Data Mining DxO Results - Absolute Camera/Lens Resolution

Started Jan 13, 2014 | Discussions
Scott McMorrow Regular Member • Posts: 366
Data Mining DxO Results - Absolute Camera/Lens Resolution
9

DxO has the largest repository of publicly available lens measurements taken across multiple Nikon camera/sensor systems. If we assume some validity to the data, we can make some interesting conclusions regarding different Cameras and Lens with regards to what DxO calls Perceptual MegaPixel (P-Mpix). Although the numbers reported are measured with reasonable measurement methods, the P-Mpix number is a weighted average that is not documented. If we assume that these numbers are relatively accurate across all the measurement made, we can form some conclusions based on the results.

In this first plot, the P-Mpix results for 54 lens measured with 7 cameras is plotted. D610 and D800 are at the top of the chart resolution, while the D700 and D7000 are down at the bottom. In between lay the D4, D3200, and D7100. Here we see that with the same lens, an upgrade to a higher resolution sensor generally achieves higher overall resolution. Clearly between a D700 and a D800 you obtain a mighty hefty overall resolution increase.

We see that most lenses (from lens 12 out to lens 41) have essentially the same P-Mpix resolution on the same camera, and as the camera is upgraded to a higher resolving sensor, the P-Mpix result increases.

We also see that below lens 41 there are a few "bad actors" that just don't resolve well on any camera system, but, they still do resolve better on a D800 than on a D700.

From lens 1 to 11 we see the truly good lenses. Some might call them "magic" lenses. They stand out from the crowd across all sensors, and have exceptional resolution. Most are not cheap. Only 3 are sub-$1K.

However, there is an interesting result when comparing the D800 and D610 results. Essentially lens 11 through 54 have the same P-Mpix resolution on both cameras. Upgrade from a 24 MP sensor to a 36 MP sensor makes no difference. It is only with the exceptional lenses that we see a difference between these two sensors. Ultimately the top 3 lenses outrun the resolving power of the D610, as seen by the plateau in the chart, yet continue to peak on the D800.

Just for fun, here are the top 10 lenses in these plots, ordered by their D800 resolution results.

  1. Carl Zeiss Distagon T* Otus 1.4/55 ZF.2 Nikon
  2. Carl Zeiss Apo Sonnar T* 2/135 ZF.2. Nikon
  3. Nikon AF-S Nikkor 200mm f/2G ED VR II
  4. Nikon AF-S Nikkor 400mm F2.8G ED VR
  5. Nikon AF-S NIKKOR 300mm f/2.8G ED VR II
  6. Sigma 35mm F1.4 DG HSM A Nikon
  7. Carl Zeiss Distagon T 25mm f/2 ZF.2 Nikon
  8. Nikon AF-S Nikkor 85mm f/1.4G
  9. Nikon AF-S NIKKOR 28mm f/1.8G
  10. Samyang 14mm f/2.8 IF ED UMC Aspherical Nikon

I would caution about drawing too many conclusions from this data. DxO does not publish their methods or their weighting factors. The results are definitely interesting. I'd be interested in other thoughts.

Nikon D3200 Nikon D4 Nikon D610 Nikon D700 Nikon D7000 Nikon D7100 Nikon D800
If you believe there are incorrect tags, please send us this post using our feedback form.
jtra Contributing Member • Posts: 855
Re: Data Mining DxO Results - Absolute Camera/Lens Resolution

Good analysis.

Btw, have you noticed that D600 often scores worse than D610?
D610 probably has weaker AA filter.
See here:
http://www.dpreview.com/forums/thread/3602255

I also guess the resolution of black and white targets is better due to how demosaicing works. So for colored targets there would be more difference in D610 and D800.

Luke Kaven Veteran Member • Posts: 5,698
DxO's "Perceptual Megapixels" is not a hard number or a real limit
2

Scott McMorrow wrote:

DxO has the largest repository of publicly available lens measurements taken across multiple Nikon camera/sensor systems. If we assume some validity to the data, we can make some interesting conclusions regarding different Cameras and Lens with regards to what DxO calls Perceptual MegaPixel (P-Mpix). Although the numbers reported are measured with reasonable measurement methods, the P-Mpix number is a weighted average that is not documented. If we assume that these numbers are relatively accurate across all the measurement made, we can form some conclusions based on the results.

[...]

