TCs or No TCs for Better Images?

[Arthur Morris]

An interesting question came up in a comment on my recent blog post, "The Central Sensor Crop." You can see the post here. (http://www.birdsasart-blog.com/2010/12/23/the-central-sensor-crop/)

The question: "Which (assuming accurate focus and perfect technique) will yield sharper images, working with a long lens with a teleconverter or working with the prime lens alone and cropping to the same subject size (as when using the TC....)"

Assuming that there is a "correct" scientific answer would it vary at all with different TC (1.4, 1.7, or 2X TC)/lens combinations?

My gut says that using the TCs will yield sharper images than using the prime lens alone and cropping (despite the 14% and 28% approximate loss of sharpness with the 1.4X and the 2X TCs respectively).

[Eric Virkler]

Artie, I wondered the same thing before I bought my TC. I ran tests as soon as I bought it to see if I'd be happy with the sharpness. My tests varied from very near subjects to the distant moon. In my case, the TC yields a sharper image than cropping w/o the TC. I'm not using a prime, my test was with the Canon 100-400 and a 1.4x TC.

Eric
Faces of Nature Photography
www.ericjvirkler.com (http://www.ericjvirkler.com)

[Roger Clark]

An interesting question came up in a comment on my recent blog post, "The Central Sensor Crop." You can see the post here. (http://www.birdsasart-blog.com/2010/12/23/the-central-sensor-crop/)

The question: "Which (assuming accurate focus and perfect technique) will yield sharper images, working with a long lens with a teleconverter or working with the prime lens alone and cropping to the same subject size (as when using the TC....)"

Assuming that there is a "correct" scientific answer would it vary at all with different TC (1.4, 1.7, or 2X TC)/lens combinations?

My gut says that using the TCs will yield sharper images than using the prime lens alone and cropping (despite the 14% and 28% approximate loss of sharpness with the 1.4X and the 2X TCs respectively).


Artie,

First, I'm assuming you mean detail on the subject, rather than pixel to pixel contrast. Adding TCs always degrades the image as you note above, but the increased magnification can give more detail on the subject despite the degradation of image quality in the sensor plane. I think detail on the subject is what photographers care about.

I can try and give the scientific answer: The short answer: it depends.

The (really) long answer:

If the the pixel sampling has already sampled all that the lens is capable of delivering, then adding a TC will not improve detail on the subject. Whether the lens can deliver more detail with a TC (even stacked TCs), depends on the lens aberrations and whether or not the pixel sampling has reached the diffraction limit.

With consumer zoom lenses, the images without a TC are somewhat soft so the benefits are low when adding a TC. Pro level zooms are better and will benefit adding a TC, but the softness will show, so the benefits are better than consumer zoom lenses, and do deliver more detail on a subject with a TC.

Super telephoto lenses have superb image quality and greatly benefit from adding TCs. I've been continuously impressed with my super telephoto lens sharpness as the DSLR pixel sizes have dropped. It seems that with the 500 f/4 and the 300 f/2.8 (the lenses I have), I can stack TCs, even two 2x TCs and still extract more detail (1D IV body).

So this says the supertelephotos are performing extremely well (so the improvement in the new lenses can't be all that much better as they are already stunningly superb optically).

With the fixed focal length telephotos, what reduces image quality (for an already well-corrected lens) is diffraction. I'll post a link to more detailed discussion about this effect below. But what happens as one approaches diffraction limits is reduction in contrast, especially in fine details.

So why aren't the superzoom P&S cameras real competitors to DSLRs with telephoto lenses? After all, the manufacturers claim (in very misleading advertising which also includes outright lies) that they have these huge "equivalent" focal lengths. For example, Canon has an SX30IS which they claim 840 mm equivalent focal length. What they don't say is that the 840 mm focal length only applies to field of view, not that the lens can deliver any detail at the pixel level at maximum zoom. it can't because the tiny lens is diffraction limited. If you average 2 pixels toogether, it would just meed the diffraction limit (0% MTF). and even averaging 3 pixels would not make much difference. Yet you can buy extenders and magnify the diffraction even more.

