Pages: [1]   Bottom of Page
 Author Topic: The Canon 50D Milestone  (Read 7883 times)
vandevanterSH
Sr. Member

Offline

Posts: 626

 « on: December 30, 2008, 11:19:31 PM » Reply

I found this statement interesting:
"The diffraction-limited resolution aperture for this sensor is around f/7! (The appropriate formula is: diffraction limited aperture equals 1600 divided by the pixel pitch in pixels per millimeter.) Using smaller apertures - say f/11 - will result in images with significant softness."

So does packing more pixels reduce the ability to increase DOF with smaller apertures?

Steve
 Logged
PierreVandevenne
Sr. Member

Offline

Posts: 512

 « Reply #1 on: December 31, 2008, 12:28:48 AM » Reply

Quote from: vandevanterSH
I found this statement interesting:
"The diffraction-limited resolution aperture for this sensor is around f/7! (The appropriate formula is: diffraction limited aperture equals 1600 divided by the pixel pitch in pixels per millimeter.) Using smaller apertures - say f/11 - will result in images with significant softness."

So does packing more pixels reduce the ability to increase DOF with smaller apertures?

Steve

You are resolution limited. Ultimately, the resolving power of a lens will depend on its diameter (that's why we make big telescopes). The bigger the diameter, the smaller the Airy disk. That's roughly the limit on your resolution. Using smaller apertures automatically reduces the maximum theoretical resolution because it increases the diameter of the Airy disk. Leving aside lens design, oversampling and Bayer matrix issues,  packing pixels beyond the resolution limit on the sensor doesn't make sense but it doesn't reduce the ultimate physical resolution.

A very nice explanation here

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

with an animated aperture/airy disk/pixel size and a useful calculator.

 Logged
jani
Sr. Member

Offline

Posts: 1604

 « Reply #2 on: December 31, 2008, 05:41:32 AM » Reply

Quote from: vandevanterSH
So does packing more pixels reduce the ability to increase DOF with smaller apertures?
No, that's not really relevant. But the greater amount of pixels may improve the apparent DoF, since they can represent greater detail.

Speaking of pixels and DoF is also quite meaningless unless you also speak about the degree of enlargement, the viewing distance, and so on.

Here's another tutorial that ought to help you get started.
 Logged

Jan
samirkharusi
Full Member

Offline

Posts: 196

 « Reply #3 on: December 31, 2008, 05:44:59 AM » Reply

I am a bit amused that somebody so nit-picking on image quality still uses zooms... Canon primes of 100mm focal or longer are superb, and should be a good match for the 50D. Wider than 100mm remains problematic but f8 should be OK for the landscape crowd, DoF considerations taken into account. Something has to be the weakest link in the imaging chain. Since I use a whole slew of primes from 14mm to 600mm, it's good to see that it's the lenses now, rather than the sensor, even for primes. About time!
 Logged

Bored? Peruse my website: Samir's Home
Marlyn
Sr. Member

Offline

Posts: 253

 « Reply #4 on: December 31, 2008, 06:50:25 AM » Reply

The cambridge colour site is a very good resource for the entire DoF / CoC / Diffraction question.

In a nutshell, yes, packing in more pixels (i.e. reducing the spacing),  does reduce the aperture at which diffraction occurs.

On a 1DSMkIII, anything over f10 is reducing capture resolution due to diffraction.  (Lens can cause other issues that affect resolutuion but it is irrelevant to diffraction.   It is pure physics between the aperture size (which determines Airy disk), and the pixel density).

On a bayer array, an airy disk of roughly twice the pixel density is the max before resolution is affected by diffraction (Main source: Cambridge colour site)   6.4um on 1DsIII,   giving 12.8um of Airy disk.  Which is achieved somewhere between f10 and f11.

Of course, you may WANT to decrease the aperture size for DoF,  which is fine, just realize that diffraction is now comming into play as well as all the other issues associated with taking a good shot.

Mark.

 Logged
NikoJorj
Sr. Member

Offline

Posts: 1063

 « Reply #5 on: December 31, 2008, 07:54:25 AM » Reply

Quote from: vandevanterSH
So does packing more pixels reduce the ability to increase DOF with smaller apertures?
If you like to dig into things a tad deeper, this page from Norman Koren's site explains well the relationship between depth of field and diffraction.
It's a part of a whole tutorial about sharpness that you may also find interesting - even if, not unlike Harold Merklinger's article, it evokes a certain nostalgia when taking examples with film (sooooo XXth...)   .
 Logged

Nicolas from Grenoble
A small gallery
PierreVandevenne
Sr. Member

Offline

Posts: 512

 « Reply #6 on: December 31, 2008, 09:02:56 AM » Reply

Quote from: Marlyn
On a bayer array, an airy disk of roughly twice the pixel density is the max before resolution is affected by diffraction (Main source: Cambridge colour site)   6.4um on 1DsIII,   giving 12.8um of Airy disk.  Which is achieved somewhere between f10 and f11.