However, there is an interesting result when comparing the D800 and D610 results. Essentially lens 11 through 54 have the same P-Mpix resolution on both cameras. Upgrade from a 24 MP sensor to a 36 MP sensor makes no difference. It is only with the exceptional lenses that we see a difference between these two sensors. Ultimately the top 3 lenses outrun the resolving power of the D610, as seen by the plateau in the chart, yet continue to peak on the D800.

Thanks for posting this interesting comparison.

One misconception about DxO's Perceptional MPix metric is that it represents a resolution limit.  It does not.  It represents a threshold in the high-frequency roll-off of the MTF.  The roll-off might be more or less gentle or steep, but it is not a hard threshold.  And in fact, there is abundant information coming through the lens beyond the P-Mpix number that could be exploited in digital signal processing.  So an upgrade from 24 to 36MP -- or beyond -- can still make a difference.

(unknown member) Senior Member • Posts: 1,053
Re: Data Mining DxO Results - Absolute Camera/Lens Resolution
4

Ken Rockwell attracts a lot of ire but I do like his remark that an obsession with sharpness indicates a lack of knowledge about photography. I'm sure your chart will engender fits of rage over THE SHARPEST.

-- hide signature --

Wait for the bright flash so I can steal your soul. Smile!

MJSfoto1956 Contributing Member • Posts: 559
Re: Data Mining DxO Results - Absolute Camera/Lens Resolution

seems there is one Nikkor that when used on the D7100 basically matches the D610 or D800! I wonder which particular lens that is? Can you tell us?

Michael

 MJSfoto1956's gear list:MJSfoto1956's gear list
Nikon Coolpix A Nikon D800 Pentax K-3 II Adobe Photoshop CS6 DxO Optics Pro Elite +3 more
OP Scott McMorrow Regular Member • Posts: 366
Re: DxO's "Perceptual Megapixels" is not a hard number or a real limit

Luke, you are correct.  Actually, an interesting comparison would be MTF50/MTF30/MTF10 across multiple lens/camera systems.  This would give us a good range of high frequency spectral data.  What you are inferring is that the slope of the MTF curve would be different. I have the ability to do this, just not the cameras.  I'm stuck with just a D800 for Imatest work.

The interesting thing in the DxO results is the correspondence of D800 and D610 measurements for a wide range of lens.  The question I have is whether the slope of the MTF curves with these lenses is in fact significantly different.

OP Scott McMorrow Regular Member • Posts: 366
Re: Data Mining DxO Results - Absolute Camera/Lens Resolution

MJSfoto1956 wrote:

seems there is one Nikkor that when used on the D7100 basically matches the D610 or D800! I wonder which particular lens that is? Can you tell us?

Michael

Michael, that is actually an interesting question. The lens is the Sigma 20mm F1.8 EX DG ASP RF. This is a case of a wide angle with soft edges and corners that resolves quite well within the DX crop range.

Clearly the DxO P-Mpix algorithm is performing some edge and corner weighting to arrive at the number that is reported. I'm guessing that is the edge/corner weighting that DxO uses that is equalizing the measurements between the D610 and the D800.  To support this theory, the other D7100 outlier at number 10 is the Samyang 14mm, that peaks at 17 P-Mpix on the D7100.

Luke Kaven Veteran Member • Posts: 5,698
Re: DxO's "Perceptual Megapixels" is not a hard number or a real limit
1

Scott McMorrow wrote:

Luke, you are correct. Actually, an interesting comparison would be MTF50/MTF30/MTF10 across multiple lens/camera systems. This would give us a good range of high frequency spectral data. What you are inferring is that the slope of the MTF curve would be different. I have the ability to do this, just not the cameras. I'm stuck with just a D800 for Imatest work.

The interesting thing in the DxO results is the correspondence of D800 and D610 measurements for a wide range of lens. The question I have is whether the slope of the MTF curves with these lenses is in fact significantly different.

Hi Scott,

It is an interesting question as to why the D800/D610 measurements line up with respect to P-Mpix.  I think you might be suggesting that the measurements are dominated by the contributions of the lenses in that category.

I think this might be true.  But the limits imposed by the lenses themselves are soft limits being measured here by hard-edged metrics.  This mistake is compounded by the suggestive use of the word "perceptual".  I'd wager that perceptual benefits of using just those lenses continue to be felt at 36MP and above.

