Diffraction 5DII and minimum aperture

[Jared-5]

I'm assuming that you could stop down a lens much more on a 5D vs 5DII before running into diffraction.

I'd like to hear from some with the 5DII or 1DsIII about minimum aperture and diffraction. I would have preferred the 5DII had only 16MP partly because of this. But how big of a deal is it in the real world?

--

• Jared -

 

[gdanmitchell]

No.

At a given aperture (with the same lens) you will get precisely the same amount of diffraction blur from either camera. If you made two exposures with identical lens, focal length, and aperture and made prints from each, there would be absolutely no difference in the amount of diffraction.

In terms of its effect on a print, diffraction on a given sensor/film size is a function of aperture only. Photosite density has no bearing on the amount of diffraction.

At slightly larger apertures and when using a very sharp lens you may get an even sharper image from the higher photosite density body, but that is Good News.

Dan

I'm assuming that you could stop down a lens much more on a 5D vs
5DII before running into diffraction.

I'd like to hear from some with the 5DII or 1DsIII about minimum
aperture and diffraction. I would have preferred the 5DII had only
16MP partly because of this. But how big of a deal is it in the real
world?

--

• Jared -

--
---
G Dan Mitchell - SF Bay Area, California, USA
Blog & Gallery: http://www.gdanmitchell.com/
IM: gdanmitchell

Gear List: Cup, spoon, chewing gum, old shoe laces, spare change, eyeballs, bag of nuts.

 

[Henrik Herranen]

I'd like to hear from some with the 5DII or 1DsIII about minimum
aperture and diffraction. I would have preferred the 5DII had only
16MP partly because of this.

This makes no sense, and I'll explain why.

Let's think of the diffraction of the plastic fantastic, the EF50mm f/1.8 (just because that's the latest lens DPReview has reviewed, and the lens really doesn't matter). Open the following review page in a separate window so you can compare and check out what I'll be writing below:
http://www.dpreview.com/lensreviews/canon_50_1p8_ii_c16/page4.asp

The Nyquist limit on the 1Ds Mark III, as well as on the 5D2 is 1870 line pairs per picture height (exactly half of vertical resolution). On the 5D1 it is 1460, and on a hypothetical 16 megapixel 5D1.5 it would have been 1630 line pairs.

Let's have a look at DPreview's widget. At f/9 lens resolution is almost exactly aligned with the 1Ds Mark III's nyquist line that visible in the image. Thus, closing aperture any more will cause resolution / contrast loss because of diffraction. This would also be true with the 5D2.

On the original 5D1, you could close down to f/16 before the same happens, and on the hypothetical 5D1.5 the number would be f/13.

So, if you are shooting at f/16, are you getting the sharpest picture from the 5D1 because it is not diffraction limited? No, you won't. You'll get a roughly equal sharpness from all images WHEN PRINTED TO THE SAME SIZE. However, if you open up the aperture to f/9, you are going to get superior results from the highest megapixel 1Ds3 and 5D2 cameras.

Conclusion: If noise and dynamic range are not compromised (this is a big "if"!), and from a pure image quality point of view, it is always good to have more pixels. The worst that can happen is that you are going to get a similar image that would have gotten with the lower megapixel body. The best is a superior picture with high resolution.

Caveat: The MTF50 measurement does not indicate that there is no information above the mentioned resolution limit, only that the contrast of it has shrunken below a certain value. This means that even if a lens/aperture combination is diffraction limited, high megapixels still might help. The corollary is that you start to slightly lose contrast even before you hit the "official" diffraction limit. E.g. I can see slight loss of microcontrast even on the 5D1 when going from f/11 to f/16 although the difference is very small. Yes, I have tried this.

Kind regards,

• Henrik

--
And if a million more agree there ain't no great society
My obligatory gallery: http://www.iki.fi/leopold/Photo/Galleria/

 

[apixel]

These types of threads come up every once in a while. After reading everything people write here, I suggest you go do some tests for yourself. I used to worry about diffraction. I don't anymore. I did my own printing tests and although diffraction was very obvious on the computer at 100%, I could barely ( if at all ) see it on print.

After doing these tests, I will never be afraid to stop down as much as I need to.

I'm assuming that you could stop down a lens much more on a 5D vs
5DII before running into diffraction.

