06-06-2010, 03:42 PM

Hi Serkan,

[quote name='PuxaVida' date='06 June 2010 - 01:41 PM' timestamp='1275824497' post='265']

Wim, thanks for the very "brief" info ... I understand that the high quality lenses produce way better resolutions than the sensor can.

But when it comes to the IQ produced, do you think the 120 lp/mm resolution of the sensor is prone to diffraction (even with high quality APO lenses), or do we still have a room for it in there? This question is asked ignoring the fact that 7D is not the best body for wide angle & closed aperture shots, but better for tele& bird shots. I'm playing my "theoretical discussion" card and try to get from you what I can ...

Regards,

PS: I'm not sure how I could manage not to post this message as a reply to Wim... but it was... anyway, gong live the new edit mode !...

Serkan

[/quote]

All mediums on which we record images and all optical lenses by definition conform to the laws of optical physics, and therefore suffer from optical diffraction. There is a difference between the way a medium behaves and how a lens behaves, but the result is really similar. It is about how much resolution you can get out of a system, at specific criteria, such as contrast percentages, which is what is basically also what is used for MTFs as published by manufacturers.

The point where diffraction starts to have an effect is defined by the Rayleigh limit. An example of a table of diffraction limits for lenses based on the Rayleigh limit is:

This is relatively simple and straightforward, and based on the formula for the Rayleigh limit, for which often the shortcut formula 1600 / N is used, where N is the numerical aperture. BTW, this table is not as accurate as the numbers may indicate. Round down to the two most significant digits, followed by zeroes if necessary. Note that for most lenses the maximum resolution is obtained about two stops down from wide open, as resolution at larger apertures is diminished or limited by aberrations. Also note that we always have diffraction limits, just that some are higher than we are capable of achieving with photographical lenses.

With recording mediums it gets a little more complicated, because now we have to look at it in a 2D way. Difraction by a more or less rounded aperture, causes a diffraction pattern that in simple terms dims the edges of a contrast transition, and adds a kind of reverberation, a circle as it were, with a slightly higher than background density. Instead of straight lines, one gets a a rounded peak, with steeply sloping sides, and a slight bump up on either side of this peak. This is the perfect situation, and a 2-dimensional representation of this is called the Airy-disk.

The Airy disk basically defines what we can see still as sharp. If we have two objects next to each other, two contrast transitions, we have to be able to clearly see the difference as it were between the two Airy disks that are generated due to diffraction. the size of the Airy disk is determined by the Rayleigh limit, and its radius is equal to 1/(Raleigh limit). In order to be able to see the transition from one to the other, they have to be spaced far enough apart to be distinguished independently, at a high enough difference of contrast. The latter is taken care of by the Rayleigh limit, because its formula takes a 9% contrast into account, therefore also in the Airy disk size. We now see calculations where diffraction based on Airy disk calculations tells us that diffraction effects are visible with the 7D from F/7.1 and smaller apertures.

However, I think I can make things simpler. A sensor can resolve a maximum of 0.5/(pixel spacing) , the Nyquist frequency IOW. I've already expressed that as approximately 120 lp/mm - line pairs take the Nyquist frequency into account; you need two contrasting lines in order to distinguish anything at all. If we now take the Rayleigh limit, 1600/N, and we work out what N is at a Rayleigh limit of 120 lp/mm, we get approximately F/13 (F/13.3). There is another way to calculate th eaperture wher ewe hit the diffraction limit, and that is N = 3.2 * (pixel spacing), where N is aperture and pixel spacing is to be given in Âµm. Calculating it this way gives a result of F/13.7, which is close enough to F/13 anyway, considering the accuracy of pixel spacing (2 digits).

Notice we now have a considerate discrepancy. Where does this come from? Well, for one the criteria are different. As I mentioned, the Rayleigh criterion is based on 9% contrast, and most calculations by lens manufacturers these days are based on lines at 50% contrast. Considering that a lens at F/7.1 manages 225 lp/mm, and that the 7D manages 120 lp/mm, we would expect a maximum system resolution, at 9% contrast, of

1/system resolution = 1/(lens resolution) + 1/(medium or sensor resolution), or approximately 78 lp/mm as system resolution. At F/13, that would be about 60 lp/mm.

So what is this diffraction limit stuff all about then? Essentially really that you are limited by diffraction, not the sensor resolution anymore. The same is true with diffraction limited lenses. When a lens is diffraction limited, it just means that you can't get a higher resolution out of it than the diffraction limit allows you. It is so good that it will resolve the diffraction limit, but due to the laws of optics it just can't resolve more. If this theoretical limit is reached at F/7.1 for the sensor in the 7D, fine. Ideally I would like a sensor to have a diffraction limit as high as possible, because it means I could potentially record the finest detail possible, be it only at large apertures, just like it is the case with lenses.

So, does this mean we can't take good and sharp pictures at smaller apertures? No of course not. It only means that the maximum resolution will be reached at that aperture, but we had to contend with that with our lenses already anyway. What is limiting is the maximum size we can enlarge to, before the picture becomes unsharp to the eye, at certain distances. So, this doesn't stop us from making sharp or good photographs at smaller apertures than F/7.1. The 7D just ups the ante a little, that is all it means. Other cameras just can't resolve as much is all this says.

Does this mean you have to use the best lenses possible in order to use the 7D? No, but if you want to get the last little detail out of this camera, you may want to use the best and sharpest lenses available. Do note that most of the prosumer lenses offered by Canon and other manufacturers, and even the consumer ones, are quite a bit better than most stuff available 30 years ago <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='' />.

Ok, another short reply <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='' />. I think I should not make a habit of this <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='' />.