So what sets DSLR telephotos apart from the little superzoom cameras? The large aperture lenses. The diffraction limit on the 300 f/2.8, 500 f/4, 400 f/2.8, 600 f/4, etc are the huge apertures. The large aperture means that the diffraction limit is well below the pixel sampling in todays DSLRs (even the small pixel 7D). That means there is plenty of contrast and resolution in the lens (assuming other aberrations are well controlled, and they are), such that one benefits from adding TCs.

Here is more information on this subject:
http://www.clarkvision.com/articles/telephoto_reach/
I started the above page as an analysis of could I do as well on my next Tanzanian safari with a 300 mm f/2.8 lens with a 1DIV compared to what I got on previous trips with a 500 mm f/4 lens and older cameras? I've convinced myself that I can, and with the lighter outfit, I can respond faster, including getting lower faster when I need to, and taking the 300 on hikes (previously I left the 500 in the lodge room when hiking).

This page shows the Moon with various lenses, including 500 f/4 plus TCs. I think it is clear that adding TCs gets more detail, and one could take the straight 500 f./4 image and enlarge it to match the TC sizes to see the detail differences better.
http://www.clarkvision.com/articles/moon-test2/
Note, that image detail with the 500 and 2x TC is becoming limited by atmospheric turbulence. Atmospheric turbulence can also be a factor in bird photography too unless the subject is very close.

So the final answer on whether adding TCs helps or not, are: they do if the lens diameter is large enough to deliver the detail (not diffraction limited), the atmospheric distance between you and the subject is steady enough, and if you can keep the camera+lens steady enough at very long focal lengths.

Roger

[Arthur Morris]

Thanks Again Roger. Yes, I was talking about detail on the subject with super-telephoto lenses). So your answer was very helpful and once again confirmed my gut feelings. I was able to follow most of what you wrote.

Is diffraction with super-telephoto lenses an issue even when working wide open? (My crude understanding of diffraction is that it becomes an issue when the lenses are stopped down and that it is never much of an issue at the wide apertures.

[Alan Lillich]

Is diffraction with super-telephoto lenses an issue even when working wide open? (My crude understanding of diffraction is that it becomes an issue when the lenses are stopped down and that it is never much of an issue at the wide apertures.

Artie,

My understanding is that the lens side of diffraction has to do with the absolute size of the aperture and not the aperture number. A 500mm lens at f/4 has a hole that is ten times as big as a 50mm lens at f/4, and thus a lot less aperture. So yes, a super-telephoto has more diffraction when stopped down than wide open, but even when stopped way down has less diffraction than a normal lens at a wide aperture.

I'm pretty sure that I first learned this is from you actually. I think you wrote a post when working on the Hummingbird Guide with Linda Robbins about why f/22 with a 500 was so sharp, no sign of diffraction fuzzing.

So when someone says a certain camera is diffraction limited at f/X they need to also say with what focal length lens.

Alan

[Alan Lillich]

A 500mm lens at f/4 has a hole that is ten times as big as a 50mm lens at f/4, and thus a lot less aperture.

I of course meant "... a lot less diffraction."

[Arthur Morris]

I of course meant "... a lot less diffraction."

Where??? Let me know exactly via e-mail and I will correct it.

[Alan Lillich]

Where??? Let me know exactly via e-mail and I will correct it.

Artie,

The aperture/diffraction mixup was mine in the previous post. There's nothing you need to correct. I did find the aperture size info I mentioned though. It was not a post, but the "Diffraction" section at the bottom of page 18 in the March 2009 update to the Hummingbird guide.

Alan

[Arthur Morris]

Thanks for the clear-a-fication:)

[Arthur Morris]

Artie,

My understanding is that the lens side of diffraction has to do with the absolute size of the aperture and not the aperture number. A 500mm lens at f/4 has a hole that is ten times as big as a 50mm lens at f/4, and thus a lot less aperture. So yes, a super-telephoto has more diffraction when stopped down than wide open, but even when stopped way down has less diffraction than a normal lens at a wide aperture.

I'm pretty sure that I first learned this is from you actually. I think you wrote a post when working on the Hummingbird Guide with Linda Robbins about why f/22 with a 500 was so sharp, no sign of diffraction fuzzing.

