the feathers of a stationary sunbird in focus shot with the R6. I shot from the terrace of my building. I have shot the same bird with my D500 at the same distance in bad light several times...

 

[sid_19911991]

Do those measurements differentiate between colour noise and luminosity noise? I find the former far more objectionable than the latter (maybe my film days) and canon to historically have noise that has more colour to it and is less randomly distributed, two factors that I feel impact how we react to noise in a picture, regardless (within reason) of how much noise there actually is.

Me too. I find colour noise a huge turn off, especially if it on the subject itself at sub 2000 ISOs. I don't mind luminosity noise so much.

 

[Tiago C]

Hi everyone,

I recently tested the Canon R6 for birding paired with Canon 100-400 USM ii.

I found the AF tracking & initial target acquisition to be the best that I have ever used. It was much faster than the D500/500 pf combo (mine) & A9i/200-600 combo (friend's).

I was able to track swallows in-flight (small birds) with great ease unlike ever before. The AI in it is extraordinary I guess.

However, I was extremely disappointed with its dynamic range performance in low light. It threw up a lot of noise even at ISO 800-1600 in low light.
The stationary sunbird that I shot had a lot color noise in its feathers.

A quick glance at DX0 rating told me R6's dynamic range is rated at 13.4 ev at ISO 100 & that it is not a BSI sensor.

1) Your thoughts?
2) How does the non BSI sensor Canon R5 fare?
3) Will the BSI sensor type in Canon's R3 up its dynamic range?

It's important to keep in mind that the same iso value may not be equivalent between brands. For example, Pentax iso 200 is sometimes equivalent to Nikon iso 100, depending on the cameras.

With that in mind, the Imaging Resource Comparometer is useful for comparing images at various iso values between different cameras: https://www.imaging-resource.com/IMCOMP/COMPS01.HTM

To my eye, the R6 and R5 images fare well against images of other brands at the same iso value.

As for the ability to recover details from shadows and highlights, this video compares the R6 with the A7III at iso 100 and 800: youtube]tXGiSIwTViY

 

[bleirer]

Do those measurements differentiate between colour noise and luminosity noise? I find the former far more objectionable than the latter (maybe my film days) and canon to historically have noise that has more colour to it and is less randomly distributed, two factors that I feel impact how we react to noise in a picture, regardless (within reason) of how much noise there actually is.

If I understand how he does the testing, he finds the iso point where the sensor still gives 6.5 stops of dynamic range, and then also converts that iso to exposure value, which is the number I gave. So for the r6 the low light iso was 5071 where the d500 was 2557, the d6 was 5722 and the a1 was 5215. No mention of color noise vs. Luminance noise.