High-resolution EOS R Camera, Where are you?
- Thread starter Canon Rumors Guy
- Start date
Didn't the EOS 5Ds R eliminate the low pass filter altogether, and put a clear piece of glass in its place? Wasn't that the only difference between it and the EOS 5Ds, which had a low pass filter?
Upvote
0
No clear glass added, just remove the low-pass filter to increase the 'sharpness' and more moire
Upvote
0
The 5DsR actually has 2 low pass filters, with one mounted in reverse to cancel the effect of the other:
Canon Professional eXchange – The Born of EOS 5DS and EOS 5DS R
Canon has been pursuing high-quality imaging by comprehensively examining products based on various assessment criteria — including resolving power, dynamic range, rich gradation, and low noise etc. Canon responds to the demand for high image resolution from Canon users by creating full frame...
www.canon.com.hk
Upvote
0
If you remove the filter, you have to compensate by adding glass or changing the flange distance. Lensrentals has a number of blog posts about the so-called optical stack and how it affects image quality.
You can observe this need by comparing the filtered and non-filtered EF-RF adapters, they have different thicknesses to account for the filter.
Upvote
0
I don't want to be a know-it-all, but I am a physicist, and therefore I know about wave optics. You may already have heard about diffraction limited aperture, "DLA". Since light has both a particle and a wave character (quantum physics), light as a wave is diffracted - sort of "bend around corners" - by the aperture blades in the lens (in fact, even the lenses itself add a bit to DLA, but the sharp aperture blades dominate this effect). That means the more I close those blades, going to higher f-stops, the more the light waves are bend at their edges. As a result, what should a tack sharp, ideally singular, point of an image on the sensor gets blurred to a tiny disc, a so-called Airy disc. Now, if this disc gets bigger than one pixel, image information starts to get lost.
Based on the sampling principles of information theory, DLA is defined as the number given by an Airy disc that exceeds 2 times the pixel pitch of the sensor. For a camera like the R7, which with its pixel pitch would equal an 85 MP FF camera, its DLA limited aperture is already f = 5.2! So if you go to higher f-stops, a growing loss of physical image sharpness creeps in. This is e.g. relevant for classical landscape images with typical f-stops around 16 and higher, to get a maximum sharpness from closest distance to infinity. So, in that case, what a high pixel density camera catches as real physical image information, may not be more than what a low MP camera delivers, depending on the setting - only hidden in fat, blown-up image files. You can of course resharpen your images digitally, but that adds only pseudo-image information (and artifacts) which tricks the eye by enhanced contrasts. You could get the same result by up-sizing images files of a lower MP camera and resharpen those images digitally.
Bryan Carnathan, still running one of the best text based photo review sites in particular for Canon users IMO, included in his in-depth R7 review a table of DLA numbers for all currently relevant EOS cameras:
Canon EOS R7 Review
Is the Canon EOS R7 right for you? Learn all you need to know in The-Digital-Picture.com's review!
www.the-digital-picture.com
There you can see that even the R5 has a DLA of only f = 7.1 already, a 30 MP R or 5D4 comes with f/8.6, and the old 22 MP 5D3 with f/10.1 (https://www.the-digital-picture.com/Reviews/Camera-Specifications.aspx?Camera=792). So for the mentioned classical landscape shooting with high f-stops, a 20 - or even a 12 - MP camera catches all physical image information that is possible with conventional camera optics. That's why I loved to use my old 5D3 for many years, since it delivered great landscape images (IF DR wasn't an issue), in lean image files that do not cram a computer with pseudo-information. Plus, about 20-24 MP on 35 mm sensors is more than sufficient for most settings, including portrait, social shootings, street etc. - and bigger pixels allow principally a higher DR range btw, because they can collect more photons until they are saturated.
Small pixels make sense e.g. if you shoot birds/wildlife with fast superteles in not-so-bright settings, like I frequently do. So, if you shoot wide open, a camera like the R7 really delivers a maximum of image information. Same with macro photography, because there higher f-stops than typically about f/6.3 (depending on the close-up distance) do not visibly enhance the depth of field (DoF) anymore - and you can work with image stacking (a very useful setting option in the R7 btw.) if you want to have more DoF.