I'd like to hear from some with the 5DII or 1DsIII about minimum
aperture and diffraction. I would have preferred the 5DII had only
16MP partly because of this. But how big of a deal is it in the real
world?

--

• Jared -

--
------------
Photo gallery at: http://www.fotophoto.net/

 

[Zevs_1000]

Tree things:

1.

The difference in pixel-pitch between 5DII and 5D is 30% - hence the difference of the airy-disc introducing diffraction is 30% ( surpose you are look at it on the screen at 100%. That is the same as the difference between f/16 and f/18 (0,3 step).

2.

If you print to the same size from 5D and 5DII, the aparture for introducing diffraction is the same for both.

3.

If you just print to A4 instead of A2, you can come away with one step smaller aperture.

Z

 

[Rick Knepper]

I'm assuming that you could stop down a lens much more on a 5D vs
5DII before running into diffraction.

On a 5D image, I see a slight progressive degradation going from f8 to f11 to f16 on 1800 x 1200 sized images, Raw or jpeg. It is slight enough though that I do not hesitate to use f16 when needed. (I should add that I do not know if the step from f8 to f11 is due to diffraction or a lens aberration but it seems to occur with every lens.)

A better lens can help by being sharper in the corners so that the photographer does not have to stop down simply to overcome poorly corrected lens aberrations. I have a Zeiss lens that can be satisfactorily charp across the frame at f5.6 in the right landscape composition whereas every Canon L lens in the 21mm FL or thereabouts would need at least f8 to f11 to match (most lenses needing f11). I eagerly await some landscape samples from the 24 II though.

I'd like to hear from some with the 5DII or 1DsIII about minimum
aperture and diffraction. I would have preferred the 5DII had only
16MP partly because of this. But how big of a deal is it in the real
world?

Although I haven't used my 5D2 for a serious trip yet (March 7-13 Big Bend NP will be my first chance), I expect diffraction to be stronger. I'm thinking DoF blending is going to be big in my future.

Lloyd Chambers wrote a pretty good article on diffraction and the 1Ds mkIII which would apply to the 5D mkII also:

http://www.diglloyd.com/diglloyd/free/Diffraction/example-1DsM3.html

 

[Rick Knepper]

Tree things:

1.
The difference in pixel-pitch between 5DII and 5D is 30% - hence the
difference of the airy-disc introducing diffraction is 30% ( surpose
you are look at it on the screen at 100%. That is the same as the
difference between f/16 and f/18 (0,3 step).

2.
If you print to the same size from 5D and 5DII, the aparture for
introducing diffraction is the same for both.

Nos. 1 & 2 seem to contradict each other. Can you clarify?

3.
If you just print to A4 instead of A2, you can come away with one
step smaller aperture.

Z

 

[Steven Noyes]

And your statement of "I wish the 5D Mk II had 16 MP because of diffraction" shows you are among them.

It is simple. All things equal like lens/light/composition/proper technique... A 21MP FF image shot at f/16 will have more detail than a 12MP FF shot at f/16. The 21MP will look better in print.

Steven

--
---
Winter 2009:
http://www.pbase.com/snoyes/images_winter_2009

2007 Paria Plateau
http://www.pbase.com/snoyes/images_spring_2007

 

[bionet]

Even at f/64, the higher pixel pitch yields an advantage (reduces "generation loss" of the optical copy, same reason film has grain size far below any DSLR sensor's pixel pitch) but of course it gets progressively smaller.

It also depends on the lens due to several characteristics (e.g. aperture blade design), my 350D kit lens for example was best at f/7.1 and from f/8 on it got worse.

I have found that with the 5D2 (and of course the same applies to the 350D and 20D), f/11 is usually at full sharpness, f/16 is visibly degraded with the Sigma 15-30, less so on the 50/1.4. At f/22, quality drops dramatically.

 

[bobn2]

I'm assuming that you could stop down a lens much more on a 5D vs
5DII before running into diffraction.

I'd like to hear from some with the 5DII or 1DsIII about minimum
aperture and diffraction. I would have preferred the 5DII had only
16MP partly because of this. But how big of a deal is it in the real
world?

