Canon to release a 100mp EOS R system camera next year [CR2]

Michael Clark

Now we see through a glass, darkly...
Apr 5, 2016
4,722
2,655
You can also quite clearly see, in DPReview's DR comparison tool, that the D850 (higher pixel density) yields better results under a +5 or +6 push than pretty much anything else, even when starting at ISO 100.

Again, one must ask the question, is what DP Review measuring actually DR, or is it more aggressive use of NR prior to analog-to-digital conversion? Just because an image is cleaner after being pushed 5 or 6 stops doesn't necessarily follow that the result is an indication of higher honest DR. (Why that is even important is beyond me - if I couldn't get closer than 5-6 stops to desired exposure when shooting, I'd have tossed all of my cameras in a dumpster a looooong time ago.) It may just be an indication of more aggressive NR. There's a reason many Sony sensors are known as "star eaters".
 
  • Like
Reactions: 1 user
Upvote 0

Michael Clark

Now we see through a glass, darkly...
Apr 5, 2016
4,722
2,655
It wasn't off by a small amount which would be the case with the type of rounding you describe. I wouldn't have mentioned it if it had been. It was off by a much larger factor but only in specific test cases. Now add the cases of extreme print sizes (small or large) where the calculator gives clearly wrong answers across all three fields. That's what leads to my suspicion of a rounding error in the formula, not in the final presentation.

Please cite me a specific test case from Cambridge's flexible DoF calculator where the differences between the sum of the front + rear DoF and the stated total DoF is of by more than a small amount attributable to rounding error. Please include all of the variables entered.

When you say the calculator "gives clearly wrong answers across all three fields" upon what authority do you base that claim? Because it doesn't agree with another DoF calculator? You do realize there are different standards for what is considered "acceptable" blur?

Back in the days when most prime lenses had DoF scales on them (of varying accuracy to, at best one or two significant digits), different manufacturers used different standards for calculating DoF. Many assumed the "standard" 8x10 print was being viewed at a distance of 12" (Imperial system countries) or 20 cm (metric system countries) [there's your first difference that will affect results - 20cm is roughly 9.5", not 12"] by a person with 20/20 vision. Zeiss (or maybe it was Leitz?), for example, assumed the viewer had 20/15 vison and thus had a more stringent standard than those that assumed 20/20. This resulted in the acceptable circle of confusion being 1/1730 of the film's diagonal, rather than 1/1500 of the film's diagonal. For the 135 format, which measures 36x24 mm, 1/1730 gives us an acceptable CoC of 0.025 mm. 1/1500 yields an acceptable CoC of 0.030 mm (actually 0.029 mm, but, hey, it's DoF, which is always at best an estimate).

Modern cameras like the Fuji X-M1 display an estimated DoF in the camera's viewfinder. Yet as many users have noted, the DoF indicated in the viewfinder and the DoF indicated by various DoF calculators are all different for the same lens on the same camera from the same distance using the same aperture. DoF comparisons only have any meaning when they are all based on the same standards and formulae. If DOF calculations are based on the what Bob Atkins referred to in the article you linked as the "simple formula", then we must also acknowledge that it will be far less accurate as subject distances approach the extremes of the hyperfocal distance or unity.
 
Upvote 0

dtaylor

Canon 5Ds
Jul 26, 2011
1,805
1,433
I've read that plenty of times. It does not reveal DxO Mark's actual testing methodology for determining the noise floor that informs how they say, "This sensor has 13.7 EV dynamic range as tested and 14.2 EV when normalized to an 8 MP size (or whatever size it is to which they normalize).

DxO is looking for a SNR of 18%.

Not to mention, as anyone can clearly see, the DoF of the second image is narrower than the first as the edges have much less contrast on the second image than they do in the first.

