Review: Canon EOS 6D Mark II by DPReview

[Aglet]

But Canon apologists live in alternate reality, where DR is not correlated to sensor size.
Keep the faith, guys. LOL

Yes, they sure do have a lot of convoluted reasoning for favoring continued mediocrity.
It's quite amusing. ;D

My 1DX MkII's have better DR than either the Nikon D5 or Sony A9, remind me again how, exactly, I am in denial?

Jeez... It's not all about the DR all the time. LOL
Canon is steeped in mediocrity all across the board!

 

[Maiaibing]

IIRC they admitted the test with the 80D was not done correctly. And promised to redo it. Well, probably they just forgot...

Oliver

They also promised to redo their rather misleading test of the 5DS/R (its noted on the review). Using the failed Adobe Lightroom color profile with crushed shadows for the review with the resultant inferior IQ without checking the results with Canon's own software was of DPR's worst blunders in a camera review IMHO. But nothing so far.

So expect to wait a couple of years. At least...

Crushing the black point doesn't necessarily result in "inferior IQ" when you want to compare a camera to others. In fact since it buries more noise under the black point, it can help a camera look better.

It does impact the impression of RAW latitude which is what DR is all about - and was a key negative of the 5DS/R according to DPR review. Here's what DPR themselves following complaints highlighted under their Dynamic RAW test which runs over 2 web pages (as we wait)... "Note: Comparisons are slightly complicated by the aggressive tone curve ACR is applying to the 5DS R files that are crushing its blacks, and potentially decreasing the levels of visible noise by making them darker. We expect a later version of ACR to fix this, at which point we will re-process and revisit these results."

 

[reef58]

But Canon apologists live in alternate reality, where DR is not correlated to sensor size.
Keep the faith, guys. LOL

Yes, they sure do have a lot of convoluted reasoning for favoring continued mediocrity.
It's quite amusing. ;D

My 1DX MkII's have better DR than either the Nikon D5 or Sony A9, remind me again how, exactly, I am in denial?

Jeez... It's not all about the DR all the time. LOL
Canon is steeped in mediocrity all across the board!

I think that is absurd. I think the frustration arises from there not being one perfect camera for me made by anyone. I will take Canon's warts for now, but have considered Sony for landscape. Canon just get bashed more because they are king of the hill. They do a lot of things well, but the sensor development on the 6d2 was not one of those things. 5d4 good sensor, 1dx2 good sensor.

 

[privatebydesign]

IIRC they admitted the test with the 80D was not done correctly. And promised to redo it. Well, probably they just forgot...

Oliver

They also promised to redo their rather misleading test of the 5DS/R (its noted on the review). Using the failed Adobe Lightroom color profile with crushed shadows for the review with the resultant inferior IQ without checking the results with Canon's own software was of DPR's worst blunders in a camera review IMHO. But nothing so far.

So expect to wait a couple of years. At least...

Crushing the black point doesn't necessarily result in "inferior IQ" when you want to compare a camera to others. In fact since it buries more noise under the black point, it can help a camera look better.

It does impact the impression of RAW latitude which is what DR is all about - and was a key negative of the 5DS/R according to DPR review. Here's what DPR themselves following complaints highlighted under their Dynamic RAW test which runs over 2 web pages (as we wait)... "Note: Comparisons are slightly complicated by the aggressive tone curve ACR is applying to the 5DS R files that are crushing its blacks, and potentially decreasing the levels of visible noise by making them darker. We expect a later version of ACR to fix this, at which point we will re-process and revisit these results."

If the RAW data was stored the same way in the different RAW file protocols then a 'same development' comparison might have a small amount of value. The trouble is the RAW information is stored very differently inside the various protocols.

Nikon RAW files have a fixed black point. Canon RAW files do not have a fixed black point, using a blacks crushing tone curve is precisely what not to do to a floating black point file. The first thing you have to do is establish a suitable black point, all this means is raising the blacks slider until a suitable area of the image is black.

This is so fundamental to getting the best out of the RAW files I can't believe anybody puts any weight on the bullsh!t DPReview 'comparisons', they are not comparisons by any reasonable measure of the word.

Flip the methodology, set the Canon black point to where it should be and dial in an entirely arbitrary value on the black slider for any Nikon RAW file then compare, is that a comparison of any value?

 

[stevelee]

75-300mm . . . I've shot some rather clear pictures of the full moon with is, so I trust I'll get some passable eclipse pictures with my T3i on the 21st is the weather cooperates. 480mm equivalent is not bad, assuming that the moon will be about the same size in the daytime as it was when I shot it at night.

