Modern Sensors Missing half the AA: in the Horizontal or the Vertical direction?

Started Jan 14, 2014 | Discussions
Jack Hogan Veteran Member • Posts: 7,671
Modern Sensors Missing half the AA: in the Horizontal or the Vertical direction?
3

An article pointed out by jtra (use google translate) confirmed the unusual Vertical/Horizontal spread in MTF50 spatial resolution measurements that I had noticed a few months ago in recent sensors. I am now wondering whether the latest generation of non-AAless sensors does not simply have weaker Anti Aliasing filters - but whether they have actually done away with 1/2 of them. Specifically in one direction only. What do I mean? Read on.

Here is a graph of MTF50 readings by MTF Mapper off the slanted edges in DPR's new studio scene for the Nikon D610 and Sony A7:

Recent FF Sensors with AA - Why such a large spread in Horizontal and Vertical MTF50 Spatial Resolution readings?

The A7 graph (same vintage sensor by the same manufacturer) looks very similar although curiously the vertical and horizontal lines are switched. On the other hand the AAless Sony Rx1R (similar sensor by the same manufacturer) vertical and horizontal readings overlay each other as expected around the 2500 lw/ph level.

FF Sensor without AA: No material Spread in Horizontal and Vertical MTF50 Spatial Resolution readings as expected

But here's the interesting thing: Allowing a small margin for lens/operator differences the two sensors with an AA filter (D610 and A7) show similar MTF50 readings as their AAless version (RX1R), albeit only in one direction (vertical in one and horizontal in the other). Could it be that in this new generation of sensors the AA was eliminated in one direction only - leaving the other intact?

It would certainly be easy to accomplish this because most modern AAs perform their function by splitting the incoming light four ways: first two-way horizontally followed by two way vertically through a so-called 4-dot beam splitter made of subsequent lithium niobate plates. Here is how Nikon puts it:

It would therefore be easy to remove entirely (or replace with a neutral plate) one of the two Low-Pass filter plates shown transforming the 4-dot beam splitter into just a 2-dot - effectively removing the anti-aliasing action in one direction only. This would explain why the D610 and A7's readings match the RX1R's in one direction only. In this scenario the D610 would have had the horizontal split removed while the A7 the vertical, also explaining the curious switch of the vertical and horizontal lines seen in the two graphs.

Here is another example where this might be happening: the Fujifilm XA1 with AA versus the XM1 without: the higher horizontal readings of the XA1 are again close to the values of the AAless XM1*. Looks like the vertical plate was removed/replaced here.

So it is possible that cameras like the D610, the A7 and the XA-1 come with only half an AA. The absent (or lower) AA strength in one direction apparently results in better spatial resolution in that direction only - but at the usual cost of increased artifacts in that direction, as shown in the article referenced above.

Thoughts on the above and related implications? Should owners of such cameras choose to shoot a scene in horizontal vs vertical orientation according to the direction of the detail contained therein?

Jack

* Excluding operator error both cameras' readings appear to be affected by shutter shock up to and including 1/250s, not an uncommon occurrence in ligher bodies with mechanical shutters.

Fujifilm X-A1 Fujifilm X-M1 Nikon D610 Sony a7 Sony RX1R
If you believe there are incorrect tags, please send us this post using our feedback form.
David Rosser
David Rosser Veteran Member • Posts: 3,491
Re: Modern Sensors Missing half the AA: in the Horizontal or the Vertical direction?

Very interesting but there is a lens in ther equation and won't this effect the results?.  Unless the readings are taken dead centre of the field Sagittal and Tangential MTF50 will be different

-- hide signature --
 David Rosser's gear list:David Rosser's gear list
Panasonic Lumix DMC-LC5 Sony RX100 III Nikon D40 Fujifilm X-Pro1 Fujifilm X-Pro2 +10 more
OP Jack Hogan Veteran Member • Posts: 7,671
Re: Modern Sensors Missing half the AA: in the Horizontal or the Vertical direction?

