Nikon D800/E Diffraction Limits

Started Apr 28, 2012 | Discussions
Great Bustard
Forum ProPosts: 20,848
Like?
Re: Same system
In reply to Marianne Oelund, May 5, 2012

Marianne Oelund wrote:

Great Bustard wrote:

If we're comparing multiple frames on one system to a single frame on another system, one would ask why take three frames with a D3s and not a few more frames from a compact.

Seriously, if we're talking comparing multiple stitched frames from one system with a single frame from another system,

You consider D3s and D800E to be different systems?

Yes. Same format, same brand, different sensor. Matter of semantics, I'm sure.

Of course, the goal in this example - and others like it - is to reduce the number of frames required. The D800E will be a huge time-saver for some of my work, but it remains true that it will require wider apertures to achieve the same diffraction-limited detail at the subject.

One frame from a D800E suffers no more from diffraction than one frame from a D3s at any given aperture.

Reply   Reply with quote   Complain
Great Bustard
Forum ProPosts: 20,848
Like?
Re: There's a world of difference...
In reply to mattr, May 5, 2012

mattr wrote:

Great Bustard wrote:

...between being able to see the effects of diffraction earlier and saying that a photo suffers from diffraction earlier.

Now you are picking nits.

Those "nits" are central to the matter at hand.

I think you could say "suffers earlier" if you "see the effects earlier" (because you are magnifying more).

No, you really can't. Saying, for example, that a D800 will "suffer ealier" than a D700 has the clear implication that at the same aperture, the D800 photo will be worse.

It is neither Bob nor I who are making statements that cause a lot of confusion --

I never blamed anybody in particular for causing the confusion. But it would be helpful if both approaches are recognized as valid and everybody would point out clearly which approach they are talking about. And BTW, I fully agree that your "image level" approach is in general far more useful and relevant for what most photographers are doing.

The "natural assumption" when comparing systems is that we are comparing photos, not a portion of one photo to a smaller portion of another photo.

I remember numerous endless and fruitless discussions where obviously both sides were talking about different approaches. Calling one approach "nonsense" and "ludicrous" turned out not to lead anywhere in these discussions.

I don't see any utility in saying that the D800 "suffers more"and/or "suffers earlier" from diffraction than a D700. For example, if the sharpest aperture for a particular lens on the D700 is f/5.6, then so it will be with the D800.

Reply   Reply with quote   Complain
mattr
Veteran MemberPosts: 3,443
Like?
Re: There's a world of difference...
In reply to Great Bustard, May 5, 2012

Great Bustard wrote:

mattr wrote:

I think you could say "suffers earlier" if you "see the effects earlier" (because you are magnifying more).

No, you really can't. Saying, for example, that a D800 will "suffer ealier" than a D700 has the clear implication that at the same aperture, the D800 photo will be worse.

"Results from my D800 suffer from diffraction earlier compared to my D700 when I view them at 100% on the screen." Nothing wrong with this statement.

It is neither Bob nor I who are making statements that cause a lot of confusion --

I never blamed anybody in particular for causing the confusion. But it would be helpful if both approaches are recognized as valid and everybody would point out clearly which approach they are talking about. And BTW, I fully agree that your "image level" approach is in general far more useful and relevant for what most photographers are doing.

The "natural assumption" when comparing systems is that we are comparing photos, not a portion of one photo to a smaller portion of another photo.

How about making fewer "natural assumptions"? Maybe it helps avoiding confusion.

Reply   Reply with quote   Complain
Great Bustard
Forum ProPosts: 20,848
Like?
Re: There's a world of difference...
In reply to mattr, May 5, 2012

mattr wrote:

I think you could say "suffers earlier" if you "see the effects earlier" (because you are magnifying more).

No, you really can't. Saying, for example, that a D800 will "suffer ealier" than a D700 has the clear implication that at the same aperture, the D800 photo will be worse.

"Results from my D800 suffer from diffraction earlier compared to my D700 when I view them at 100% on the screen." Nothing wrong with this statement.

But there is. "I notice the diffraction earlier with my D800 than with the D700 when I view the photos at 100%" is more to the point. The photos don't "suffer more" because they still resolve more.

