D800 and E... poor for landscape use?

[Bart_van_der_Wolf]

Thanks for your pointers on choosing the proper radius to determine the resolution from your chart. I did notice the reddish case at f/5.6. What causes that and how would it be corrected?

Hi Bill,

You're welcome. The cause is a mystery to me, but it is more prominent with ACR/LR than with Capture One or RawTherapee, so it must be something in the way the demosaicing is done. Maybe it is a by-product of something they do to avoid false color artifacts (it doesn't happen with the D800). Just guessing.

I did use your tool to determine optimal sharpening radii, but did not post the results since the values for f/2.8 and f/4.0 seem too high and I intend to repeat the analysis when I get time. The tool is quite helpful and many thanks for publishing it.

Besides residual aberrations kicking in at wider apertures, the most likely reason for significantly higher radii at wider apertures is defocusing. As one of the examples in the help for the introduction section shows, even at f/5.6 with a slightly longer focal length than you used, the differences in focus are picked up by my tool even within that DOF zone. So a good method for aquiring an aperture range is to concentrate on the best score for a wide open aperture, and use that distance setting for the other apertures. Of course, once you know what is possible or what to expect, its easier to know when the focus was spot on. Contrast detect focusing should also allow to do well on a flat plane.

Adjusting the radius according to the Gaussian radius is a departure from the conventional sharpening advice according to the Jeff Schewe/Bruce Fraser model where they suggest a small radius for high frequency images such as landscapes and a larger radius for low frequency images such as portraits.

He, I'm all for breaking with conventions if it produces better results . No, seriously, in general that is not bad advice, but it is primarily based on a film workflow with postprocessing once the image was digitized. In that book there is also a constant battle against the halo effects of unsharp masking and very little about deconvolution.

As my tool detects and demonstrates, the effect of aperture alone on the required Capture sharpening will require significantly different radius settings to optimally compensate for the blur characteristics, and that is within the same scene (assuming high frequency detail is present).

There are better deconvolution tools than ACR/LR, but I prefer parametric  editing since 16 bit TIFFs need for a stand alone produce get rather large when one is using the D800. The extra effort of using a standalone tool would be worth the trouble for one's most important images that need to be printed large.

I agree, but it should be relatively simple for Adobe to do a better job on the sharpening, as they should for resampling. Bi-cubic with a twist, come on, this is 2012.  Maybe proving that they are dragging their feet in this area will wake them up a bit, although I won't be holding my breath. I remember discussing HDR with one of the programmers in the Photoshop team some 10 years ago, and he didn't see the need ...

Okay, that doesn't look not too bad and it allowed you to find/confirm a sweetspot near f/5.6 or f/8 (which would shift more towards f/5.6 if there was some defocus involved), but as you said it's a bit on the high side at the wide open aperture side but not impossible (the tonecurve also has an influence). It does show how much difference there should be made in Capture sharpening depending on the actual aperture used (again, simple for Adobe to implement since they have the EXIF data). I know it complicates the life of folks, but don't blame me for revealing the truth .

Cheers,
Bart

[Ray]

Not interested in your semantic discussions Ray. The point is that at f/16 the D800 cannot reach its maximum effective resolution, and that means you cannot enjoy its full 36Mpx because of diffraction (among other possible factors).

Then you will likely be more confused than you otherwise would be, Guillermo. Semantic issues are to help clarify concepts and help us say exactly what we mean. This thread seems to be full of confusion about the significance of driffaction on cameras which have a high pixel-count such as the D800. I see this confusion resulting, in part, from the sloppy concept that a camera's sensor can be diffraction limited. It can't.

The sensor is there to record whatever the lens throws at it.

Your statement:

The point is that at f/16 the D800 cannot reach its maximum effective resolution, and that means you cannot enjoy its full 36Mpx because of diffraction (among other possible factors)

simply reinforces the confusion. The facts are, the D800 cannot reach its maximum resolution at any aperture, whether F16, F8 or F5.6, unless you think we have reached the stage of producing the perfect lens. People sometimes spend thousands of additional dollars to get a lens which is noticeably sharper, at its sharpest aperture, than another cheaper lens of the same focal length. Such additional sharpness from the more expensive lens can be seen when the image is recorded on sensors with a much lower pixel-count than the D800 has.

With any camera, whatever its pixel count, there is a trade-off in resolution at the plane of focus, when attempting to increase the DoF of the scene by using a narrower aperture than the aperture at which a particular lens is sharpest. That sharpest aperture is usually around F4 to F5.6, but can range from F2.8 to F11, depending on the quality of the lens.

Those who are a bit obsessed with resolution, as I am, and many others on this forum it would seem, naturally try to avoid using a narrower aperture than is necessary to achieve the desired DoF because we all know that stopping down reduces resolution at the plane of focus, whatever the camera we are using and however slight such loss may be.