This is both false and an indication that you might not actually understand what DoF is. Worse, you are still insisting you are right despite two real world examples, any number of online DoF calculators, and countless online tutorials (if you search) talking about the DoF advantages of framing macro with smaller formats (Which is the same as backing up and cropping with a high density FF sensor.) Is the whole world, including the laws of physics in my bathroom, wrong?

I'm done. Either try it yourself, or keep believing that moving back and cropping does not increase DoF.
 
Last edited:
Upvote 0

dtaylor

Canon 5Ds
Jul 26, 2011
1,805
1,433
Again, one must ask the question, is what DP Review measuring actually DR, or is it more aggressive use of NR prior to analog-to-digital conversion?

It's not a measurement of anything, it's the actual files pushed and loaded so you can visually compare them at different pushes (+0 through +6ev). The industry is not using special secret analog NR that's undetectable by PtP in high density sensors but ignoring it in low density sensors.
 
  • Haha
Reactions: 1 user
Upvote 0

justaCanonuser

Grab your camera, go out and shoot!
Feb 12, 2014
1,016
913
Frankfurt, Germany
The highest DR 35mm sensors at DxO and PtP are high pixel density sensors. At DxO the top 9 spots are held by high density sensors.



This has not been my experience at 50mp, and I don't hear 90D or M6 mark II owners complaining at even higher pixel density.



And for a few stops beyond that you'll still see IQ gains.
Well, I am not sure about the secret alchemy behind DxO rankings, they recently had to admit that they gave the 1D-X III a too bad ranking. What you summarize is what I said from another perspective. I said that a high MP camera delivers about the same results like a 20 MP camera in most settings. This doesn't exclude that in some settings with a lot of light and a fast, open lens, a high MP camera certainly can deliver much more detail (e.g. in lab tests with optimum conditions). IMO on-sensor pixel binning would be a smart solution to such sensor designs, so you could switch between higher and lower resolutions on the hardware level.

I should say that I am editor of a German physics magazine, and we recently published a two parts series about smartphone camera technology. One of the authors was an engineer from Zeiss. I just summarized in my posting what I learned from this series about the limitations of sensors with very small pixels - and as a physicist, the core messages didn't surprise me. If you imagine a pixel like a bucket, and light like water, smaller buckets have less capacity until they are filled. That's an easy to understand image for the dynamic range limitations of small pixels, for instance. Basically, those authors state, that high MP smartphone cameras are designed as selling point for marketing.

Now, camera makers go the same way, but of course they follow the demand of a share of the camera market, so it is economically logic. If Canon makes enough customers happy - why not? Fortunately, and wisely, they also offer with the R6 a camera that is with 20 MP close to a very sweet spot of 35mm sensors. I've seen recently lab test results of different new ML FF cameras in a German photozine, and the R6 boasted with quite impressive DR and low light performance.
 
Last edited:
Upvote 0

dtaylor

Canon 5Ds
Jul 26, 2011
1,805
1,433
Well, I am not sure about the secret alchemy behind DxO rankings,

While PtP has a different standard and therefore reports a different final value for "photographic dynamic range", their results confirm DxO's (similar graphs and relative performances). And both result sets line up pretty well with what you can visually inspect in DPReview's comparison tool with RAW converted/pushed samples.

I said that a high MP camera delivers about the same results like a 20 MP camera in most settings.

It delivers better results (more detail and superior sharpness) in most settings. There have to be severe limiting factors at play to bring a higher resolution sensor 'down to' a 20-24mp one. Whether or not the improvements are relevant for a particular purpose (subject, view size, audience) is an entirely separate question. But the improvements are there most of the time.

I should say that I am editor of a German physics magazine, and we recently published a two parts series about smartphone camera technology. One of the authors was an engineer from Zeiss. I just summarized in my posting what I learned from this series about the limitations of sensors with very small pixels - and as a physicist, the core messages didn't surprise me. If you imagine a pixel like a bucket, and light like water, smaller buckets have less capacity until they are filled.

If you are an editor at a German physics magazine then you know the following: observation trumps theories, expectations, stories, narratives, analogies, hopes, dreams, and desires. Observation always trumps those things.

