Do we really need the AA filter?

[giellaleafapmu]

I was just curios to ask all of you who own a recent camera without the AA filter how often did you really experience the Moire effect that the filter should avoid. I remember that the old professional Kodak cameras had it quite badly but I have never seen any produced by a recent camera. Right now I only own the Pentax Q with a very weak AA filter (or is it without?) and have used a few times the D800e (not mine, just rented) and I never experienced any problem whatsoever but still every time I read a review of a camera of that type the famous Moire effect is mentioned as something which can happen very often and should make these models kind of specialty ones. So Pentax Q, E-5e, Ricoh A12 module, Ricoh GRD, D800e (did I forget any camera?) did any of you ever see Mr Moire in real life?

GLF

 

[3rdtrick]

AA filter, so 20th century...

I have a D800E and have had an M9 since they first came out and have not noticed moire in 10,000 or more images. I feel that the extreme resolution of these cameras is much finer than anything that could cause visible moire.
Pete

 

[jsrockit]

I notice it right away if there is a window screen or TV in focus in a photo. Generally at smaller apertures.

 

[johannielscom]

Results of my Ricoh GXR-M with quality glass show that the sharpness greatly improves while the moiré effect doesnt show if the filter is absent.

 

[Aristophanes]

More megapixels plus software correction in-camera makes it a problem that can be mitigated (not solved) without a filter. Bigger sensors help in that regard.

 

[jschrader]

D800 would not have problems with Moire because the lenses do not resolve enough details. Only if you resolve lines as fine as the pixels, you get the effect.

That would be around 90 lines per mm in the case of the D800

 

[j j]

Feathers provide another problem area. I sometimes see it in my bird photos even with an AA filter. Usually a small change in file size sorts it (whether increase or decrease).

This potential problem gets too much space these days. An exaggerated hangover from smaller-resolution sensor days (much like the too many pixels fears were until cameras such as the D800 appeared).

 

[giellaleafapmu]

D800 would not have problems with Moire because the lenses do not resolve enough details. Only if you resolve lines as fine as the pixels, you get the effect.

That would be around 90 lines per mm in the case of the D800

I got confused, this makes perfect sense to me but then a smaller sensor such as in the Ricoh, Pentax Q or even just an APS size sensor should help because would put that number even higher, why somebody else wrote that larger sensors would help and what is the truth?

GLF

 

[willie_901]

It is a fact, much as gravity is a fact, that aliasing artifacts are always present when a Bayer color filter array is used to discretely model a non-discrete phenomenon.

The issue really is under what conditions are the artifact levels due to aliasing less than the artifacts introduced by AA filtering.

Moiré artefacts not affect color but they affect luminance as well. The total level of image degradation is not always as obvious as color artifacts associated with highly ordered closely spaced detail. Luminance aliasing artifacts compromise detail just as the AA filter does. The effects are real, but they are not as obvious as false color patterns.

There is a two page informative thread on Luminance Landscape from 2009 that exhaustively discusses the question asked by the OP.

http://www.luminous-landscape.com/forum/index.php?topic=37660

There is no simple answer. Many factors combine to determine the artifact levels. Like many compromises, different photographers have different tolerances for artifact levels and filtering effects. In appropriate sharpening can degrade IQ to the point where it becomes the prime factor instead of aliasing.

It seems in the early years of Bayer imaging, camera manufacturers used AA filters that were too strong. This could have been due to erring on the side of caution to minimize consumer complaints or perhaps stronger filter technologies were cheaper back then.

 

[jschrader]

I got confused, this makes perfect sense to me but then a smaller sensor such as in the Ricoh, Pentax Q or even just an APS size sensor should help because would put that number even higher, why somebody else wrote that larger sensors would help and what is the truth?

GLF

Aliasing occurs, when the signal frequency is close to the sampling frequency. CDs have 44.2 kHz samplig rate, so a tone of 22.1 kHz has exacly 2 sampling points to describe a full sinus wave. A tone of 21 kHz produces a 1.1 kHz artifact tone (22.1 - 21). Read about the Nyquist theorem on Wikipedia.

In cameras, it is the same. Calculate, how many pixels You have per mm. If you have a "signal" (e.g. feathers of a bird) that fine, you get artifacts. But, if the pixels are fine enough, the lens is your AA filter; it does not resolve them and any detail the lens reproduces is large enough to hit 2 or 3 pixels.

Why cameras with smaller sensors should have less problems: their pixels are even smaller. A D7100 (?) with 24 MPixels would need an even better lens, because the pixels are smaller thatn in the D800.

 

[giellaleafapmu]

Why cameras with smaller sensors should have less problems: their pixels are even smaller. A D7100 (?) with 24 MPixels would need an even better lens, because the pixels are smaller thatn in the D800.

Yep, that's exactly what it should be (and it all agrees with my empirical knowledge with the Q and the old Kodaks). It is also interesting to see that an in-camera ND filter is very common, even in cheap point and shot but an in-camera AA filter which can be used or not used is unknown of. Well, that's probably a good way to have people buy the more expensive versions without a filter (sic!).

GLF

 

[giellaleafapmu]

Read about the Nyquist theorem on Wikipedia.

I like the fact that my real work is research in abstract harmonic analysis, Fourier analysis, wavelets and in this "did you see Moire" trend you send me back to the Nyquist-Shannon Theorem, maybe life makes sense after all...

GLF

 

[jschrader]

I like the fact that my real work is research in abstract harmonic analysis, Fourier analysis, wavelets and in this "did you see Moire" trend you send me back to the Nyquist-Shannon Theorem, maybe life makes sense after all...

GLF

I didn't want to sound rude, sorry. I should have said something like "if you want to understand the theory behind it, check..."

But basically, it is nothing more than that.

