Canon R6 Mark III High ISO and Dynamic Range – Good, but not Class Leading

[Richard CR]

Dpreview today has put out their studio scene with the EOS R6 Mark III, which gives us our first side-by-side look at how the EOS R6 Mark III compares to its relatives and also its competitors. One thing about sensor design is that there is no such thing as a free lunch. Something has to […]

See full article...

 

[Maximilian]

Thanks for that comparison, @Richard CR.
What do you think about the performance with mech shutter compared to electronic? (I don't need 40 fps most of the time)

 

[Richard CR]

i honestly was fine around 10EV for a sensor - ie: the M6 Mark II I felt that unless i really screwed up, had more lattitude than I needed.

but if you are thinking mech vs electronic? It's going to be around 1EV and probably close to 1.5EV difference depending on how extreme you are pushing. do you need the 11.2EV or whatever it turns out or are you perfectly happy with 10?

Everyone there is different. I mean, I had someone argue with me that they needed more than 11EV in a damned studio shooting still life.

I'm going to add in a comparison between the shutters.

 

[ThomasTH]

but if you are thinking mech vs electronic? It's going to be around 1EV and probably close to 1.5EV difference

Because of the 14 bits vs. 12 bits?

 

[Richard CR]

Because of the 14 bits vs. 12 bits?

yes. actually i just added it in there. It's closer to 1.5-1.8 EV. The difference of 2 bits is 2EV difference assuming 100% efficiency and 0 noise. So in this case, around 1.8 I think. very similar to the R6 Mark II with and without ES.

 

[Maximilian]

i honestly was fine around 10EV for a sensor - ie: the M6 Mark II I felt that unless i really screwed up, had more lattitude than I needed.

but if you are thinking mech vs electronic? It's going to be around 1EV and probably close to 1.5EV difference depending on how extreme you are pushing. do you need the 11.2EV or whatever it turns out or are you perfectly happy with 10?

Everyone there is different. I mean, I had someone argue with me that they needed more than 11EV in a damned studio shooting still life.

I'm going to add in a comparison between the shutters.

Thanks for that reply, Richard!
Having the R6m2 and being pretty pleased with it it was mainly about the higher MP count vs. high ISO performance.
And better AF as well.
I shoot a lot of wildlife action, dragonflies in flight, etc. so high ISO (>3200-6400) and good s/n there is what I'm looking for.

 

[GMCPhotographics]

This is some very interesting reading. It's looking like the R6iii is about a 1/3 stop worse in high Iso noise vs the R6ii....which is impressive! Where as a R5 vs R6ii comparison has the R5 about a whole stop worse. If some one shoots identical images on both cameras (R6iii and R6ii) at high iso and then prints them to the same size, say A2...the R6iii would show less noise due to the higher pixel count.

Does anyone know the R6iii's sensor readout speed yet? Ah...I see them measured on the main DPreview R6iii review page specs:
R6iii ~13.5ms vs R6ii ~14.7
About 1ms faster than the R6ii

 

[Richard CR]

This is some very interesting reading. It's looking like the R6iii is about a 1/3 stop worse in high Iso noise vs the R6ii....which is impressive! Where as a R5 vs R6ii comparison has the R5 about a whole stop worse. If some one shoots identical images on both cameras (R6iii and R6ii) at high iso and then prints them to the same size, say A2...the R6iii would show less noise due to the higher pixel count.

Does anyone know the R6iii's sensor readout speed yet? Ah...I see them measured on the main DPreview R6iii review page specs:
R6iii ~13.5ms vs R6ii ~14.7
About 1ms faster than the R6ii

it's faster with 17% more lines to read. Canon did a bang up job.. AND it is still FSI. bonkers.

 

[Tom R5R7]

I noticed with the e-shutter comparison, at +6, the Nikon does lose some detail as part of the noise reduction compared to the R6 III. That's something to be considered.

Overall, the increase in pixel density and faster readout (slightly) are a good tradeoff.

 

[GMCPhotographics]

it's faster with 17% more lines to read. Canon did a bang up job.. AND it is still FSI. bonkers.

Amusingly, it's slightly faster than Sony's semi stacked sensor. Qudos to Canon for matching and exceeding Sony's design without stacking.
I wrote a comment here about 4 years ago about how sensor readout speeds will be the big new feature over the next decade. Everyone laughed at me and mocked me....but here we are. The megapixel war is well and truely over....the sensor read out speed war is in full flight. The difference in real world shooting between 24mp through to 45mp on Canon is suprisingly slight. But, being able to shoot at 30/40 fps in 14 bit Electronic shutter mode without readout artifacts can make or break a picture.
With the R7 and R5, the max of 20 fps ES compared to 15/12fps FCS isn't enough to justify the risks of high ES frame rates. The drop in DR and the artifacts aren't worth the bother...especially with the R7's REALLY slow read out speed. The R6ii 's EC can function at 40fps and that's a whole different ball game. The diffrence between 12fps and 40fps is massive.
I'm wondering if in the future (R6iv?) we'll see a ES mode with 14 bit, but isn't stacked but limited to say 20 or 30 fps.
The R5ii's stacked sensor is amazing with it's read out speed in the same region as the R6ii/iii's 1st Curtain shutter. It's just seems to loose some of it's resolution advantage and some DR over the original R5. Once all the tests are availablem it's possible we could be seeing that the new 32mp R6iii sensor actually matching the current R5ii's resolution. The R5ii's sensor is resolving notibly less detail that the original R5.

 

[Zadak]

It's just seems to loose some of it's resolution advantage and some DR over the original R5

Huh?

