09-27-2011, 05:38 PM
[quote name='Brightcolours' timestamp='1317022095' post='11857']
Your post does not make sense in two ways. First of all, about my 28mm f1.8. I said it is diffraction limited at f2.8. Which it is. And the figures I quoted were for a longer wave length (e-line), hence the lower diffraction limit number (536lpmm). I did not say it was diffraction limited at f1.8.
I did also not say that diffraction limits are dependent on something else than the aperture... The figures are related to the wavelength, though.
The 2nd part that makes no sense is your f11 figure for Nikkor 1 10 mp. The f-value where diffraction already takes its toll in resolution is demonstrably f8 for a Canon EOS 5D mk II. If we make the pixel pitch higher, we only will see the softening caused by the diffraction more clearly. Simply put: the projected image at f8 is softer than the projected image at f5.6 due to diffraction. And the 5D mk II has enough resolving power to detect this.
Those are the facts... diffraction effects getting to be measurable at f8 already for a 5D mk II with 6.4µm pixel pitch. And lower for 3.2µm pixel pitch.
It is basically talking about two different things then... In relation to Nikon 1 10mp, it will start to suffer from softer results due to diffraction at f4, and if we use smaller f-values, the image will resolve less and less. And with a 5D mk II we see a softer image at f8 than at f5.6 already.
Those are the numbers that matter, for us photographers.
[url="http://web.ncf.ca/bf250/diffraction.html"]http://web.ncf.ca/bf...iffraction.html[/url]
[/quote]
Hi BC,
Well, you are using monochromatic light, while the standard used to determine such things, is the Rayleigh criterion at 0.000512 mm. This results in 571 lp/mm as the diffraction limit. It's really a bit of a moot point, I think, however. And I did get distracted in my discourse, too, yes.
The Nikon V1/J1 will get softer earlier on than the 5D II, yes, it does. I provided the values used for this, the same with the 5D II: those are the values calculated based on the Rayleigh criterion and based on the requirements for being able to distinguish between two adjacent Airy disks on their sensors (Nyquist freq., closeness of Airy disks, etc.). Essentially, the quoted apertures are those where diffraction is equal to or slightly more than the sensor itself can resolve, or rather, is resolved in Raw, at the pixel pitch for those cameras, using standard criteria for being able to resolve two adjacent lines from a diffracted image (which an image always is as I mentioned).
And it is not so much that the sensor will see softened images, it is actually because of the lens / camera (or rather sensor or medium) combination. Furthermore, as I stated, diffraction works at ALL apertures. Just that at larger apertures, diffraction is less than that at smaller ones.
Considering the numbers I provided, it is generally recommended, for better sharpness, to use a 1 stop leeway beween the diffraction limit for a particular lens / camera combination, and the actual diffraction limit for such a system (IOW, use F/14 for the 5D II, and F8 for the Nikon).
As to what is useful for photographers, this is useful to me, not the articles on calculations based on the type of sensel distribution. Why not you may ask? Well, simply because although the image may get softer, and we see diffraction effects, that is always the case. That is the whole thing with diffraction in the first place. My point here really is that effectively, for daily use, we don't notice these effects to a degree where they deteriorate the image, until one reaches values where diffraction based on the Rayleigh criterion becomes the limit. IOW, that is a very practical limit, even for Bayer sensors. There never were, well, let me rephrase that, I never saw, any articles on the diffraction limit of film. However, film grain also has its diffraction limits, just like a sensor, and even more equal because film grain generally is very standard in size for that particular film, plus it has a few layers (colour negative film anyway), where each of the primary colour receptors don't really overlap, due to the density of the silver halide particles, IOW, it also has a kind of pattern, and the primary colours are offset, prior to developing the film. It is more naturally distributed than sensor wells, i.e, not in a rectangular or hexagonal pattern, but there is a limit to the possibilities. IOW, film and sensors are more equal than most people think.
Considering we get way sharper images from a sensor at above 6MP than we ever got from comparable prints from film, already is an indicator IMO for the rather theoretical approach to sensor only based diffraction limits. The Rayleigh criterion was created because it is a standard based on actual practice, and because the Raw image we get is really already a processed image, I don't think we really need to consider and / or wory about these articles.