Now that I think about it, better yet would be: "Diffraction cuts into the resolution advantage of my D800 earlier than it did with the D700".

The "natural assumption" when comparing systems is that we are comparing photos, not a portion of one photo to a smaller portion of another photo.

How about making fewer "natural assumptions"? Maybe it helps avoiding confusion.

Because it would make communication more difficult. Just think of the number of "natural assumptions" that are made when talking about comparing sensors, not the least of which is that we are using a lens that is equally sharp and are using a shutter speed high enough that neither motion blur nor camera shake is not an issue at the same light levels.

What are the situations that make spelling out the "natural assumption" of comparing photo to photo each and every time so that people don't make the mistake of thinking we were comparing a smaller portion of one photo to a larger portion of another photo, or three stitched photos from one camera to a single photo from another cameras?

Reply   Reply with quote   Complain
Pierre Sottas
Contributing MemberPosts: 785
Like?
Re: Nikon D800/E Diffraction Limits
In reply to bobn2, May 5, 2012

bobn2 wrote:

but the rule is always the same whatever the camera - 'if you want the maximum resolution find the f-number where the lens peaks and use that'.

yes, but if we enter DOF in the equation this gets more complicated. In macrophotography we often shoot at f16-f32 and at f32 there is no advantage in resolution to use a D3x or D800 over a D3. The decrease in MTF with f-number is higher for the sensors that have smaller pixels and the MTF between the D3x and D3 become equal at f32.

Pierre

Reply   Reply with quote   Complain
Marianne Oelund
Veteran MemberPosts: 4,296Gear list
Like?
FOV change
In reply to Great Bustard, May 5, 2012

Great Bustard wrote:

One frame from a D800E suffers no more from diffraction than one frame from a D3s at any given aperture.

Yes, if both frames cover the same FOV.

No, if the D800E frame covers 3x the subject area; it will need a wider aperture.

  • If it is required that diffraction softening is to be kept constant at the subject in absolute linear measure, i.e., not proportional to FOV.

Reply   Reply with quote   Complain
Great Bustard
Forum ProPosts: 20,848
Like?
Re: FOV change
In reply to Marianne Oelund, May 5, 2012

Marianne Oelund wrote:

Great Bustard wrote:

One frame from a D800E suffers no more from diffraction than one frame from a D3s at any given aperture.

Yes, if both frames cover the same FOV.

Another of my "natural assumptions". Correct me if I'm wrong, but I frame the scene the same when using a camera regardless of the pixel count of the sensor. I guess I represent a minority of photographers?

No, if the D800E frame covers 3x the subject area; it will need a wider aperture.

Which prompts the question as to why we are shooting a scene that covers three times the area with the D800E than we would with the D3s.

  • If it is required that diffraction softening is to be kept constant at the subject in absolute linear measure, i.e., not proportional to FOV.

What type of photography is that, that requires "constant diffraction softening in absolute linear measure"?

Reply   Reply with quote   Complain
Great Bustard
Forum ProPosts: 20,848
Like?
Re: Nikon D800/E Diffraction Limits
In reply to Pierre Sottas, May 5, 2012

Pierre Sottas wrote:

bobn2 wrote:

but the rule is always the same whatever the camera - 'if you want the maximum resolution find the f-number where the lens peaks and use that'.

yes, but if we enter DOF in the equation this gets more complicated. In macrophotography we often shoot at f16-f32 and at f32 there is no advantage in resolution to use a D3x or D800 over a D3. The decrease in MTF with f-number is higher for the sensors that have smaller pixels and the MTF between the D3x and D3 become equal at f32.

There's still a detail advantage with the D800 over D3x, and with a D3x over a D3, but it is severely mitigated to the point where the detail advantage is trivial at such DOFs. However, if you are shooting such photos, it prompts the question as to why you are using FF instead of a smaller format, even a compact.