Just to confirm what everyone else has said, diffraction is the same for both cameras using the same lens - the 5D renders whatever the lens produces better. !6MP would be somehwere in between, but crucially the results would be worse than the 21MP of the 5DII at any aperture, although when the lens is in the diffraction limited regime the difference will be marginal.

What's less often realised, is that the diffraction blur in a print of a given size caused by setting the aperture for a given depth of field (which is presumably why people want to use small apertures) is the same for any size sensor, any pixel density. That's the rule, if you want extended DoF you pay for it with loss of resolution. No get outs, no exceptions. The best way out is DoF stacking.

--
Bob

 

[Zevs_1000]

Nos. 1 & 2 seem to contradict each other. Can you clarify?

1.

If you look at 21Mp-file at 100% and a 12p-file also at 100% at your screen, the real-estate-size accordin to the hole picture will be different - hence the airy-disk of a certain size will cover more pixels in the picture with the smallest pixels (5DII).

2.

If you print a 21Mp-file and a 12Mp-file to the same size (let´s say A2), the grain from each pixel will of cause be smaller in the print from the biggest fill. But because the airy-disk have the same proportional size to the frame in both cases, it will also have the same affect printet to the same papersize.

Z

 

[Henrik Herranen]

1.
The difference in pixel-pitch between 5DII and 5D is 30% - hence the
difference of the airy-disc introducing diffraction is 30% ( surpose
you are look at it on the screen at 100%. That is the same as the
difference between f/16 and f/18 (0,3 step).

This is not quite right.

When you add linear resolution by 28.6% (5616 vs 4368 horizontal pixels), you need to have a 28.6% larger linear aperture to get diffraction smaller enough to look the same on the pixel level. Thus, the difference is as the difference between f/16 and f/(16*1.286) = f/21, or 2/3 of a stop.

Other than that, your article applies.

Kind regards,

• Henrik

--
And if a million more agree there ain't no great society
My obligatory gallery: http://www.iki.fi/leopold/Photo/Galleria/

 

[gdanmitchell]

Although I haven't used my 5D2 for a serious trip yet (March 7-13 Big
Bend NP will be my first chance), I expect diffraction to be
stronger. I'm thinking DoF blending is going to be big in my future.

It won't be. It will be exactly the same.

Of course, when you look at, say, a 400 x 400 pixel 100% crop on the screen you will be LOOKING MORE CLOSELY AT A SMALLER SECTION OF THE IMAGE so it will appear larger. But in your print or other final product it will be no different whatsoever at the apertures we are considering here.

Let me also add that the "diffraction smaller than pixel pitch" business is irrelevant - except in the astonishingly unlikely even that a sub-pixel size element of your photograph lines up perfectly with the center of single pixel or, even most unlikely, that a perfectly horizontal or vertical line with a projected width of less than one pixel lines up perfectly with the center of a line of vertical or horizontal photosites. Of course that would also require the use of a lens with zero barrel/pincushion distortion used a a camera perfectly aligned to the scene. In other words, it ain't gonna' happen.

In the real world sub-pixel sized subjects projected on the sensor rarely line up in this way, and values "spill across" adjacent photosites.

It would be accurate - but probably irrelevant - to say the the sensor with the higher photosite density could form a more accurate image of the diffraction blur...

... but there is no "more diffraction."

Dan

Dan

--
---
G Dan Mitchell - SF Bay Area, California, USA
Blog & Gallery: http://www.gdanmitchell.com/
IM: gdanmitchell

Gear List: Cup, spoon, chewing gum, old shoe laces, spare change, eyeballs, bag of nuts.

 

[Rags_and_Old_Iron]

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

 

[bobn2]

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

Don't use this calculator unless you understand what it means, you'll just confuse yourself.
--
Bob

 

[gdanmitchell]

+1!

It is simplistic formulas like this, applied without understanding the context, that lead to common misconceptions like this one about diffraction and photosite density.

Dan

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

Don't use this calculator unless you understand what it means, you'll
just confuse yourself.
--
Bob

--
---
G Dan Mitchell - SF Bay Area, California, USA
Blog & Gallery: http://www.gdanmitchell.com/
IM: gdanmitchell

Gear List: Cup, spoon, chewing gum, old shoe laces, spare change, eyeballs, bag of nuts.