They may be true, but funnily enough we get higher resolution results from a sensor, even at or near lens diffraction limits, than we did with film. This mostly because, I think, of the way a sensor works, as compared to film - much more an abrupt cut-off of frequencies than film does (amongst others due to the thickness of a film emulsion), and because where we used to get 20-40 lp/mm on a negative (20 for the average amateur, and 40 for the very good ones), we now get close to the maximum from what the sensor can produce. However, this is a completely different subject, maybe something for another topic.
In short, yes, we see diffraction effects at F/4, F/5.6, even at F/1. The question really is how important those are, and from which POV? From my practical experience, I would conclude that I don't care about medium diffraction limits, whether the medium is sensor or film, I do care about lens and system diffraction limits, because at the lens diffraction limit not only do images become distinctly mushy, but the deeper DoF at the (lens) diffraction limit and beyond doesn't provide additional sharpness due to this mushiness. I don't care about medium diffraction limits, because those are inherent problems of the medium, whether film or sensor, which we do not and cannot control directly (ok, a littlel bit, by "development"). It is a given, and we try to work around this. BTW, the figure used in the quoted article, 1.4 X pixel pitch, is for a single line, or single pixel) while I used 3.2 X (standard used for film), for a line pair, which is actually more strict than 1.8X, because it puts the Airy disks further away from each other than the 1.4X value indicates <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='
' />. In short, I get a larger f-stop value, smaller aperture, because I use more strict criteria for distinction between two Airy disks for diffraction limits than the quoted article <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='' />. And yes, this is counter intuitive, as diffraction is anyway.
Finally, I'd like to repeat once more, diffraction occurs at all apertures, just that the effects get much more distinct at smaller apertures than at larger ones. And in the end it is up to the photographer/artist, to use the absence or presence of these effects creatively.
HTH, kind regards, Wim
P.S.: You can do an interesting exercise on the MTF figures provided here on Photozone. Calculate lens lp/mm for the values provided at different apertures (lw/ih), and see what you get. Rather interesting exercise, which may make you worry less about sensor diffraction as well.
Your post does not make sense in two ways. First of all, about my 28mm f1.8. I said it is diffraction limited at f2.8. Which it is. And the figures I quoted were for a longer wave length (e-line), hence the lower diffraction limit number (536lpmm). I did not say it was diffraction limited at f1.8.
I did also not say that diffraction limits are dependent on something else than the aperture... The figures are related to the wavelength, though.
The 2nd part that makes no sense is your f11 figure for Nikkor 1 10 mp. The f-value where diffraction already takes its toll in resolution is demonstrably f8 for a Canon EOS 5D mk II. If we make the pixel pitch higher, we only will see the softening caused by the diffraction more clearly. Simply put: the projected image at f8 is softer than the projected image at f5.6 due to diffraction. And the 5D mk II has enough resolving power to detect this.
Those are the facts... diffraction effects getting to be measurable at f8 already for a 5D mk II with 6.4µm pixel pitch. And lower for 3.2µm pixel pitch.
It is basically talking about two different things then... In relation to Nikon 1 10mp, it will start to suffer from softer results due to diffraction at f4, and if we use smaller f-values, the image will resolve less and less. And with a 5D mk II we see a softer image at f8 than at f5.6 already.
Those are the numbers that matter, for us photographers.
[url="http://web.ncf.ca/bf250/diffraction.html"]http://web.ncf.ca/bf...iffraction.html[/url]
[/quote]
Hi BC,
Well, you are using monochromatic light, while the standard used to determine such things, is the Rayleigh criterion at 0.000512 mm. This results in 571 lp/mm as the diffraction limit. It's really a bit of a moot point, I think, however. And I did get distracted in my discourse, too, yes.
The Nikon V1/J1 will get softer earlier on than the 5D II, yes, it does. I provided the values used for this, the same with the 5D II: those are the values calculated based on the Rayleigh criterion and based on the requirements for being able to distinguish between two adjacent Airy disks on their sensors (Nyquist freq., closeness of Airy disks, etc.). Essentially, the quoted apertures are those where diffraction is equal to or slightly more than the sensor itself can resolve, or rather, is resolved in Raw, at the pixel pitch for those cameras, using standard criteria for being able to resolve two adjacent lines from a diffracted image (which an image always is as I mentioned).
