Go Back   Steve's Digicams Forums > Digital SLR and Interchangeable Lens Cameras > Canon EOS dSLR

Thread Tools Search this Thread
Old Dec 13, 2004, 1:08 PM   #1
Join Date: Aug 2004
Posts: 40

A buddy of mine picked up a EF180 macro lens this weekend. To make a long story short he has found that from f16, f22 and on the sharpness of his images takes a dive. He confirmed this on another friends 20D with 100mm Macro, and he also confirmed the problem exists on all his other lenses (EF 24-70L, EF 17-40L, EF 70-200L F4). He did some test shots of a line grid on paper, the edge transitions get worse and worse the more one stops down.

He called Canon and they had nothing to say but send the camera back.

Any one else see or confirm this problem?
dcl2q is offline   Reply With Quote
Sponsored Links
Old Dec 13, 2004, 10:25 PM   #2
Senior Member
Join Date: Dec 2002
Posts: 5,803

Most lenses (special cases below) get better as you stop them down.. in the beginning. And then at some point they start to get worse. I would expect all the leses you listed as "his other lenses" would be worse at f16 and up vs. f8 or f11. It's how most lenses work.

Now there are two exceptions to this rule.

1) Really expensive lenses. I'm talking things like the 500 f4 and 600 f4 (maybe others?) They start out at their almost best, get only a little bit better as you stop them down then they stay good for a little bit and then they get worse. Most lenses don't do this, but a few do. That is part of what you get for $7,000USD.

2) Macro lenses. A macro lense should still be very good out at f16-f32. That is one of the reasons they are so expensive... they have high f-stops and they work well at them. You need this because of the shallow DOF that you get in macro work.

So there are two situation being described in your post.

The 180macro (a great lens, I've used it and would like to buy it) should have been very good at f16 & f22. So something is clearly wrong and he should send it back.

The other lense you list will get worse when you stop them down that far (if they are even capable of being stopped down that far!) Maybe they shouldn't get as bad as you think. But they will get worse.

Do you (or your friend) know how you read an MTF chart? They are a chart describing the sharpness of a picture produced at a specific fstop. Canon posts the MTF charts for all their lenses on www.canoneos.com. Now, I've been told by smarter people than me that their charts lie and their lenses aren't as good as the charts show. Either way, they still show you what happens to lenses when you stop them down.

eric s is offline   Reply With Quote
Old Dec 13, 2004, 11:56 PM   #3
Join Date: Aug 2004
Posts: 40

Well as it turns out, I remember several years ago when the 10D came out someone posting about defraction limitations with the 10D causing it to be less than ideal at f16 and smaller (actually f13). So I did some searching to revisit what I read. The problem is a Physics one and has to do with what is known as airy disks. One will find a lot written about this in regards to telescopes and microscopes. What happens is that as you stop down a lens the point size of the image (assuming a point light soure) becomes smaller than the size of a focal plane pixel (or film grain) and then the defraction waves around the point are seen by neighboring sensors. In other words a point source seen through a pinhole will have hallows or rings of light around them. This is probable one reason the new 1Dmk2 has a mode where you can reduce the resolution by half, thus increasing "grain" size or area. This will allow for greater DOF at the expense of resolution, but at least the image will look good. This is allrelated to sensor size, dept of field and airy disks or circle of confusion.Do a google on Airy disks for more info. More links below

Here is an excerpt from a thread Note the pixel size of the 20D is around 8.6um.

Lastly, I promised to explain more about why the "4/3"-thing is bound
to fail, so I will try to do so here:

One very real and very definite reason is diffraction, which will not
be eliminated by any "technological progress." Some of you should know
this, others do not, and it merits inclusion in the discussion because
it helps understanding the shortcomings of the "4/3"-system.

Without exception, every optical system, be it a telescope or a camera
lens, has a finite aperture. In camera lenses this can be adjusted,
but only within limits, and there may be a conflict between the
/useful/ aperture and the /possible/ aperture. The finiteness of
aperture means that any optical instrument, in our case a camera lens,
can only collect a part of the multitudinous waves entering it. The
result is an inevitable deviation of light rays when these pass
through the opening (aperture). This is what we mean when we say that
an optical system is (and must be) diffraction limited.

In other words, it is the /diffraction/ of the incoming light which
ultimately limits the resolution.

Since every optical system (here: camera lens) is diffraction limited,
it does not project light from the subject perfectly, but as what we
call a diffraction disk, also known as Airy disk which is a very small
spot surrounded by a fuzzy edge. When two such diffraction disks are
very close, they gradually overlap, making it impossible to separate
them. When that point has come, the physical limit of resolution has
come. As long as they do /not/ overlap, the optical system is said to
satisfy the so-called Raleigh's criterion.

Now, the precise area of this Airy disk can be used to determine the
precise resolution of any given optical instrument, for example a
camera lens, since any subject is resolved only insofar the size of
the Airy disks do not overlap in the final image, destroying resolution.

With digital cameras, the ultimate diffraction limiting factor is the
diffraction at the sensor sites, the individual pixels. No picture can
be sharp and well-defined unless its pixels satisfy Raleigh's
criterion. To find out this there is a formula which first need to be
calculated, and it is as follows:

Diameter Airy disk = N * 0.00135383mm at the visible light spectrum.
This is valid in all photographic lenses, and N is then the relative

Thus, the diameter of the Airy disk at f/4 is:

4 * 0,00135383 = 5.4332 (in micron)

For a photographically maximally sharp result, the individual pixel
size cannot be much smaller in micron than the Airy disk at the sensor
at which the system is to be used, in order for the camera/lens
combination to satisfy Raleigh's criterion.


