Posts: 6,715
Threads: 236
Joined: Apr 2010
Reputation:
22
[quote name='PuxaVida' timestamp='1288684930' post='3904']
Hi Wim,
the reference for my statements were:
http://www.cambridgeincolour.com/tutorials/digital-camera-sensor-size.htm and mostly:
http://www.cambridgeincolour.com/tutorials/diffraction-photography.htm
The cornerstone of their statement, which compares the size of the airy disk (diffraction pattern) with pixel size of the sensor came logical to me. Because if the pixel size is smaller than the airy disk, then the diffused photons will be collected by the neighbour pixel, which might alter the real image because of the color recognition concept of the bayer filter. Can this be a problem also for the film?
The amount and cause of diffraction is based on the size of the entrance pupil, that's a fact. But beyond that, I'm trying to understand the effects of diffraction on IQ given that sensors might have different pixel sizes and areas. If you think that this is not the place to discuss this, please send me the links you have so that I can read and learn.
PS: Thanks for the correction regarding the macro comparison.
Kind regards,
[/quote]
You seem to mix things up a bit. Photons do not get diffused, they just end up in other areas of the projected image. Which has no influence on how the projected image gets sampled, and which will not change anything about the captured colours.
For every APS-C 1.5x crop camera, diffraction softening effects on the image will be the same. No matter how high the pixel density is. For every APS-C 1.6x crop camera diffraction softening effects on the image will be the same, no matter how high or low the pixel density is. The same goes for full frame 135 format, whether one uses film or electronic sensors.
The smaller the pixel, the more you zoom in to the image when you look at it at "100% pixel size". This just means you magnify the diffraction softening. But the diffraction is the same as with sensors with lower resolutions.
Airy disks just try to explain the nature of diffracted light, and have less to do with pixels. Circles of confusion do not exist, they merely are a number to try and explain the effect of magnifications to the observer. Magnifications by among other things the sensor size, the print size, the viewing distance.
According to me there is only one thing worthwhile to understand about diffraction: For a given sensor size, diffraction softening at a certain f-value is always the same, no matter what pixel density. And when you compare different sized sensors, equivalent f-values will also give equivalent diffraction softening. So use the crop factor to determine the equivalents.
genotypewriter
Unregistered
[quote name='Brightcolours' timestamp='1288706873' post='3918']
According to me there is only one thing worthwhile to understand about diffraction: For a given sensor size, diffraction softening at a certain f-value is always the same, no matter what pixel density.
[/quote]
Hmm... this is starting to sound more and more like the typical "f/2 is f/2 regardless of the sensor size"-type arguments that [url="http://forums.dpreview.com/forums/read.asp?forum=1041&message=36804039"]you often hear in Olympus forums[/url] <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='B)' />
Yes, diffraction is an optical thing... plain and simple. But that's not a complete model of the system... the system also includes the sensor.
And yes, what you said about diffraction effects being the same regardless of the pixel density, for a particular sizes sensor is true... but you forgot to mention one important thing (or maybe you did but it wasn't highlighted enough): reproduction magnification also has to be the same.
What I think the other party is trying to get at is, to obtain the highest possible amount of information (image detail) from a particular sensor, you need to also take in to consideration the size of the airy disks (and by extension the f-number, because they're closely related).
In practice, it's not easy to say which f-number gives the best results for all lenses on a particular sensor. Just that line itself sounds wrong... different lenses behave differently when stopped down. Maybe the supposed optimal f-number for the maximum theoretical resolution might unfortunately also be the f-number where the lens exhibits the highest CA. So, as I said, this is a system... we have to think about all variables as much as we can (including the side of the bed you got off on the day you went out shooting <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/smile.gif' class='bbc_emoticon' alt='  ' /> )
GTW
[quote name='Brightcolours' timestamp='1288706873' post='3918']
According to me there is only one thing worthwhile to understand about diffraction: For a given sensor size, diffraction softening at a certain f-value is always the same, no matter what pixel density.