 

[Maximilian]

Huh?

@GMCPhotographics was talking about R5m2 vs. R5. Just look at this:

Photographic Dynamic Range versus ISO Setting

Noticeably from ISO50 to 800, slightly from ISO800 to max.

 

[AlanF]

Huh?

He's right on resolution too. R5ii has slightly less resolution than the R5. They have been compared by optyczne.pl, and we discussed them in another thread.

 

[DocInfoSci]

Thanks! Very interesting to see this. Overall quite positive and not a crazy sacrifice for the pixels gained. Super happy to see this, and maybe with the R6 iv they'll keep the pixels but bump the ISO back up -- which would be a killer combo for this tier.

That stated, I think that the first comparison must always be pixel to pixel -- not downsized. If I buy a matrix of pixels then I want to use all of those pixels -- downsizing is a silly comparison. If I downsize of my R6 20mp images to 10mp they're amazing even at 52k ISO -- but why would I do that? I wouldn't. Crop, yes, but downsize for quality? Nope. I use all of those pixels to the best that I can and each pixel must stand on its own.

 

[scyrene]

That stated, I think that the first comparison must always be pixel to pixel -- not downsized.

You're welcome to feel that way, but it's no more correct.

why would I [downsize]? I wouldn't.

Unless your display device magically got bigger, than you will do that at least some of the time.

 

[snapshot]

He's right on resolution too. R5ii has slightly less resolution than the R5. They have been compared by optyczne.pl, and we discussed them in another thread.

do you have a link to the other thread? i felt the ES wasnt worth much to me on the R5 (20FPS for fast action is cool, but fast action produces rolling shutter artifacts - not cool). ES on R52 is barely ok for tennis and another factor of 2 in read out speed would be great. In terms of resolution in EFC, i dont know that i see a difference between R5 and R52. I believe you are a much more critical eye than i am, have you seen anything in your pictures?

 

[AlanF]

do you have a link to the other thread? i felt the ES wasnt worth much to me on the R5 (20FPS for fast action is cool, but fast action produces rolling shutter artifacts - not cool). ES on R52 is barely ok for tennis and another factor of 2 in read out speed would be great. In terms of resolution in EFC, i dont know that i see a difference between R5 and R52. I believe you are a much more critical eye than i am, have you seen anything in your pictures?

I can't find where. But, go to optyczne.pl and look at the sections on camera tests where they measure the resolution of each. https://www.optyczne.pl/457.4-Test_aparatu-Canon_EOS_R5_Rozdzielczość.html and https://www.optyczne.pl/510.4-Test_aparatu-Canon_EOS_R5_Mark_II_Rozdzielczość.html. There's not much in it but the R5 has a small edge on the pixel peeping level.

 

[DocInfoSci]

You're welcome to feel that way, but it's no more correct.

It isn’t more correct, but it provides an important understanding of the device.

Unless your display device magically got bigger, than you will do that at least some of the time.

Sure, downsizing happens all the time in the real world. But I’d be surprised if it’s more often for quality purposes — I suspect other concerns are more typically being addressed, such as saving on storage for the purpose of speed (network, printer processing relative to maximum effective resolution) or cost (less disk on a server).

For example, the maximum pixel peeping resolution in pixel terms for a Canon Pixma Pro 200 is 600ppi. Sending 1,200ppi won’t make the output sharper / cleaner — it just takes more time.

All things being equal. A smaller PNG is cheaper to store in the cloud than a larger PNG.

But , can it improve perceived quality in terms of noise due to the average that occurs on downsizing? Yeah. Is it wrong for this purpose? No. Is it useful for comparing sensors? Only if the target output is the same and in reach of both. Otherwise, it just hides pragmatic performance.

 

[snapshot]

I can't find where. But, go to optyczne.pl and look at the sections on camera tests where they measure the resolution of each. https://www.optyczne.pl/457.4-Test_aparatu-Canon_EOS_R5_Rozdzielczość.html and https://www.optyczne.pl/510.4-Test_aparatu-Canon_EOS_R5_Mark_II_Rozdzielczość.html. There's not much in it but the R5 has a small edge on the pixel peeping level.

i looked but i am disappointed with the test procedure. is this really seeing a RF50 1.2L vs RF28-70 2.8L @ F4+ comparison? if so i am impressed honestly that the RF28-70 competes very favorably. to bad they didnt show the RF50 1.2L at 2.8 also. is the idea really that noise reduction in the raw files significantly impact camera resolution? if so i would advise them to find a high resolution lens they can adapt to any major system and manually focus it. as it is i see a fun curve peaking at F/4 and tailing off.

 

[Richard CR]

Thanks! Very interesting to see this. Overall quite positive and not a crazy sacrifice for the pixels gained. Super happy to see this, and maybe with the R6 iv they'll keep the pixels but bump the ISO back up -- which would be a killer combo for this tier.

That stated, I think that the first comparison must always be pixel to pixel -- not downsized. If I buy a matrix of pixels then I want to use all of those pixels -- downsizing is a silly comparison. If I downsize of my R6 20mp images to 10mp they're amazing even at 52k ISO -- but why would I do that? I wouldn't. Crop, yes, but downsize for quality? Nope. I use all of those pixels to the best that I can and each pixel must stand on its own.

I think it's the reverse, actually, unless you are printing 100" wide prints and want people to camp out 6" away from the print and compare side to side, it's a little irrelevant. There's a diminishing return to more mega pickles.

You aren't paying x dollars per megapixel, you're paying for the camera on the whole.

In this case, objectively, just downsizing is the worst-case scenario, as I mentioned in the article.