Of course, you may WANT to decrease the aperture size for DoF,  which is fine, just realize that diffraction is now comming into play as well as all the other issues associated with taking a good shot.

Mark.

Just did a quick (but hopefuly clean) test with 3 lenses on a 5DMKII at F11 and F22 on a tripod. The 24-70mm is borderline, the 70-200 much better, the 100 prime beats them all (and that's the only one I put on-line)Correlates nicely with everyhting in this thread.

http://www.datarescue.com/life/f11f22/

(warning the page is close to 3MB)

F11 is noticeably sharper, but from a composition point of view, one might still want to sacrifice some resolution for DOF.

Since the 100 prime still has 12 elements, I am going to try with a top quality apochromatic triplet over the week-end.

 Logged
Ray
Sr. Member

Offline

Posts: 8948

 « Reply #7 on: December 31, 2008, 09:46:11 AM » Reply

One has to be careful about defining diffraction limits on resolution based on mathematical calculations of Airy disc size in relation to pixel size.

In practice, a camera such as the 50D will continue to provide more resolution than the 40D all the way down to F16, when images are compared at 200%, representing a 6ftx9ft print. However, at F22 the increase in detail is so small, one could consider it as non-existent.

The fact is, diffraction is a property of the lens, not the sensor. The recorded image is always something less than the lens can provide. When a good prime lens is used at its sharpest aperture, say F5.6, the recorded detail is likely to be significantly less than what the lens actually transmits, at least in the centre of the image. As one stops down, all lenses lose contrast in fine detail to the point where the detail becomes too faint to be recorded. At this point there is no advantage to the sensor with the higher pixel count.

One could argue that a 50D image at F16 would have very slightly less (shallower) DoF than a 40D shot at F16 on the basis that the plane of focus would be very marginally sharper in the 50D shot. But the difference would be so small it would be completely drowned out by more significant factors such as print size and viewing distance, as Jani mentioned.

At F5.6 to F8 however, the differences in resolution at the plane of focus might be sufficient to create the perception that the 50D shot had a slightly shallower DoF.

 Logged
vandevanterSH
Sr. Member

Offline

Posts: 626

 « Reply #8 on: December 31, 2008, 11:27:33 AM » Reply

Thanks for the references!!!! I need to do some reading...but:

Ray statement: "The fact is, diffraction is a property of the lens, not the sensor" seems to be the reflect what is being said in the comments but seems to contradict the statement that:
"The diffraction-limited resolution aperture for this sensor is around f/7! (The appropriate formula is: diffraction limited aperture equals 1600 divided by the pixel pitch in pixels per millimeter.)"

I suspect that this is the explanation:
"One has to be careful about defining diffraction limits on resolution based on mathematical calculations of Airy disc size in relation to pixel size."

Steve
 Logged
PierreVandevenne
Sr. Member

Offline

Posts: 512

 « Reply #9 on: December 31, 2008, 11:45:02 AM » Reply

Quote from: vandevanterSH
Thanks for the references!!!! I need to do some reading...but:

Ray statement: "The fact is, diffraction is a property of the lens, not the sensor" seems to be the reflect what is being said in the comments but seems to contradict the statement that:
"The diffraction-limited resolution aperture for this sensor is around f/7! (The appropriate formula is: diffraction limited aperture equals 1600 divided by the pixel pitch in pixels per millimeter.)"

There's always the physical reality, and then what is applicable in practice for photographers. For example, while I agree 100% with Ray, there are other factors to consider

- diffraction by the diaphragm blades and their edges.