Upvote
0
The 5Ds has the same two ‘filters’. Each one is a sheet of lithium niobate crystal, which splits incoming light into two points in a defined direction. In the 5Ds (and other traditional AA filter sensors) the filters are oriented orthogonally, so for example the first filter would shift one point ‘right’ and the second would shift it ‘down’. One point of light has become four.The 5DsR actually has 2 low pass filters, with one mounted in reverse to cancel the effect of the other:
View attachment 213148
Canon Professional eXchange – The Born of EOS 5DS and EOS 5DS R
Canon has been pursuing high-quality imaging by comprehensively examining products based on various assessment criteria — including resolving power, dynamic range, rich gradation, and low noise etc. Canon responds to the demand for high image resolution from Canon users by creating full frame...www.canon.com.hk
In the 5DsR, the filters are oriented at 180°, e.g., the first as ‘right’ the second filter as ‘left’. That keeps the stack the same thickness but cancels the AA effect. One point of light split in two then combined back to one.
Put simply, Canon’s new AA filters use more layers in more orientations, both straight and diagonal. That yields 16 points instead of four.
Upvote
0
Informative post, and it backs up some observations I’ve made in the six years I’ve been using 5DS cameras. That amount of usage hasn’t convinced me about high mp relative to format at all. Whilst I accept the niche uses, mainly severe cropping, for the rest of the time, if you’re shooting at smaller apertures, you’re much better off with increasing the format size rather than increasing the number of pixels.I don't want to be a know-it-all, but I am a physicist, and therefore I know about wave optics. You may already have heard about diffraction limited aperture, "DLA". Since light has both a particle and a wave character (quantum physics), light as a wave is diffracted - sort of "bend around corners" - by the aperture blades in the lens (in fact, even the lenses itself add a bit to DLA, but the sharp aperture blades dominate this effect). That means the more I close those blades, going to higher f-stops, the more the light waves are bend at their edges. As a result, what should a tack sharp, ideally singular, point of an image on the sensor gets blurred to a tiny disc, a so-called Airy disc. Now, if this disc gets bigger than one pixel, image information starts to get lost.
Based on the sampling principles of information theory, DLA is defined as the number given by an Airy disc that exceeds 2 times the pixel pitch of the sensor. For a camera like the R7, which with its pixel pitch would equal an 85 MP FF camera, its DLA limited aperture is already f = 5.2! So if you go to higher f-stops, a growing loss of physical image sharpness creeps in. This is e.g. relevant for classical landscape images with typical f-stops around 16 and higher, to get a maximum sharpness from closest distance to infinity. So, in that case, what a high pixel density camera catches as real physical image information, may not be more than what a low MP camera delivers, depending on the setting - only hidden in fat, blown-up image files. You can of course resharpen your images digitally, but that adds only pseudo-image information (and artifacts) which tricks the eye by enhanced contrasts. You could get the same result by up-sizing images files of a lower MP camera and resharpen those images digitally.
Bryan Carnathan, still running one of the best text based photo review sites in particular for Canon users IMO, included in his in-depth R7 review a table of DLA numbers for all currently relevant EOS cameras:
Canon EOS R7 Review
Is the Canon EOS R7 right for you? Learn all you need to know in The-Digital-Picture.com's review!
There you can see that even the R5 has a DLA of only f = 7.1 already, a 30 MP R or 5D4 comes with f/8.6, and the old 22 MP 5D3 with f/10.1 (https://www.the-digital-picture.com/Reviews/Camera-Specifications.aspx?Camera=792). So for the mentioned classical landscape shooting with high f-stops, a 20 - or even a 12 - MP camera catches all physical image information that is possible with conventional camera optics. That's why I loved to use my old 5D3 for many years, since it delivered great landscape images (IF DR wasn't an issue), in lean image files that do not cram a computer with pseudo-information. Plus, about 20-24 MP on 35 mm sensors is more than sufficient for most settings, including portrait, social shootings, street etc. - and bigger pixels allow principally a higher DR range btw, because they can collect more photons until they are saturated.
Small pixels make sense e.g. if you shoot birds/wildlife with fast superteles in not-so-bright settings, like I frequently do. So, if you shoot wide open, a camera like the R7 really delivers a maximum of image information. Same with macro photography, because there higher f-stops than typically about f/6.3 (depending on the close-up distance) do not visibly enhance the depth of field (DoF) anymore - and you can work with image stacking (a very useful setting option in the R7 btw.) if you want to have more DoF.
I’m also not sure if the Bayer Array principle of defining colour follows the logic of more pixels on target equals more data for colour interpolation, especially in low light or heavy shadow situations.
One thing I am sure about is that my future FF cameras will be considerably less than 50mp, given current tech.