For example:  I have the 28/1.8G, but I also kept and still use my 28/2 AI.  The 28/2 I've used both on a 24MP D3x and a 36MP D800.  To my eye, it delivers more on the D800, and that includes some outright pixel-level detail (e.g., a strand of hair 20 feet away).  The lens clearly emphasizes spatial middle-frequencies, and emphasizes less in the high frequencies.  This is part of its desirable character.  But whatever the P-Mpix numbers might suggest, those fine-grained details are still -- pretty much -- all there.

Hasa Contributing Member • Posts: 685
Re: Data Mining DxO Results - Absolute Camera/Lens Resolution

Great overview, thank you for your hard work!

Did you leave out the zooms because they may perform differently at different focal lenghts, or?

One of the reasons I got the Nikkor AF-S NIKKOR 70-200mm f/4G ED VR was the analysis on DxO (21Pmpix on D800), Photozone.de and elsewhere: never a bad word about this lens.

-- hide signature --

Smile and the world smiles back!

 Hasa's gear list:Hasa's gear list
Nikon D800 Nikon AF Nikkor 20mm f/2.8D Nikon AF-S Teleconverter TC-17E II Sigma 15mm F2.8 EX DG Diagonal Fisheye Carl Zeiss Distagon T* 2/35 +16 more
OP Scott McMorrow Regular Member • Posts: 366
Re: Data Mining DxO Results - Absolute Camera/Lens Resolution

Did you leave out the zooms because they may perform differently at different focal lenghts, or?

That is correct.  I figured that measurements on prime lenses would be reasonably consistent across multiple sensor.  Zooms add another dimension of variability that is not controlled, since DxO does not publish P-Mpix numbers for each focal length.

MJSfoto1956 Contributing Member • Posts: 559
Re: Data Mining DxO Results - Absolute Camera/Lens Resolution

Scott McMorrow wrote:

MJSfoto1956 wrote:

seems there is one Nikkor that when used on the D7100 basically matches the D610 or D800! I wonder which particular lens that is? Can you tell us?

Michael

Michael, that is actually an interesting question. The lens is the Sigma 20mm F1.8 EX DG ASP RF. This is a case of a wide angle with soft edges and corners that resolves quite well within the DX crop range.

uggh, not a lens I enjoyed shooting with -- hands down had the worse coma of any lens I have ever measured. I would say this is a great portrait lens (think street photography) but worthless for astrophotography or anything with specula's. Thanks for compiling this -- lots of manual search, manual data entry I can tell.

Michael

 MJSfoto1956's gear list:MJSfoto1956's gear list
Nikon Coolpix A Nikon D800 Pentax K-3 II Adobe Photoshop CS6 DxO Optics Pro Elite +3 more
OP Scott McMorrow Regular Member • Posts: 366
Re: Data Mining DxO Results - Absolute Camera/Lens Resolution

uggh, not a lens I enjoyed shooting with -- hands down had the worse coma of any lens I have ever measured. I would say this is a great portrait lens (think street photography) but worthless for astrophotography or anything with specula's. Thanks for compiling this -- lots of manual search, manual data entry I can tell.

Michael

Not my favorite either, Michael.  But surprising the resolution it has in the center stopped down a bit. I had pretty much discarded the Sigma 20 for quite a while after I started shooting a D700 and then a D800.  Accidentally placed it in my bag on one outing and then saw the results.

The compilation was not too hard.  Cut and paste from a DxO search into Excel for each camera.  Then some cleanup, additional cut and paste, manipulation, and plotting.

OP Scott McMorrow Regular Member • Posts: 366
Re: DxO's "Perceptual Megapixels" is not a hard number or a real limit
4

Luke Kaven wrote:

It is an interesting question as to why the D800/D610 measurements line up with respect to P-Mpix. I think you might be suggesting that the measurements are dominated by the contributions of the lenses in that category.

I think this might be true. But the limits imposed by the lenses themselves are soft limits being measured here by hard-edged metrics. This mistake is compounded by the suggestive use of the word "perceptual". I'd wager that perceptual benefits of using just those lenses continue to be felt at 36MP and above.