So when someone says a certain camera is diffraction limited at f/X they need to also say with what focal length lens.

Alan

Thanks for the telephoto/diffraction explanation Alan!

[Charles Glatzer]

In actual use I find most of the big guns produce much sharper res images out-of-camera at 5.6 - f/8, rather than shooting wide open. When necessary I typically will shoot at 4.5 rather than wide open, as I see a difference stoping down just a third of a stop.

This has been brought up numerous times, but I will mention it again... I find my out-of-camera images are always sharper when stopping down approx one stop from max aperture when using converters.

If i shoot at or beyond a smaller aperture than /11, as I find necessary when using a 2x, diffraction starts to become an issue. Especially, with full frame 1Ds bodies.

As always, the whole thing is a compromise...as Roger mentioned above... proper stabilization becomes a critical component, as does atmospheric conditions as we increase focal length. Longer focal length is not always better... I see quite a few photog's shooting 800mm images of subjects at great distance complaining the images are not as sharp. The more inclement the weather, particulate matter in the air, etc the shorter the focal length I use. Many factors effect apparent image sharpness. Luckily, we can negate quite a few of the problems associated with apparent image sharpness in post-production by increasing contrast, using deconvolution, etc, but this is different from utilizing a lens resolving power at its max capability.

Of course, when possible, I still find it optimum to get closer to the subject and not use converters. Personal safety and animal welfare is always first priority.

I plan to upgrade my lenses when the new big guns come out, and I should receive the new converters prior to heading off to the Falklands on the 29th. I will let you know my findings regarding the new converter sharpness with the current 300 and 500 lenses upon my return home.

Best in your quest and Happy Holidays to all,

Chas

[Roger Clark]

Is diffraction with super-telephoto lenses an issue even when working wide open? (My crude understanding of diffraction is that it becomes an issue when the lenses are stopped down and that it is never much of an issue at the wide apertures.

Artie,
The answer to your question is absolutely, but most people do not realize it! I did some calculations to show the effects. See the attached plot (I will add this to my telephoto reach article on clarkvision). Diffraction is constant in linear size in the focal plane for a given f/ratio.
The gray line on the plot shows where contrast (MTF) has been reduced 50%. All f/4 lenses fall below the 50% contrast loss (in the finest details) on all but the largest pixel cameras, like the Canon 1DII and Nikon D3. Now, in all these lines, I have not accounted for the effects of the blur filter, which only makes things worse.

The point at which there is no additional resolved detail is the Dawes limit. This is where people really see the effects of diffraction, but the loss of contrast is happening well before that.

The 7D hits the Dawes limit around f/6 to f/7 (the blur filter smears things more so you won't notice as much effect until f/8 or a little slower).

The 1D4 hits the Dawes limit around f/9 and again you'll notice the effects aro0und f/11 or a little slower (again due to the blur filter).

Large pixel cameras, like the Nikon D3 and Canon 1DII don't hit the Dawes limit until after f/13 (more like f/16+ before the effects become greater than the blur filter).

At the other end, P&S superzoom cameras hit the Dawes limit around f/2.5, and since at the full zoom end, they are not f/2.5, they have poor resolution on the subject, and not at all like DSLRs with big telephoto lenses.

As DSLR pixel sizes become smaller, we need faster and faster lenses to deliver the contrast from pixel to pixel. Those lenses will be larger, bulkier, heavier, and more expensive. 600 f/2.8 anyone?

Roger

[Arthur Morris]

Artie, The answer to your question is absolutely, but most people do not realize it! I did some calculations to show the effects. See the attached plot (I will add this to my telephoto reach article on clarkvision). Roger

Roger, Thanks for your incredibly detailed explanation. As folks might have guessed, you lost me (at least) thereafter (except for understanding the need for a 600/f/2.8 and an 800 f/4.

Here is the key question for me: can the effects (image degradations) of diffraction be noticed with the human eye?

I was in engineering school (Brooklyn Polytechnic Institute) on a full academic scholarship and I am sure that with a few days of study I could likely master the lexicon and fully grasp the concepts. But... :)

Thanks again.

