PetaPixel poops on the 6D2 sensor

[9VIII]

Or download the files and process them optimally and see how good or bad it actually is. No, why would we do that when we can spend so much time shouting at each other?

Anyway, I know half of you will pull this to pieces but....



1st shot. So here is a crop of the 100iso shot from the 6D MkII vs the D750 both lifted 5 stops, and it's link.

https://www.dpreview.com/articles/3416153698/canon-eos-6d-mark-ii-dynamic-range

2nd shot. I downloaded the 6D MkII file and processed it close to optimally and this is the result. I took five minutes to do this, given a few hours I could do better and work out profiles for each iso and lift amount that could be saved as presets in LR.

You cannot apply the same 'standard' settings to different cameras and say look at the differences. You have to process each file optimally!

When you do you get very different results. When will people stop eating this sh!t up?

I hate to say it, but your processing doesn't help, part of the bottle is still missing because it was so drowned in noise, the details have still been destroyed.

Yes, the noise pattern is gone and it's a smooth image, but you haven't necessarily saved the picture.

The original 6D is a better low light camera.
(Which is not to say that the 6D2 is "bad" or that thousands upon thousands of people wouldn't find it the best camera for their purposes, but in terms of sensor development I don't think there's going to be very much positivity from anyone.)

 

[Jopa]

The original 6D is a better low light camera.

I thought we're talking about DR, not low light performance? The same image would look great if shot at ISO 3200 (5 stops above ISO 100).

 

[Aglet]

Or download the files and process them optimally and see how good or bad it actually is. No, why would we do that when we can spend so much time shouting at each other?

Anyway, I know half of you will pull this to pieces but....



1st shot. So here is a crop of the 100iso shot from the 6D MkII vs the D750 both lifted 5 stops, and it's link.

https://www.dpreview.com/articles/3416153698/canon-eos-6d-mark-ii-dynamic-range

2nd shot. I downloaded the 6D MkII file and processed it close to optimally and this is the result. I took five minutes to do this, given a few hours I could do better and work out profiles for each iso and lift amount that could be saved as presets in LR.

You cannot apply the same 'standard' settings to different cameras and say look at the differences. You have to process each file optimally!

When you do you get very different results. When will people stop eating this sh!t up?

I hate to say it, but your processing doesn't help, part of the bottle is still missing because it was so drowned in noise, the details have still been destroyed.

Yes, the noise pattern is gone and it's a smooth image, but you haven't necessarily saved the picture.

The original 6D is a better low light camera.
(Which is not to say that the 6D2 is "bad" or that thousands upon thousands of people wouldn't find it the best camera for their purposes, but in terms of sensor development I don't think there's going to be very much positivity from anyone.)

I both agree and disagree...
The processing methods he used will provide a clean looking image that many people will still like.
But having seen the other camera's rendering... I'd take the Nikon version as the faint reflection is still there and there's no disembottled cap floating there.

 

[privatebydesign]

Or download the files and process them optimally and see how good or bad it actually is. No, why would we do that when we can spend so much time shouting at each other?

Anyway, I know half of you will pull this to pieces but....



1st shot. So here is a crop of the 100iso shot from the 6D MkII vs the D750 both lifted 5 stops, and it's link.

https://www.dpreview.com/articles/3416153698/canon-eos-6d-mark-ii-dynamic-range

2nd shot. I downloaded the 6D MkII file and processed it close to optimally and this is the result. I took five minutes to do this, given a few hours I could do better and work out profiles for each iso and lift amount that could be saved as presets in LR.

You cannot apply the same 'standard' settings to different cameras and say look at the differences. You have to process each file optimally!

When you do you get very different results. When will people stop eating this sh!t up?

I hate to say it, but your processing doesn't help, part of the bottle is still missing because it was so drowned in noise, the details have still been destroyed.

Yes, the noise pattern is gone and it's a smooth image, but you haven't necessarily saved the picture.

The original 6D is a better low light camera.
(Which is not to say that the 6D2 is "bad" or that thousands upon thousands of people wouldn't find it the best camera for their purposes, but in terms of sensor development I don't think there's going to be very much positivity from anyone.)

