DoF question

ty_mickan · March 9, 2012

Diffraction differs for every lens/focal length. Keep your lens size to around 4mm (give or take) , and you can't go wrong.

This is the simple answer. As G Dan alluded to, this is an involving subject that really goes beyond the scope of the

thread. Mertlinger is a good read.

rubo · March 9, 2012

John,

I remember when i just started learning photography in the 80's with an old russian Zenith SLR and an old russian Leica knock off, the lenses had a lot more detailed information on them.

But on the lenses i have now the info is pretty basic, so if i need a critical focus i just go to Live View @ 10x magnification :-)

Ty,

You are correct, but it's irelevant to what i'm talking about.

It does not matter what kind of lens you put on 60D, past f/11 the image is losing sharpnes. Once you get past f/16 it's already too soft for my taste.

I have tested this with Canon 10-22mm, 15-85mm, 24-105mm L, 28-135mm, Sigma 30mm, Sigma 150mm Macro, Sigma 70-200. With all of them the smaller the aperture (past f/11) the softer the resulting image.

g dan mitchell · March 9, 2012

Michael Young wrote:

"In any case, I'll stand by my earlier statements, and brush off GDan's bland accusations of ignorance in everyone but himself. Diffraction effectively limits the detail that can be captured. If the diffraction limits the detail that can be captured on a 12 MP camera to equivalently 6 MP, the usable print size is effectively the same as that captured by a 6 MP camera. This is the diffraction limiting that he denies exists."

</blockquote>

First, about "everyone buy himself," you might notice the string of posts agreeing with my points in this thread. If anyone is largely agreeing only with himself, I'm afraid that it is you, not me.

As for the rest of your point, wow. Hard to know where to begin. Yes, diffraction blur limits the amount of detail that can be capture - but it does so in precisely the same way regardless of sensor photo site density (or size of film grain, for that matter).

Nowhere do I "deny" that something called "diffraction limit" exists. Let me state, yet again, that diffraction blur exists, it increases as you stop down, photographers who shoot at small apertures and make big prints should think about it, etc. However, your notion that it must be worse on a higher MP count sensor is just not born out by the facts.

Your point about print size has nothing to do with the difference in the amount of diffraction blur in your hypothetical 6MP and 12MP cameras. When you suggest that we will see more diffraction if we print that 12MP image larger than we would be interested in printing the 6MP image, you are making a point about print size limitations based on pixel size, not diffraction. If you did print that 6MP image at the same larger size you would have precisely the same amount of diffraction in the 12MP and 6MP based prints. (You might also see the "jaggies," but that isn't diffraction blur.)

You are close to getting an important concept related to large print sizes, though you are still fixated incorrectly on the diffraction blur issue and are erroneously connecting that to photo site density. The truth that you are close to is that if you attempt to make very large prints from very small originals, you must pay very careful attention to all factors that affect resolution: careful focus, lens quality, aperture choice, camera stability, careful and skillful sharpening, and so on.

But, no, there is not more diffraction blur with higher photo site density. And no amount of posting will change that. The reason I'm being so direct about this issue is that there is unfortunately a common and widespread misconception about this that is only encouraged when incorrect information about this issue is shared without challenge.

There are plenty of fine reasons to choose a lower MP sensor camera, but avoidance of diffraction blur is not one of them.

Take care,

Dan

g dan mitchell · March 9, 2012

By the way, in terms of general guidelines...

If you shoot a cropped sensor camera and intend to push the upper boundaries of print size, I would be extremely cautious about stopping down beyond about f/8, and in many circumstances I might shoot at something more like f/5.6 depending upon the lens I was using and the nature of the subject.

If you shoot a full frame sensor camera, there is nothing at all to worry about when shooting at f/11 and in my experience producing relatively large prints, shooting at f/16 will produce so little diffraction blur that you can essentially disregard it. If you don't need f/16, you will probably shoot at a larger aperture, but if your shot requires f/16 you can use this aperture without concern. In my experience, f/22 is a different issue. I rarely shoot at f/22, though I will occasionally do so if a longer shutter speed is more important than maximum resolution.

