D60's ( APS-C )sensor is too small to stop down below f/11?

line pairs per millimeter

Mike,
I'm amazed this thread is still going. All credit to you. Since I last contributed to this thread, I've acquired a 20D and 5D and taken a few thousand shots with both. The 5D was bought later. If I'd known the 5D was round the corner, I wouldn't have bought the 20D. However, none of us have reliable crystal balls (except psychics of course).

A major issue for me has been whether to sell the 20D. I figured it might provide a worthwhile benefit with my longest telephoto, the 100-400 IS zoom, so I kept it. However, the 100-400 is not a particularly sharp lens and I was doubtful whether or not it would really provide any benefit with a fine-pixel-pitch camera such as the 20D which, according to you, won't serve any purpose beyond f11 or so (let's not get into small fractions of an f stop).

Recently, I found the time to do some rigorous testing using the 100-400 with and without 1.4x extender, switching bodies between the 5D and 20D.

The results were surprising. I found that image quality was very much dependent upon choice of large f stop numbers, larger than one would normally even think of using on a 35mm camera, never mind a sub-35mm camera.

The outcome is, I shall keep the 20D because at f22 with the 560mm lens (400mm + 1.4x extender) it produces a worthwhile improvement in sharpness and definition over the 5D with the same lens at the same aperture. With both bodies, image quality was best at f22 and worst at f57. F8 and f45 seemed to be on a par at the point of focus (with both bodies), but as to be expected, the foreground was significantly sharper at f45. If movement of subject is not an issue and I have a tripod, I would prefer to use f45 than f8 with this lens, with either body.

These tests were carried out using a tripod, remote cord and MLU in case the shutter speed became slow, as it does at f45.

The subject was a brick, tiled-roof house in a rural setting, about 300m away. The resolution of the target varied from brickwork (at that distance, a bit like the printed page) to fine wisps of uncut grass.

The comparison between a 5D image cropped to the same FoV as a 20D image is similar to a D30 image compared to a D60 image, although the difference in pixel pitch in the former is not quite as great.

I won't pretend that my mathematical understanding of the impact of Airy Disks is as great as yours, but I would be interested in hearing your explanation for these results.

Regards, Ray

I await Sven-Goran's explanation

lol im with you ray guess we dont have perfect eyes . some of you people are like my partner you put way to much thinking into something and really it hurts you more than helps you just my 2 cents worth nothing

Photoshooter,
There's nothing wrong with our eyes. If you need glasses, then get them and all will be well. (Excluding serious visual impairment, of course).

Putting too much thinking into something of no practical consequence is counterproductive, I agree. But working out the effects of various apertures on image quality is a 'must' for every serious photographer.

That my 100-400 zoom at 400mm with extender is sharpest at f22, and sharper at this aperture with the 20D rather than the 5D, is something I need and want to know as a photographer.

Having a curious and inquisitive mind, I'm also interested in the reasons.

No explanation? How sad! I guess this thread can now be truly put to rest. I was sort of hoping Mike Davis might publicly confess, 'Hey! I got it wrong'. But it seems it's not to be.

Rayz,

don't want to drag this out too much, but if you want him to answer, you might provide a link with sample images first.

Best regards,
Andy

Andy,
That's a good point. However, the results I've described do not make complete sense even to me (though I'm opposed to Mike Davis's view on this topic). I'm considering a completely new set of tests with even more rigor, to be sure to be sure, you understand.

But, for what it's worth, my test images are on Luminous Landscape.

Here are a couple of links you might find interesting.

http://luminous-landscape.com …x.php?showtopic​=11351&hl=

http://luminous-landscape.com …x.php?showtopic​=11396&hl=

Frankly, short of necessity, shooting at f/22 isn't the ideal situation with any body unless you need the expanded depth-of-field (macro comes to mind, as does a situation where the lens+converter is less-than-ideal). It isn't horrible, as my own tests have shown, but f/16 is noticeably better with most lenses, and f/8-f/11 better still.

F/45 isn't very good on any format from 35 mm down.

Technically, small apertures should be worse (when viewing 100% crops) on images resulting from smaller pixels. Science says so, even if the OP's original conclusions are quite extreme. Situations where increased (and possibly noticeable) diffraction are preferred include macro shooting, situations where extreme DOF is needed, and situations where the lens/teleconverter combination might result in softer images at wider apertures. (for example, I routinely stack teleconverters on my 300/2.8, and find that despite increasing diffraction, f/18-f/22 is best when 3 telecons are stacked).

In other words, sometimes, stopping down gets better results when the effects of soft glass are stronger than the effects of diffraction. YMMV.

Ray,

thank you for the link to the images and discussion on luminous landscape.

My take on all this is the following:

- There are issues with diffraction and using small apertures on the 20D and other digital cameras.
I have seen enough samples (and taken some myself, where the images @ f/22 were clearly a lot worse than those taken @ f/11) to believe this statement.