Good point: lens, microlens, filters, 'Fill Factor', plate alignment and other imperfections would all suggest some residual directionality in spatial resolution measurements - although a 20% difference?

And why would the outlandishly large difference practically get back to normal when read off an AAless sensor with presumably the same lens (as in the XA1 vs XM1 charts above)? For reference here are also the D4 and D800 with (presumably) the same Nikkor 85mm:1.8G used on the D610.

Regular Strength AA on 7.8 micron photosites

Weaker AA on 4.8 micron photosites

David Rosser
David Rosser Veteran Member • Posts: 3,491
Re: Modern Sensors Missing half the AA: in the Horizontal or the Vertical direction?

Jack Hogan wrote:

Good point: lens, microlens, filters, 'Fill Factor', plate alignment and other imperfections would all suggest some residual directionality in spatial resolution measurements - although a 20% difference?

I have in front of me EROS 200 Lens Test output for my 55mm f/3.5 micro Nikkor.  Back in the 1970s R.G.Lewis Ltd. had an EROS 200 lens testing setup and tested all the lenses they sold on this optical bench.  The test results for my lens were taken at f/8 at an angle of 15deg off axis. The readings of spatial frequency at MTF50 are

radial (sagittal) 48 cycles/mm

tangential       61 cycles/mm

Something over 20% difference here.

For reference the on axis MTF50 rerading is 72 cycles/mm.  Remember these readings are for the lens alone no film or digital sensor involved.

-- hide signature --
 David Rosser's gear list:David Rosser's gear list
Panasonic Lumix DMC-LC5 Sony RX100 III Nikon D40 Fujifilm X-Pro1 Fujifilm X-Pro2 +10 more
OP Jack Hogan Veteran Member • Posts: 7,671
Re: Modern Sensors Missing half the AA: in the Horizontal or the Vertical direction?

David Rosser wrote:

Jack Hogan wrote:

Good point: lens, microlens, filters, 'Fill Factor', plate alignment and other imperfections would all suggest some residual directionality in spatial resolution measurements - although a 20% difference?

I have in front of me EROS 200 Lens Test output for my 55mm f/3.5 micro Nikkor. Back in the 1970s R.G.Lewis Ltd. had an EROS 200 lens testing setup and tested all the lenses they sold on this optical bench. The test results for my lens were taken at f/8 at an angle of 15deg off axis. The readings of spatial frequency at MTF50 are

radial (sagittal) 48 cycles/mm

tangential 61 cycles/mm

Something over 20% difference here.

For reference the on axis MTF50 rerading is 72 cycles/mm. Remember these readings are for the lens alone no film or digital sensor involved.

Cool, I did not realize the differences were so marked.  Although I don't think this applies in our context because of my previous post and because the images were taken on-axis with fairly central, tangential readings in both directions:

Crop of center of capture showing the location of the MTF50 horizontal and vertical slanted edge readings

Scott McMorrow Regular Member • Posts: 366
Re: Modern Sensors Missing half the AA: in the Horizontal or the Vertical direction?

There are several other measurement error possibilities that I can think of.

1) The number of pixels sampled by the slanted edges could be too small and result in resolution error

2) the illumination in the two directions could be different

3) the camera sensor plane and test card plane may not be totally parallel planar, resulting in poor focus for one direction over the other.

OP Jack Hogan Veteran Member • Posts: 7,671
Re: Modern Sensors Missing half the AA: in the Horizontal or the Vertical direction?

Scott McMorrow wrote:

There are several other measurement error possibilities that I can think of.

Hi Scott, also good points.

1) The number of pixels sampled by the slanted edges could be too small and result in resolution error

MTF Mapper needs between 50 and 100 pixels to do its job well - but it takes into consideration up to 400. The slanted edges fed to it are always more than 400 pixels long.