[/quote]
Of course the amount of softening is the same no matter the pixel density, but how this given amount of diffraction affects your ability to exploit the sensor's resolution is absolutely not the same.
Well, I must say I'm a bit confused, that's absolutely correct... To be more precise let me explain the cases which I have in mind:
1) Two 1,5x cropped sensors with different pixel densities (e.g. 12mp vs. 18mp). Two images are shot at F13 (same lens) and they are printed on a 60x90 paper. Compare the effects of diffraction on both prints.
2) One 1,5x cropped sensor, one FF sensor with the same pixel sizes. Two images are shot at F13 (different DoF, same lens, different subject distance to provide the same FoV). Print on 60x90 and check the effects of diffraction.
3) One 1,5x cropped sensor, one FF sensor (same sensor resolution, FF has bigger pixels). Two images are shot at F13. Same DoF, same lens, different subject distance to provide the same FoV. Print on 60x90 and check the effects of diffraction.
4) One 1,5x cropped sensor, one FF sensor (same sensor resolution, FF has bigger pixels). Two images are shot at F12 (1,5x) and F18 (FF). Same DoF, same lens, different subject distance to provide the same FoV. Print on 60x90 and check the effects of diffraction.
Kind regards,
Serkan
Posts: 1,597
Threads: 22
Joined: Apr 2010
Reputation:
8
Hi Serkan,
I'll keep it short... <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt=' ' />.
[quote name='PuxaVida' timestamp='1288769977' post='3932']
Well, I must say I'm a bit confused, that's absolutely correct... To be more precise let me explain the cases which I have in mind:
1) Two 1,5x cropped sensors with different pixel densities (e.g. 12mp vs. 18mp). Two images are shot at F13 (same lens) and they are printed on a 60x90 paper. Compare the effects of diffraction on both prints.[/quote]
Diffraction effects exactly the same.
Quote:2) One 1,5x cropped sensor, one FF sensor with the same pixel sizes. Two images are shot at F13 (different DoF, same lens, different subject distance to provide the same FoV). Print on 60x90 and check the effects of diffraction.
Less diffraction with the FF prints.
Quote:3) One 1,5x cropped sensor, one FF sensor (same sensor resolution, FF has bigger pixels). Two images are shot at F13. Same DoF, same lens, different subject distance to provide the same FoV. Print on 60x90 and check the effects of diffraction.
Diffraction advantage (less diffraction) to FF.
Quote:4) One 1,5x cropped sensor, one FF sensor (same sensor resolution, FF has bigger pixels). Two images are shot at F12 (1,5x) and F18 (FF). Same DoF, same lens, different subject distance to provide the same FoV. Print on 60x90 and check the effects of diffraction.
You can't multiply apertures by a factor. You first have to work out what the factor means in f-stops.
However, using an f-stop on FF equivalent to the f-stop on APS-C to get the same DoF and the same framing, they will both show the same amount of diffraction.
Quote:Kind regards,
Serkan
Do note, BTW, that diffraction is always present, basically because you are shooting through an aperture. However, this is an exponentially gliding scale, as in, diffraction gets exponentially worse with closing down the aperture, and sensor size, or rather, CoC, plays an important factor. CoC is determined on the magnification factor required to see an image with sharp details, an image of A4 size for ease of reference, at a distance of 30 cm, with a resolution of 6-8 lp/mm, which is apparently the acuity of the human eye at that distance. In order to get that sharpness, you need a CoC of 0.3 mm for FF, and about 0.2 mm for APS-C, which is why with the latter the diffraction hits sooner.
What does it mean that diffraction "hits"? In simple terms, that the effects of diffraction overshadow the gain in DoF you would get from stopping down further. IOW, roughly from F/13 on APS-C and from F/18 on FF you don't get added benefit from stopping down further, because the fuzziness caused by diffraction will hide the additional gain in DoF. However, it also depends on the size of the print. Pritn smaller, and the effective sharpness increases, etc.