Whatever part FWC has played in the past, whatever part it plays at smartphone scales, we observe that it is not the driving factor of dynamic range in large ILC sensors at this time, water buckets be cursed. Perhaps it will be the driving factor again at some point in the future.

Now, camera makers go the same way, but of course they follow the demand of a share of the camera market, so it is economically logic. If Canon makes enough customers happy - why not? Fortunately, and wisely, they also offer with the R6 a camera that is with 20 MP close to a very sweet spot of 35mm sensors. I've seen recently lab test results of different new ML FF cameras in a German photozine, and the R6 boasted with quite impressive DR and low light performance.

The R6, while good, has observably worse DR and high ISO than the R5, D850, and A7r IV. Now perhaps the lab in question got different results, and perhaps that difference is worth investigating. But keep in mind the results I speak to are replicated 3x and concur.
 
Upvote 0

jolyonralph

Game Boy Camera
CR Pro
Aug 25, 2015
1,423
944
London, UK
www.everyothershot.com
A lot of folks expect they understand all facets of issues that they don't deal with themselves. They think they know all of the answers when they don't even have a clue what half of the questions are.

How many customers have a contract with you that specifically states that you will maintain such copies and provide them if needed for a certain period of time? One year? Three years? Five years? Ten years?

All of this is standard stuff. You build all of this time and expense into your operating costs. And, as we've said before right at the beginning, if the big files don't work for you, then don't use it.

I have two backups of important images on two separate raid arrays, with the critical images (processed, 'keepers') on two different cloud systems.

For me the biggest cost and expense is sorting out which images to keep and which to throw. The cost equation on this is generally such now that It's simpler to keep almost everything.

Of course there will be people who will find an APS-C body better suits their requirements than using a FF camera with crop. But all I'm saying is over time that group of people will be smaller and smaller until a point sometime in the future (we're not there yet) that it's not economically viable for Canon to produce them any more.

I mean, there are still people who claim they're never going to move away from a mirror.
 
  • Like
Reactions: 1 user
Upvote 0
Jan 29, 2011
10,675
6,121
I have to be honest I was intrigued as to what does actually happen at macro distances for dof, framing, perspective etc etc and the differences between crop cameras backed up and ff cameras at 1:1 in actual real life use. As anybody that has a passing interest in macro knows the dof calculators break down at close focusing distances, further, subject to sensor distances do not follow 'the rules' because the focal length of most macro lenses changes a lot at such short distances.

I used a 1DX II and a Canon 100mm L Macro for all the shots, for the 'crop' camera shots I very carefully measured and or cropped to get correct framing etc for the relevant comparison. I used 10x Live View manual focus, no IS, sturdy tripod, 1/200 second ambient (zero light in the exposure) and a single flash bounced off the ceiling in manual mode.

So, common wisdom, take a picture with a ff camera, move back 1.6 times and then use a crop camera, the dof should be 1.6 times the amount, the perspective will be different for three dimensional subjects but the framing should be the same.

Here is a FF shot at 1:1, with the setup I have that is basically 300mm from sensor plane to plane of focus.

FF1.jpg


So the 'theory' is if I move to 300 x 1.6 = 480mm and use a crop camera (or crop a ff camera it is exactly the same) I should get identical framing. This is what I actually got after the crop and enlargement.

Crop-1.jpg

Framing and enlargement are nothing like the same!

So I then wondered how far do I have to move back to get the same framing? As I hadn't set up for that I changed my arrangement so I could very accurately measure the distance I moved the camera back.

Here is the second FF shot.
FF2.jpg

And here is the crop framed shot to get the same framing, I had to move the camera back only 51mm!


Crop2.jpg

Here is the plane of focus from the FF camera at 100%
1617851117392.png

Here is the plane of focus from the 'crop' camera at the same magnification.