- diffraction introduced by the CCD structures themselves in front-illuminated CCDs (I believe, maybe wrongly, that all current digital camera sensors are of the front illuminated type)

http://www.dalsa.com/public/corp/PDFs/pape...%20abstract.pdf

But we are beyond pixel peeping here ;-)
 Logged
samirkharusi
Full Member

Offline

Posts: 196

 « Reply #10 on: December 31, 2008, 01:00:56 PM » Reply

Quote from: PierreVandevenne
Since the 100 prime still has 12 elements, I am going to try with a top quality apochromatic triplet over the week-end.
Nobody puts in extra lens elements for no good reason. At one time I was also an astro enthusiast believeing that the premium astro APOs would trounce camera lenses any time. Yes, I believed the ads. Until I tested two against Canon primes, on stars. There is no way a 3-element APO, i.e. without some flattener at the rear, can even hope to cover a sensor of any significant acreage, eg even an APS-C. If all the elements are at the front, you have field curvature in the image field. At least I am unaware as to how one can avoid it. The astro 3-element APOs do a good job on-axis, but you'd better have a matched flattener to do any sensible pixel peeping off-axis. The two OTAs I pitted against Canon lenses were both from TeleVue and were both supposed to have flat fields (ie superior to 3-element APOs). One was a modified Petzval 4-element, the other had a built-in flattener. Both had superb reviews published previously that any manufacturer would be proud of. Both comparos are at my website below. get ready to splurge on a flattener
 Logged

Bored? Peruse my website: Samir's Home
PierreVandevenne
Sr. Member

Offline

Posts: 512

 « Reply #11 on: December 31, 2008, 01:38:58 PM » Reply

Quote from: samirkharusi
One was a modified Petzval 4-element, the other had a built-in flattener. Both had superb reviews published previously that any manufacturer would be proud of. Both comparos are at my website below. get ready to splurge on a flattener

Hi Samir - Great site you have there! I remember you, we exchanged a few forum messages about the colour of double stars a few years ago. I agree a field flattener is necessary. But I am jut curious about the eventual resolution advantage at the center of the field. I've done comparisons as well, between a TMB apo and, among other things, a 300mm 2.8L Canon. The Canon indeed has a flatter field (and that is fortunate for photography) but suffers from coma on the edges and chromatic aberration.

I have the ultimate weapon though: a Lichtenknecker 8" 2.7 Flat Field Camera. Unfortunately, a bit heavy for every day photographic use ;-)
 Logged
barryfitzgerald
Sr. Member

Offline

Posts: 609

 « Reply #12 on: December 31, 2008, 01:42:19 PM » Reply

So the bottom line is he feels 10mp is enough for most on APS.
Interesting read though. (I am ignoring the 35mm = megapixel comments!)
Sad but true, some folks are looking forward to 20mp + on APS sensors, I guess they won't like the lens bill that goes with it.
 Logged
Marlyn
Sr. Member

Offline

Posts: 253

 « Reply #13 on: December 31, 2008, 04:51:42 PM » Reply

Quote from: Ray
One has to be careful about defining diffraction limits on resolution based on mathematical calculations of Airy disc size in relation to pixel size.

I agree ,  it is only PART of the story of good image making, and as with all things, one needs to use these various technichal limitations with a serious dose of common sense.

Quote from: Ray
In practice, a camera such as the 50D will continue to provide more resolution than the 40D all the way down to F16, when images are compared at 200%, representing a 6ftx9ft print. However, at F22 the increase in detail is so small, one could consider it as non-existent.

Having played with this reasonably extensivley lately, I can't actually agree with you Ray. I can see noticable difference in large prints betwen an F10 shot and an F16 or an F22 Shot.  The F10 ones contain far more micro detail than the F22 ones.  (For reference, I'm shooting real world things, not test charts.  My favourite candidate is the Opera House, with its tiny tiles.  at F10 they are just visible, at F16 they are mushy, and F22 all gone).  Like anything though YMMV.  (And I'm not using either a 40D or a 50D, but a 1dsIII so yes, I'm not comparing the same thing).

Quote from: Ray
The fact is, diffraction is a property of the lens, not the sensor.

Diffraction is a physics property of Light encountering an obstacle such as a small hole (the Apeture in this case).  It only has passing association with the Lens in the fact that the apeture in inside the lens.  It dosn't have anything to do with quality of the glass etc.     HOWEVER that is ONLY Diffraction I'm referring to. Captured resolution definatly has a lot to do with the Glass, and only in the 'sharpest' lenses does all this matter anyway, as other properties in the Lens cause loss of resolution and sharpness. (In this case, I belive Refraction is the biggest culprit, along with , Dispersion, Reflection(Flaring), etc. All of which then cause loss of sharpness, loss of effective resolution, and loss of contrast)

It is only related to the sensor in that, the sensor defines an absolutle limit on capture resolution,  when the size of the effect of the diffraction is greater than the size of the capture medium's resolution, then detail will be lost due to diffraction. (i.e. the effect on our photo's is a combination of the two).    This is one of the reason bigger pixels (and bigger sensors for those pixels) = good.