For example: I have the 28/1.8G, but I also kept and still use my 28/2 AI. The 28/2 I've used both on a 24MP D3x and a 36MP D800. To my eye, it delivers more on the D800, and that includes some outright pixel-level detail (e.g., a strand of hair 20 feet away). The lens clearly emphasizes spatial middle-frequencies, and emphasizes less in the high frequencies. This is part of its desirable character. But whatever the P-Mpix numbers might suggest, those fine-grained details are still -- pretty much -- all there.

Luke

I'm with you. Part of the problem is that the DxO weighting obviously averages across the entire frame, which is why some lenses show better relative numbers on DX cameras. The D7100 measurements are quite interesting, and get to some of what you are say. If we do a little hand waving and normalize the P-Mpix numbers from the D7100 to average lines/mm, we actually get some pretty incredible average spatial resolution numbers. If we do it for all the Cameras, we get the figure below.

The high frequency spatial content is there in the DX frame. If we take even the 58mm f/1.4G on the D7100, we still see 181 lines/mm of average resolution. With a little manipulation, this comes out to 20.7 P-Mpix on a D800. The P-Mpix reported are 18 and 16 for the D800 and D610 respectively. So actually this is not too far off. A little bit of edge and corner lower resolving average could account for this. But there's still more gas left in this lens. If we adjust the D7100 result for a 48MP FF sensor, the overall resolution would be 27.6 MP.

I think several things are happening. First, this fairly large cluster of lenses have essentially the same raw resolution. This is evident since the spread of measured resolution is about the same for the D700, D4, D610, an D800. There is not much to distinguish any of them on the resolution front. .

Second, I think the D610 has a weaker AA-filter than the D800, which is serving to equalize the results. Measurements from a D800e would show a difference.

Third, the D7100 resolution plots seem to show that the "average lens" resolves about 180 to 200 lines/mm on the sensor.  For this to occur, the optical lens resolution (if my calculation is correct) is 250 to 320 lines/mm.  Lets say an average of 300 lines/mm.  if placed on a camera with a 100 MP sensor without AA-Filter, the final system resolution would be 71 MP, at least in the center of the frame.

The Zeiss APO lenses are just plain in another league.  They appear to be essentially transparent to even a 100 MP sensor.  Just the D7100 results show that the Otus will resolve around 48MP on a 54MP AA-less sensor, and the Zeiss 135 APO f/2 will resolve around 46MP.

As a result, I can say unequivocally (as much as it pains me) that Reilly Diefenbach is right.  Even the lowly 50mm f/1.8G will continue to increase in photographic resolution with further increases in sensor resolution.  The "magic" lenses will just outpace it (and the other "average" lenses) by about 2:1, when flatness of field is taken into account.

Oh, and here is the secret lens decoder ring for those that want to look up a particular prime lens on either chart.