About DOF, I'm not a big fan of DOF calculators and the like. They only offer a subjective estimate of the depth of field and are based on a whole range of assumptions that may not match up to your purposes. I deal with it in several ways in actual shooting:

<ul>

<li>When I use live view - which is probably the majority of the time in my work - I zoom in to 10x magnification, press the DOF preview button and scan around the screen to check sharpness directly.</li>

<li>When working quickly, I tend to stop down a bit more than I might think I need to - a stop or two perhaps - since I'd rather make that error when keeping objects at varying distances in focus is important.</li>

<li>I keep in mind that the "amount of DOF" needed depends a lot on intended output. If you are only going to share online jpg images or print small, you can probably use a larger aperture more in line with the traditional DOF calculations. On the other hand, if you are going to print large, you will likely find that you want to stop down just a bit more if keeping objects at different distances in sharp focus is important.</li>

</ul>

Virtually everything in photography involves some sort of compromise - the task is often not selection the perfect, but to select the best compromise. When it comes to aperture selection, no single parameter necessarily determines the choice. I may want more (or less!) depth of field than I would get at the supposed "sharpest aperture," so I'll sacrifice some theoretical sharpness in order to produce a photograph that looks better. This is also a topic about which much, much more could be said... but I'll stop here.

Dan

scott_ferris · March 9, 2012

Seconded.

Now on the subject of Merklinger, he was very dismissive of the hyper focal focusing technique, in summary he said using it makes sure nothing is sharp, a bit over simplified but if you read his various papers it makes sense.

In practice his suggestion is to focus on what is important, your subject, seems obvious but if that subject is the distant mountain then focus at infinity, don't focus on some mathematical suggestion of where optimum should be, that will make sure your mountains are not optimal!

Having hit diffraction limits so often with my prints and 21 mp cameras, I have moved to tilt lenses for considered landscapes. Using tilt opens up a world of possibilities for improvements to image quality, being able to use optimal aperture for practically any subject depth of field is one cornerstone, not working through diffraction issues, just working around them. With simple tilt I now get f8 system sharpness and resolution, and depth of field from foreground to background.

michael_young3 · March 9, 2012

GDan, as I wrote earlier, it's a very good place to start a conversation. We all know and accept as unqualified fact that diffraction affects image quality. The questions remaining are how much does diffraction degrade an image, and how can we best express that loss.

You have it half right about diffraction not caring about sensor size or pixel density. Diffraction is a function of the aperture size and focal length alone, and has nothing at all to do with the sensor. But that's only half right. You now need to turn the same statement around and finish the thought.

For some given amount of diffraction, there is a minimal pixel spacing wide enough to not be affected by that diffraction. Sensors with pixel spacing smaller than this minimum are degraded by the diffraction. Sensors with pixel spacing at or larger than this minimum do not detect the diffraction.

This is simply a regurgitation of the DLA formula. Rather than solving for the limiting aperture for a given pixel distance, we solve for pixel distance for a given aperture.

Here is the surprising part. I thought it was oversimplified when I first came across it, and I suspect you'll gag on it and reject it as well.

Applying the minimum diffraction limited spacing calculated above over your sensor area implies some number of megapixels. For example, if the calculated diffraction limited spacing is 100 pixels/mm, a hypothetical fullframe sensor with that pixel spacing has 8.64 MP.

Intuitively, we expect that a 12 MP sensor should capture no less than 8.64 MP of detail. This is the argument that you, Scott, and others have been making. I also agree. The optic path, diffraction and all, can deliver that much detail to the sensor, and the sensor is capable of imaging all of it. (Or as much of it as it would have if diffraction were not a factor.)

But what about the other side? Can we capture more detail by packing more pixels onto the sensor? If a 12.7 MP camera can capture 8.64 MP of diffracted detail, does a 21 MP 5D2 or 36 MP D800 capture any more detail in those same conditions?

The answer here is resoundingly NO.Packing more pixels into the optically limited path does not produce any more detail. This is the fundamental basis of the megapixel debate, and I think needs no further elucidation in this audience. The result, initially surprising, is that we can capture no more than the 8.64 MP of detail implied by DLA.

In summation then, the single number DLA is both the upper and lower bound on diffraction limited detail.

This answers the second part: how much does diffraction degrade the image. The remaining question is how can we best express that loss. I choose to think of it as equivalent megapixels.This has direct meaning for me, having used various generations of Canon APS-C cameras through the past almost 10 years.

bengt_rehn · March 9, 2012

Since larger sensors contains a larger number of pixels at any given number of pixel spacing, the effect of diffraction on resolution will be less on a larger sensor.