2) the illumination in the two directions could be different

Polarized, I assume you mean. I don't think it is nor that it should matter: DPR says that they use calibrated lights for this purpose. I also wonder whether polarization would survive diffuse reflection and even then whether it would make a difference on the MTF 50 readings - which do not depend on light intensity (within limits of course)

3) the camera sensor plane and test card plane may not be totally parallel planar, resulting in poor focus for one direction over the other.

Yes, operator error could result in some areas being in better focus than others. Although I think that it would be unusually coincidental to have this error happen only when the sensors with AA are being measured and not when the AAless ones are. And also that when an unusually large spread is observed in the H/V readings of a sensor with AA, the unusually high reading is very similar to that of the same/similar sensor without the AA.

Which together with the article pointed out by jtra is what got me thinking this way in the first place. Earlier I suspected an ovally shaped 2D filtering action in the newer sensors as a means to weaken the AAs. Now I suspect it is linear

xpatUSA
xpatUSA Forum Pro • Posts: 18,917
Re: Modern Sensors Missing half the AA: in the Horizontal or the Vertical direction?

Irresistible!

Here's a cross-check with QuickMTF for y'alls consideration. I downloaded the RAW DSC_0162.NEF. Opened it with FastStone Viewer (my versions of ACR and DCraw didn't like it). Cropped and saved as 8-bit TIFF. Opened with QuickMTF:

Horizontal MTF:

Vertical MTF:

MTF's different but not greatly so.

Horizontal edge:

Vertical edge:

The horizontal shows slightly less sharp than the vertical.

-- hide signature --

"Engage Brain before operating Keyboard!"
Ted

 xpatUSA's gear list:xpatUSA's gear list
Panasonic Lumix DMC-LX1 Sigma SD9 Sigma SD15 Panasonic Lumix DC-G9 Panasonic Leica DG Macro-Elmarit 45mm F2.8 ASPH OIS +14 more
Allan Olesen Veteran Member • Posts: 3,391
Re: Modern Sensors Missing half the AA: in the Horizontal or the Vertical direction?

Scott McMorrow wrote:

There are several other measurement error possibilities that I can think of.

1) The number of pixels sampled by the slanted edges could be too small and result in resolution error

2) the illumination in the two directions could be different

3) the camera sensor plane and test card plane may not be totally parallel planar, resulting in poor focus for one direction over the other.

All of this can be tested by taking a portrait shot and a landscape shot of the same test target.

Then you will now if the increased resolution in one direction is caused by the scene or the camera.

Allan Olesen Veteran Member • Posts: 3,391
Re: Modern Sensors Missing half the AA: in the Horizontal or the Vertical direction?

David Rosser wrote:

Very interesting but there is a lens in ther equation and won't this effect the results?. Unless the readings are taken dead centre of the field Sagittal and Tangential MTF50 will be different

If you for example use the test targets generated by MTF Mapper, it will not matter.

These targets have a lot of slanted edges with varying orientation all over the paper. So you get to test Sagittal and Tangential MTF all over the surface of the sensor.

If the difference in horisontal and vertical MTF on the sensor is simply caused by the lens having different Sagittal and Tangential MTF, you will get the same Sagittal and Tangential MTF numbers at the same distance from the centre of the sensor, no matter if you look above, below, to the right or to the left of the centre.

But if the difference is truly caused by the sensor having different horisontal and vertical resolution, you will see the Sagittal and Tangential MTF values below and above the sensor centre being different from the values to the right and left of the sensor centre.

(You should of course test with the camera in both landscape and portrait orientation to make certain that there is not a difference in the scene which causes the observations.)

xpatUSA
xpatUSA Forum Pro • Posts: 18,917
Re: Modern Sensors Missing half the AA: in the Horizontal or the Vertical direction?

Did anybody look at the Siemens stars?

Seems like there's more color artifacts vertically than horizontally in the D610 raw image.

We 'foveonistas' call it Bayeritis - just kidding, just kidding . . .