1 Carl Zeiss Distagon T* Otus 1.4/55 ZF.2 Nikon

2 Carl Zeiss Apo Sonnar T* 2/135 ZF.2 Nikon

3 Nikon AF-S Nikkor 200mm f/2G ED VR II

4 Nikon AF-S Nikkor 400mm F2.8G ED VR

5 Nikon AF-S NIKKOR 300mm f/2.8G ED VR II

6 Sigma 35mm F1.4 DG HSM A Nikon

7 Carl Zeiss Distagon T 25mm f/2 ZF.2 Nikon

8 Nikon AF-S Nikkor 85mm f/1.4G

9 Nikon AF-S NIKKOR 28mm f/1.8G

10 Samyang 14mm f/2.8 IF ED UMC Aspherical Nikon

11 Nikon AF Micro-Nikkor 200mm f/4D ED-IF

12 Nikon AF-S NIKKOR 85mm f/1.8G

13 Tamron SP 90mm F/2.8 Di MACRO 1:1 VC USD Nikon

14 Tokina AT-X M100 AF PRO D AF 100mm f/2.8 Nikon

15 Carl Zeiss Distagon T 28mm f/2 ZF2 Nikon

16 Nikon AF Micro-Nikkor 60mm f/2.8D

17 Nikon AF-S Nikkor 58mm f/1.4G

18 Sigma 85mm F1.4 EX DG HSM Nikon

19 Carl Zeiss Distagon T 35mm f/1.4 ZF2 Nikon

20 Carl Zeiss Distagon T 35mm f/2 ZF2 Nikon

21 Carl Zeiss Distagon T* 2.8/15 ZF.2 Nikon

22 Carl Zeiss Makro-Planar T 100mm f/2 ZF2 Nikon

23 Carl Zeiss Makro-Planar T 50mm f/2 ZF2 Nikon

24 Carl Zeiss Planar T 85mm f/1.4 ZF2 Nikon

25 Nikon AF DC-Nikkor 105mm f/2D

26 Nikon AF Nikkor 50mm f/1.8D

27 Nikon AF Nikkor 85mm f/1.8D

28 Nikon AF-S Nikkor 24mm f/1.4G ED

29 Nikon AF-S NIKKOR 35mm f/1.4G

30 Samyang 85mm f/1.4 Aspherique IF Nikon

31 Sigma 105mm F2.8 EX DG Macro Nikon

32 Sigma 105mm F2.8 EX DG OS HSM Nikon

33 Sigma 150mm F2.8 EX DG OS HSM APO Macro Nikon

34 Sigma 20mm F1.8 EX DG ASP RF Nikon

35 Sigma 28mm F1.8 EX DG ASP Macro Nikon

36 Nikon AF Nikkor 35mm f/2D

37 Nikon AF Nikkor 50mm f/1.4D

38 Nikon AF Nikkor 85mm f/1.4D IF

39 Nikon AF-S Nikkor 50mm f/1.4G

40 Nikon AF-S NIKKOR 50mm f/1.8G

41 Nikon AF-S VR Micro-Nikkor 105mm f/2.8G IF-ED

42 Sigma 50mm F1.4 EX DG HSM Nikon

43 Carl Zeiss Distagon T 25mm f/2.8 ZF2 Nikon

44 Nikon AF Nikkor 20mm f/2.8D

45 Nikon AF Nikkor 28mm f/2.8D

46 Samyang 35mm F1.4 AS UMC Nikon

47 Carl Zeiss Distagon T 21mm f/2.8 ZF2 Nikon

48 Tamron SP AF 90mm F/2.8 Di MACRO Nikon

49 Carl Zeiss Distagon T 18mm f/3.5 ZF2 Nikon

50 Carl Zeiss Planar T 50mm f/1.4 ZF2 Nikon

51 Nikon AF Nikkor 14mm f/2.8D ED

52 Sigma 70mm F2.8 EX DG Macro Nikon

53 Nikon AF Nikkor 180mm f/2.8D IF-ED

54 Sigma 14mm F2.8 EX Aspherical HSM Nikon

Flashlight Veteran Member • Posts: 6,778
Re: DxO's "Perceptual Megapixels" is not a hard number or a real limit

Scott McMorrow wrote:

As a result, I can say unequivocally (as much as it pains me) that Reilly Diefenbach is right. Even the lowly 50mm f/1.8G will continue to increase in photographic resolution with further increases in sensor resolution.

This is a 100% crop of an image I made with a 50mm f/1.4D @f/4 and the Raspberry Pi Camera, which is a tiny (cropfactor=9.8x) sensor with a 5MP resolution. Next to it is the size a comparable shot with a D800 would look like @ 100% (click image for full resolution):

-- hide signature --

Philip

Reilly Diefenbach
Reilly Diefenbach Forum Pro • Posts: 11,257
DxO's "Perceptual Megapixels"
3

One of the many pleasures of shooting a D800e is that almost regardless of lens, it makes a picture which is at long last sharp enough and clean enough.  No eye/brain strain is required to attempt to piece together fine foliage detail, etc., which was certainly the case with lesser cameras I've owned.  Effortless, airy rendition almost free of crunchy artifacts and noise is the order of the day.  Now I can relax and point the camera in the right direction and get the results for which I've been hoping for over half a century shooting everything from a Brownie to 4X5 and pretty much everything in between.

Fine lenses work better than average ones, so stipulated.  I've got a few.

My main complaint with all this lens twiddling is that the upper end of lens design and production has been completely taken over by the one eye in focus brigade.  DXO follows along rating lenses high or low at max ap, where I never shoot.  What wouldn't I pay for a small, light f2.8 35mm with deep dish color and highly flare resistant nano and top shelf optics a la 60G Micro? Or a 14-24 without the giant blob prone front element?  The manufacturers are certainly making bank on the wafer-thin dof craze, so I'm not holding my breath.