I have a Hasselblad 50/4,0 that locks much better at 22 than my EF 50/1,4 at 22. The HBL lens even goes to f/32. There are of course a difference of FOV.

scott_ferris · March 9, 2012

That is because for the same image size you are enlarging the smaller one, and the diffraction, more. At the same

reproduction ratio, I.e. a 15x enlargement, the diffraction would be the same, but the print sizes different.

The diffraction isn't worse, you are enlarging it more.

g dan mitchell · March 10, 2012

"For some given amount of diffraction, there is a minimal pixel spacing wide enough to not be affected by that diffraction."

You apparently do not understand how digital sampling works, and you also fail to understand that subjects are not composed of purely vertical and horizontal lines that neatly line up perfectly with the rows of photo sites on your sensor.

That is all.

rubo · March 10, 2012

Ok, let's see if i car word it correctly. The higher the pixel density/MP count for a given sensor size, the more "visibly apparent" become the effect of a "diffraction blur" for a given apperture (past a certain apperture).

Would this be a correct way to word it?

john_wheeler6 · March 10, 2012

Hi Rubo

I bet you cold get a lot of different answers to your last question (different ways to say the same thing).

First, there is a prerequisite that the print being made is large enough and viewed close enough to even have the eye distinguish such fine level or sharpness. Assuming such a large print and close viewing distance:

I like to think of the camera system sharpness as several factors working in parallel e.g.:

- lens resolution (or lack of resolution blur)

- Camera vibration or subject movement blur

- focus accuracy

- DOF for elements not in the plane of focus

- Sensor resolution

- Lens diffraction

- ISO Noise (though not technically a blur can get in the way of sharpness)

- Wide aperture aberrations

- I may have left one or two other factors out

In a simplified approach (not showing the math on how to combine the above blurs) the sharpness is basically determined by the weakest link in the above categories. To the first order, if you try to improve a category of blur and you area already limited by another source of blur, you won't get improvement.

Lets take the example that sensor resolution blur is worse than diffraction blur and that all other factors are much better than the diffraction blur. In that specific case, if you continued to improve the sensor resolution while the diffraction blur remains the same, at first the image looks sharper. As you increase the sensor resolution beyond the diffraction blur, all that happens is that the sharpness stops improving. It's like hitting a blur wall which is the next weakest link in the chain. If that is what you mean by "apparent" then the answer would be yes. If you expect the diffraction blur to be easily distinguishable for other blurs, then I would say that is wishful thinking and probably not. This is all IMHO of course.

rubo · March 10, 2012

As you increase the sensor resolution beyond the diffraction blur, all that happens is that the sharpness stops improving. It's like hitting a blur wall which is the next weakest link in the chain. If that is what you mean by "apparent" then the answer would be yes.

</blockquote>

That's exactly what i mean :-) With the 60D + 28-135mm on a very good tripod, focused using live view, using remote release and mirror lockup, i'm starting to see a progressive drop in sharpness when going past f/11. Past f/16 "diffraction blur" is "apparent" enough for me not to shoot @ those appertures. When i had Rebel XTi (using the same lens, same scene, same composition) the resulting images @ f/16 (although not as sharp as @ f/11) did not "appear" to be as affected by a "difrraction blur" and look much shrper.

I know the anti-aliasing filter on the 60D is stronger compared to XTi, but again shooting @ f/8 the images look almost indistinguishable from the two bodies.

Ty,

That's exactly what i do :-)

Conditions permiting, i always try to shoot between f/8 and f/11, when i shoot landscapes.

But some times you need more DoF than that and that's when i start to think i may need to get an MF film camera, but don't have enough $ to do that right now :-)

ty_mickan · March 10, 2012

Ty,

You are correct, but it's irelevant to what i'm talking about.

It does not matter what kind of lens you put on 60D, past f/11 the image is losing sharpnes. Once you get past f/16 it's

already too soft for my taste.

I have tested this with Canon 10-22mm, 15-85mm, 24-105mm L, 28-135mm, Sigma 30mm, Sigma 150mm Macro, Sigma

70-200. With all of them the smaller the aperture (past f/11) the softer the resulting image.