So today my eclipse filter came in the mail. It's rated for 18 stops, or something ridiculous like that. When held up to the light, it looks opaque. The sun has been popping in and out today. During a short period of sunlight, I tried taking its picture through the filter. I'm glad I tried practicing rather than waiting for the eclipse to try the filter. Focusing is a challenge, trying to see the dim image on the screen in bright sunlight. I'm glad to have the T3i's flippy screen and am more convinced that I will be glad to have one on the 6D2. I guess I need a black cloth to put over my head and the screen during the partial phases of the eclipse, though having the horns of the moon's shadow to focus on should help. The 480 equivalent gives a small image, but I might not want it a lot bigger when I'm photographing corona. I guess I should crank up screen brightness, too. I'm using features that I don't normally use enough to remember where they are, or even that they do exist.

 

[bclaff]

...Nikon RAW files have a fixed black point. Canon RAW files do not have a fixed black point, using a blacks crushing tone curve is precisely what not to do to a floating black point file. ...

Perhaps you can clarify this.
I've never seen a Canon raw file with a non-integral BlackLevel (black point).
There's four values, one for each osition in the Color Filter Array (CFA); but they are always integers.
Older Nikon cameras used to always use 0 but this is no longer true.

 

[LonelyBoy]

The D5 is a highly-specialized sports camera. There are different expectations for that camera.

Did DPR say that about the 7D, 7D2, and 1DX too? Or does that only apply for high-specialized Nikon sports cameras?

 

[privatebydesign]

...Nikon RAW files have a fixed black point. Canon RAW files do not have a fixed black point, using a blacks crushing tone curve is precisely what not to do to a floating black point file. ...

Perhaps you can clarify this.
I've never seen a Canon raw file with a non-integral BlackLevel (black point).
There's four values, one for each osition in the Color Filter Array (CFA); but they are always integers.
Older Nikon cameras used to always use 0 but this is no longer true.

Probably not to your satisfaction, and I certainly can't talk to your level of understanding. I can point you to some links and add that I find Canon files to be far more sensitive to black slider input than anything else I have processed, and I do process a variety of camera outputs for printing.

Now if I have misspoken on a technical term I apologize in advance, but like I say, whatever the reason, I have always found Canon files to respond better once the black levels slider is moved. That might be an incomplete work around for a far deeper technical difference, or it might not, but it makes the image output better so call it what you will.

http://ninedegreesbelow.com/photography/floating-point-dcraw.html
http://www.openphotographyforums.com/forums/showthread.php?t=12280

If you find those old flawed and irrelevant,

I certainly trust Roger Clarks understanding of the underlying physics and maths.
http://www.clarkvision.com/articles/canon.raw.processing1/

 

[bclaff]

...Nikon RAW files have a fixed black point. Canon RAW files do not have a fixed black point, using a blacks crushing tone curve is precisely what not to do to a floating black point file. ...

Perhaps you can clarify this.
I've never seen a Canon raw file with a non-integral BlackLevel (black point).
There's four values, one for each osition in the Color Filter Array (CFA); but they are always integers.
Older Nikon cameras used to always use 0 but this is no longer true.

... I certainly can't talk to your level of understanding. ...

Now if I have misspoken on a technical term I apologize in advance, but like I say, whatever the reason, I have always found Canon files to respond better once the black levels slider is moved. That might be an incomplete work around for a far deeper technical difference, or it might not, but it makes the image output better so call it what you will.

...

OK. So perhaps you erred in offering your rationale for what you observed.
I can certainly see how different sensors could respond differently to changes in black level during post processing.
This has nothing to do with the BlackLevel in the raw data.
The most likely technical explanation is that the Photon Transfer Curve (PTC) slopes are different in that region so moving a slider will have more/less effect depending on that slope.

 

[Talys]

Actually, theoretical dynamic range scales the exact same way too.
Did you skip physics classes by any chance? LOL

That's ridiculous. The primary benefit of a larger sensor to dynamic range is that you get less noise, and less noise means higher potential dynamic range. That should be all you "expect" out of going Full Frame.

Hmm, I don't get it. You to fully agree with me ... after you've slammed my statement as ridiculous ??

To increase the theoretical dynamic range of a camera's ability to record, you need to mess with the analog to digital converter and that isn't a trivial exercise.

That's not theoretical DR. What you are saying is implementation-specific.
Let's not confuse theory with implementation.

Neuro's original argument was that DR doesn't scale with sensor size - when in fact it does.
That's the theory and it's correct.