[Alan Lillich]

Artie,

There are some nice photos showing diffraction in this Luminous Landscape article:
http://www.luminous-landscape.com/tutorials/understanding-series/u-diffraction.shtml
The text says these are unprocessed, so maybe careful processing could mitigate some of what is visible.

Roger,

Now I'm a bit confused also. I've quoted below the explanation about large lenses that I mentioned earlier. (I hope Linda Robbins and Artie consider this non-infringing. It is from Linda's Hummingbird Guide update - a wonderful book available at the BAA store.) In it George Lepp explains that large lenses show diffraction less. It made sense to my 35 years back college physics, recalling that the ratio of opening to wavelength affects diffraction strength. I guess George is among the "most people don't realize it"?

From the Hummingbid Guide March 2009 update:

After seeing a George Lepp Tech Tips article in Outdoor Photographer dealing with unsharpness caused by diffraction when using small and tiny apertures, Artie wrote George as follows:

Hey George, Linda Robbins has written a guide to high speed flash hummingbird photography and taught me to make some great images using six, seven, and now eight flashes. We normally work at apertures between f/16 and f/29, yet the digital files look perfectly sharp with great feather detail there does not seem to be any loss of sharpness due to diffraction. We are using either the 500 f/4L IS or more recently, the 400mm f/4 IS DO lens (which we both love and feel is great for the hummers because of its incredibly fast initial focusing acquisition.) I am puzzled as to why the hummingbird images made at small or tiny apertures look so sharp. Are we missing something? I read your recent Tech Tips piece on diffraction and you mention that diffraction will be evident in prints as small as 81/2 x 12 inches. Does this imply that diffraction will not be evident in the digital files at the native resolution?

George graciously replied:

“You are getting away with the small f/stops on the hummingbirds because of the relatively large lenses and the fact that the f/stops on the big glass aren’t as physically small as they are on a shorter lens like the 24-105mm. Diffraction is caused by the number of light rays that have to be bent to form the image. With a small lens the aperture opening for f/22 is a lot smaller physically than that of the aperture f/22 on the 500mm lens. This is why the old guys with the 4x5 and 8x10 cameras and the big optic on the front of the field camera got away with f/64. Hence the Ansel Adams group thought that f/64 was the ideal f/stop for landscapes. They didn't seediffraction at f/64. Do that with a 35mm based camera and a 50mm lens and the resulting image will be junk. Watch that f/29 with the 400mm, as you are getting close to a problem. Shoot some at f/16 and others at f/29 or f/32 and you might see a difference in sharpness.”

[Don Nelson]

Diffraction occurs at the edge of any iris in the optical system. The effect is due to the bending of the light waves around the edge, and can be computed by taking every point on the wave againt the iris edge. The calculation is very simple using Fourier Optics. The amount of diffraction is correleated only to the diameter of the iris to the wavelength of light, not to the dimensionless f-stop. Larger openings have less diffraction compared to smaller openings, but its always present.

The f-number system is dimensionless. It has no relationship to the wavelength of light. It merely represents the number of iris opening diameters divided into the focal length of the lens (which, for a non-telephoto design can simply be found by focusing on an object an "infinite" distance away, and measuring the distance between the lens nodal point and the film plane.

This distance is quite close to the focal length stamped on the lens (for instance, an 800mm lens may actually be 802mm when measured on an optical bench). For retrofocus designs (think wide angle) and telephoto designs, its a little more complex, so just use the lens focal length). And you can easily figure the diameter of the iris opening for any lens.

(Again, we will simplify for telephoto and wide angle lense -- the actual iris diameter is actually modified slightly to become an effective aperature. You could calculate this based upon the difference in the apparent iris dimesnions when you look through both ends of the lens.).

So quite simply, if I use f8
with a 800mm lens, the iris will be 100mm opening
With a 400mm lens, f8 will be 50mm opening
With a 100mm lens, f8 will be 12.5mm opening
With a 24mm lens, f8 will be 3mm opening
(again, note that this is the effective aperature. For telephoto designs the actual aperature will be physically smaller)

Since the amount of diffraction is determined by the relation of the iris opening to the wavelength of the light:
Diffraction will be considerably worse for comparable enlargements for the 24mm lens compared to the 800mm lens.

