Diffraction Myth Dispelled

Moderated by: guardian

Visit the FM Store · Image Upload · Buy & Sell
FM Forum Rules
Canon SLRs, primes, and zooms lenses reviews

FM Forums | Canon-mount SLRs | Join Image Upload
1 2 3

5 end

Photon
Offline
Buy and Sell: On

p.4 #1 · Diffraction Myth Dispelled

Chuck Gardner's suggestion of shooting through the entire range of f stops brings up an issue that hasn't been discussed directly and explicitly:
Diffraction increases gradually as the aperture is stopped down, but some lens aberrations decrease gradually at the same time.
The optimal aperture for a lens (aside from any need for depth of field) will be the point where increasing diffraction negates any advantage in control of coma, CA, etc. One reason for saying that a lens is "diffraction limited" at a particular stop is that the resolution is not going to increase past that point. However, in some cases closing down one more stop or so will not decrease the resolution - the diffraction and aberration variables will essentially balance each other in that range. So there may be no measurable performance difference for a certain lens between 5.6 and 8, and maybe between 8 and 11 for another.

Currently, the closest thing to a "high pixel density" body I have is a 1DIII. Pixel peeping with an 85L, the degradation between f/11 and f/16 is quite subtle to me.

Oct 10, 2008 at 01:05 AM

beewee
Offline
Image Upload: Off

p.4 #2 · Diffraction Myth Dispelled

There's a nice series of lectures on optics, sensors and their intermingled relationships by Canon and Panavision here.

One of the key points is that each part of the optical path, including the sensor, imparts its own MTF that will affect perceived/recorded sharpness.

Oct 10, 2008 at 04:03 AM

thw2
Online
Image Upload: Off

p.4 #3 · Diffraction Myth Dispelled

And the final proof to shut the mouths of fools and pseudo-engineers:

http://forums.dpreview.com/forums/read.asp?forum=1019&message=29641192

"The 50D's higher resolution allows you to see detail at f/16 that you can't see on a 40D file shot at f/5.6. Yes, I know, you can see the loss of sharpness as you hit f/16 but regardless, even at this aperture the higher resolution of the 50D carries the day."

Oct 10, 2008 at 10:20 AM

thw2
Online
Image Upload: Off

p.4 #4 · Diffraction Myth Dispelled

n0b0 wrote:
Why are you telling me about the pixel density stuff? I don't remember talking about that at all here. What have you been smokin mate?

Sorry for the digression.

Oct 10, 2008 at 10:25 AM

Pixel Perfect
Offline
Image Upload: On

p.4 #5 · Diffraction Myth Dispelled

Oct 10, 2008 at 10:47 AM

thw2
Online
Image Upload: Off

p.4 #6 · Diffraction Myth Dispelled

Pixel Perfect wrote:
If the 50D at f/16 is showing more detail than the 40D at f/5.6 based on those shots, I need a double corneal transplant. It's such a small difference it's not even worth mentioning.

Maybe. But the very fact that you still get a significant advantage at f/11 is enough to shut many fools up.

Oct 10, 2008 at 10:54 AM

brainiac
Online
Image Upload: On

p.4 #7 · Diffraction Myth Dispelled

thw2 wrote:
And the final proof to shut the mouths of fools and pseudo-engineers:

http://forums.dpreview.com/forums/read.asp?forum=1019&message=29641192

"The 50D's higher resolution allows you to see detail at f/16 that you can't see on a 40D file shot at f/5.6. Yes, I know, you can see the loss of sharpness as you hit f/16 but regardless, even at this aperture the higher resolution of the 50D carries the day."

Thanks for that - it's a great test which illustrates the point perfectly. The 40D file at f5.6 is clearly sharper than the 50D file at f16 BUT RESOLVES LESS. For instance, on the 50D f16 file you can see the even spacing between all of the strips, whereas on the 40D f5.6 file the spacing is often wrongly shown as uneven, and the strips wrongly appear to have varying thicknesses, like a bar code. With appropriate edge sharpening the 50D shot at f16 will produce a more accurate picture, in terms of detail, than the 40D shot at f5.6 or f8. Sharpness is not the goal, as that can be added later if required. Accurate resolution of pictorial information is the goal. This all reminds me of the M8 v 5D debate. High contrast between pixels is seductive, but it is not the same thing as resolution of detail.