In practice, Canon's off-chip ADC implementation makes the theory look wrong.
But if you look at Sony sensors, FF sensors generally have better DR than crop sensors - in line with the theory.

But Canon apologists live in alternate reality, where DR is not correlated to sensor size.
Keep the faith, guys. LOL

Omg. How do I put this in easy words?

Bigger sensor does not mean more DR at base ISO.

Bigger sensor DOES mean less noise at higher ISO. So at higher ISO, you have less noise, meaning photos shot at high ISO will take a smaller hit. But who shoots ISO 3200 for high DR? Still, 6DII at ISO 3200 will have less noise than 80D, and therefore, the DR will suffer less as ISO goes up.

There are plenty of small sensors (even smartphone sized) that have excellent DR at base ISO. So going from a Galaxy S8 or iPhone 6S to some point and shoot, or even a $150 APSC will probably be a DR downgrade at base ISO, even though the sensor is bigger. But even versus cheap APSC, the smartphone will have crappy low light performance.

Regarding premium product expectations: the 6DII is not a premium FF camera. It is an entry level FF camera, that is barely more expensive than top-end APSCs, and 40% less than Canon's premium FF (5DIV).

 

[privatebydesign]

...Nikon RAW files have a fixed black point. Canon RAW files do not have a fixed black point, using a blacks crushing tone curve is precisely what not to do to a floating black point file. ...

Perhaps you can clarify this.
I've never seen a Canon raw file with a non-integral BlackLevel (black point).
There's four values, one for each osition in the Color Filter Array (CFA); but they are always integers.
Older Nikon cameras used to always use 0 but this is no longer true.

... I certainly can't talk to your level of understanding. ...

Now if I have misspoken on a technical term I apologize in advance, but like I say, whatever the reason, I have always found Canon files to respond better once the black levels slider is moved. That might be an incomplete work around for a far deeper technical difference, or it might not, but it makes the image output better so call it what you will.

...

OK. So perhaps you erred in offering your rationale for what you observed.
I can certainly see how different sensors could respond differently to changes in black level during post processing.
This has nothing to do with the BlackLevel in the raw data.
The most likely technical explanation is that the Photon Transfer Curve (PTC) slopes are different in that region so moving a slider will have more/less effect depending on that slope.

So in your terminology what does this statement mean "Nikon's raw files have no bias (no offset). Canon's do,"? Because Roger Clark made it about Canon cameras as recent as the 6D.

Further, I base my understanding (though complete technical lack thereof) on this statement and example of post processing

"Raw image data from Canon's cameras have more dynamic range than what people commonly show on the internet. Canon raw data needs to be processed differently than Nikon raw files. Nikon's raw files have no bias (no offset). Canon's do, so that one can do proper statistics at the low end. Thus, to show proper dynamic range the offsets need to be properly managed."

Combined with my personal experience of using RAW files from many cameras his comments 100% align with my personal observations.

Also would you agree that my basic point is valid? Using a "blacks crushing tone curve" will give an entirely distorted impression of a cameras capabilities if said camera file was particularly sensitive to a boosted blacks level slider?

 

[bclaff]

...

Bigger sensor does not mean more DR at base ISO.

...

This depends on the context of what is meant by "DR".
If you're talking DR that has been normalized to a consistent print size and viewing distance, like Photographic Dynamic Range (PDR) at PhotonsToPhotos or Landscape Score like at DxOMark then yes a larger sensor area will produce a higher DR (in general).
If you're talking about at the pixel level, well, that's "apples to oranges" and all bets are off.

 

[daphins]

Funny! I am having a similar experience. I'm taking a trip to NZ ina. Fee months, and a huge goal of mine is to take photographs of their cave glow worms. I've been waiting for the mkii to come out for awhile. I've been practicing in my basement at night trying to focus on LED's from our smoke alarm. Focusing on near dark is HARD. DPAF is useless, have to go manual and work to get it even visible. It's easy to be so out of focus that you don't even see the LED

75-300mm . . . I've shot some rather clear pictures of the full moon with is, so I trust I'll get some passable eclipse pictures with my T3i on the 21st is the weather cooperates. 480mm equivalent is not bad, assuming that the moon will be about the same size in the daytime as it was when I shot it at night.