(I quite often use this on old unmarked brass lenses missing waterhouse stops -- I can find the focal length by infinitiy focusing, measure this. Then measure the diameter of the lens elements (minimum), divide this into the focal length of the lens to get the f number of the lens (wide open). Knowning these, you can quite easily fabricate waterhouse stops out of darkened brass).


Large format photographers can create their own depth-of-field charts, which are based upon how much the eye can see in an enlargement at a given distance. Do a search for "circle of confusion" online and and you'll find a few depth of field calculators. The popular one that Artie's webpage shows allows you to select only a format, which makes some assumptions about a target print size to arrive at the differences. With regards to those in F64 -- some of the members, such as Edward Weston, didn't enlarge. For those of us using Ultra Large Format (say 7x17 or 12x20 or 20x24 inch film), with contract printing (like Weston) means you don't see any of the diffraction as its too small for the eye to resolve. My 550mm or 1100mm Schneider Fine Art XXL lens can quite easily be used at f256 with no visible diffraction effects at all. I wouldn't be using it more than f45 if I was enlarging an 8x10 negative made with these lenses.

And for telephoto/retrofocus designs such as all of our Canon white lenses, the iris opening diameter is computed as effective. Actual dimensions will be difference, actually somewhat smaller, due to telephoto design.

[Don Lacy]

Roger, Will a higher mega pixel camera like a 7D record less resolution or detail then a smaller mega pixel camera with the same sized sensor like a 40D due to diffraction at smaller apertures using the same lens all else being equal.

[Arthur Morris]

Thanks Don for the detailed comment. I am slightly less lost than I was with Roger's comments yet I have a similar question. With a big IS super-telephoto lens be can the human eye note the differences in diffraction at the wide open aperture as compared say to either f/8 or f/16?

[Charles Glatzer]

Thanks Don for the detailed comment. I am slightly less lost than I was with Roger's comments yet I have a similar question. With a big IS super-telephoto lens be can the human eye note the differences in diffraction at the wide open aperture as compared say to either f/8 or f/16?

I do not think diffraction is a perceptual problem when shooting wide open, but with my 500 and 600 images are rendered sharper when closed down slightly.

Chas

[Arthur Morris]

Thanks Chas. My oft-professed poor eye for fine detail keeps me from seeing those differences. But what I am trying to understand is whether you can ever see the effects of diffraction with big telephotos...

If the answer is no, then why all the charts and formulas? :)

[Roger Clark]

Diffraction occurs at the edge of any iris in the optical system. The effect is due to the bending of the light waves around the edge, and can be computed by taking every point on the wave againt the iris edge. The calculation is very simple using Fourier Optics. The amount of diffraction is correleated only to the diameter of the iris to the wavelength of light, not to the dimensionless f-stop. Larger openings have less diffraction compared to smaller openings, but its always present.

The f-number system is dimensionless. It has no relationship to the wavelength of light. It merely represents the number of iris opening diameters divided into the focal length of the lens (which, for a non-telephoto design can simply be found by focusing on an object an "infinite" distance away, and measuring the distance between the lens nodal point and the film plane.

This distance is quite close to the focal length stamped on the lens (for instance, an 800mm lens may actually be 802mm when measured on an optical bench). For retrofocus designs (think wide angle) and telephoto designs, its a little more complex, so just use the lens focal length). And you can easily figure the diameter of the iris opening for any lens.

(Again, we will simplify for telephoto and wide angle lense -- the actual iris diameter is actually modified slightly to become an effective aperature. You could calculate this based upon the difference in the apparent iris dimesnions when you look through both ends of the lens.).

So quite simply, if I use f8
with a 800mm lens, the iris will be 100mm opening
With a 400mm lens, f8 will be 50mm opening
With a 100mm lens, f8 will be 12.5mm opening
With a 24mm lens, f8 will be 3mm opening
(again, note that this is the effective aperature. For telephoto designs the actual aperature will be physically smaller)

Since the amount of diffraction is determined by the relation of the iris opening to the wavelength of the light:
Diffraction will be considerably worse for comparable enlargements for the 24mm lens compared to the 800mm lens.