Reply   Reply with quote   Complain
Mikael Risedal
Senior MemberPosts: 2,927
Like?
Link to raw files D800 D800e color artifact F-8
In reply to Mikael Risedal, May 5, 2012
Reply   Reply with quote   Complain
noirdesir
Forum ProPosts: 10,562
Like?
Technically you are comparing different sized sensors
In reply to Marianne Oelund, May 5, 2012

Marianne Oelund wrote:

PCB documentation example

I have a 4.25" x 7.25" PCB which will be delivered as a prototype unit, and I wish to take a documentation photo of it, which will record detail approaching 1mil. I could use my D3s, and take a set of 3 frames, or I could use the D800E and take just one frame.

The D3s will allow me to stop down to f/11 without significant diffraction softening. For the same degree of diffraction softening - referred to the subject - with the D800E setup, I can only stop down to f/6.3. (Note this also brings DOF for the two cases to about the same.) If I use the D800E at f/11, its result will be noticeably softer than the stitched D3s result.

I think we all have always argued that the f-stop for a given amount of diffraction is a function of the sensor size and thus if you stitch images you are simulating a larger sensor and thus a different f-stop applies there compared to smaller sensors.

Thus your example shows exactly what we have been preaching all along: (1) diffraction for the same DOF is always the same regardless of sensor or pixel size (with the caveat that lower resolution sensors add their own sampling blur) and (2) the f-stop for a given amount of DOF (and thus diffraction) is a function of sensor size only.

Reply   Reply with quote   Complain
noirdesir
Forum ProPosts: 10,562
Like?
Re: There's a world of difference...
In reply to mattr, May 5, 2012

mattr wrote:

Great Bustard wrote:

No, you really can't. Saying, for example, that a D800 will "suffer ealier" than a D700 has the clear implication that at the same aperture, the D800 photo will be worse.

"Results from my D800 suffer from diffraction earlier compared to my D700 when I view them at 100% on the screen." Nothing wrong with this statement.

What you are doing with this is grading on a scale. It is like saying a car with a quit engine suffers more from wind noise than a car with a noisy engine. It is not suffering more, wind noise is simply more noticeable.

There is only one real use case where your statement provides relevant information:

When the question is for which f-stop range (and shutter speed) do I achieve at least 80% of the maximum performance.

Reply   Reply with quote   Complain
mattr
Veteran MemberPosts: 3,443
Like?
Re: There's a world of difference...
In reply to Great Bustard, May 5, 2012

Great Bustard wrote:

mattr wrote:

"Results from my D800 suffer from diffraction earlier compared to my D700 when I view them at 100% on the screen." Nothing wrong with this statement.

But there is. "I notice the diffraction earlier with my D800 than with the D700 when I view the photos at 100%" is more to the point. The photos don't "suffer more" because they still resolve more.

Now that I think about it, better yet would be: "Diffraction cuts into the resolution advantage of my D800 earlier than it did with the D700".

No, you are misunderstanding my statement. "Results" are what I see on the screen at 100%. "Suffer" refers to comparing different f-ratios with the same sensor and observing a loss of contrast with fine detail. "Earlier" means at wider f-ratios (e.g. f/11 with the D800 and f/16 with the D700).

The "natural assumption" when comparing systems is that we are comparing photos, not a portion of one photo to a smaller portion of another photo.

How about making fewer "natural assumptions"? Maybe it helps avoiding confusion.

Because it would make communication more difficult. Just think of the number of "natural assumptions" that are made when talking about comparing sensors, not the least of which is that we are using a lens that is equally sharp and are using a shutter speed high enough that neither motion blur nor camera shake is not an issue at the same light levels.

What are the situations that make spelling out the "natural assumption" of comparing photo to photo each and every time so that people don't make the mistake of thinking we were comparing a smaller portion of one photo to a larger portion of another photo, or three stitched photos from one camera to a single photo from another cameras?

Well, we are going in circles here. I guess I said in this thread already what I had to say.

Reply   Reply with quote   Complain
mattr
Veteran MemberPosts: 3,443
Like?
Re: There's a world of difference...
In reply to noirdesir, May 5, 2012

noirdesir wrote:

What you are doing with this is grading on a scale. It is like saying a car with a quit engine suffers more from wind noise than a car with a noisy engine.

No, that's not like what I'm saying (see above).

It is not suffering more, wind noise is simply more noticeable.

Yes, that's actually more like what I'm saying.