So today my eclipse filter came in the mail. It's rated for 18 stops, or something ridiculous like that. When held up to the light, it looks opaque. The sun has been popping in and out today. During a short period of sunlight, I tried taking its picture through the filter. I'm glad I tried practicing rather than waiting for the eclipse to try the filter. Focusing is a challenge, trying to see the dim image on the screen in bright sunlight. I'm glad to have the T3i's flippy screen and am more convinced that I will be glad to have one on the 6D2. I guess I need a black cloth to put over my head and the screen during the partial phases of the eclipse, though having the horns of the moon's shadow to focus on should help. The 480 equivalent gives a small image, but I might not want it a lot bigger when I'm photographing corona. I guess I should crank up screen brightness, too. I'm using features that I don't normally use enough to remember where they are, or even that they do exist.

 

[bclaff]

...Nikon RAW files have a fixed black point. Canon RAW files do not have a fixed black point, using a blacks crushing tone curve is precisely what not to do to a floating black point file. ...

Perhaps you can clarify this.
I've never seen a Canon raw file with a non-integral BlackLevel (black point).
There's four values, one for each osition in the Color Filter Array (CFA); but they are always integers.
Older Nikon cameras used to always use 0 but this is no longer true.

... I certainly can't talk to your level of understanding. ...

Now if I have misspoken on a technical term I apologize in advance, but like I say, whatever the reason, I have always found Canon files to respond better once the black levels slider is moved. That might be an incomplete work around for a far deeper technical difference, or it might not, but it makes the image output better so call it what you will.

...

OK. So perhaps you erred in offering your rationale for what you observed.
I can certainly see how different sensors could respond differently to changes in black level during post processing.
This has nothing to do with the BlackLevel in the raw data.
The most likely technical explanation is that the Photon Transfer Curve (PTC) slopes are different in that region so moving a slider will have more/less effect depending on that slope.

So in your terminology what does this statement mean "Nikon's raw files have no bias (no offset). Canon's do,"? Because Roger Clark made it about Canon cameras as recent as the 6D.

...

This statement was once true but is now outdated (was outdated at the time Roger wrote that.)
The following Nikon cameras, at a minimum, no longer have a Black Level of zero:
D3300, D5, D500, D5300, D5500, D7200, D750, D7500, D810, D810A.

 

[privatebydesign]

...Nikon RAW files have a fixed black point. Canon RAW files do not have a fixed black point, using a blacks crushing tone curve is precisely what not to do to a floating black point file. ...

Perhaps you can clarify this.
I've never seen a Canon raw file with a non-integral BlackLevel (black point).
There's four values, one for each osition in the Color Filter Array (CFA); but they are always integers.
Older Nikon cameras used to always use 0 but this is no longer true.

... I certainly can't talk to your level of understanding. ...

Now if I have misspoken on a technical term I apologize in advance, but like I say, whatever the reason, I have always found Canon files to respond better once the black levels slider is moved. That might be an incomplete work around for a far deeper technical difference, or it might not, but it makes the image output better so call it what you will.

...

OK. So perhaps you erred in offering your rationale for what you observed.
I can certainly see how different sensors could respond differently to changes in black level during post processing.
This has nothing to do with the BlackLevel in the raw data.
The most likely technical explanation is that the Photon Transfer Curve (PTC) slopes are different in that region so moving a slider will have more/less effect depending on that slope.

So in your terminology what does this statement mean "Nikon's raw files have no bias (no offset). Canon's do,"? Because Roger Clark made it about Canon cameras as recent as the 6D.

...

This statement was once true but is now outdated (was outdated at the time Roger wrote that.)
The following Nikon cameras, at a minimum, no longer have a Black Level of zero:
D3300, D5, D500, D5300, D5500, D7200, D750, D7500, D810, D810A.

So most modern RAW files are sensitive to where the black slider is? So in your estimation how relevant is a comparison between RAW files that simply have an exposure slider change?

And again, would you agree that my basic point is valid? Using a "blacks crushing tone curve" will give an entirely distorted impression of a cameras capabilities if said camera file was particularly sensitive to a boosted blacks level slider?

 

[bclaff]

...Nikon RAW files have a fixed black point. Canon RAW files do not have a fixed black point, using a blacks crushing tone curve is precisely what not to do to a floating black point file. ...

Perhaps you can clarify this.
I've never seen a Canon raw file with a non-integral BlackLevel (black point).
There's four values, one for each osition in the Color Filter Array (CFA); but they are always integers.
Older Nikon cameras used to always use 0 but this is no longer true.

... I certainly can't talk to your level of understanding. ...

Now if I have misspoken on a technical term I apologize in advance, but like I say, whatever the reason, I have always found Canon files to respond better once the black levels slider is moved. That might be an incomplete work around for a far deeper technical difference, or it might not, but it makes the image output better so call it what you will.