(I quite often use this on old unmarked brass lenses missing waterhouse stops -- I can find the focal length by infinitiy focusing, measure this. Then measure the diameter of the lens elements (minimum), divide this into the focal length of the lens to get the f number of the lens (wide open). Knowning these, you can quite easily fabricate waterhouse stops out of darkened brass).


Large format photographers can create their own depth-of-field charts, which are based upon how much the eye can see in an enlargement at a given distance. Do a search for "circle of confusion" online and and you'll find a few depth of field calculators. The popular one that Artie's webpage shows allows you to select only a format, which makes some assumptions about a target print size to arrive at the differences. With regards to those in F64 -- some of the members, such as Edward Weston, didn't enlarge. For those of us using Ultra Large Format (say 7x17 or 12x20 or 20x24 inch film), with contract printing (like Weston) means you don't see any of the diffraction as its too small for the eye to resolve. My 550mm or 1100mm Schneider Fine Art XXL lens can quite easily be used at f256 with no visible diffraction effects at all. I wouldn't be using it more than f45 if I was enlarging an 8x10 negative made with these lenses.

And for telephoto/retrofocus designs such as all of our Canon white lenses, the iris opening diameter is computed as effective. Actual dimensions will be difference, actually somewhat smaller, due to telephoto design.

Don,
You forgot one very important factor: focal length. The angle that the light is bent by diffraction passing the opening is magnified by the focal length. So in the focal plane, the effects of larger aperture are canceled by the corresponding greater focal length such that the physical size of a diffraction spot is constant for all optical systems with the same f/ratio. Thus the diffraction spot diameter is the same size, as a linear measurement in the focal plane, whether the lens is a 600 mm focal length f/4 or a 10 mm f/4, at a given wavelength.

Roger

[Charles Glatzer]

Thanks Chas. My oft-professed poor eye for fine detail keeps me from seeing those differences. But what I am trying to understand is whether you can ever see the effects of diffraction with big telephotos...

If the answer is no, then why all the charts and formulas? :)

I do wonder the same most times. Like you I go with what I see from experience and images taken in the field in real world conditions the past 28 years. I say no...at least I cannot see a visual difference from diffraction when shooting wide open, but I can see it for sure when shooting closed down f/11 and greater. That said, I do find f/5.6 - f/8 provides greater image sharpness with my 500/600.

BTW- Happy Holidays to you and yours,

Chas

[Arthur Morris]

Thanks Chas and ditto. I did not realize that we have both been at this for 28 years :) Actually, August 7, 2011 will be 28 for me, 28 years since I purchased the old Canon 400 f/4.5 FD manual focus lens.

[Roger Clark]

Artie,

There are some nice photos showing diffraction in this Luminous Landscape article:
http://www.luminous-landscape.com/tutorials/understanding-series/u-diffraction.shtml
The text says these are unprocessed, so maybe careful processing could mitigate some of what is visible.

Roger,

Now I'm a bit confused also. I've quoted below the explanation about large lenses that I mentioned earlier. (I hope Linda Robbins and Artie consider this non-infringing. It is from Linda's Hummingbird Guide update - a wonderful book available at the BAA store.) In it George Lepp explains that large lenses show diffraction less. It made sense to my 35 years back college physics, recalling that the ratio of opening to wavelength affects diffraction strength. I guess George is among the "most people don't realize it"?

From the Hummingbid Guide March 2009 update:

After seeing a George Lepp Tech Tips article in Outdoor Photographer dealing with unsharpness caused by diffraction when using small and tiny apertures, Artie wrote George as follows:

Hey George, Linda Robbins has written a guide to high speed flash hummingbird photography and taught me to make some great images using six, seven, and now eight flashes. We normally work at apertures between f/16 and f/29, yet the digital files look perfectly sharp with great feather detail there does not seem to be any loss of sharpness due to diffraction. We are using either the 500 f/4L IS or more recently, the 400mm f/4 IS DO lens (which we both love and feel is great for the hummers because of its incredibly fast initial focusing acquisition.) I am puzzled as to why the hummingbird images made at small or tiny apertures look so sharp. Are we missing something? I read your recent Tech Tips piece on diffraction and you mention that diffraction will be evident in prints as small as 81/2 x 12 inches. Does this imply that diffraction will not be evident in the digital files at the native resolution?