1617851198623.png

Personally I do not see a real world dof advantage in these images for the crop camera. I also now know that if I had a crop camera I'd only have 51mm of working distance advantage to get the same framing as the ff camera at 1:1, not the 160mm 'the rules' would imply I should have.
 

Attachments

  • FF2.jpg
    FF2.jpg
    81.5 KB · Views: 118
Last edited:
  • Like
  • Love
Reactions: 8 users
Upvote 0

dtaylor

Canon 5Ds
Jul 26, 2011
1,805
1,433
Personally I do not see a real world dof advantage in these images for the crop camera. I also now know that if I had a crop camera I'd only have 51mm of working distance advantage to get the same framing as the ff camera at 1:1, not the 180mm 'the rules' would imply I should have.

I should have noted when I posted my samples that A) The impact seems less than 1.6 (but certainly not equal, i.e. there is a DoF gain), and B) I'm not sure about the practical advantage. I suppose if you're stacking it might reduce the number of frames you have to take? Hence the comment that started this side discussion.
 
  • Haha
Reactions: 1 user
Upvote 0
Jan 29, 2011
10,675
6,121
I should have noted when I posted my samples that A) The impact seems less than 1.6 (but certainly not equal, i.e. there is a DoF gain), and B) I'm not sure about the practical advantage. I suppose if you're stacking it might reduce the number of frames you have to take? Hence the comment that started this side discussion.
In your examples I understood you moved back 1.6 times for the crop camera shots, and then just enlarged the resulting image to match the magnification of the ff camera.

My methodology demonstrated that is not a valid way to get accurate results considering the original premise was to move back with a crop camera to get the same framing as a ff camera and hypothesized you’d get 1.6 times the dof. The 1.6x focus distance is implied by dof calculators that break down at macro distances. Your methodology results in exaggerated differences in the crop cameras favor.

For it to be a valid comparison you need to get the framing the same and in that case the crop camera is not moved back anywhere near 1.6 times, in my case just 28% of that! Or 1.17 x focus distance. This would imply differences in dof to be significantly less than suggested by calculators and that is very much in line with my empirical results.
 
  • Love
Reactions: 1 user
Upvote 0
In your examples I understood you moved back 1.6 times for the crop camera shots, and then just enlarged the resulting image to match the magnification of the ff camera.

My methodology demonstrated that is not a valid way to get accurate results considering the original premise was to move back with a crop camera to get the same framing as a ff camera and hypothesized you’d get 1.6 times the dof. The 1.6x focus distance is implied by dof calculators that break down at macro distances. Your methodology results in exaggerated differences in the crop cameras favor.

For it to be a valid comparison you need to get the framing the same and in that case the crop camera is not moved back anywhere near 1.6 times, in my case just 28% of that! Or 1.17 x focus distance. This would imply differences in dof to be significantly less than suggested by calculators and that is very much in line with my empirical results.
Thanks for your tests, interesting stuff. Some questions: Would focus breathing of the chosen lens be influencing your results? Perhaps why you didn’t need to move back very far? Would the result be different with different lenses?
 
Upvote 0
Jan 29, 2011
10,675
6,121
Thanks for your tests, interesting stuff. Some questions: Would focus breathing of the chosen lens be influencing your results? Perhaps why you didn’t need to move back very far? Would the result be different with different lenses?
Oh there is definitely focus breathing and lens breathing going on, but there is also the fact that the common dof calculations (and therefore calculators) break down at close focus distances. Yes different lenses will have different results.

This isn’t new, the main issue is that at normal focus distances the offset of the lens to the sensor plane is relatively insignificant when compared to the lens to subject distance. But at macro distances the lens to sensor distance can be a significant percentage of the lens to subject distance, indeed it is often greater than the lens to subject distance, this causes the more basic dof calculations to fail.

Another issue is the calculators consider the lens to be a simple lens, that is a single element at the focal distance. Again at normal focus distances the differences between that and the actual nature of a complicated focusing compound lens are insignificant, but at macro distances those differences can be significant percentages of the calculation.