Edited on Oct 10, 2008 at 11:26 AM · View previous versions

Oct 10, 2008 at 11:01 AM

n0b0
Offline
Image Upload: Off

p.4 #8 · Diffraction Myth Dispelled

Both the 50D and 40D files have been interpolated to 4230x2820 pixels to simulate prints of equal paper size, and sharpened (after the interpolation) using Photoshop's (version 8.0) USM set to 200,0.2,0.

It looks like the 40D image has been enlarged and sampled up while the 50D image was sampled down. Wouldn't that put the 50D image at an advantage? I always thought up sampling an image would make it less accurate in terms of details since a computer basically enlarge it and fill in the blank as best as it could.

Oct 10, 2008 at 11:19 AM

rockitman
Offline
Image Upload: Off

p.4 #9 · Diffraction Myth Dispelled

Thank you brainiac. You have been successful in dispelling this issue and the supposed terrible high iso perfromance of the 1Ds3. Cheers !

Oct 10, 2008 at 11:21 AM

brainiac
Online
Image Upload: On

p.4 #10 · Diffraction Myth Dispelled

n0b0 wrote:
>Both the 50D and 40D files have been interpolated to 4230x2820 pixels to simulate prints of equal paper size, and sharpened (after the interpolation) using Photoshop's (version 8.0) USM set to 200,0.2,0.

Whoops - missed that. That makes this comparison entirely bogus, as by downsizing the 50D file, some resolution has probably been thrown away.

>It looks like the 40D image has been enlarged and sampled up while the 50D image was sampled down. Wouldn't that put the 50D image at an advantage? I always thought up sampling an image would make it less accurate in terms of details since a computer basically enlarge it and fill in the blank as best as it could.

No. Uprezzing and then sharpening appropriately should have no perceptible negative effect on the quantity of detail in the file, at least every time I have tried to discern it.

Oct 10, 2008 at 11:34 AM

n0b0
Offline
Image Upload: Off

p.4 #11 · Diffraction Myth Dispelled

brainiac wrote:
No. Uprezzing and then sharpening appropriately should have no perceptible negative effect on the quantity of detail in the file, at least every time I have tried to discern it.

I think that depends on how much bigger you uprez the image. If it's only an extra 10% or so, you'd probably won't be able to tell. At 50% uprez, I'd say you'll notice it.

Oct 10, 2008 at 11:44 AM

brainiac
Online
Image Upload: On

p.4 #12 · Diffraction Myth Dispelled

n0b0 wrote:
brainiac wrote:
No. Uprezzing and then sharpening appropriately should have no perceptible negative effect on the quantity of detail in the file, at least every time I have tried to discern it.

I think that depends on how much bigger you uprez the image. If it's only an extra 10% or so, you'd probably won't be able to tell. At 50% uprez, I'd say you'll notice it.

No. If anything uprezzing by a small amount produces a less accurate upsample than uprezzing a lot. The more you uprez it, the less interference between the two pixel grids will matter.

Oct 10, 2008 at 12:05 PM

Alex Nail
Offline
Buy and Sell: On

p.4 #13 · Diffraction Myth Dispelled

So basically this is just a thread to sa that shooting at f16 isnt the end of the world, which anybody who had done any sort of test before would have known anyway. I have never heard someone say that shooting at f16 is doomsday or whatever. I do shoot at f16 when necessary, but I do also know that it does effect sharpness of my imagesbecause I have done the tests. That said the sharpness loss at f16 is relatively low.

Alex

Oct 10, 2008 at 01:11 PM

danmitchell
Offline
Image Upload: Off

p.4 #14 · Diffraction Myth Dispelled

Etadam wrote:
brainiac wrote:
Diffraction will affect a sensor with small light wells and large interpixel gaps no differently than it affects a sensor with the same pixel density ('pitch') with large microlenses and no gaps.

I don't agree. The more diffraction there is, the larger the airy disk (diffraction pattern) is. Thus, for the same pixels density, pixels with larger gaps are less diffraction-sensitive since a larger airy disk (more diffraction) is needed to cover more than 1 pixel (in average).

Just a simple response. If you presume that all airy discs are centered exactly in the center of photosites your assumption might have a certain kind of logic.

But they aren't. If such a spot is centered on a point between two photosites (just as likely) the effect of that "spot" will extend in a _larger_ distance from its center point.

Bottom line: sensor size and photosite density are essentially irrelevant to the "amount" of diffraction you will see.

Dan

Oct 10, 2008 at 04:03 PM

Etadam
Offline
Image Upload: On

p.4 #15 · Diffraction Myth Dispelled

danmitchell wrote:
Just a simple response. If you presume that all airy discs are centered exactly in the center of photosites your assumption might have a certain kind of logic.