That may be so, but why are you using f16 on a 30mm lens? For the ideal balance of diffraction limited spot size,

sharpness, and DOF, you may find f8 or f9.5 more agreeable.

michael_young3 · March 10, 2012

You apparently do not understand ...

</blockquote>

And that leaves very little mystery how you can remain so ignorant fully two years after we started this conversation. I'll leave you to stew in your filth with this final thought: the Hubble telescope's sensor array has a pixel spacing of 15 microns, which equates to a DLA of f/24, exactly matching its f/24 optics. What a wild, woolly, cosmic coincidence. Eh?

But some times you need more DoF than that and that's when i start to think i may need to get an MF film camera...

</blockquote>

Rubo, sometimes you can't even get out of the game. Old-school says "f/16 and be there," presumably in reference to film's DLA.

Increasing the focal length to keep the same field of view means you increase the aperture number proportionally to keep the same DoF. f/16 on 645 is more akin to f/8 or f/11 on your 35mm-size gear. (I believe this brings us full circle back to the OP question. What a happy way to end a thread.)

But maybe also check with GDan to see what effect "sampling" on film's non-orthogonal grid might have on diffraction effects. I'm dying to hear a cogent answer.

rubo · March 10, 2012

Michael,

i see what you are saying about small format film.

BUT what do you think of something like 8x10 view camera @ f/64 ?

Surely Ansel Adams and the rest of group f/64 weren't fools for shooting @ those appertures.

I know even with the biggest digital back today we are no were near that, and lugging around a view camera is not a very practical solution, but i can dream, right :-)

michael_young3 · March 10, 2012

Hi Rubo. Yes, you're right. Different numbers apply for the larger format film because the enlargement ratio can be much smaller.

michael_young3 · March 10, 2012

... But, on second thought, the normal lens for 8x10 is 300mm, compared to 50mm for 35mm format. DoF at f/64 is shallower still compared to f/16 on your present gear. (Proving yet again that we can't get out of this game.)

scott_ferris · March 10, 2012

Everybody still seems to be talking cross purposes here.

Rubo, let me try and cut to your specifics.

"When i had Rebel XTi (using the same lens, same scene, same composition) the resulting images @ f/16 (although not as sharp as @ f/11) did not "appear" to be as affected by a "difrraction blur" and look much sharper."

</blockquote>

To explain this statement there are several scenarios.

<ul>

<li>First, the diffraction in both is identical. This should be obvious.</li>

<li>At the same print size, both will be identical. Any differences are not due to diffraction.</li>

<li>At the same % value on screen the 60D will look worse, but that is only because you are looking at it bigger, you are enlarging the 60D image 80% more so the diffraction is displayed 80% bigger. So is any other detail in your image. You must compare like with like, so fit to screen is good, actual pixels or a % value views are not equal.</li>

<li>The AA filter is different on each camera so differences seen at the same reproduction ratio will be due to that and processing differences, not the diffraction.</li>

<li>To make a meaningful comparison you must process both files optimally, not the same way and not out of camera jpegs. Comparing out of camera jpegs won't work because the camera is tuned to process files differently.</li>

</ul>

So how are you viewing your 10mp XTi images and your 18mp 60D images, at 100% or the same reproduction ratio? And now to your original question.

<blockquote>"So here is my question: After i get the 5d Mkiii, what would be the right f/stop to get the same DoF i'm getting right now lets say @ f/8 on a 60D, assuming the framing is the same , so instead of 30mm on a 60D (which is 48mm equivalent on FF) it would be 50mm?"</blockquote>

Again, the issue we are talking about is reproduction ratio. To make the same sized print you enlarge the 5D MkII image less than the 60D image, this means you enlarge its diffraction less, this means for the same sized print you can use bigger blur so you can stop down the 5D MkII image more than the 60D image to get the same result. The theory of equivalence that I linked to very early on covers all these factors.

<ol>

<li>Same reproduction size (print size) from different sensor sizes. Pixel density is irrelevant.</li>

<li>Adjusted aperture to match DOF and diffraction across different sensor sizes and magnification ratios.</li>

<li>Adjusted focal length to match field of view differences.</li>

<li>Adjusted iso numbers to account for same generation sensor noise.</li>

</ol>

So the answer to your question of how to get an identical image from a 5D MkII and a 60D is simple.