Kind regards, Wim

[quote name='PuxaVida' date='06 June 2010 - 01:41 PM' timestamp='1275824497' post='265']

Wim, thanks for the very "brief" info ... I understand that the high quality lenses produce way better resolutions than the sensor can.

But when it comes to the IQ produced, do you think the 120 lp/mm resolution of the sensor is prone to diffraction (even with high quality APO lenses), or do we still have a room for it in there? This question is asked ignoring the fact that 7D is not the best body for wide angle & closed aperture shots, but better for tele& bird shots. I'm playing my "theoretical discussion" card and try to get from you what I can ...

Regards,

PS: I'm not sure how I could manage not to post this message as a reply to Wim... but it was... anyway, gong live the new edit mode !...

Serkan

[/quote]

All mediums on which we record images and all optical lenses by definition conform to the laws of optical physics, and therefore suffer from optical diffraction. There is a difference between the way a medium behaves and how a lens behaves, but the result is really similar. It is about how much resolution you can get out of a system, at specific criteria, such as contrast percentages, which is what is basically also what is used for MTFs as published by manufacturers.

The point where diffraction starts to have an effect is defined by the Rayleigh limit. An example of a table of diffraction limits for lenses based on the Rayleigh limit is:

This is relatively simple and straightforward, and based on the formula for the Rayleigh limit, for which often the shortcut formula 1600 / N is used, where N is the numerical aperture. BTW, this table is not as accurate as the numbers may indicate. Round down to the two most significant digits, followed by zeroes if necessary. Note that for most lenses the maximum resolution is obtained about two stops down from wide open, as resolution at larger apertures is diminished or limited by aberrations. Also note that we always have diffraction limits, just that some are higher than we are capable of achieving with photographical lenses.

With recording mediums it gets a little more complicated, because now we have to look at it in a 2D way. Difraction by a more or less rounded aperture, causes a diffraction pattern that in simple terms dims the edges of a contrast transition, and adds a kind of reverberation, a circle as it were, with a slightly higher than background density. Instead of straight lines, one gets a a rounded peak, with steeply sloping sides, and a slight bump up on either side of this peak. This is the perfect situation, and a 2-dimensional representation of this is called the Airy-disk.

The Airy disk basically defines what we can see still as sharp. If we have two objects next to each other, two contrast transitions, we have to be able to clearly see the difference as it were between the two Airy disks that are generated due to diffraction. the size of the Airy disk is determined by the Rayleigh limit, and its radius is equal to 1/(Raleigh limit). In order to be able to see the transition from one to the other, they have to be spaced far enough apart to be distinguished independently, at a high enough difference of contrast. The latter is taken care of by the Rayleigh limit, because its formula takes a 9% contrast into account, therefore also in the Airy disk size. We now see calculations where diffraction based on Airy disk calculations tells us that diffraction effects are visible with the 7D from F/7.1 and smaller apertures.

However, I think I can make things simpler. A sensor can resolve a maximum of 0.5/(pixel spacing) , the Nyquist frequency IOW. I've already expressed that as approximately 120 lp/mm - line pairs take the Nyquist frequency into account; you need two contrasting lines in order to distinguish anything at all. If we now take the Rayleigh limit, 1600/N, and we work out what N is at a Rayleigh limit of 120 lp/mm, we get approximately F/13 (F/13.3). There is another way to calculate th eaperture wher ewe hit the diffraction limit, and that is N = 3.2 * (pixel spacing), where N is aperture and pixel spacing is to be given in Âµm. Calculating it this way gives a result of F/13.7, which is close enough to F/13 anyway, considering the accuracy of pixel spacing (2 digits).

Notice we now have a considerate discrepancy. Where does this come from? Well, for one the criteria are different. As I mentioned, the Rayleigh criterion is based on 9% contrast, and most calculations by lens manufacturers these days are based on lines at 50% contrast. Considering that a lens at F/7.1 manages 225 lp/mm, and that the 7D manages 120 lp/mm, we would expect a maximum system resolution, at 9% contrast, of

1/system resolution = 1/(lens resolution) + 1/(medium or sensor resolution), or approximately 78 lp/mm as system resolution. At F/13, that would be about 60 lp/mm.

So what is this diffraction limit stuff all about then? Essentially really that you are limited by diffraction, not the sensor resolution anymore. The same is true with diffraction limited lenses. When a lens is diffraction limited, it just means that you can't get a higher resolution out of it than the diffraction limit allows you. It is so good that it will resolve the diffraction limit, but due to the laws of optics it just can't resolve more. If this theoretical limit is reached at F/7.1 for the sensor in the 7D, fine. Ideally I would like a sensor to have a diffraction limit as high as possible, because it means I could potentially record the finest detail possible, be it only at large apertures, just like it is the case with lenses.

So, does this mean we can't take good and sharp pictures at smaller apertures? No of course not. It only means that the maximum resolution will be reached at that aperture, but we had to contend with that with our lenses already anyway. What is limiting is the maximum size we can enlarge to, before the picture becomes unsharp to the eye, at certain distances. So, this doesn't stop us from making sharp or good photographs at smaller apertures than F/7.1. The 7D just ups the ante a little, that is all it means. Other cameras just can't resolve as much is all this says.

Does this mean you have to use the best lenses possible in order to use the 7D? No, but if you want to get the last little detail out of this camera, you may want to use the best and sharpest lenses available. Do note that most of the prosumer lenses offered by Canon and other manufacturers, and even the consumer ones, are quite a bit better than most stuff available 30 years ago <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='' />.

Ok, another short reply <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='' />. I think I should not make a habit of this <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 ....