But they aren't. If such a spot is centered on a point between two photosites (just as likely) the effect of that "spot" will extend in a _larger_ distance from its center point.

Bottom line: sensor size and photosite density are essentially irrelevant to the "amount" of diffraction you will see.
Agreed, this is why I added "in average", meaning with "more diffraction" the blur will be "more likely" to be observable. And yes, diffraction depends only on aperture (and wavelength of light) ; however the discussion was "does more or less gap between pixels (or photosites) will show more or less diffraction [in average!]?", and my answer was yes. ("More or less gap" for the same pixels/photosites density).

Oct 10, 2008 at 05:42 PM

mh2000
Offline
Image Upload: Off

p.4 #16 · Diffraction Myth Dispelled

hmm, I'm both an artist and an engineer. :0

>>And the final proof to shut the mouths of fools and pseudo-engineers

Oct 10, 2008 at 07:13 PM

danmitchell
Offline
Image Upload: Off

p.4 #17 · Diffraction Myth Dispelled

Etadam wrote:
danmitchell wrote:
Just a simple response. If you presume that all airy discs are centered exactly in the center of photosites your assumption might have a certain kind of logic.

But they aren't. If such a spot is centered on a point between two photosites (just as likely) the effect of that "spot" will extend in a _larger_ distance from its center point.

Bottom line: sensor size and photosite density are essentially irrelevant to the "amount" of diffraction you will see.
Agreed, this is why I added "in average", meaning with "more diffraction" the blur will be "more likely" to be observable. And yes, diffraction depends only on aperture (and wavelength of light) ; however the discussion was "does more or less gap between pixels (or photosites) will show more or less diffraction [in average!]?", and my answer was yes. ("More or less gap" for the same pixels/photosites density).

Still not right. The "average" will be no different. You will get no more or less diffraction blur by changing the number of photosites (or photosite density).

Bottom line: Diffraction blur is related to focal length, aperture, and sensor size. Diffraction is not related to photosite density or photosite size.

Dan

Oct 10, 2008 at 08:27 PM

beewee
Offline
Image Upload: Off

p.4 #18 · Diffraction Myth Dispelled

How does sensor size come into play?

If we're talking about perceived diffraction then by your claim, a sensor with an arbitrary size (say 10x10mm in size) that consists of a single pixel, should be able to detect diffraction since you claim that photosite density or size have no bearing on detecting diffraction effects. Clearly, that's impossible.

If we're talking about physical diffraction (not perceived diffraction) then according to the "Cambridge Handbook of Physical Formulas", diffraction for a circular aperture is a function of the initial intensity of the light ray, wavelength of incoming light and aperture diameter.

Oct 11, 2008 at 01:31 AM

danmitchell
Offline
Image Upload: Off

p.4 #19 · Diffraction Myth Dispelled

beewee wrote:
How does sensor size come into play?

If we're talking about perceived diffraction then by your claim, a sensor with an arbitrary size (say 10x10mm in size) that consists of a single pixel, should be able to detect diffraction since you claim that photosite density or size have no bearing on detecting diffraction effects. Clearly, that's impossible.

If we're talking about physical diffraction (not perceived diffraction) then according to the "Cambridge Handbook of Physical Formulas", diffraction for a circular aperture is a function of the initial intensity of the light ray, wavelength of incoming light and aperture diameter.

There are two ways to measure diffraction. One is what I'll call the "absolute size" of the diffraction blur - imagine measuring it with a very small (imaginary) ruler placed on the sensor. It is true that this measurement would be the same on different size sensors given the same lens and aperture.

But that isn't particularly relevant to photography.

The second way to measure diffraction is in terms of its "relative size" - the relationship between its size the the size (let's use the horizontal dimension) of the sensor. This IS relevant to photography.

Diffraction of some "absolute size" will be "smaller" relative to the overall size of the larger sensor, and thus be less visible in a print.

To use a goofy and over the top example, imagine some lens/aperture combination that produced diffraction blur that is "1mm across." This diffraction blur would cover 10% of the image from a 10mm wide sensor, but it would only cover 1% of the image from a 100mm wide sensor. (Yes, these sensor sizes are bogus and invented solely for this example, but the principle is the same.)

Dan

Edited on Oct 11, 2008 at 02:19 AM · View previous versions

Oct 11, 2008 at 02:16 AM

Photon
Offline
Buy and Sell: On

p.4 #20 · Diffraction Myth Dispelled

I can hardly wait until we have sensels so much smaller than the wavelengths of light that we'll be able to pixel peep the airy disks.

