Idaho Airships, Inc.

Moire is essentially a spatial patterning defect in digital images. It is caused by the deconstructive misregistration of image details, which itself is due to the way CMOS and CCD sensors are configured with respect to Red, Green, and Blue collection sites.

That’s actually the easy part. If you’re looking to resolve moire-not simply mask it to some extent-this post may be of some value.

Next time you have the chance take two small pieces of screen and align them perfectly so that you can see through them. Look at something that has a patterning or detail regularity through the nicely “registered” screens. Looks fine, if a bit obscured. Then, offset them and look again. You should notice a moire.

The irregularity of your sight sampling through the misregistered screens and the regularity of the detail in your sighted subject simply aren’t going to get along.

In a DSLR (we’ll use Canon for the purposes of this discussion) certain filters are used to prevent some ranges of electromagnetic energy (of which visible band light is simply a subspecies) from reaching the sensor. Canon employs a type of filter that actually blurs the image to combat moire. It works-of course-but why spend $8K on long Canon L primes if Canon itself is blurring your images to obviate moire?

Canon certainly recognizes this problem-and their solution is simply genius. The filter they employ displaces light a specified amount (that is calculated on the basis of the associated sensor site pitch), and when the light arrives at the sensor it is captured and the offset is compensated for in the camera’s processing. When I say that the DSLR’s we shoot are computers, I’m not exaggerating!

The process, however, is imperfect, and moire will still manifest-particularly if you’re shooting something with brutally tight and regular detail. Interestingly, because moire boils down to matters such as Airy Disks, issues such as your focus accuracy and aperture may have a profound effect upon it. Thus, if you’re in a studio shooting a fabric and have the ability to immediately view your images on a monitor, simply changing your aperture may resolve much of the problem. If you’re in a helicopter traveling at 90 miles an hour shooting a chain link fence…well…tough. You’re not going to be able to judge the moire on your camera display for a variety of reasons.

Now for the good news. Photoshop has the answer, but I’ll forewarn you that it’s not for the faint of heart. Welcome to Channels.

Haz U Teh Channelz Skilz?

The solution I’m trying to refine is not new-I learned about it back in CS2 though I don’t remember where. Nor is it perfect-it requires both an in-depth knowledge of Channel manipulations and Histogram statistics, but it is effective and reliable across a broad sample of moires that I’ve had to deal with. Nonetheless, my experiences are anecdotal, and if you have the chance to try my technique (and improve upon it, please), I’d be very grateful.

For many reasons, let’s presume we’re experiencing Red & Blue moire on a rather neutrally colored surface:

Split the Image Channels into standalone Documents

From the Green (or least affected) Channel, select and copy an area of interest

Paste and register (align) the G import into the other Channel Documents

Match the Histogram of the imported area to the native area

Flatten the Channel Document

Import the standalone Documents back into a new RGB Document (via Merge Channels).

Split Channels and Merge Channels are available in the Channels Palette Menu. When Merging the Channels back into a single RGB Document, ensure that RGB-not “Multichannel” Mode is selected. Or you’ll be sorry.

Now, on to the details on matching up the imported area to the native area in the two diseased Channels. If you are using standardized Selections (such as those available to users of Alpha Channels or Layer Masks) you should be able to use Lab Curves, Equalize…heck, if you’re really good you can even use Gradient Maps…to match up the Histograms. That part about Gradient Maps is a joke. You can add noise, Draw/Paint, or any number of other things that you can’t do in the Channel while it’s sitting there in its native RGB Document (such as vary transparency and Blending Modes for the import).

The one drawback is that you can’t use Match Color on Grayscale Documents. But! You can convert Grayscale Documents to RGB, use Match Color, then convert them back (I’m talking, obviously, about separate Documents containing only the selections of the areas of interest).

So, there are a myriad of potential ways to match the moire areas, and I’m very certain that I have only scratched the surface.

If anyone does come up with a more effective way to “match the patches” please let me know. I can post it here (with due credit, obviously) or link out to your site, with appreciation.

Just some things to explore: Histogram Statistics (Kurtosis and Skewness quantification is left up to you) in blue and Histogram localization in green. Mousing over the Histogram or a range on the Histogram will provide Level, Count, and Percentile data. Perhaps we could talk Adobe into Color Sampler Tools-like readers for Histograms?

L