- Your introduction of a 1.4x converter adds some more variables into the testing methodology that have influence on certain characteristics of the test and offset some of the findings observed in all the other tests. What about AF performance? From my knowledge, AF on the 20D with that combination can only work with using 'the tape trick' and might not yield optimal results.

- For that reason, I personally do not see your tests as a valid falsification of the hypotheses voiced in this thread here. Not unless you are able to replicate your findings (better image quality @ f/22 than f/11) with a good prime or zoom lens without using teleconverters.

- I do see your test as good and valuable for all those who want to use a 100-400 or similar lens with teleconverter and want to get the sharpest overall image where the effects of a soft lens and of diffraction are compounded in the best way.

Best regards,
Andy

Any and Tom,
I have enough experience to realise that f22 on most normally good lenses produces a slightly softer image than at f16 to f8. In this example with the extender, it appears the 560mm lens is best (or at least equal to f16) at f22.

The point that's relevant to this thread is that the smaller pixel-pitched 20D is not at a disadvantage, compared to the 5D, in making the most of whatever resolution is available at f22.

I thought Mike Davis might like to comment on this aspect.

Thinking about this in a larger sense, it isn't just small cameras. All a small camera does is estabilish a different (and tougher) standard on the effects of diffraction.

Diffraction effects occur at all apertures. There, I said it. But the more of the image that is formed by light that passes close to the edge of the diaphragm, the more effect diffraction will have. So, the loss of resolution caused by diffraction continuously grows as you stop down.

On the other hand, optical faults are worst wide open, where more of the fault lens is being used to make the image. The smaller the aperture, the less visible those effects will be.

And apparent sharpness (as opposed to acutance) will increase as you stop down, just because things on either side of the plane of sharp focus will be smeared over a smaller piece of sensor.

So, every lens has a sweet spot between the effects of lens faults and the effects of diffraction. For some lenses, that spot might be 5.6, and for others, it might be 16 or even 22. The point is that the smaller the aperture at the sweet spot, the more optical faults are figuring into lens performance. Because diffraction is not a fault--it's a fact of life.

I can see where a lens at f/22 might be sharper than at f/11. But it's probably a pretty poor lens. Tom's example of a 300 with a couple of stacked teleconverters is an example. Of course, telephoto lenses have less diffraction effect, because the aperture is physically larger at a given f-stop.

Resolution does not sell images, in my opinion. I have seen photographers and non-photographers alike gask with surprise at the "sharpness" of images displayed at 600 pixels wide on a computer screen. I've seen them gasp at the sharpness of portraits where the eyes are pretty sharp, and much sharper than the background, but where they would not stand the test of comparison against an image that is really sharp.

Shoot a little large format, and you'll know what I mean. Rarely did I make an image that made me gasp, even with 4x5 negatives printed at a paltry 11x14.

And I again repeat the story I've told often of how Ansel Adams's Dogwoods and Merced River is noticeably sharper in the reproduction in Yosemite and the Range of Light than in the photographic print I have made by Bob Ross from the negative. The screening process used in the lithography creates edges.

What I want is an image that looks three-dimensional, and draws the viewer in. That requires a smoothness in the tonality, and faithfulness to the texture of what is being photographed (and not the film or sensor). Bigger photosites gather more light, and bigger samples are always more accurate, even if they are not more precise. Bigger sensors collect more light, and present a totality of information greater than smaller sensors, no matter what the size of the photosites. Poor sensors or post-processing may mask it, but it's true just as it was true with film.

When I was in college, I was a dedicated 35mm shooter, using a Canon F-1. Then I started comparing prints made by friends, and it became clear to me that the prints made on 120 roll film had something more than what the 35mm prints had. That "something" was not sharpness--my prints were sharp. There was something else. I wll know why some get drawn into making 11x14 negatives, and one of these days I'm going to spring for portraits made on a Polaroid 20x24 camera.

Format is king, once you clear away any masking effects.

Rick "who hasn't read the older parts of the thread and is probably repeating himself" Denney

Man this is the thread that could span decades!

Rick,
I don't see this as being the case. You are correct that the physical aperture at f22 on a telephoto lens will be wider than the physical aperture at f22 on a shorter lens (Aperture Diameter= Focal length/F Stop), but unfortunately the lens can not selectively enlarge parts of the image we like and exclude other parts we are not so keen on. With a telephoto lens, the whole image is enlarged, including Airy disks. The diffraction spot size at f22 on a 560mm lens might begin its journey very small, but by the time it reaches the sensor, it's the same size as any lens at f22 would produce. Furthermore, the diffractions spot size at f22 and 560mm will be the same size on any camera, from a small P&S 2/3rds format to the largest format field camera that's ever been made. The thing that varies is the 'field of view'.

Okay, I buy that.

Rick "considering it more carefully, but not seeing a change in conclusions" Denney