You can adjust it however you personally want. My point was, and it hasn't been contested yet, is that it is showing a completely different result to the DPReview crap. Yes I could take more time and make it marginally 'better', yes the Nikon is still better, but not by close to the amount DPReview are implying and certainly enough for me to use on any areas I needed to do a 5 stop lift.

I disagree on you saying all detail has been lost, the writing on the label is comparable.

 

[9VIII]

You can adjust it however you personally want. My point was, and it hasn't been contested yet, is that it is showing a completely different result to the DPReview crap. Yes I could take more time and make it marginally 'better', yes the Nikon is still better, but not by close to the amount DPReview are implying and certainly enough for me to use on any areas I needed to do a 5 stop lift.

I disagree on you saying all detail has been lost, the writing on the label is comparable.

You're correct that the writing on the left looks good, there certainly is still a usable exposure range even with such a noisy base image, but the shaded side of the gold foil on the other bottle is heavily deteriorated.

I still consider the wi-fi utilities worth more than an extra stop of DR, but Canon is going to have a hard time explaining this one.

The original 6D is a better low light camera.

I thought we're talking about DR, not low light performance? The same image would look great if shot at ISO 3200 (5 stops above ISO 100).

A bit of both, I'm probably carrying ideas between threads. With the Nikon D5 a compromise was made in order to boost high ISO performance, and that has traditionally been the advantage to Canon sensors (in the D800 era the 6D was alwas held as the low light king),
logically we should expect Canon to actually make progress in low light performance if they're still taking the same approach. Thus far there is nothing to indicate that being the case though.

 

[edoorn]

I'm as Canon fanboy as it can get, but I would not touch this sensor with a 10 ft pole for wedding and event work when you have the option to pick alternatives that are not at least 5 years behind. Did they just went cheap, picked an older production line to make as much profit as possible?

 

[Pippan]

I'm as Canon fanboy as it can get, but I would not touch this sensor with a 10 ft pole for wedding and event work when you have the option to pick alternatives that are not at least 5 years behind. Did they just went cheap, picked an older production line to make as much profit as possible?

Perhaps if you're doing wedding and event work you should use a camera designed for it. AFAIK the 5D MkIV is just such camera.

 

[SecureGSM]

Right! Would you still hesitate if the price was, say: 2 for $2,500.00 ? I wouldn't

I'm as Canon fanboy as it can get, but I would not touch this sensor with a 10 ft pole for wedding and event work when you have the option to pick alternatives that are not at least 5 years behind. Did they just went cheap, picked an older production line to make as much profit as possible?

 

[StudentOfLight]

So the points I'm trying to make are:

- if you optimize the processing for Canon and an ABC body... does the gap in final IQ at least narrow enough to make it worth the effort?

- I don't even want to spend 5 seconds in post if I can avoid it and I do that by starting with a cleaner file

- Optimized the Nikon is better, but not by enough to worry me for the occasions I need to globally push files by a true 5 stops.

- If you learn your craft and don't stubbornly stay in M mode because you think that's what pro's do you don't need to spend any time in post. I have never had a camera in any auto mode come close to five stops under exposed. The example is entirely artificial but if I needed to save an image I had underexposed by five stops I'd look for a new job knowing the 6D MkII is better than me anyway.....

With Canon's RGB Live View histogram I find it very easy to "pre-chimp" and adjust for my white point (and Kelvin Temp) when I change venues. This makes it relatively easy to get consistent in-camera results in full manual both in terms of exposure and color. #TapLV

 

[snoke]

You can adjust it however you personally want. My point was, and it hasn't been contested yet, is that it is showing a completely different result to the DPReview crap. Yes I could take more time and make it marginally 'better', yes the Nikon is still better, but not by close to the amount DPReview are implying and certainly enough for me to use on any areas I needed to do a 5 stop lift.