I guess the reason we all talk about Diffraction so much is it is the only one we as the shooter can control in the process of taking a shot.  Refraction, Dispersion, Reflection etc are soley in the hands of the lens manufacturers, and we can only control this by our purchases of different lenses (Prime vs Zoom,  good quality vs poor etc)

Quote from: Ray
The recorded image is always something less than the lens can provide. When a good prime lens is used at its sharpest aperture, say F5.6, the recorded detail is likely to be significantly less than what the lens actually transmits, at least in the centre of the image. As one stops down, all lenses lose contrast in fine detail to the point where the detail becomes too faint to be recorded. At this point there is no advantage to the sensor with the higher pixel count.

True.  And all this dosn't matter for many smaller prints when you head into issues of viewing distance,  etc etc.  Just to be clear I'm refering to maximising the Captured resolution,  and not reducing that in the field.    It is also important to realise, that (IMO) even when a shot is diffraction limited, more pixels DO help.  for exampe they provide a greater 'blend' of tones between one value and the next.  Just the amount they help is reduced.

Huge caveat with all of this.  It is in a way, a significant form of 'pixel peeping'.  My Personal intent when finding out and testing it, was to ensure I was maximising my CAPTURE resolution, having just spent a small fortune on the 21MP camera.  This came about when I was not happy with the initial results over my 5D or Velvia on the 4x5.  (shots at f22 for example).
Having now changed my shooting to accomodate these factors, I find the technical quality of the shots to be significantly better.

Asthetics is a whole 'nother ball game entirely

Merry Christmass and a Happy new Year to All.

Regards

Mark Farnan
 « Last Edit: December 31, 2008, 04:55:56 PM by Marlyn » Logged
Ray
Sr. Member

Offline

Posts: 8948

 « Reply #14 on: December 31, 2008, 05:38:36 PM » Reply

Quote from: Marlyn
Having played with this reasonably extensivley lately, I can't actually agree with you Ray. I can see noticable difference in large prints betwen an F10 shot and an F16 or an F22 Shot.  The F10 ones contain far more micro detail than the F22 ones.  (For reference, I'm shooting real world things, not test charts.  My favourite candidate is the Opera House, with its tiny tiles.  at F10 they are just visible, at F16 they are mushy, and F22 all gone).  Like anything though YMMV.  (And I'm not using either a 40D or a 50D, but a 1dsIII so yes, I'm not comparing the same thing).

I guess I wasn't clear on that point, Mark. I was addressing the concept that at a particular aperture there is a cut-off regarding any increase in resolution with a camera such as the 50D which has an extremely high pixel density for a DSLR.

In practice, I've found this not to be the case. The differences in resolution between the 40D and 50D will of course be most obvious when the lens is used at its sharpest aperture. But such differences continue to exist all the way to F22, but just get less significant with each stop downwards. I'm referring here to comparisons between same F stop numbers.

However, if one wishes to match different F stop numbers, one might find that F16 with the 50D (in terms of detail captured at the plane of focus) is very close to detail at F11 on the 40D, and that F11 on the 50D is very close to detail captured at F5.6 with the 40D, depending on lens quality.

This fact I find encouraging. It's so easy to get the impression from some comments on this issue that there would be little point in using a 50D at F11 because diffraction would rob it of any advantage over the 40D. This is not so.
 Logged
Marlyn
Sr. Member

Offline

Posts: 253

 « Reply #15 on: December 31, 2008, 05:46:17 PM » Reply

Quote from: Ray
I guess I wasn't clear on that point, Mark. I was addressing the concept that at a particular aperture there is a cut-off regarding any increase in resolution with a camera such as the 50D which has an extremely high pixel density for a DSLR.

In practice, I've found this not to be the case. The differences in resolution between the 40D and 50D will of course be most obvious when the lens is used at its sharpest aperture. But such differences continue to exist all the way to F22, but just get less significant with each stop downwards. I'm referring here to comparisons between same F stop numbers.

However, if one wishes to match different F stop numbers, one might find that F16 with the 50D (in terms of detail captured at the plane of focus) is very close to detail at F11 on the 40D, and that F11 on the 50D is very close to detail captured at F5.6 with the 40D, depending on lens quality.

This fact I find encouraging. It's so easy to get the impression from some comments on this issue that there would be little point in using a 50D at F11 because diffraction would rob it of any advantage over the 40D. This is not so.