Upvote
0
Thank you very much!I don't want to be a know-it-all, but I am a physicist, and therefore I know about wave optics. You may already have heard about diffraction limited aperture, "DLA". Since light has both a particle and a wave character (quantum physics), light as a wave is diffracted - sort of "bend around corners" - by the aperture blades in the lens (in fact, even the lenses itself add a bit to DLA, but the sharp aperture blades dominate this effect). That means the more I close those blades, going to higher f-stops, the more the light waves are bend at their edges. As a result, what should a tack sharp, ideally singular, point of an image on the sensor gets blurred to a tiny disc, a so-called Airy disc. Now, if this disc gets bigger than one pixel, image information starts to get lost.
Based on the sampling principles of information theory, DLA is defined as the number given by an Airy disc that exceeds 2 times the pixel pitch of the sensor. For a camera like the R7, which with its pixel pitch would equal an 85 MP FF camera, its DLA limited aperture is already f = 5.2! So if you go to higher f-stops, a growing loss of physical image sharpness creeps in. This is e.g. relevant for classical landscape images with typical f-stops around 16 and higher, to get a maximum sharpness from closest distance to infinity. So, in that case, what a high pixel density camera catches as real physical image information, may not be more than what a low MP camera delivers, depending on the setting - only hidden in fat, blown-up image files. You can of course resharpen your images digitally, but that adds only pseudo-image information (and artifacts) which tricks the eye by enhanced contrasts. You could get the same result by up-sizing images files of a lower MP camera and resharpen those images digitally.
Bryan Carnathan, still running one of the best text based photo review sites in particular for Canon users IMO, included in his in-depth R7 review a table of DLA numbers for all currently relevant EOS cameras:
Canon EOS R7 Review
Is the Canon EOS R7 right for you? Learn all you need to know in The-Digital-Picture.com's review!
There you can see that even the R5 has a DLA of only f = 7.1 already, a 30 MP R or 5D4 comes with f/8.6, and the old 22 MP 5D3 with f/10.1 (https://www.the-digital-picture.com/Reviews/Camera-Specifications.aspx?Camera=792). So for the mentioned classical landscape shooting with high f-stops, a 20 - or even a 12 - MP camera catches all physical image information that is possible with conventional camera optics. That's why I loved to use my old 5D3 for many years, since it delivered great landscape images (IF DR wasn't an issue), in lean image files that do not cram a computer with pseudo-information. Plus, about 20-24 MP on 35 mm sensors is more than sufficient for most settings, including portrait, social shootings, street etc. - and bigger pixels allow principally a higher DR range btw, because they can collect more photons until they are saturated.
Small pixels make sense e.g. if you shoot birds/wildlife with fast superteles in not-so-bright settings, like I frequently do. So, if you shoot wide open, a camera like the R7 really delivers a maximum of image information. Same with macro photography, because there higher f-stops than typically about f/6.3 (depending on the close-up distance) do not visibly enhance the depth of field (DoF) anymore - and you can work with image stacking (a very useful setting option in the R7 btw.) if you want to have more DoF.
Now I've finally understood why DLA occurs and what it is about.
And you even succeded in explaining to a total physics dummy!
Upvote
0
Agreed; in Europe a brand new R5 is a little over 3000€, and an used one between 2000€ and 2500€.
An used 5D IV is 800/1000€ tops, so if you're lucky you can buy almost three 5D IV for the price of a good used R5, and almost four for the price of a new one. 5D III, which is still a beautiful camera, can be bought under 500€, that's an even better bargain.
Upvote
0
Great explanation!
For landscape, I haven't used smaller that f13 and preferred F7.1-f10 except when I needed a longer shutter speed without filters and was prepared to accept some softness. That said, if there were closer foreground elements then I would always focus stack. The only issue I found that was when the lens wasn't close to parfocal and the field of field changed from MFD etc.
For 1:1 macro, I have used narrower apertures (f16 or higher) and focus stacked and they seemed to be okay. Would using f6.3 as you suggest just mean more images to be stacked to achieve the same level of sharpness?
Also, how does DLA impact the hyperfocal distance?
Upvote
0
Thanks! But in fact that's old knowledge you can find in good classic film photography textbooks, you only have to think "pixel size" when you read "grain size".
Stacking depends on the area of depth of field (DoF) that you still consider as "sharp" and the distance you want to have sharp by stacking. I never did any lab test with test charts, so I can only say from my own practical experience that with a 100mm macro and a crop camera you don't win much more visibly sharp DoF of you go to much higher f-stops than f/6.3. So I guess you wouldn't need much more images if you close down to 6.3. I'd recommend to try it with a stable test object: take the same number of images with with f/16 and f/6.3 and check out if and how many more intermediate steps you may need to get a thoroughly sharp looking stacked result.
