On the topic of Moire...

[dreed]

For some time I've been of the opinion that moire isn't a sensor problem and to a certain extent, I think Michael's photographs showing the 100% magnified grid without and the down-sized grid with demonstrate that.

This makes me believe that the AA filter in front of sensors isn't there to prevent moire being recorded but rather to make it easier for the JPEG rendering engine in the camera (and for some outside of it) to present an image that is free of moire.

Or to put it differently, moire is a result of using individual red/green/blue pixels on the sensor and the jpeg rendering engines being buggy and the easiest solution to moire is to degrade the quality of the captured image.

Similarly, with foveon sensors, there should be no moire at 100% magnification.

Is the above accurate?

[Schewe]

Is the above accurate?

No...moiré is a real problem when the subject frequency interferes with the sensor frequency. Plain and simple...color moiré is pretty easy to deal with in LR4/ACR7 but luminance moiré is tougher (read impossible).

Whether or not moiré is going to be a problem for YOU depends entirely what you are shooting, the sensor and angle you are shooting your subject. I've seen fabric weaves produce totally unfixable moiré that required reshooting (at great expense because the models were from NYC).

Unless you've been bitten by real moiré I suggest you quit suggesting what you THINK you know, ok? Moiré is a very serious problem, when it happens. When it doesn't, it doesn't matter at all.

BTW, LR4 and ACR 7.x deals with color moiré very well. It handles luminance moiré ok...

[dreed]

No...moiré is a real problem when the subject frequency interferes with the sensor frequency. Plain and simple...color moiré is pretty easy to deal with in LR4/ACR7 but luminance moiré is tougher (read impossible).

Whether or not moiré is going to be a problem for YOU depends entirely what you are shooting, the sensor and angle you are shooting your subject. I've seen fabric weaves produce totally unfixable moiré that required reshooting (at great expense because the models were from NYC).

Unless you've been bitten by real moiré I suggest you quit suggesting what you THINK you know, ok? Moiré is a very serious problem, when it happens. When it doesn't, it doesn't matter at all.

BTW, LR4 and ACR 7.x deals with color moiré very well. It handles luminance moiré ok...

So does that mean in digital photography that moire can occur both at capture time and also at render time?

[bill t.]

The insidious thing about moire is that it sometimes sneaks past people who should have been more attentive.  AA filters might be a busy photographer's, or editor's, best friend.

Check the fabric near the pretty girl's left elbow.  (caution: 20mb jpeg).

[Schewe]

So does that mean in digital photography that moire can occur both at capture time and also at render time?

It happens at capture...it may or may not be visible at render/output. But if it's visible and obvious at capture, there's not a lot you can do about it in post (short of doing the moiré brush in ACR/LR or C1).

Color moiré can be easily fixed (in post), luminance moiré not so much...

[dreed]

It happens at capture...it may or may not be visible at render/output. But if it's visible and obvious at capture, there's not a lot you can do about it in post (short of doing the moiré brush in ACR/LR or C1).

Color moiré can be easily fixed (in post), luminance moiré not so much...

Ok.

Does the moire in the reduced size image of the mesh in Micahel's D800 story fit into either of those categories?

Or is that "pattern moire"?

I did a little bit of searching and I found this page:
http://photo.stackexchange.com/questions/11909/what-is-moire-how-can-we-avoid-it
... at #15 it includes a section on how the anti-aliasing filter works (it didn't really sink home what it was doing until I saw this page and discovered that its function is the same as anti-aliasing when rendering graphics on screen.)

[Bart_van_der_Wolf]

Does the moire in the reduced size image of the mesh in Micahel's D800 story fit into either of those categories?

Hi,

Moiré is one type of manifestation of aliasing, false color demosaicing and 'jaggies' (stairstepped edges and lines) are others. Moiré occurs when there are two regular patterns with a small difference in size and/or orientation.

Aliasing is allways present in digital photography, when the subject detail has adequate contrast and its size (as projected on the sensor) is smaller than 2x the sensel pitch (the sampling grid). That aliasing may, or may not, be very visible but it will be there nevertheless when those conditions are met.

Downsampling of images by definition causes aliasing, because the detail again becomes smaller than the sampling grid density with which it will be represented. I've dedicated a full page on the effects of different downsampling methods alone.

The most efficient way of reducing aliasing is by the use of an Optical Low-Pass Filter (OLPF, or anti-aliasing filter), but defocusing and/or narrow aperture induced diffraction also reduce the aliasing risk, but at a higher cost of resolution loss and with variable effectivity. An OLPF has always the same effect, and the loss of micro-contrast is easier to (partially) correct. It's a trade-off.


Cheers,
Bart

[hjulenissen]

For some time I've been of the opinion that moire isn't a sensor problem and to a certain extent, I think Michael's photographs showing the 100% magnified grid without and the down-sized grid with demonstrate that.

This makes me believe that the AA filter in front of sensors isn't there to prevent moire being recorded but rather to make it easier for the JPEG rendering engine in the camera (and for some outside of it) to present an image that is free of moire.

Or to put it differently, moire is a result of using individual red/green/blue pixels on the sensor and the jpeg rendering engines being buggy and the easiest solution to moire is to degrade the quality of the captured image.

Similarly, with foveon sensors, there should be no moire at 100% magnification.

Is the above accurate?

Your CD player contains a smoothing filter in its D/A converter, and the recording studio used a smoothing filter in front of the A/D converter when they recorded the audio. It has nothing to do with limitations in mp3 coding, and everything to do with that the CD format can play back any waveform limited to <22050Hz extremely well, but not beyond. This bandwidth limiting means that you will not have slopes or steps in the output waveform that are infinetely steep.

The phenomenon of aliasing can be described in a theoretical context, and happens whenever you sample a (continous) waveform using a finite set of samples whenever that waveform contains frequency components that are >= (sampling rate)/2

I think it makes sense to either read or accept this theory as a fundament, then adding all of the nitty gritty details of photographic scenes, color filters, human vision etc.

-h