Reply   Reply with quote   Complain
Great Bustard
Forum ProPosts: 20,848
Like?
Re: There's a world of difference...
In reply to mattr, May 5, 2012

mattr wrote:

"Results from my D800 suffer from diffraction earlier compared to my D700 when I view them at 100% on the screen." Nothing wrong with this statement.

But there is. "I notice the diffraction earlier with my D800 than with the D700 when I view the photos at 100%" is more to the point. The photos don't "suffer more" because they still resolve more.

Now that I think about it, better yet would be: "Diffraction cuts into the resolution advantage of my D800 earlier than it did with the D700".

No, you are misunderstanding my statement. "Results" are what I see on the screen at 100%.

Ah. For me, and I suspect most, "results" means the resulting photo they get.

"Suffer" refers to comparing different f-ratios with the same sensor and observing a loss of contrast with fine detail. "Earlier" means at wider f-ratios (e.g. f/11 with the D800 and f/16 with the D700).

"Suffer", to me, means "does worse" for the same photo.

Well, we are going in circles here. I guess I said in this thread already what I had to say.

Sure, sure. I'll close by saying by asking a rhetorical question: what's the point of saying that the D800 "suffers" more from diffraction when it resolves more detail at any given aperture?

Reply   Reply with quote   Complain
Great Bustard
Forum ProPosts: 20,848
Like?
Actually...
In reply to noirdesir, May 5, 2012

noirdesir wrote:

mattr wrote:

Great Bustard wrote:

No, you really can't. Saying, for example, that a D800 will "suffer ealier" than a D700 has the clear implication that at the same aperture, the D800 photo will be worse.

"Results from my D800 suffer from diffraction earlier compared to my D700 when I view them at 100% on the screen." Nothing wrong with this statement.

What you are doing with this is grading on a scale. It is like saying a car with a quite engine suffers more from wind noise than a car with a noisy engine. It is not suffering more, wind noise is simply more noticeable.

...it's more like saying that a Ferrari at top speed suffers from more wind noise than a Civic at top speed.

There is only one real use case where your statement provides relevant information:

When the question is for which f-stop range (and shutter speed) do I achieve at least 80% of the maximum performance.

But the fact remains that if a lens peaks at, say, f/5.6 on a D700, then it will peak at f/5.6 on a D3X, peak at f/5.6 on a D800, and peak at f/5.6 no matter how many pixels the sensor has (so long as it is the same size sensor).

Reply   Reply with quote   Complain
mattr
Veteran MemberPosts: 3,443
Like?
Re: There's a world of difference...
In reply to Great Bustard, May 5, 2012

Great Bustard wrote:

mattr wrote:

"Results from my D800 suffer from diffraction earlier compared to my D700 when I view them at 100% on the screen." Nothing wrong with this statement.

But there is. "I notice the diffraction earlier with my D800 than with the D700 when I view the photos at 100%" is more to the point. The photos don't "suffer more" because they still resolve more.

Now that I think about it, better yet would be: "Diffraction cuts into the resolution advantage of my D800 earlier than it did with the D700".

No, you are misunderstanding my statement. "Results" are what I see on the screen at 100%.

Ah. For me, and I suspect most, "results" means the resulting photo they get.

Why was that not clear when I said "..when I view the results at 100% on the screen"?

"Suffer" refers to comparing different f-ratios with the same sensor and observing a loss of contrast with fine detail. "Earlier" means at wider f-ratios (e.g. f/11 with the D800 and f/16 with the D700).

Sure, sure. I'll close by saying by asking a rhetorical question: what's the point of saying that the D800 "suffers" more from diffraction when it resolves more detail at any given aperture?

The point is that I can see diffraction effects earlier at 100% on the screen... sigh. See, I even answered your rhetorical question..

And BTW, you keep distorting what I'm saying by omitting essential qualifications ("when viewed at 100% on the screen"). I never said "the D800 suffers more from diffraction" without qualification. You keep trying to put this into my mouth. Oh, and I said "earlier" and not "more", which also affects the meaning.

Reply   Reply with quote   Complain
Great Bustard
Forum ProPosts: 20,848
Like?
Very cool.
In reply to DigVis, May 6, 2012

Interesting model. However, as you note, the model presumes a perfectly sharp lens. It would be great if you could make a series of charts to show what happens as the lens deviates more and more from perfect.