And it is not so much that the sensor will see softened images, it is actually because of the lens / camera (or rather sensor or medium) combination. Furthermore, as I stated, diffraction works at ALL apertures. Just that at larger apertures, diffraction is less than that at smaller ones.
Considering the numbers I provided, it is generally recommended, for better sharpness, to use a 1 stop leeway beween the diffraction limit for a particular lens / camera combination, and the actual diffraction limit for such a system (IOW, use F/14 for the 5D II, and F8 for the Nikon).
As to what is useful for photographers, this is useful to me, not the articles on calculations based on the type of sensel distribution. Why not you may ask? Well, simply because although the image may get softer, and we see diffraction effects, that is always the case. That is the whole thing with diffraction in the first place. My point here really is that effectively, for daily use, we don't notice these effects to a degree where they deteriorate the image, until one reaches values where diffraction based on the Rayleigh criterion becomes the limit. IOW, that is a very practical limit, even for Bayer sensors. There never were, well, let me rephrase that, I never saw, any articles on the diffraction limit of film. However, film grain also has its diffraction limits, just like a sensor, and even more equal because film grain generally is very standard in size for that particular film, plus it has a few layers (colour negative film anyway), where each of the primary colour receptors don't really overlap, due to the density of the silver halide particles, IOW, it also has a kind of pattern, and the primary colours are offset, prior to developing the film. It is more naturally distributed than sensor wells, i.e, not in a rectangular or hexagonal pattern, but there is a limit to the possibilities. IOW, film and sensors are more equal than most people think.
Considering we get way sharper images from a sensor at above 6MP than we ever got from comparable prints from film, already is an indicator IMO for the rather theoretical approach to sensor only based diffraction limits. The Rayleigh criterion was created because it is a standard based on actual practice, and because the Raw image we get is really already a processed image, I don't think we really need to consider and / or wory about these articles.
They may be true, but funnily enough we get higher resolution results from a sensor, even at or near lens diffraction limits, than we did with film. This mostly because, I think, of the way a sensor works, as compared to film - much more an abrupt cut-off of frequencies than film does (amongst others due to the thickness of a film emulsion), and because where we used to get 20-40 lp/mm on a negative (20 for the average amateur, and 40 for the very good ones), we now get close to the maximum from what the sensor can produce. However, this is a completely different subject, maybe something for another topic.
In short, yes, we see diffraction effects at F/4, F/5.6, even at F/1. The question really is how important those are, and from which POV? From my practical experience, I would conclude that I don't care about medium diffraction limits, whether the medium is sensor or film, I do care about lens and system diffraction limits, because at the lens diffraction limit not only do images become distinctly mushy, but the deeper DoF at the (lens) diffraction limit and beyond doesn't provide additional sharpness due to this mushiness. I don't care about medium diffraction limits, because those are inherent problems of the medium, whether film or sensor, which we do not and cannot control directly (ok, a littlel bit, by "development"). It is a given, and we try to work around this. BTW, the figure used in the quoted article, 1.4 X pixel pitch, is for a single line, or single pixel) while I used 3.2 X (standard used for film), for a line pair, which is actually more strict than 1.8X, because it puts the Airy disks further away from each other than the 1.4X value indicates <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='
' />. In short, I get a larger f-stop value, smaller aperture, because I use more strict criteria for distinction between two Airy disks for diffraction limits than the quoted article <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='' />. And yes, this is counter intuitive, as diffraction is anyway.Finally, I'd like to repeat once more, diffraction occurs at all apertures, just that the effects get much more distinct at smaller apertures than at larger ones. And in the end it is up to the photographer/artist, to use the absence or presence of these effects creatively.
HTH, kind regards, Wim
P.S.: You can do an interesting exercise on the MTF figures provided here on Photozone. Calculate lens lp/mm for the values provided at different apertures (lw/ih), and see what you get. Rather interesting exercise, which may make you worry less about sensor diffraction as well.
Gear: Canon EOS R with 3 primes and 2 zooms, 4 EF-R adapters, Canon EOS 5 (analog), 9 Canon EF primes, a lone Canon EF zoom, 2 extenders, 2 converters, tubes; Olympus OM-D 1 Mk II & Pen F with 12 primes, 6 zooms, and 3 Metabones EF-MFT adapters ....