Oct 11, 2008 at 02:16 AM

beewee
Offline
Image Upload: Off

p.4 #21 · Diffraction Myth Dispelled

danmitchell wrote:
To use a goofy and over the top example, imagine some lens/aperture combination that produced diffraction blur that is "1mm across." This diffraction blur would cover 10% of the image from a 10mm wide sensor, but it would only cover 1% of the image from a 100mm wide sensor. (Yes, these sensor sizes are bogus and invented solely for this example, but the principle is the same.)

The problem with this is, if you had a 10x10mm sensor of four pixels (2x2 matrix), each 5mm in length and width, if that 1mm diffraction blur falls within one entire pixel, you wouldn't detect it at all. All of the energy from that one single light ray is picked up by a single pixel. In effect, you will have one perfectly sharp point even though the diffraction blur is 1mm. This is why pixel spacing and photosite size come into play.

The very reason we see blurring caused by diffraction is because single parallel rays of light are turned into a cone of light by diffraction whereby, when its projected onto a flat surface, perpendicular to the direction of the lightray, it forms an Airy disc. When energy from the airy disc falls within a single pixel, then that diffraction blur is not detected by the sensor and it appears perfectly sharp. However if the pixels are smaller and denser, that airy disc will distribute energy over multiple pixels. This is what we know as blur.

To take this concept further, the whole idea of depth of field simply boils down to what's considered "acceptibly sharp" since theoretically speaking, a focal plane is infinitismally thin. However, depending on the aperture setting, objects within a given distance from the focal plane can appear to be in focus since film/sensors cannot resolve the apparent out of focus blur, or at least not to a point where most people would deem it to out of focus. Why can't it resolve the out of focus blur, it's because although these light rays are no longer parallel and they diverge, when it hits the sensor, the pixels still capture most or all of the energy of each light ray thereby recording a sharp image.

Oct 11, 2008 at 04:35 AM

mh2000
Offline
Image Upload: Off

p.4 #22 · Diffraction Myth Dispelled

one pixel cannot appear "perfectly sharp," is it just a value. The diffraction blur will average out to some imperfect value with a larger pixel, and with more smaller pixels some will be more accurate and some will be less accurate (how do they average when you look at the photo from 3 feet?). Remember, the image is made from infinite point sources which all get superimposed, not by single point sources that act independantly... as I said, you can devise thought experiments that support either side to this, but images are random in nature and will not fall precisely within any...

If you are using a camera and lens that are not able to give you good photos you have to look for a new camera or lens.

Oct 11, 2008 at 06:14 PM

RDKirk
Offline
Image Upload: Off

p.4 #23 · Diffraction Myth Dispelled

It's no different than during film days, it's just not many people could afford to blow up images at 100x150 cm and inspect them at 20 cm distance for diffraction. The moral of the story, more MP do not produce more diffraction, they just allow you to enlarge more and you enlarge everything at the same time, hand-shake blur, small focusing errors, lens imperfections and diffraction. This is another internet myth.

A lot of these myths seem to come from people who think photography only came to being after the invention of the CCD.

We are only now getting 24x36mm sesors that rival the sharpest thin-emulsion black and white films, and all these issues have been known and dealt with. We've had sharper "sensors" before and found it a great place to be.

Oct 11, 2008 at 10:56 PM

RDKirk
Offline
Image Upload: Off

p.4 #24 · Diffraction Myth Dispelled

How does sensor size come into play?

The sensor size comes into play when you enlarge the image to fit a predetermined final display image size.

That's why someone shooting in large format can use f-stops as small as f/45 and f/64 and still produce 16x20 images of superior sharpness to anything coming from 24x36mm...even given the greater diffraction and inferior "sensor" plane flatness of the larger format.

Diffraction at f/22 that matter when the original image is enlarged 16 times and viewed at one meter doesn't matter when the original image is enlarged only four times and viewed at the same distance.

Edited on Oct 12, 2008 at 04:51 PM · View previous versions

Oct 11, 2008 at 11:06 PM

RobertLynn
Offline
Buy and Sell: On

p.4 #25 · Diffraction Myth Dispelled

This is a recent shot, 70mm on a crop (don't know how much that affects the argument or facts) f/22

Oct 12, 2008 at 12:12 AM


FM Forums \| Canon-mount SLRs \| Join Image Upload 1 2 3 4 5 end