DLA creeps in with higher f-stops, so you expand the hyperfocal distance with higher f-stops but lose sharpness on the pixel level. Finally you have two opposite effects that you have to balance out for a result you are happy with.
Upvote
0
Thank you, I am happy I could help you. Well, I work as a science journalist and edit a physics magazine, so I am used to explain things as simple to understand as possible. Luckily, wave optics is easier to explain than quantum computingThank you very much!
Now I've finally understood why DLA occurs and what it is about.
And you even succeded in explaining to a total physics dummy!
Upvote
0
Thank you. I do hope that Canon will follow its current strategy of offering a low pixel MP camera like the R6 (or the bigger but very expensive pro body cameras) besides higher MP cameras, because many tech geeks want high MP just because of the bigger ... erm ... number
Upvote
0
Speaking of hyperfocal focusing, there's one thing I noticed. Digital cameras are far less tolerant than film. I ruined several pictures relying on the lens barrel indications, the focus area being more limited than it should have been.
Since I was using a brand new lens (Zeiss ZM 2,8/25), I contacted Zeiss for an explanation (was something wrong with the WA?). They said that if I wanted to use for example hyperfocal at f11, I should rely on f5,6 lens barrel indications (2 diaphrams wider) and so on. Properly focused, the Zeiss 25mm is maybe the sharpest lens I have.
Strange is that I never had any issues with original Leica M lenses in similar situations...
Last edited:
Upvote
0
A way around this problem is focus bracketing, but in art, styles change for various reasons and there is no reason more photographers can't use wider apertures for landscapes - an advantage is directing the eye to an area.I don't want to be a know-it-all, but I am a physicist, and therefore I know about wave optics. You may already have heard about diffraction limited aperture, "DLA". Since light has both a particle and a wave character (quantum physics), light as a wave is diffracted - sort of "bend around corners" - by the aperture blades in the lens (in fact, even the lenses itself add a bit to DLA, but the sharp aperture blades dominate this effect). That means the more I close those blades, going to higher f-stops, the more the light waves are bend at their edges. As a result, what should a tack sharp, ideally singular, point of an image on the sensor gets blurred to a tiny disc, a so-called Airy disc. Now, if this disc gets bigger than one pixel, image information starts to get lost.
Based on the sampling principles of information theory, DLA is defined as the number given by an Airy disc that exceeds 2 times the pixel pitch of the sensor. For a camera like the R7, which with its pixel pitch would equal an 85 MP FF camera, its DLA limited aperture is already f = 5.2! So if you go to higher f-stops, a growing loss of physical image sharpness creeps in. This is e.g. relevant for classical landscape images with typical f-stops around 16 and higher, to get a maximum sharpness from closest distance to infinity. So, in that case, what a high pixel density camera catches as real physical image information, may not be more than what a low MP camera delivers, depending on the setting - only hidden in fat, blown-up image files. You can of course resharpen your images digitally, but that adds only pseudo-image information (and artifacts) which tricks the eye by enhanced contrasts. You could get the same result by up-sizing images files of a lower MP camera and resharpen those images digitally.
Bryan Carnathan, still running one of the best text based photo review sites in particular for Canon users IMO, included in his in-depth R7 review a table of DLA numbers for all currently relevant EOS cameras:
Canon EOS R7 Review
Is the Canon EOS R7 right for you? Learn all you need to know in The-Digital-Picture.com's review!
There you can see that even the R5 has a DLA of only f = 7.1 already, a 30 MP R or 5D4 comes with f/8.6, and the old 22 MP 5D3 with f/10.1 (https://www.the-digital-picture.com/Reviews/Camera-Specifications.aspx?Camera=792). So for the mentioned classical landscape shooting with high f-stops, a 20 - or even a 12 - MP camera catches all physical image information that is possible with conventional camera optics. That's why I loved to use my old 5D3 for many years, since it delivered great landscape images (IF DR wasn't an issue), in lean image files that do not cram a computer with pseudo-information. Plus, about 20-24 MP on 35 mm sensors is more than sufficient for most settings, including portrait, social shootings, street etc. - and bigger pixels allow principally a higher DR range btw, because they can collect more photons until they are saturated.
Small pixels make sense e.g. if you shoot birds/wildlife with fast superteles in not-so-bright settings, like I frequently do. So, if you shoot wide open, a camera like the R7 really delivers a maximum of image information. Same with macro photography, because there higher f-stops than typically about f/6.3 (depending on the close-up distance) do not visibly enhance the depth of field (DoF) anymore - and you can work with image stacking (a very useful setting option in the R7 btw.) if you want to have more DoF.
Upvote
0
Similar threads
- Contains 1 staff post(s)