 

[Rags_and_Old_Iron]

Yep, I guess. The OP had asked about "real world", and this calculator is pretty abstract in its application of bizarre tangibles one might relate to the real word. Sorry for wasting anyone's time with the link. I only make small prints, only to 24 inches, so I guess I don't have anything useful to add for real word, but I have used this calculator for a guideline to good practices for my FF, 1.3, 1.6, 2/3 sensor sized cameras, as well as calibrating the software to the sensors, and it is funny, the lab people can't really tell which camera recorded what images when they are printed this small and viewed from arms length. Maybe if I hadn't started with the cambridge in colour site to guide me with the transition from film to digital my results would have been different, but it all seems pretty well written and easy to understand for me, so that is why I recommended it, but i am not super-pro so, and I am mostly concerned with how actual prints look so... yep, sorry.

http://www.cambridgeincolour.com/tutorials.htm

 

[Ben_Egbert]

I use f16 regularly on my 1DS-mk3. I have no problem printing to 16x24 other than poor IQ in the edges and corners on my 17-40 which I am unable to correct in post or by stopping down further.

My other lenses don't exhibit this problem. 35 f1.4, Zeiss 50 f1.4 and 70-200 f2.8IS.

I have a hard time seeing any degradation from f8 to f16. F22 is pretty easy to see however.

--
http://www.pbase.com/roserus/root

Ben

 

[bobn2]

Yep, I guess. The OP had asked about "real world", and this
calculator is pretty abstract in its application of bizarre tangibles
one might relate to the real word. Sorry for wasting anyone's time
with the link. I only make small prints, only to 24 inches, so I
guess I don't have anything useful to add for real word, but I have
used this calculator for a guideline to good practices for my FF,
1.3, 1.6, 2/3 sensor sized cameras, as well as calibrating the
software to the sensors, and it is funny, the lab people can't really
tell which camera recorded what images when they are printed this
small and viewed from arms length. Maybe if I hadn't started with
the cambridge in colour site to guide me with the transition from
film to digital my results would have been different, but it all
seems pretty well written and easy to understand for me, so that is
why I recommended it, but i am not super-pro so, and I am mostly
concerned with how actual prints look so... yep, sorry.

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

The thing is, as I posted elsewhere, the sensor size and pixel pitch has no effect on the diffraction blur for a given DoF, so it's hardly surprising that people couldn't tell which camera was in use. There is no difference between film and digital so far as DoF and diffraction is concerned, save that no-one ever pixel peeped with film, so these misconceptions didn't arise.

People really are befuddled about diffraction. They should stop worrying about it, and just understand that DoF comes at the cost of blur. Full stop.

--
Bob

 

[rrcphoto]

diffraction rears it's head mostly in pixel peeping.

someone above had it right - diffraction with respects to print size will be the same regardless of the Mp density of the sensor.

that's because the image size is the same, and the observer distance is also the same.

Noticable areas of sharpness or in focus always depend on observer distance and image size. diffraction simply occurs when the airy disc of light being diffracted around the aperture blades is larger than the circle of confusion.

when you view at 100% on a monitor, you are changing the image size, however, leaving the observer distance equal - thus a greater degree of magnification.

this is of course all assuming you don't crop.

with diffraction, the rule of thumb (assuming you do this for fine art prints) - is to know what aperture you can use for your related print size.

with the higher density print - you will also get finer resolving of micro detail - the "perceived" depth, tonality of the print will be much finer with higher density images - in other words, nothing scientific that pixel peepers can look at - but in essense - it "feels" better. Just like looking at prints from MF verus 35mm ... they feel better.

why? because elements that are resolved into a squarish pixel with a lower density sensor are resolved more round and natural on a higher density sensor. Also the tonality color in the is much more accurate in diffused areas simply because you have more pixels in the bayer matrix interpreting the colors.

so would it have been better with a 16Mp camera? no.

this is the fundamental reason why canon feels that 30, 40 and 60Mp FF sensors can be achieved - the pixel peeping will drive people nuts, but the prints will look objectively better with higher Mp's .. with the same size print being the cavaet.

there is a complete and final image magnification size that 35mm DSLR's will reach - that will be when no matter what aperture, the diffraction is greater than perceived circle of confusion as related to the observer distance.