The high pixel-count of the D800 doesn't change that broad principle. However, it is useful to know, and also comforting for those who have moved up to the D800 from a camera with fewer pixels, such as the D3X, that such lucky owners of a D800 can now use F16 with their D800 and expect to get at least the same resolution at the plane of focus that they would have got using F11 with the D3X, or other equivalent camera, but also get the advantge of that increase in DoF that F16 provides.

I only believe in charts and graphs that can be confirmed with real-world testing. When DXOMark make a claim that camera A has, for example, 2EV more dynamic range than camera B, I'll test that for myself, if possible, if I own the cameras in question, or can at least get access to them without too much trouble. Whenever I've done this (the Canon 5D versus the Nikon D3, the Canon 50D versus the Nikon D7000, the Nikon D700 versus the D7000 etc), I've always found that the DXOMark results very closely match mine, with respect to DR differences.

My advice to some of you guys is, instead of waffling on, ad nauseum, about imaginary concerns, go out and take the shots for comparison purposes, using your best technique. Or stay in and photograph a newspaper taped to the wall. The comparative legibility of fine text is always a very meaningful result regards image resolution.

[Fine_Art]

Then you will likely be more confused than you otherwise would be, Guillermo. Semantic issues are to help clarify concepts and help us say exactly what we mean. This thread seems to be full of confusion about the significance of driffaction on cameras which have a high pixel-count such as the D800. I see this confusion resulting, in part, from the sloppy concept that a camera's sensor can be diffraction limited. It can't.

The sensor is there to record whatever the lens throws at it.

Your statement:  simply reinforces the confusion. The facts are, the D800 cannot reach its maximum resolution at any aperture, whether F16, F8 or F5.6, unless you think we have reached the stage of producing the perfect lens. People sometimes spend thousands of additional dollars to get a lens which is noticeably sharper, at its sharpest aperture, than another cheaper lens of the same focal length. Such additional sharpness from the more expensive lens can be seen when the image is recorded on sensors with a much lower pixel-count than the D800 has.

I doubt guillermo said the sensor can be diffraction limited. The system can be diffraction limited.

You must have made a mistake in your tests. I've tested my camera with a similar pixel pitch (APS-C) at near nyquist. So have others with different cameras. When I look at 100% pixels on the D800 images provided at several sites, I see good detail. In other words these systems are not lens limited. That is assuming primes or high quality zooms.

If all you are saying is that kit zooms cant handle the D800 I dont see what the point is. Everyone else knows that too. I am sure you have good pictures with fine detail off your D7000. What is the full frame equivalent MP? What lenses are you using?

[Ray]

I doubt guillermo said the sensor can be diffraction limited. The system can be diffraction limited.

You must have made a mistake in your tests. I've tested my camera with a similar pixel pitch (APS-C) at near nyquist. So have others with different cameras.

Excellent! Show us all the results. I have to tell you, one of the great joys in my life is being proved wrong, because then I feel I have learned something. Life is a process of learning.

When I look at 100% pixels on the D800 images provided at several sites, I see good detail. In other words these systems are not lens limited.

Total nonsense. All camera systems are both lens limited and sensor limited, whatever the quality of the lens or sensor.

If all you are saying is that kit zooms cant handle the D800 I dont see what the point is. Everyone else knows that too. I am sure you have good pictures with fine detail off your D7000. What is the full frame equivalent MP? What lenses are you using?

An excellent example of confusion. The lens doesn't handle the sensor. The sensor handles the lens. The higher the pixel-count of the sensor, the better it handles the lens, period.

[Fine_Art]

Already did, it's in Bart's thread on his Siemens star chart.

[Fine_Art]

Total nonsense. All camera systems are both lens limited and sensor limited, whatever the quality of the lens or sensor.

An excellent example of confusion. The lens doesn't handle the sensor. The sensor handles the lens. The higher the pixel-count of the sensor, the better it handles the lens, period.

Limited in terms of recording reality, yes. Limited in terms of the system no. The camera system has a weakest link. IMO it is still the sensors.

False. Do you know how small the pixels are on P&S cameras? On any crop sensor you extrapolate the resolution of the camera up to FF. Why because the sensor is cropping out the middle of the lens image.

[Ray]

Limited in terms of recording reality, yes. Limited in terms of the system no. The camera system has a weakest link. IMO it is still the sensors.

Of course resolution is limited in terms of the system. Resolution depends on both the quality of the lens and the pixel quality and pixel count of the sensor. Maximum quality results from maximum quality of both sensor and lens.

False. Do you know how small the pixels are on P&S cameras? On any crop sensor you extrapolate the resolution of the camera up to FF. Why because the sensor is cropping out the middle of the lens image.