And on this, I totaly agree with you Ray.

Even when things start being 'limited' by say, Diffraction, or other factors, more resolution is good, and will provide (IMO) a better result.  More graduations accross the image, (i.e. better blending of tones, etc).     There is NO Absolute 'step' in any of this,  all the steps are gradual when it comes to factors that affect the propagation of light, and more pixels, as a general rule, is good.  Just the rule of diminishing returns may start to apply (as you say, the gap may get smaller).

All in all, its nice to have such toys to play with and compare

Regards

Mark.

 Logged
iliosgallery
Newbie

Offline

Posts: 17

 « Reply #16 on: December 31, 2008, 09:55:45 PM » Reply

Quote from: vandevanterSH
I found this statement interesting:
"The diffraction-limited resolution aperture for this sensor is around f/7! (The appropriate formula is: diffraction limited aperture equals 1600 divided by the pixel pitch in pixels per millimeter.) Using smaller apertures - say f/11 - will result in images with significant softness."

So does packing more pixels reduce the ability to increase DOF with smaller apertures?

Steve

The answer to your question is 'yes'. There's an easy way to visualize this without getting into formulas. Rays of light emanate from a point source on the subject. Their various wavelengths diverge when hitting the edge of the iris and land in a circular pattern on the recording plane, in this instance the sensor.  The smaller the aperture the larger this circular pattern or 'airy disc'. When the diameter of the airy disc is larger than the diameter of a single photo site this visual data spills onto neighboring sites. The end result is that we have photo sites receiving their information as well as their neighbor's data. This overlapping data causes loss of resolution. Obviously, then, the smaller the diameter of the photo site the more susceptible it is to stopping down. The smaller the photo site the smaller the airy disc needs to be to prevent it from spilling onto neighboring sites.
 Logged
NikosR
Sr. Member

Offline

Posts: 622

 « Reply #17 on: January 01, 2009, 03:20:17 AM » Reply

Quote from: vandevanterSH
So does packing more pixels reduce the ability to increase DOF with smaller apertures?

Steve

You have to be careful when comparing across sensors with different pixel pitch. The fact that in sensor A with pixel pitch a using lens L diffraction becomes noticeable at aperture F11 (i.e. F11 provides less real resolution than F8) and in sensor B with pixel pitch b using lens L diffraction becomes noticeable at F8 (i.e. F5.6 appears sharper than F8) does not necessarily mean that sensor A at F11 provides more resolution than sensor B at F11.

The diffraction discussion is only applicable in a staightforward way when we want to find out the aperture wich provides max. resolution in a given sensor / lens combination.

DOF being a highly perceptual phenomenon (although originating due to physical properties of light) is equally complex to discuss across different sensors and lenses. The more sharp the object appears at the plane of focus the more limited the DOF appears to be regardless of what the formulas predict.
 « Last Edit: January 01, 2009, 03:22:05 AM by NikosR » Logged

Nikos
jani
Sr. Member

Offline

Posts: 1604

 « Reply #18 on: January 01, 2009, 10:03:17 AM » Reply

Quote from: iliosgallery
The answer to your question is 'yes'. There's an easy way to visualize this without getting into formulas. Rays of light emanate from a point source on the subject. Their various wavelengths diverge when hitting the edge of the iris and land in a circular pattern on the recording plane, in this instance the sensor.  The smaller the aperture the larger this circular pattern or 'airy disc'. When the diameter of the airy disc is larger than the diameter of a single photo site this visual data spills onto neighboring sites. The end result is that we have photo sites receiving their information as well as their neighbor's data. This overlapping data causes loss of resolution. Obviously, then, the smaller the diameter of the photo site the more susceptible it is to stopping down. The smaller the photo site the smaller the airy disc needs to be to prevent it from spilling onto neighboring sites.
What you're speaking of here is theoretically true under a set of very strict conditions which do not have a real world relevance, as shown by Ray's photographic tests.

You're missing a few important parameters, such as whether the AA filters are "equal", whether the lower-resolution sensor had the optimal resolution, and so on.

If you have images of a scene taken with e.g. a 40D and a 50D, and the former has an apparent greater depth of field than the 50D at the same magnification and viewing distance, then it is most likely because the 40D's image also has visibly less detail within the field.

I'd say that this was a sign that your viewing distance for the 40D's image was too short, or that the magnification was too great, for the subject matter you chose.
 « Last Edit: January 01, 2009, 10:04:16 AM by jani » Logged

Jan
 Pages: [1]   Top of Page