Johan1967 Regular Member • Posts: 383
Re: DxO's "Perceptual Megapixels" is not a hard number or a real limit
6

Very funny that the Carl Zeiss Distagon T 21mm f/2.8 ZF2 Nikon is on place 47. It is perhaps the sharpest lens that I have. And the other lenses I have are all mentioned in the first 25.

Something tells me that these ratings are BS.

Luke Kaven Veteran Member • Posts: 5,698
Re: DxO's "Perceptual Megapixels" is not a hard number or a real limit
1

I agree, as you noted, flatness of field is not taken into account in these full-frame measurements.  The DPR reviewer noted the 58/1.4g "noct nova" had significant field curvature, but that the corners were perfectly sharp otherwise.  The P-Mpix metric does not appear to reflect this.

Considering the uses of a lens such as the 58/1.4g, favored frequently at f/1.4-f/2.8, one would expect only a very small portion of the frame to be in focus at all.

With your contribution, these numbers are made more meaningful.  My major complaint again is in the use of the word "perceptual" in this suggestively-named metric: this is not a thesis about perception, and it does not purport to measure limits on perceptual performance or competency.

Weegee
Weegee Senior Member • Posts: 1,652
I hardly ever use very limited depth of field.

I usually need all the depth of field I can get. That's why I used to use a 4 x 5 camera ( and 8 x 10 ) with swings and tilts. Most commercial photography ( yes, the making money kind ) is about maximum sharpness across the board.

Since I do alot of stock photography, I shoot mostly with my Olympus OMD-E5 because it's easier to get good depth of field. So I'm with Reilly on that.

An example above.

Luke Kaven Veteran Member • Posts: 5,698
Re: DxO's "Perceptual Megapixels"
2

Reilly Diefenbach wrote:

One of the many pleasures of shooting a D800e is that almost regardless of lens, it makes a picture which is at long last sharp enough and clean enough. No eye/brain strain is required to attempt to piece together fine foliage detail, etc., which was certainly the case with lesser cameras I've owned. Effortless, airy rendition almost free of crunchy artifacts and noise is the order of the day. Now I can relax and point the camera in the right direction and get the results for which I've been hoping for over half a century shooting everything from a Brownie to 4X5 and pretty much everything in between.

Fine lenses work better than average ones, so stipulated. I've got a few.

My main complaint with all this lens twiddling is that the upper end of lens design and production has been completely taken over by the one eye in focus brigade. DXO follows along rating lenses high or low at max ap, where I never shoot. What wouldn't I pay for a small, light f2.8 35mm with deep dish color and highly flare resistant nano and top shelf optics a la 60G Micro? Or a 14-24 without the giant blob prone front element? The manufacturers are certainly making bank on the wafer-thin dof craze, so I'm not holding my breath.

Remember the war of the "faster than f/2" standard 50mm lenses in the 1970s?  f/1.9, f/1.8, and wow, f/1.7.  Sooo tantalizingly close to f/1.4!  An f/1.4 lens cost seemingly as much as an engagement ring.

But I loved just walking the streets with a Nikon F and 50 f/2, straight prism finder, which I could carry in one palm with my thumb and fingers wrapped around the lens mount.  A small 35/2.8, or 24/2.8 was perfect.  I've never held a camera since that had that kind of working feel.

OP Scott McMorrow Regular Member • Posts: 366
Re: DxO's "Perceptual Megapixels" is not a hard number or a real limit
4

There is a problem in the lens test community.  No one (except Roger Cicala) has published their methods fully.  Most use Imatest, which has quite a few features, and many different ways to determine and publish results. It is extremely accurate, extremely sensitive, but needs some engineering discipline to set up properly documented testing methodologies.

DxO are the other guys, and have the most extensive set of measurements across a wide range of lenses and camera systems.  I have no doubt that their actual measurements are good.  But the they way they have post processed them and dumbed them down for the average consumer has complicated good comparisons.

I need to get some time myself, away from my normal engineering consulting, to rent a D610, a D7100, a D800E, and compare them to my D800.  I have the full Imatest suite, along with their largest test chart, which is suitable for D800 measurements.  I've already correlated the raw unsharpened measurements to Roger Cicala at Lens Rental. (He was kind enough to allow me to check some of my assumptions with him).  I've also used it to dial in optimal sharpening with Lightroom. Similar to what the guys at Falk Lumo did in their paper on the D800 AA-filter.

Keyboard shortcuts:
FForum MMy threads