-- hide signature --

"Engage Brain before operating Keyboard!"
Ted

 xpatUSA's gear list:xpatUSA's gear list
Panasonic Lumix DMC-LX1 Sigma SD9 Sigma SD15 Panasonic Lumix DC-G9 Panasonic Leica DG Macro-Elmarit 45mm F2.8 ASPH OIS +14 more
fvdbergh2501 Contributing Member • Posts: 582
Re: Modern Sensors Missing half the AA: in the Horizontal or the Vertical direction?

Jack Hogan wrote:

Scott McMorrow wrote:

2) the illumination in the two directions could be different

Polarized, I assume you mean. I don't think it is nor that it should matter: DPR says that they use calibrated lights for this purpose. I also wonder whether polarization would survive diffuse reflection and even then whether it would make a difference on the MTF 50 readings - which do not depend on light intensity (within limits of course)

I think what Scott meant was not polarization, but rather uneven illumination, something like a vertical illumination gradient. This will definitely affect the slanted edge method, but my gut feeling is that a gradient large enough to effect a 20% H/V difference would be visible to the naked eye.

I will see if I can play around with a few of these images to see if there is any evidence of minor gradients.

fvdbergh2501 Contributing Member • Posts: 582
Re: Modern Sensors Missing half the AA: in the Horizontal or the Vertical direction?

xpatUSA wrote:


MTF's different but not greatly so.

The horizontal shows slightly less sharp than the vertical.

Thanks for cross-checking with QuickMTF. You appear to be seeing a comparable horizontal/vertical difference of >20%.

OP Jack Hogan Veteran Member • Posts: 7,671
Double Check MTF50 Procedure

xpatUSA wrote:

Irresistible!

Here's a cross-check with QuickMTF for y'alls consideration. I downloaded the RAW DSC_0162.NEF. Opened it with FastStone Viewer (my versions of ACR and DCraw didn't like it). Cropped and saved as 8-bit TIFF. Opened with QuickMTF:

Hey Ted, excellent cross check. To compare undemosaiced/unsharpened apple-to-apple results for the purpose of this thread I would use dcraw 9.19 on file DSC_0171.NEF (which at ISO800 is still clean enough and at 1/500s is away from shock territory) as follows:

1) For simplicity put the dcraw.exe executable in the same directory as the NEF
2) Open the Windows Command Line Prompt and navigate to that directory
3) Once there enter the following command line: dcraw -d -4 -T -w DSC_0171.NEF

This will generate an untagged 16-bit TIFF copy of the virgin raw data - after just zero subtraction (N/A for the D610) and white balance. Now open the file DSC_0171.tiff with photoshop or other well behaved editor (not all will open an untagged grayscale image), crop as shown in the earlier post and save the two edges separately as 16-bit unlayered TIFFs: h.tif and v.tif. Feed them to QuickMTF and compare. For that file I get the following MTF50 readings:

Horizontal = 0.249717 cycles/pixel
Vertical = 0.312464 cycles/pixel

Let me know what you get.

Horizontal MTF:

Vertical MTF:

It is interesting to see the D610 Imatest graphs in the 'Screen Resolution' section of the opticzne.pl articlementioned in the OP (as translated by google). They used a better lens. But given the massive difference shown between V and H readings I'd be curious to know the shutter speed of the relative NEF before checking other potential culprits. I would guess around 1/30-1/250s, possibly 1/60th?

Jack

fvdbergh2501 Contributing Member • Posts: 582
Re: Modern Sensors Missing half the AA: in the Horizontal or the Vertical direction?
2

I quickly grabbed the D610's ISO 100 raw (DSC_0162.NEF, same one that xpat used above).