Of course I do. Do you think I am a complete nincompoop? One of the highest resolving sensors at the moment is the Nokia PureView 808 with 41mp on a sensor which is only 11mm diagonally. Full frame 35mm is about 44mm diagonally, so we might expect some time in the future a 164mp full-frame 35mm sensor (41x4).

On any crop sensor you extrapolate the resolution of the camera up to FF. Why because the sensor is cropping out the middle of the lens image.

Nonsense! One extrapolates the image to any size one wants to print it, whether it's a full-frame or cropped format image. I have a 24x36 inch print from a 6mp Canon D60 on my wall, which looks quite okay from an appropriate viewing distance. I've also made 5x7inch prints from a 12mp full-frame DSLR which one probably could not distinguish from prints of shots from a 3mp P&S camera.

Get real!

[Petrus]

Do you think I am a complete nincompoop? One of the highest resolving sensors at the moment is the Nokia PureView 808 with 41mp on a sensor which is only 11mm diagonally. Full frame 35mm is about 44mm diagonally, so we might expect some time in the future a 164mp full-frame 35mm sensor (41x4).

Or maybe a 41x16 MPix = 656 MPix FF sensor?

 

[Ray]

Or maybe a 41x16 MPix = 656 MPix FF sensor?

 

Well done! I'm glad at least someone is alert. 

[Fine_Art]

So you want to believe those cheap P&S cameras are made with Miracle Glass TM.

No, top 35mm lenses are ground to the same precision.

[dimapant]

There is a point which is, in my personal opinion, very important, it is a fundamental one, and is not, still in my  personal opinion, properly addressed in these discussions: in photography, what cannot be seen, it does not matter, and it does not matter at all.

At the end of the day the only real judgment for the quality of the picture is the eye and the brain of the viewer of the picture.

And this fact plays a fundamental role in the selection of the sharpening parameters, a role which overwhelming in respect of just the theoretical radius calculation based on diffraction airy disk at a certai F number and pixel pitch 

Two different pictures, shot with  the same equipment, lens and camera, settled at the same parameters, like same F number of the lens, may needs different, completely different sharpening, according to the different subject, different paper, even different location of the print, even different light on the print.

In other worlds, in my personal opinion, for a quality picture, you cannot define the sharpening radius based on a simple count of diffraction airy disk at a certain F number and pixel pitch, it does not work in the real world, provide you just an indication, but it does not work in the real world, and it does not work at all, at least for my print (360 PPI, in general Glossy Paper, A3 size)

Coming from D 700 and D 7000, I did some test, with 60 mm Micro AFS F 2,8, the same lens you used for the test, and also with 14 – 24 at high F number, with sharpening (Smart Sharpen, Lens Blur, More Accurate) and  what I found it is that using the radius coming from an estimation of the radius when the airy disk of diffraction cover 1/3 of the nearby pixels, and from what I seen, the radius coming from that calculation ( radius= Airy Disk Diameter/ 4,88/3) is too high for the sharpening, it is tooo high in basically all kind of picture.

So, what I did with D 800, is to forget about calculation and go back to use the same parameters used with D 7000 and, different, for D 700 for capture sharpening, and a different  capture sharpening from picture to picture, according to the subject, by the eye, and still by just the eye for the Artistic Sharpening and the same, exactly the same Print Sharpening.

Compliments for your deep analisys and thank you for sharing to us!

Best regards.
Alessandro

[Bart_van_der_Wolf]

There is a point which is, in my personal opinion, very important, it is a fundamental one, and is not, still in my  personal opinion, properly addressed in these discussions: in photography, what cannot be seen, it does not matter, and it does not matter at all.

At the end of the day the only real judgment for the quality of the picture is the eye and the brain of the viewer of the picture.

And this fact plays a fundamental role in the selection of the sharpening parameters, a role which overwhelming in respect of just the theoretical radius calculation based on diffraction airy disk at a certai F number and pixel pitch 

Two different pictures, shot with  the same equipment, lens and camera, settled at the same parameters, like same F number of the lens, may needs different, completely different sharpening, according to the different subject, different paper, even different location of the print, even different light on the print.

In other worlds, in my personal opinion, for a quality picture, you cannot define the sharpening radius based on a simple count of diffraction airy disk at a certain F number and pixel pitch, it does not work in the real world, provide you just an indication, but it does not work in the real world, and it does not work at all, at least for my print (360 PPI, in general Glossy Paper, A3 size)

Hi Alessandro,

That's why the distinction is made between Capture sharpening (to eliminate blur that occurs during the image acquisition/Raw conversion) which can be quite accurate, and Creative sharpening (where certain image features are more, or less, accentuated). Finally there is output sharpening (where the effects of the output medium and the viewing distance are optimized).

Cheers,
Bart