If DPR did post work like remove noise then the review become a review of their skill in post and of software to remove the noise and not (only) camera. ISO invariant test almost inappropriate but accept only because it isn't skilled. The review isn't what you can make image look like but how the camera works and what in its files. Post processing to remove noise destroys the detailed evidence of what the camera recorded.

 

[Mikehit]

You can adjust it however you personally want. My point was, and it hasn't been contested yet, is that it is showing a completely different result to the DPReview crap. Yes I could take more time and make it marginally 'better', yes the Nikon is still better, but not by close to the amount DPReview are implying and certainly enough for me to use on any areas I needed to do a 5 stop lift.

If DPR did post work like remove noise then the review become a review of their skill in post and of software to remove the noise and not (only) camera. ISO invariant test almost inappropriate but accept only because it isn't skilled. The review isn't what you can make image look like but how the camera works and what in its files. Post processing to remove noise destroys the detailed evidence of what the camera recorded.

I can see your point, but no-one is ever looking at the data coming off the sensor, they are looking at a processed version of that data so yes, it is about how they make the image look. So how do you define what your standard process should be? One that suits (even unintentionally) Nikon? Or Sony? Or Canon?
What use is the information from DPR if the output on which they base their assessment of the camera bears little relationship to what you or I can produce? There isn't an easy answer to this which is why standardisation is a bitch and why standardised tests can be so polarising.

 

[YuengLinger]

You can adjust it however you personally want. My point was, and it hasn't been contested yet, is that it is showing a completely different result to the DPReview crap. Yes I could take more time and make it marginally 'better', yes the Nikon is still better, but not by close to the amount DPReview are implying and certainly enough for me to use on any areas I needed to do a 5 stop lift.

If DPR did post work like remove noise then the review become a review of their skill in post and of software to remove the noise and not (only) camera. ISO invariant test almost inappropriate but accept only because it isn't skilled. The review isn't what you can make image look like but how the camera works and what in its files. Post processing to remove noise destroys the detailed evidence of what the camera recorded.

Agreed. Showing what could be done to salvage the image was a pretty desperate attempt to apologize for Canon.

Canon's customers have every right to complain about subpar products. Hearing over and over, reflexively, from self-appointed defenders of the faith, that, because Canon is the current market leader our concerns are irrelevant and futile doesn't make the criticism of disappointing products less valid.

 

[Mikehit]

You can adjust it however you personally want. My point was, and it hasn't been contested yet, is that it is showing a completely different result to the DPReview crap. Yes I could take more time and make it marginally 'better', yes the Nikon is still better, but not by close to the amount DPReview are implying and certainly enough for me to use on any areas I needed to do a 5 stop lift.

If DPR did post work like remove noise then the review become a review of their skill in post and of software to remove the noise and not (only) camera. ISO invariant test almost inappropriate but accept only because it isn't skilled. The review isn't what you can make image look like but how the camera works and what in its files. Post processing to remove noise destroys the detailed evidence of what the camera recorded.

Agreed. Showing what could be done to salvage the image was a pretty desperate attempt to apologize for Canon.

Canon's customers have every right to complain about subpar products. Hearing over and over, reflexively, from self-appointed defenders of the faith, that, because Canon is the current market leader our concerns are irrelevant and futile doesn't make the criticism of disappointing products less valid.

I saw PBD comments not as an apologist but pointing out that after a 5-stop push the difference to other marques is far less than the basic DPR image would suggest. I have no problem people saying the sensor is disappointing but if they are going to reject the 6D2 as an option base that decision on what the sensor is actually capable of, not based on some non-optimal standardised process.

 

[Loswr]

Canon's customers have every right to complain about subpar products. Hearing over and over, reflexively, from self-appointed defenders of the faith, that, because Canon is the current market leader our concerns are irrelevant and futile doesn't make the criticism of disappointing products less valid.

Complain all you want, that's perfectly valid. But if you're complaining about a camera that is meeting Canon's sales expectations, then yes, your complaints are futile and irrelevant to Canon. That's reality. It's unfortunate that having reality explained upsets you, but really, that's your own problem.