I performed a quick-and-dirty analysis on the un-demosaiced raw file straight from dcraw (i.e., dcraw dcraw -D -T -4 -c DSC_0162.NEF > DSC_0162.tif). Then I clipped out the two regions using the GDAL tools (grab the FWTools package after googling it), using the following commands:

gdal_translate.exe -srcwin 3450 1680 228 660 DSC_0162.tif horizontal.tif

gdal_translate.exe -srcwin 2700 2400 660 228 DSC_0162.tif vertical.tif

Next, I passed them through MTF Mapper 0.4.16 using the command:

mtf_mapper.exe horizontal.tif hr -arefb --bayer green

This extracted only the green photosites, completely ignoring red and blue, so wavelength-depended diffraction is avoided. (I did check the red and blue channels too, and focus was definitely optimized for green).

Got about 0.2356 cycles per pixel for MTF50 in the vertical direction, and 0.345 cycles per pixel in the horizontal direction (note about conventions: horizontal means the edge is vertical, so this seems to be opposite to other posts in this thread). Without looking any further, I can already tell you that 0.345 cycles per pixel indicates a lack of an OLPF in that direction (or some manipulation of the data before the raw was saved in the camera).

Here are the MTF curves:

Note the lp/mm horizontal scale. Horizontal MTF50=58.4 lp/mm, vertical MTF50=39.9 lp/mm.

And the PSF curves:

There is a slight difference in the white level between the horizontal and vertical targets, but only about 4% (ratio of median ESF on the "right end"):

So I do not think there is a significant illumination gradient here.

Incidentally, the overall resolution is somewhat low. Simulated results at f/5.6 indicates that we could see values of up to 62 lp/mm with an OLPF, and about 86.3 lp/mm without an OLPF.. The 62 lp/mm is consistent with what I saw on real photos captured with a  D7000 (4.73 micron photosite pitch, vs ~5.09 micron pitch on the D610). I will look into the D610 images a bit more to seek possible explanations for the lower-than-expected MTF50 values (in the vertical direction, specifically).

fvdbergh2501 Contributing Member • Posts: 582
Re: Modern Sensors Missing half the AA: in the Horizontal or the Vertical direction?

I re-ran the experiment from my previous post on the ISO800 D610 sample image:

The MTF50 values are now 58.3 lp/mm and 44.4 lp/mm for horizontal and vertical resolution, respectively. (or 0.344 and 0.262 cycles per pixel).

This seems to close the gap between horizontal and vertical resolution somewhat, giving us a difference of about 31%. It still looks like an OLPF that blurs only in the vertical direction.

OP Jack Hogan Veteran Member • Posts: 7,671
Illumination Gradients and Edge Spread Functions

fvdbergh2501 wrote:

Jack Hogan wrote:

Scott McMorrow wrote:

2) the illumination in the two directions could be different

Polarized, I assume you mean. I don't think it is nor that it should matter: DPR says that they use calibrated lights for this purpose. I also wonder whether polarization would survive diffuse reflection and even then whether it would make a difference on the MTF 50 readings - which do not depend on light intensity (within limits of course)

I think what Scott meant was not polarization, but rather uneven illumination, something like a vertical illumination gradient. This will definitely affect the slanted edge method, but my gut feeling is that a gradient large enough to effect a 20% H/V difference would be visible to the naked eye.

Hi Frans , thanks for chiming in. Would the effect of a a vertical light gradient on the vertical edge be limited to effectively weigh more the contribution of pixels in the brighter area while leaving the shape of the edge spread function unchanged - or would it actually change it? If you had enough pixels and the gradient were not too severe I would guess the former. What about a vertical light gradient on the horizontal edge?

DPR new Studio Scene raw captures come in 'Daylight' and 'Low Light' versions. The Low Light illumination creates a clearly visible gradient as follows:

Yet MTF50 readings are fairly consistent, if we exclude the above mentioned shock-prone shutter speeds (and operator error, always potentially lurking in the background):

Shutter Speeds 1/5th to 1/13th of a second from 'Low Light' illuminated NEFs; from 1/30th to 1/4000th from 'Daylight' illuminated NEFs

Jack

Edit: I see you were busy while I was writing

OP Jack Hogan Veteran Member • Posts: 7,671
Re: Modern Sensors Missing half the AA: in the Horizontal or the Vertical direction?

fvdbergh2501 wrote:

This extracted only the green photosites, completely ignoring red and blue, so wavelength-depended diffraction is avoided. (I did check the red and blue channels too, and focus was definitely optimized for green).