All this to say it is well known that dof calculators and equivalence theories break down at macro distances because of known limitations to the simple calculations. Because I had the gear and time I was just able to actually measure that difference to get a decent comparison.

I wasn’t surprised the example illustrated the problem, which is why I made my initial comment of ‘anybody that thinks moving a crop camera back to get the same framing will give them 1.6 times dof’. I was surprised at the minimal difference in distance you have to move the crop camera back to get that same framing though.
 
Last edited:
  • Like
Reactions: 1 users
Upvote 0

Michael Clark

Now we see through a glass, darkly...
Apr 5, 2016
4,722
2,655
All of this is standard stuff. You build all of this time and expense into your operating costs. And, as we've said before right at the beginning, if the big files don't work for you, then don't use it.

I have two backups of important images on two separate raid arrays, with the critical images (processed, 'keepers') on two different cloud systems.

For me the biggest cost and expense is sorting out which images to keep and which to throw. The cost equation on this is generally such now that It's simpler to keep almost everything.

Of course there will be people who will find an APS-C body better suits their requirements than using a FF camera with crop. But all I'm saying is over time that group of people will be smaller and smaller until a point sometime in the future (we're not there yet) that it's not economically viable for Canon to produce them any more.

I mean, there are still people who claim they're never going to move away from a mirror.

Of course you do. I certainly understand that. But the one to whom I was replying, who keeps insisting that file sizes don't matter because "storage is cheap" doesn't seem to have a clue that all of the other even exists.
 
Upvote 0

Michael Clark

Now we see through a glass, darkly...
Apr 5, 2016
4,722
2,655
Oh there is definitely focus breathing and lens breathing going on, but there is also the fact that the common dof calculations (and therefore calculators) break down at close focus distances. Yes different lenses will have different results.

This isn’t new, the main issue is that at normal focus distances the offset of the lens to the sensor plane is relatively insignificant when compared to the lens to subject distance. But at macro distances the lens to sensor distance can be a significant percentage of the lens to subject distance, indeed it is often greater than the lens to subject distance, this causes the more basic dof calculations to fail.

Another issue is the calculators consider the lens to be a simple lens, that is a single element at the focal distance. Again at normal focus distances the differences between that and the actual nature of a complicated focusing compound lens are insignificant, but at macro distances those differences can be significant percentages of the calculation.

All this to say it is well known that dof calculators and equivalence theories break down at macro distances because of known limitations to the simple calculations. Because I had the gear and time I was just able to actually measure that difference to get a decent comparison.

I wasn’t surprised the example illustrated the problem, which is why I made my initial comment of ‘anybody that thinks moving a crop camera back to get the same framing will give them 1.6 times dof’. I was surprised at the minimal difference in distance you have to move the crop camera back to get that same framing though.

With a simple lens, at unity (1:1 reproduction ratio or 1.0X MM) the lens will be 2X the focal length from the imaging plane and 2X the focal length from the subject. In other words, the lens will be exactly halfway between the subject and the imaging plane. The subject will be 4X the lens' focal length, measured when focused to infinity, from the film/sensor plane.

So yes, there is considerable breathing going on.
 
Upvote 0

Michael Clark

Now we see through a glass, darkly...
Apr 5, 2016
4,722
2,655
I should have noted when I posted my samples that A) The impact seems less than 1.6 (but certainly not equal, i.e. there is a DoF gain), and B) I'm not sure about the practical advantage. I suppose if you're stacking it might reduce the number of frames you have to take? Hence the comment that started this side discussion.

Look at the correctly produced examples with the same reproduction ratio in both. There is no difference. None.

With the same angle of view and the same final magnification ratio DoF will also be the same.

If actual subject size is AS and displayed subject size is DS, and if the final ratio of AS: DS is the same for both images, DoF will also be the same.
 
Upvote 0