If we take a look at the pixel size in the "4/3" sensor chip, it can
be maximum 6.8 micron at 5 Mp. That is not too bad, but let us
remember that even the Canon D60/300D/10D have 7.4 micron, and the
Canon 1Ds has a pixel size of 8.8 micron even at 11 Mp. (Or more
precise 8.8x8.8 horozontally/vertically)

As for the "Four Thirds" sensor, its pixel sizes at the following
resolutions are:

At 10 megapixels: 4.4 micron
At 20 megapixels: 3.3 micron
At 30 megapixels: 2.8 micron

Further, since we have the formula N * 0.00135383mm it means that we
can make an accurate list of Airy disk sizes in micron for a number of
common apertures:

f/1.4 = 1.9 micron
f/1.8 = 2.4 micron
f/2 = 2.7 micron
f/2.8 = 3.7 micron
f/4 = 5.4 micron
f/8 = 10.8 micron

Thus we see that even the present six-megapixel DSLR cameras as well
as the top-of-the-line Canon 1Ds are suboptimal when it comes to
diffraction at apertures below f/8. However, there is always some
leeway, and matters do not become serious because these cameras still
have /relatively/ large sensor and pixel sizes, thus the Airy disks do
not overlap so rapidly.

With the "4/3"-sensor size, the picture is not nearly as glorious. The
1Ds has a pixel size of 8.8 micron at 11 megapixels, and the Four
Thirds will have 4.4 micron pixel size at that same resolution. And so
its individual pixels will start to become /smaller/ than the Airy
disk before f/4 is reached. With decreasing pixel size one has
progressively less leeway as the Airy disks become smaller in relation
to the individual pixels, and a "4/3"-system with 10 megapixels will
need to be used at large apertures to deliver optimum results, with
all that implies of reduced possibilities of controlling depth of field.

Also, imagine what kind of optical resolution such lenses will need to
have in order to be useful on a Four Thirds system and at large
apertures all the time. We certainly understand why "4/3"-lenses
cannot be exactly cheap.

On top of this come the lesser dynamic range and much increased
noise levels from the smaller pixels.

In short, the Four Thirds system seems doomed to remain a /high/
price, /low/ performance system.

Per Inge Oestmoen, Norway

More Links of interest:



dcl2q is offline   Reply With Quote
Old Dec 14, 2004, 6:26 PM   #4
Senior Member
Join Date: Dec 2002
Posts: 5,803

I still need to sit down and read what you quoted. What I skimmed looks interesting.

But the fact remains that everything I said was true when there was only film. I'm trying to correct something which you (or your friend) seem to believe. That those lenses you listed (EF 24-70L, EF 17-40L, EF 70-200L F4) should be as good (or better?) at f16 then at something more reasonable like f8.

This is completely incorrect. The size of the photosites might effect it in some way, might even increase the problem or create another one.... but lenses were not sharper that far stoped down even when used with film.

eric s is offline   Reply With Quote
Old Dec 15, 2004, 9:58 AM   #5
Join Date: Aug 2004
Posts: 40

I spent some mort time reading Norman Koren's site, I recommend the tutoral Understanding Image Sharpness. LOTS of good info there. esp in parts 1,1A & 6. Here is a summary of defraction discussion:


This indicates that for the 20D with a pixes sspacing of 6.4um the optimum fstop should be one stop greater than 3.2 * 6.8 = 20.46, so f14.25or bigger is optimum.

I'll try to get my friends test photos to post.

The whole upshot is that one can get more DOF at smaller aperature size, but there will be a softening of the image. Or to quote Norman,

"Now here's the rub. If it weren't for diffraction, you could stop down a lens as much as you needed to get the depth of field you desired. But in the real world you reach a point where diffraction starts degrading the image more than misfocus. There is an optimum aperture that results in the best sharpness over a range of distances. But how to find that optimum isn't exactly common knowledge."

Happy reading
dcl2q is offline   Reply With Quote
Old Dec 15, 2004, 10:04 PM   #6
Senior Member
Join Date: Dec 2003
Posts: 575

That is really interesting stuff! It is always cool to read/understand teh science behind how stuff works.

Gandalf065 is offline   Reply With Quote
Old Dec 16, 2004, 9:15 AM   #7
Join Date: Aug 2004
Posts: 40

YesI agree, ant from an artistic point of view it allows one to know how to get the most out of one's medium.

Happy shooting
dcl2q is offline   Reply With Quote
Old Dec 16, 2004, 10:14 AM   #8
Join Date: Aug 2004
Posts: 40

OK I hope this works. I have photos of a steel ruler taken from f8 to f32 with the EF 180mm Macro lens. Notice how sharpness and contrast decrease as aperature size is decreased beyond the limits of defraction.

looks like one photo per message
Attached Images
dcl2q is offline   Reply With Quote
Old Dec 16, 2004, 10:15 AM   #9
Join Date: Aug 2004
Posts: 40

Attached Images
dcl2q is offline   Reply With Quote
Old Dec 16, 2004, 10:16 AM   #10
Join Date: Aug 2004
Posts: 40

Attached Images
dcl2q is offline   Reply With Quote

Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off

All times are GMT -5. The time now is 1:46 PM.