 

[Hector1970]

I'm not in the market for a 6DII as I have the 5D IV already.
I think a big chunk of the 6DII market is upgrade from APS-C group of photographers.

I think the camera has a number of good improvements over the original 6D.
The frame rate, focus points, and flippy screen are great improvements.

I think Canon made a mistake not using the 5D IV sensor in the 6D II.
Having a high dynamic range is very useful to a generalist shooter.
They will shoot in a lot of high contrast situations.
I'm sure the 6D II sensor is probably not that bad and most people wouldn't notice but its a factor once highlighted takes away from the camera if you were going to invest alot of your hard earned money on it.

If you are not over invested in Canon you may well consider a switch to Nikon / Sony.
Someone like me is highly entangled with Canon and have to hope each generation of the cameras keep improving to maximise the glass that I have (which still remains top class).

The 6DII will be interesting to watch over time as to whether it's a success or not. It will do well initially with Pre-orders from the initial announcement. If the reviews are poor in magazines I'm not sure it will do well in the long run. It's still basically a good camera but might not look so good against it's direct competitors.

 

[hbr]

Check this out:

http://www.imaging-resource.com/news/2017/07/20/canon-6d-mark-ii-first-shots-production-level-lab-samples

 

[Mikehit]

Check this out:

http://www.imaging-resource.com/news/2017/07/20/canon-6d-mark-ii-first-shots-production-level-lab-samples

Interesting.
The 6D2 has more texture in the cloth than both other cameras but seems to have a bit less detail and richness in the dark areas. SO I can see why the 6D2 has been touted as the best possible sensor but it is not on all aspects of the performance.

 

[Luds34]

You can adjust it however you personally want. My point was, and it hasn't been contested yet, is that it is showing a completely different result to the DPReview crap. Yes I could take more time and make it marginally 'better', yes the Nikon is still better, but not by close to the amount DPReview are implying and certainly enough for me to use on any areas I needed to do a 5 stop lift.

If DPR did post work like remove noise then the review become a review of their skill in post and of software to remove the noise and not (only) camera. ISO invariant test almost inappropriate but accept only because it isn't skilled. The review isn't what you can make image look like but how the camera works and what in its files. Post processing to remove noise destroys the detailed evidence of what the camera recorded.

I can see your point, but no-one is ever looking at the data coming off the sensor, they are looking at a processed version of that data so yes, it is about how they make the image look. So how do you define what your standard process should be? One that suits (even unintentionally) Nikon? Or Sony? Or Canon?
What use is the information from DPR if the output on which they base their assessment of the camera bears little relationship to what you or I can produce? There isn't an easy answer to this which is why standardisation is a bitch and why standardised tests can be so polarising.

I agree, it's tough. And heaven forbid one company started cooking the RAWs a little bit and all of a sudden they start looking really good in some of these basic tests.

 

[chrysoberyl]

Dynamic Range turns into Dynamic RAGE

DR = dramatic ranting 8)

Ha ha ha ha. Good one.

Canon really messed up - I just bought a 5DIV instead of a 6DII.

 

[monkey44]

...The majority of buyers are not pros, they are Jon and Jean Public ... they want a camera and lens that comes with no hassles, fits without adapters, and shoots memories...

The majority of pros want a camera and lens that comes with no hassles, fits without adapters and shoots whatever the client needs shot.

Hardly any pros care about the small differences in dynamic range that so many people on this forum obsess over.

Yes, exactly my point taken to the next level. Most of us shoot well enough in the field to pretty much print what we need with minimal post processing - and least that's true for my work. And, most applications for our print in whatever media we need works quite well without tweaking it in areas most viewers will never notice.

I suspect very high end commercial work or gallery work might require a bit more, but in general public and in the majority of pro work, my experience tells me major post processing with all the little tweaks is more a personal perception of what an individual photographer wants to see in his/her work, than what might be required for the production image or print.

Is extra tweaking necessary sometimes to pull the shadow detail, or some other hidden factor? Sure, but in general, when it requires that much additional tweaking, I'll just go to a different image in the series.