Hmm. Are you saying that the differences in the various channels are mainly due to diffraction effects? In this case by measuring MTF50 of the 3 channels individually and comparing them one might be able to calculate/extract the impact of diffraction vs aberrations. Interesting.

And what effect does only measuring the green channel have on MTF50 readings - since we are effectively cutting the sampling rate in half?

So I do not think there is a significant illumination gradient here.

Agreed

Incidentally, the overall resolution is somewhat low. Simulated results at f/5.6 indicates that we could see values of up to 62 lp/mm with an OLPF, and about 86.3 lp/mm without an OLPF.. The 62 lp/mm is consistent with what I saw on real photos captured with a D7000 (4.73 micron photosite pitch, vs ~5.09 micron pitch on the D610).

The D610's pixel pitch is about 5.9 microns. Plus the 85mm:1.8G is a good lens, not an outstanding one. And there is always the possibility of operator error (say incorrect focus peaking? That's clearly the case in the D5300 captures for instance). Although results are fairly consistent with the similar-sensored A7 and RX1R (different lenses though).

Jack

PS Yikes, it looks like I have been unconventional in my vertical/horizontal naming for months!

fvdbergh2501 Contributing Member • Posts: 582
Re: Modern Sensors Missing half the AA: in the Horizontal or the Vertical direction?

Jack Hogan wrote:

fvdbergh2501 wrote:

This extracted only the green photosites, completely ignoring red and blue, so wavelength-depended diffraction is avoided. (I did check the red and blue channels too, and focus was definitely optimized for green).

Hmm. Are you saying that the differences in the various channels are mainly due to diffraction effects? In this case by measuring MTF50 of the 3 channels individually and comparing them one might be able to calculate/extract the impact of diffraction vs aberrations. Interesting.

And what effect does only measuring the green channel have on MTF50 readings - since we are effectively cutting the sampling rate in half?

Well, there are many possible causes for observing different MTF curves in different wavelengths. Diffraction is always present; I measure 53.63 lp/mm in green, and 55.65 lp/mm in blue if I generate synthetic D610 images. The single largest factor is probably the lens design (achromatic, apochromatic): different wavelengths focus at difference distances from the focal plane (the sensor). The hope is that you can bring all the wavelengths close enough to perfect focus so that no one notices, but even an apochromatic design can one really have three wavelengths in focus, with varying degrees of defocus on all the wavelengths in between.

Of course, the expected increased resolution at blue wavelengths (lower diffraction) can be countered by focus (i.e., green is more in focus than blue), which appears to be the case with these D610 images. I do have actual D7000 photos where the blue channel produces slightly higher resolution than the green.

In terms of noise, yes, a using only the green photosites cuts the number of samples in half. That would put the minimum edge length at about 60 pixels (with 100 pixels being more comfortable), but the DPR charts have >600 pixels along the edge, so in this case it works well even for the blue or red photosites.

So here is the trade-off: we know how noise affects MTF measurements, i.e., the mean value over many individual measurements is unbiased, but the deviation from the mean can be quite large for any single measurement (at high noise levels). A difference between the blue and green (for example) focal plane position relative to the sensor would be systematic, i.e., it would not decrease with repeated measurements. This means that an un-demosaiced image (dcraw -D) followed by white balancing will produce an edge that is a blend of three individual edges (red, green and blue). White balancing, in itself, does not affect edge sharpness, so a perfectly white-balanced mosaiced edge image will probably produce a weighted MTF curve (25% red curve, 50% green curve, 25% blue curve).

If absolute accuracy is important, and multiple images are available, then repeated measurements followed by single-channel analysis (e.g., mtf mapper's "--bayer green" option) would be the best strategy.