<ul>

<li>60D image, 30mm-1/125 sec-f8-100iso</li>

<li>5D MkII image, 48mm-1/125 sec-f16-200iso</li>

</ul>

At any given print size these two images will display virtually identical fov, dof, diffraction and, in theory, noise. In practice the noise will be less in the 5D MkII image due to the rounding up of the figures.

"f/16 and be there," is a misquote, the long talked about phrase is "f8 and be there," I think you are confusing it with the "Sunny 16" rule which is all about exposure. Neither have anything to do with diffraction! The old-timer aperture expression is the f64 club, this relates to optimum aperture on an 8x10, and bigger, plate camera. If you grasp the equivalence concept it is obvious that you enlarge the bigger diffraction of the bigger negative much less for any given print size, so even though the diffraction is worse, it is no more visible.

scott_ferris · March 10, 2012

Typo to above...........

<ul>

<li>60D image, 30mm-1/125 sec-f8-100iso</li>

<li>5D MkII image, 48mm-1/125 sec-f16-200iso</li>

</ul>

Should read

<ul>

<li>60D image, 30mm-1/125 sec-f8-100iso</li>

<li>5D MkII image, 48mm-1/125 sec-f11-200iso</li>

</ul>

michael_young3 · March 10, 2012

That's a fair summation, Scott. I'll add only that "same print size" is a completely foreign concept. Who does that? A wedding photographer might when putting together an album My interest in more pixels is to print larger. I'm hanging pictures on my wall, and printing as much detail as large as I can.

rubo · March 10, 2012

Well Scott, that pretty much answered my question :-)

Michael, i agree, "who does that?" :-). i honnestly don't remember the last time i printed any of my images myself, but my friends sometime do.

This is just a hobby for me and something i like to do when i am finaly able to get out of NYC when fishing, hiking, camping etc.

I wish i could make a living doing this, but so far i have to found a way to do so :-)

scott_ferris · March 10, 2012

Thank you gentlemen, I am glad I seem to have brought some unity.

My point about "same print size", whilst being an outmoded concept for some, was that it forces people to compare like for like, as opposed to % values when viewed on a screen, I did say 'fit to screen' also works though :-)

As to who does that? Well I do, and I know G Dan has said he does. Maybe that is why we have the understanding of the issues that screen viewing clouds.

As for being interested in printing larger, there are many ways to increase print sizes to a higher quality, some easy, and some not so easy, or cheap. I have found, after years of film use in sizes up to 6x9, that 135 format digital, when using the correct gear and good technique easily capable of very high quality prints to 24"x36".

g dan mitchell · March 10, 2012

Michael Young helpfully offered:

"I'll leave you to stew in your filth..."

</blockquote>

Classy.

Dan

g dan mitchell · March 10, 2012

Scott: Good summary, and thanks for joining the folks who, like me, are trying to explain why there is no more diffraction blur with sensors with higher photo site density.

I have found that people who read this fall generally into two categories. One group reads and understands and says, "Oh, that makes sense." The other group has a hard time letting go of this notion, and they all too often get quite angry about it.

The second group also has a tendency to confuse the question of how much diffraction blur there is with the separate issue of how things that affect sharpness become more critical when you make larger prints. In this thread, we see one poster still coming back to that point. If you print larger with your 21MP sensor camera than you would have with your 12MP camera, the issue is that anything that lessens resolution becomes more visible at larger print sizes: diffraction blur, outer limits of DOF, camera stability, lens quality, precise focus. It isn't that there is more diffraction blur than you had at 12MP - it is that you never printed this large at 21MP because other factors limited the print size.

Dan

michael_young3 · March 10, 2012

Dan, DLA on your 5D is f/13. 5D owners really are among the least qualified of everyone to speak of diffraction from first hand experience. Really and truly, without rehashing everything that was already written in just this one thread, diffraction effects are quite real and very noticeable. The 7D (and 60D) gets a double whammy from the smaller crop sensor AND the very fine pixel pitch. My perception of the issue is necessarily different from yours. You have very little idea how ignorant you look when you diss everyone who corrects you on the topic.

Sign In

DoF question

Recommended Posts

Link to comment

Share on other sites

Top Posters In This Topic

Popular Days

Top Posters In This Topic

Popular Days

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Create an account or sign in to comment

Create an account

Sign in