...

OK. So perhaps you erred in offering your rationale for what you observed.
I can certainly see how different sensors could respond differently to changes in black level during post processing.
This has nothing to do with the BlackLevel in the raw data.
The most likely technical explanation is that the Photon Transfer Curve (PTC) slopes are different in that region so moving a slider will have more/less effect depending on that slope.

So in your terminology what does this statement mean "Nikon's raw files have no bias (no offset). Canon's do,"? Because Roger Clark made it about Canon cameras as recent as the 6D.

...

This statement was once true but is now outdated (was outdated at the time Roger wrote that.)
The following Nikon cameras, at a minimum, no longer have a Black Level of zero:
D3300, D5, D500, D5300, D5500, D7200, D750, D7500, D810, D810A.

... would you agree that my basic point is valid? Using a "blacks crushing tone curve" will give an entirely distorted impression of a cameras capabilities if said camera file was particularly sensitive to a boosted blacks level slider?

I think it's circular to say that a sensor is "sensitive" to boosting black based on how the result appears to you.
You haven't proven any cause and effect.

That said. This is one of the primary reasons I only do measurements on raw data (prior to any tone curve).

 

[testthewest]

I evade the biggest bs there is, if I can. That's true.
Perhaps you simply open canonrumors main site and just look at the topics presented there instead of the forums, which have next to no exposure in comparison. The main page is where its at.

I see. In that case, perhaps you should simply restrict your participation in this site to viewing the main page. You won't be missed here on the forums.

These threads are linked on the main page for discussing the gear and topics presented there. That's what I do. If you can't take someone being critical about Canons offereing, how about YOU just stop reading here? Make your own echo chamber, if you really can't take it. Just put your 21K posts on rotation on your screen, perhaps then your fine.
And wheter someone is missed or not, is not yours to decide btw. I think alot of readers here could live without your input as well.

 

[BillB]

...

Bigger sensor does not mean more DR at base ISO.

...

This depends on the context of what is meant by "DR".
If you're talking DR that has been normalized to a consistent print size and viewing distance, like Photographic Dynamic Range (PDR) at PhotonsToPhotos or Landscape Score like at DxOMark then yes a larger sensor area will produce a higher DR (in general).
If you're talking about at the pixel level, well, that's "apples to oranges" and all bets are

Right you are. But apples to oranges discussions are what keep the posts coming. At this point, DR means pretty much whatever the poster wants it to mean. Signal to noise ratio is independent of sensor size. Apparent noise in a print of a given size (viewed at a given distance) is related to sensor size, at least in most cases. How significant is all this? Don't even ask. It would spoil the fun, and might reduce the level of those all important clicks.

 

[neuroanatomist]

...

Bigger sensor does not mean more DR at base ISO.

...

This depends on the context of what is meant by "DR".
If you're talking DR that has been normalized to a consistent print size and viewing distance, like Photographic Dynamic Range (PDR) at PhotonsToPhotos or Landscape Score like at DxOMark then yes a larger sensor area will produce a higher DR (in general).
If you're talking about at the pixel level, well, that's "apples to oranges" and all bets are

Right you are. But apples to oranges discussions are what keep the posts coming. At this point, DR means pretty much whatever the poster wants it to mean. Signal to noise ratio is independent of sensor size. Apparent noise in a print of a given size (viewed at a given distance) is related to sensor size, at least in most cases. How significant is all this? Don't even ask. It would spoil the fun, and might reduce the level of those all important clicks.

+1

I agree that a larger sensor will have less noise and therefore more DR, all else being equal. But when is all else equal? Outside comparing a cropped image to an uncropped image from the same sensor, pretty much never.

 

[jeffa4444]

Part of the issue is the 'Limited" testing reviewers do. To get an accurate performance reading on the 6D MKII or any other camera testing one is unrealistic. We tested 10 Sony F55 cameras and found variances that affected dynamic range and for certain visual effects shots apparently that matters.
Senscore tested one each the Canon 5DS and the 5DSr the only difference should have been resolution but they rated the dynamic range greater on the 5DSr. Ive been told on this forum that its not an issue but I challenge that statement because batch testing cameras from some pretty serious high end manufacturers we have seen the opposite i.e. variances.

Our test equipment and testing conditions replicate major manufacturers as does the software we use to interpret results. So many components can affect the final output and regardless of what we think human interpretation that I believe batch testing is the only true measure and the only source regularly on here that does that is lens rentals.