George graciously replied:

“You are getting away with the small f/stops on the hummingbirds because of the relatively large lenses and the fact that the f/stops on the big glass aren’t as physically small as they are on a shorter lens like the 24-105mm. Diffraction is caused by the number of light rays that have to be bent to form the image. With a small lens the aperture opening for f/22 is a lot smaller physically than that of the aperture f/22 on the 500mm lens. This is why the old guys with the 4x5 and 8x10 cameras and the big optic on the front of the field camera got away with f/64. Hence the Ansel Adams group thought that f/64 was the ideal f/stop for landscapes. They didn't seediffraction at f/64. Do that with a 35mm based camera and a 50mm lens and the resulting image will be junk. Watch that f/29 with the 400mm, as you are getting close to a problem. Shoot some at f/16 and others at f/29 or f/32 and you might see a difference in sharpness.”

The simple answer is George is incorrect. All lenses at the same f/ratio show the same diffraction effects because the diffraction spot size is the same linear size in the focal plane.
I'm not sure when the above was written, but if it was a while back, cameras had larger pixels, thus the diffraction had to be worse (slower f/stops) for it to become evident. Try the same experiment today with a Nikon D3 versus a Canon 7D and you will see different results, because the smaller pixels have finer sampling in the focal plane.

So the answer today is more complex because the different size pixels in digital cameras. Larger pixels will show less diffraction effects than cameras with smaller pixels. That is why there are multiple horizontal lines on my plot for different cameras.

Roger

[Don Nelson]

For those wanting an excellent website showing the effects of dffraction effects on pixel size -- see

http://www.cambridgeincolour.com/tutorials/diffraction-photography.htm

check about halfway down -- you can select an f-number and a camera (unfortunately not updated for 7d and Mark iv) to select the fstop and digital camera (sensor). It's a lot more effective than a series of graphs presented in some sites as you can see how the Airey disk sits over a sensor pixel site.

And note that the pixel does not cover the entire area of the selected site. Further some pixels are not square - for instance some Nikon imagers have rectangular shape.

And most camera imagers (exception Foveon sensors of Sigma Cameras) use a Bayer pattern of 2 green and 1 red and 1 blue filter in 4 imaging sites, with interpolation made for actual color. With half the sites taken by green, you can see that diffraction effects will be seen first in green (and in luminousity calculations)

[Roger Clark]

Roger, Thanks for your incredibly detailed explanation. As folks might have guessed, you lost me (at least) thereafter (except for understanding the need for a 600/f/2.8 and an 800 f/4.

Here is the key question for me: can the effects (image degradations) of diffraction be noticed with the human eye?

I was in engineering school (Brooklyn Polytechnic Institute) on a full academic scholarship and I am sure that with a few days of study I could likely master the lexicon and fully grasp the concepts. But... :)

Thanks again.

Artie,
Like many perceptions in life, sometimes our perceptions cloud our view. For example, when we are first learning photography we think our pictures are sharp, including those made hand held at slow shutter speed with a low cost consumer zoom lens. Only after we see what can be made with quality lenses and a tripod do we see our perception was wrong.

People think diffraction makes images look soft. While true in the extreme, the first effect is
loss of contrast. So how much contrast is lost before we notice? If we had something to compare to it would be obvious. For example, photograph some bird feathers at close range so that the finest detail is small (near the pixel level) with a 300 f/2.8 lens (or any f/2.8 lens). Record images at f/2.8 and and slower down to f/22 ot f/32. When examining the images, you'll first notice a loss of contrast. You can also see the loss of contrast on a sharp bright/dark boundary. It takes 2 to 3 pixels to transition from bright to dark on a very good f/2.8 lens, and as you stop down the transition will take more pixels. At some point we judge the loss in contrast as loss of detail.