Finally, thsi si really only a guide. Having shot a lot of macro over the past 38 years or so, I always found that in a lot of cases extra DoF does trump diffraction. This is caused really by the fact that at macro level we have really nothing to compare with, and that even very low resolutions with smooth transitions look better than nothing at all. And thsi also varies from lens to lens. A lens with extremely good microcontrast, f.e., creating the illusion of sharpness essentially, may actually provide better pictures at F/22 than another lens which doesn't. There are really many factors at play in this.
It is possible to determine absolute values for diffraction by aperture, as they essentially are dependent only on aperture. However, you need to take the magnification into account when looking at prints, and this is where the difficulty, subjectivity sensor size, magnification used, etc., come into play. The more you enlarge, the more you will see the effects of diffraction (note that for a FF sensor you have to magnify 1.5 or 1.6X less, linear, than with APS-C). Thus, the maximum resolution of a lens at a certain diffraction limited lens, when used on FF vs APS-C, will show up diffraction effects sooner on APS-C than on FF, in principle by a straight factor of 1.5 or 1.6X, although that has to be converted to f-stops to make comparisons possible at all (i.e, convert to area first, convert that to f-stops, and add that, or rather subtract that when going from FF to APS-C).
Oops... I said I would keep it short... Better stop now <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='  ' />.
Kind regards, Wim
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 ....
Posts: 6,715
Threads: 236
Joined: Apr 2010
Reputation:
22
[quote name='BG_Home' timestamp='1288766475' post='3927']
Of course the amount of softening is the same no matter the pixel density, but how this given amount of diffraction affects your ability to exploit the sensor's resolution is absolutely not the same.
[/quote]
That is the wrong way to look at it, really.
It is the projected image sharpness that takes a hit. So it is the print size that gets "exploited" or not, totally regardless of sensor resolution.
The only thing sensor resolution can do is: when sensor resolution is too low, it starts to inhibit print size even more than the projected image softness.
It really is that simple.
Posts: 6,715
Threads: 236
Joined: Apr 2010
Reputation:
22
[quote name='genotypewriter' timestamp='1288765828' post='3926']
Hmm... this is starting to sound more and more like the typical "f/2 is f/2 regardless of the sensor size"-type arguments that [url="http://forums.dpreview.com/forums/read.asp?forum=1041&message=36804039"]you often hear in Olympus forums[/url] <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt=' ' />
[/quote]
Oh dear... Olympus forums!!!
One wonders if Olympus uses highly radioactive glass. Only that can be an explanation for the collective madness. <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt=' ' />
[quote name='Brightcolours' timestamp='1288789398' post='3939']
That is the wrong way to look at it, really.
It is the projected image sharpness that takes a hit. So it is the print size that gets "exploited" or not, totally regardless of sensor resolution.
The only thing sensor resolution can do is: when sensor resolution is too low, it starts to inhibit print size even more than the projected image softness.
[/quote]
So, the impact of diffraction is obviously NOT independent of sensor resolution. At a given aperture, for some sensor resolutions, diffraction does not impose additional constraints regarding print size (or whatever), while at other sensor resolutions, it will impose additional constraints beyond the constraints already imposed by the resolution of the sensor.
Posts: 6,715
Threads: 236
Joined: Apr 2010
Reputation:
22
[quote name='BG_Home' timestamp='1288811888' post='3945']
So, the impact of diffraction is obviously NOT independent of sensor resolution. At a given aperture, for some sensor resolutions, diffraction does not impose additional constraints regarding print size (or whatever), while at other sensor resolutions, it will impose additional constraints beyond the constraints already imposed by the resolution of the sensor.
[/quote]
No, that is nonsense.
The diffraction softening may place a constraint on print size, regardless of resolution of the sensor. For ALL same size sensors that is the same. Even with FILM of the same size, the same, regardless of ISO, regardless of film grain.
It may however be that the sensor has such a low resolution that the resolution puts a bigger constraint on print size.
Don't twist it around, as that is nonsensical.
[quote name='Brightcolours' timestamp='1288820462' post='3946']
Don't twist it around, as that is nonsensical.
[/quote]
Ya, and your opinion is gospel, eh?
|