DigVis wrote:

jvora wrote:

Has anyone had the opportunity to test the D800/E for diffraction limits - At which f-stop does it begin to show in the images ?

I tried to simulate this using a simplified model.

The resolution is inversely proportional to the "width" of the total point spread function (PSF).

The total point spread function is the result of a convolution of a number components (of sources of blur), major ones being:

  • Atmospheric effects

  • Lens optical "softness"

  • Lens defocus blur

  • Diffraction

  • OLPF (AA-filter)

  • Sensor element integration area

In general, the variance of the total PSF is the sum of the variance of each of these components:

total PSF variance = atmospheric variance + optical blur variance + defocus blur variance + diffraction variance + OLPF variance + integration area variance

The resolution of the sensor is proportional to 1/sqrt(total PSF variance). To simplify, I will disregard everything but sensor resolution, OLPF and diffraction.

The sensor element integration function can be modeled as a box function with width w . The variance becomes w ²/12.

The variance of the diffraction airy disc can be approximated as (0.42λ N )², where N is the ƒ-number, and λ the wavelength (we can use λ=550nm to model green light).

Let's also add an optical low pass filter assuming ±0.4 pixel splitting. The variance becomes 0.16 w ².

The D800 has a sensor element size of w =4.7µm, and the D700 has w =8.4µm. Here is a plot of the resulting proportional resolution for these cameras at various ƒ-stops:

This assumes an ideal lens perfectly sharp at all apertures (such a beast doesn't exist). The worse the lens, the higher the ƒ-stop until a certain proportion of the resolution is lost, and the less the difference between the two cameras.

Speaking about noticeable effects, according to this model, the first stop where 20 % of the resolution is lost due to diffraction is for the D800, ƒ/8, and for the D700, ƒ/16.

Now, how accurate this model is, I don't know.

Reply   Reply with quote   Complain
noirdesir
Forum ProPosts: 10,562
Like?
Re: There's a world of difference...
In reply to mattr, May 6, 2012

mattr wrote:

noirdesir wrote:

What you are doing with this is grading on a scale. It is like saying a car with a quit engine suffers more from wind noise than a car with a noisy engine.

No, that's not like what I'm saying (see above).

Let me expand my analogy: "A car with a more quit engine suffers earlier from wind noise (when increasing the speed of the car) than a car with a louder engine."

Don't you feel that that sentence creates the impression that the wind noise is a bigger problem, in absolute terms, with the car with the quieter engine?

Reply   Reply with quote   Complain
noirdesir
Forum ProPosts: 10,562
Like?
Re: Actually...
In reply to Great Bustard, May 6, 2012

Great Bustard wrote:

There is only one real use case where your statement provides relevant information:

When the question is for which f-stop range (and shutter speed) do I achieve at least 80% of the maximum performance.

But the fact remains that if a lens peaks at, say, f/5.6 on a D700, then it will peak at f/5.6 on a D3X, peak at f/5.6 on a D800, and peak at f/5.6 no matter how many pixels the sensor has (so long as it is the same size sensor).

Yes, but for some people the peak is not the information they want, they want the range where it is close enough to the peak performance. That is a valid question if you look at the behaviour of a single camera (not though when you compare cameras).

Reply   Reply with quote   Complain
mattr
Veteran MemberPosts: 3,443
Like?
Re: There's a world of difference...
In reply to noirdesir, May 6, 2012

noirdesir wrote:

mattr wrote:

noirdesir wrote:

What you are doing with this is grading on a scale. It is like saying a car with a quit engine suffers more from wind noise than a car with a noisy engine.

No, that's not like what I'm saying (see above).

Let me expand my analogy: "A car with a more quit engine suffers earlier from wind noise (when increasing the speed of the car) than a car with a louder engine."

Don't you feel that that sentence creates the impression that the wind noise is a bigger problem, in absolute terms, with the car with the quieter engine?

Actually, no. It just means that I hear wind noise earlier because the engine is quieter.

Reply   Reply with quote   Complain
Keyboard shortcuts:
FForum MMy threads