With all f/4 supertelephotos and all cameras with pixel sizes less than about 6.5 microns (e.g. Canon 5DII, 1DIV, 50D, 60D, 7D, Nikon D300) have all lost at least 50% of the contrast of the real scene at the finest details (over a couple of pixels). This is usually recovered by application of a little unsharp mask. So do we "notice" the loss in f/4 supertelephotos? Not really, but if we had a quality 500 f/2.8 to compare with the 500 f/4, we would see the difference. And as we stop down, or add TCs, the f/ratio gets larger and the contrast loss in fine details is greater. People often think focal length is key to telephoto reach, but aperture is actually as much or more important.

I often find images from a 7D have less contrast in the fine details compared to cameras with larger pixels. The pixel sampling throws another variable into an already complex topic.

Roger

[Don Nelson]

Artie
I'd not seen the TC II + lens MTF's posted anywhere before(not currently on the canon product site, nor do I recall previously seeing them even in the EOS lens book), but note that for the new lenses (400 II and 300II) on the Canon site that TC1.4III + 300II and TC2III + 300II MTF curves are now plotted. (as are the TCIII + 400II MTF curves).

Has anyone seen TCII + any lens MTF curves posted anywhere? Anyone have a copy of they could post for comparision (now if Canon would just post TCII + 300II or 400II MTF's, we could do the comparison)

[Roger Clark]

Roger, Will a higher mega pixel camera like a 7D record less resolution or detail then a smaller mega pixel camera with the same sized sensor like a 40D due to diffraction at smaller apertures using the same lens all else being equal.

Don,
Given the same lens, the sensor with smaller pixels will record more detail on the subject with lower signal-to-noise ratio per pixel. This will be true up to the point where the Dawes limit is reached (0% MTF). So at some point as one closes the f/stop, the smaller pixel detail on the subject will be the same as that recorded by the larger pixel camera.

Roger

[Don Lacy]

Don,
Given the same lens, the sensor with smaller pixels will record more detail on the subject with lower signal-to-noise ratio per pixel. This will be true up to the point where the Dawes limit is reached (0% MTF). So at some point as one closes the f/stop, the smaller pixel detail on the subject will be the same as that recorded by the larger pixel camera.

Roger
Thanks for the reply Roger that was my understanding also.

[Arthur Morris]

Artie,
Like many perceptions in life, sometimes our perceptions cloud our view. For example, when we are first learning photography we think our pictures are sharp, including those made hand held at slow shutter speed with a low cost consumer zoom lens. Only after we see what can be made with quality lenses and a tripod do we see our perception was wrong.

People think diffraction makes images look soft. While true in the extreme, the first effect is
loss of contrast. So how much contrast is lost before we notice? If we had something to compare to it would be obvious. For example, photograph some bird feathers at close range so that the finest detail is small (near the pixel level) with a 300 f/2.8 lens (or any f/2.8 lens). Record images at f/2.8 and and slower down to f/22 ot f/32. When examining the images, you'll first notice a loss of contrast. You can also see the loss of contrast on a sharp bright/dark boundary. It takes 2 to 3 pixels to transition from bright to dark on a very good f/2.8 lens, and as you stop down the transition will take more pixels. At some point we judge the loss in contrast as loss of detail.

With all f/4 supertelephotos and all cameras with pixel sizes less than about 6.5 microns (e.g. Canon 5DII, 1DIV, 50D, 60D, 7D, Nikon D300) have all lost at least 50% of the contrast of the real scene at the finest details (over a couple of pixels). This is usually recovered by application of a little unsharp mask. So do we "notice" the loss in f/4 supertelephotos? Not really, but if we had a quality 500 f/2.8 to compare with the 500 f/4, we would see the difference. And as we stop down, or add TCs, the f/ratio gets larger and the contrast loss in fine details is greater. People often think focal length is key to telephoto reach, but aperture is actually as much or more important.

I often find images from a 7D have less contrast in the fine details compared to cameras with larger pixels. The pixel sampling throws another variable into an already complex topic.

Roger

Thanks again Roger. I am glad that I came back to this as I almost missed your reply. I can actually follow most of your reasoning above (as opposed to some of the other comments :)) I have two additional comments: #1: my oft-professed inability to note fine detail surely comes into play here. #2: I have long had some reservations about 7D images especially when viewed at high magnification. It seems that your last comment may have hit that nail on the head....

[Don Nelson]

Message moved to correct thread!

[Arthur Morris]

Please provide a link Don :)