If multiple images are not available, then edge length will dictate your choice: if the edge is sufficiently long (e.g., 200 pixels for green-only analysis), use single-channel analysis, otherwise use the white-balanced analysis.

The worst possible method is, of course, to use a demosaiced image, because nobody knows exactly how the MTF curve could be distorted by some non-linear interpolation scheme that the demosaicing algorithm might employ.

Incidentally, the overall resolution is somewhat low. Simulated >>results at f/5.6 indicates that we could see values of up to 62 >>lp/mm with an OLPF, and about 86.3 lp/mm without an OLPF.. The >>62 lp/mm is consistent with what I saw on real photos captured >>with a D7000 (4.73 micron photosite pitch, vs ~5.09 micron >>pitch on the D610).

The D610's pixel pitch is about 5.9 microns. Plus the 85mm:1.8G >is a good lens, not an outstanding one. And there is always the >possibility of operator error (say incorrect focus peaking? >That's clearly the case in the D5300 captures for instance). >Although results are fairly consistent with the similar-sensored >A7 and RX1R (different lenses though).

Whoops! Good catch. I see that I used 5.9 in generating the simulated images, but 5.09 when reporting some of the lp/mm ratings. So the simulated results should be 53.63 lp/mm with an OLPF, and 74.5 lp/mm without an OLPF. The 44 lp/mm at ISO 800 (vertical resolution) is still quite a bit below the expected 53.63 lp/mm. Of course, it is entirely possible that my simulated OLPF effect is not quite perfect yet.

Employing some dodgy extrapolation tells us that the D7000's 4.73 micron pitch is about 25% smaller than the D610's 5.9 micron pitch, so if we add 25% to the 44 lp/mm, we should only be able to get 55 lp/mm out of a D7000. My D7000 measurements came in above 60 lp/mm, and that agrees with other sources (DxO also measured this with the Sigma 17-55 mm f/2.8 lens on this body). That means the D610 + 85 mm images coming from DPReview are not quite as sharp as they potentially could be, but this would also depend on where the resolution is measured (relative to the absolute centre of the lens where my results hail from). At any rate, the lower-than-expected resolution issue is mostly irrelevant to the current discussion, and certainly we are seeing figures that are at least in the right ballpark.

Jack

PS Yikes, it looks like I have been unconventional in my vertical/horizontal naming for months!

Scott McMorrow Regular Member • Posts: 366
Re: Modern Sensors Missing half the AA: in the Horizontal or the Vertical direction?

fvdbergh2501 wrote:

Jack Hogan wrote:

Scott McMorrow wrote:

2) the illumination in the two directions could be different

Polarized, I assume you mean. I don't think it is nor that it should matter: DPR says that they use calibrated lights for this purpose. I also wonder whether polarization would survive diffuse reflection and even then whether it would make a difference on the MTF 50 readings - which do not depend on light intensity (within limits of course)

I think what Scott meant was not polarization, but rather uneven illumination, something like a vertical illumination gradient. This will definitely affect the slanted edge method, but my gut feeling is that a gradient large enough to effect a 20% H/V difference would be visible to the naked eye.

I will see if I can play around with a few of these images to see if there is any evidence of minor gradients.

Yes, I meant uneven illumination.  i'm just throwing out possible sources of error that I would look at first to eliminate other possible contributions to MTF reduction.  Honestly.  since Jack did not have control over the test apparatus, I think the most prudent thing to do would be to replicate the experiment and control it.  I think the hypothesis is a good one.

I can think of two possibilities if the results hold up.  Either the AA-filter is as Jack describes, there is some type of asymmetry in the microlens array, or there is an asymmetry in the sensor layout.

If I were going to test for these things, I'd start with a more telecentric lens design like the Zeiss 135mm APO. And then repeat with several other lenses to eliminate the lens from the equation.

Keyboard shortcuts:
FForum MMy threads