Now thinking about it, my digital experience is all from a D700 and I never saw any moire. But the D700 is in fact a "very large pixel" camera and probably has an effective AA filter

 

[giellaleafapmu]

I didn't want to sound rude, sorry. I should have said something like "if you want to understand the theory behind it, check..."

But basically, it is nothing more than that.

Now thinking about it, my digital experience is all from a D700 and I never saw any moire. But the D700 is in fact a "very large pixel" camera and probably has an effective AA filter

Uh, why are you apologizing, it is completely true what I wrote about by work, despite this I didn't think for a second about the Nyquist-Shannon Theorem at all (sleepy? stupid? I don't know) and I was both amazed and pleased to read you comment. Anyway, after reading what I already suspected, that is that with current Mpx the Moire effect is not something to worry about, I am dangerously attracted both to the Pentax K-05iie and the Ricoh A12 module. The only thing which annoys me is that no AA filter means more expensive camera. Maybe I should just grab one of my very old cameras and try to learn how to remove the AA filter if this is at all possible, but then again, maybe I should use my time for something more interesting and just follow the classical "Shut up and take my money"...

GLF

 

[semilog]

If you value image accuracy over apparent sharpness, use good lenses, shoot subjects with high spatial frequencies, and have immaculate technique (well-damped tripod, very careful focus, etc.), you want an AA filter or better be prepared to deal with moiré in post.

If you don't do all of those things, it does not matter, most of the time, becaue you're not getting the fully resolution that the sensor is capable of and under those conditions the presence or absence of an AA filter makes no difference.

Overblown issue. See Willie901's post above for an accurate but accessible description of why.

If I were choosing between a D800 and a D800E, I'd take the D800. No contest. Every other current Nikon DSLR uses an AA filter because Nikon understands that for most purposes, that's a technically equivalent or superior solution. Canon doesn't pander to the no-AA filter market at all.

 

[Noll]

Do we really need all these megapixels that we're getting these days ? Probably not, unless you know you need it. Ask yourself what your max print size is and that should lead you to your answer.

 

[giellaleafapmu]

If you value image accuracy over apparent sharpness, use good lenses, shoot subjects with high spatial frequencies, and have immaculate technique (well-damped tripod, very careful focus, etc.), you want an AA filter or better be prepared to deal with moiré in post.

If you don't do all of those things, it does not matter, most of the time, becaue you're not getting the fully resolution that the sensor is capable of and under those conditions the presence or absence of an AA filter makes no difference.

Overblown issue. See Willie901's post above for an accurate but accessible description of why.

If I were choosing between a D800 and a D800E, I'd take the D800. No contest. Every other current Nikon DSLR uses an AA filter because Nikon understands that for most purposes, that's a technically equivalent or superior solution. Canon doesn't pander to the no-AA filter market at all.

Mmmmh, actually that was not exactly my question, I was asking if anyone had an actual example of what you say, i.e. Moiré in a picture. You seem to be just replicating a dogma which I read a billion time but I have never seen in real life recently (yep, my tripod is not very sturdy and when I focus at F8.0 an object on a still-life table probably AF goes crazy, that's possibly why I never saw it). As for the Canikon talk and what they understand about market it seems a bit contradictory what you write, if Nikon understands that a camera with an AA filter is technical superior why do they...ok ok, never mind.

GLF

 

[giellaleafapmu]

yeah, me too!

With the great resolution provided by mid level cameras being enough for a lot of professional work, amateurs seem obsessed with maximum resolution - when not using the best lenses and with the final product a social media image that could easily be made on a $500 camera + kit lens. This has got to be about gear prestige/bragging and not about photography.

Some smart guy recently pointed out that both Penn and Avedon could have built their entire carriers with a Nikon D5000 + kit lens and some lighting gear.

I don't see the relation with the original question. Yep, you can still sell 5Mpx pictures, hence use as your unique working tool a D5000, so? What does it have to do with the curiosity of knowing how different sensor configurations perform? I immagine that, since you can only eat rice, chicken and salade and live for a long time you have never tasted fish and chips right?

GLF

 

[giellaleafapmu]

If you can't tell the difference, why worry?

http://blog.phaseone.com/2012/05/10/removing-color-moire-with-capture-one-6/

I think that one of the basic condition for having a conversation (albeit a virtual one) is to understand each other, now if by reading that I have never seen Moiré in a recent pincture from a camera without AA filter you deduced that I saw no difference between a picture taken with a camera with AA filter and a camera sporting the same sensor but without the filter we have a problem. Thank you again for you attempts to answer but it seem that you don't have an example of I was asking and the conversation is getting rather boring.

GLF

 

[willie_901]

People have replicated the dogma a billion times that the diffraction of light is due to contructive/destructive wave interference. That doesn't mean diffraction broadening is unimportant.

How many different ways can spatial aliasing be explained anyway? Just because a scientific fact is repeated conistantly it doesn't mean it's not important. Physics is not propaganda.

To repeat myself, besides the false color artifacts which are obvious (just like diffraction broadening is obvious) aliasing cause luminance artifacts which degrade image quality and are much less obvious. But the degradation is real.

As far a camera manufacturers go, early digital cameras used strong filters and early photographers did not understand sharpening. Eventually the AA filters became weaker. Photographers became more sophisticated about sharpening. All filtering is a compromise. All filtering degrades information content. You trade whatever benefit the filter provides for loss of information.

But here's what I don't understand. You say you have never seen aliasing artifacts. You also imply you may use techniques that do not produce images sharp enough to observe aliasing artifacts. So why do you even care if your camera has an AA filter?

If you want to loose your spatial aliasing observational virginity, Google

"examples of aliasing artifacts in digital photography" and select the images view .

Other people see aliasing artifacts all the time.