Understanding Moiré Patterns in Digital Photography

OWYOUNG, TODD. "Understanding Moiré Patterns in Digital Photography." ishootshows.com. ishootshows. Web. 9 April, 2012.

You’re probably seen it before. Strange waves or rings of color and tone rippling over fabric, or perhaps weird, maze-like squiggles instead of parallel lines when you’re zoomed in at 100% in Photoshop. This is called a moiré pattern.

Moiré patterns are artifacts that have a fixture of digital imaging, but which have been most recently thrown into contrast recently with introduction cameras like the Nikon D800E and Leica M9, which are more prone to producing moiré patterns.

Here’s an in-depth look at the causes of moiré, what it looks like and what it may mean to your photography.

THE ANTI-ALIASING PREDICAMENT

The Nikon D800E is one of the first mainstream DSLRs with no anti-aliasing applied at the sensor level. Traditionally, anti-aliasing filters are used with almost all DSLR sensors to reduce instances of moiré and false color artifacts.

With no anti-aliasing applied, the D800E offers exceptional detail, but also a higher chance of producing these moiré artifacts. It also has many photography deciding whether the Nikon D800 or D800E is the right camera for them.

While cameras like the Leica M9 and Sigma’s line of DSLRs (which use non-Bayer sensors) don’t feature anti-aliasing filters in the optical chain, the D800 and D800E are unique in that they offer choice, which thrusts the issue of moiré into more stark scrutiny.

WHAT IS MOIRÉ?

In essence, moiré occurs when two patterns are overlaid and result in a new, third pattern. With digital photography, these artifacts result when the frequency of detail in a scene exceeds the sensor’s pixel pitch and ability to resolve “real” information.

Moiré wasn’t an issue with film because the photo sensitive grains in film are arranged in a much more random, organic way.

Moiré artifacts can be an issue with any digital sensor, but particularly with the Bayer-type sensor of individual red, green and blue pixels, as the spacing between like-color pixels can result in even more errors. (New cameras like the Fuji X-Pro1 claim to address moiré with a different, more random arrangement of the RGB pixels, which should reduce instances of moiré.)

When information (scene detail) can’t be accurately recorded distinctly by one pixel or another, errors can result. These errors can take the form the wrong value (luminance) or color (chrominance) for pixels.

In contrast to pixels errors that result in digital noise, which are random, moiré artifacts are distinct and localized to the area of the “offending” detail.

More Pixel, Less Moiré

In general, the finer the pixel pitch and/or resolution of a sensor, the fewer instances of moiré should be rendered. This is logical, since a “coarse” capture of information is precisely what fosters the errors that produce moiré.

With higher resolution sensors, not only is more information captured, it’s captured more precisely. This is one reason why high resolution medium format digital backs lack anti-aliasing filters and record exceptional detail.

COLOR MOIRÉ

Color moiré is one of the most common instances of moiré and occurs when inaccurate color information is recorded. It often looks like a rainbow or a rippling of weird color over fabric. This happens most readily with very fine patterns or fabric with a high level of sheen.

Thankfully, color moire it’s generally one of the easier types of moiré to reduce, as only the color channel is affected, not the luminance channel.

Here’s an example of some mild color moiré produced by the tight diagonal pattern of a shirt’s fabric at 200%:

You can clearly see a small but distinct rainbow pattern created in between the black bars of this pattern. Thankfully, it’s easily removed if shooting RAW.

Reducing Color Moiré

In this next sample, we see the same region that’s been treated with the Adjustment Brush to reduce moiré in Adobe Lightroom 4. This is a new feature to Adobe Lightroom – it’s as easy as “painting on” the effect to the problem areas.

At 50% strength with the moiré reduction brush, we see pretty much a 100% elimination of the color moiré. Since this is a black and white fabric with basically no color information, it would be possible to eliminate these color artifacts by desaturating in Photoshop.

With colored areas, the adjustment brush still works quire well, but might not entirely remove all instances of color moiré.

MAZE ARTIFACTS

Maze artifacts are less common than color moiré, but they’re also more difficult to remove, since the errors that occur are in the luminance channel. Unlike color moiré, where it’s possible to shift the color back to normal, maze artifacts affect the value of the pixels, so it’s a much more “structural” kind of error. If color moiré were like accidentally paining one wall of a house the wrong color, then a maze artifact is like having a wall in completely different place in the house than it was originally intended.

Maze artifacts often occur when there are parallel lines at the limit of the sensor’s ability to resolve detail. Because the detail doesn’t fall precisely over one sensor pixel or another, the sensor essentially has to “guess” at what’s right, which can result in errors. These errors can take the form of connection between the parallel lines, hence the name “maze” given to these artifacts.

Here’s one typical example maze artifacts, created by the repeating lines of a metal fence, visible in the center of the frame:

Since these errors occur on the pixel level, they’re often very small, as in this sample. At 200%, the artifacts become more noticeable.

Again, the errors are subtle. For most types of photography, these very minor types of moiré aren’t a huge issue. However, with cameras like the Nikon D800E or Leica M9, with no anti-aliasing filters, these effects would be even more pronounced.

In the below 100% crop, you can see another example of these types of maze artifacts on the left of the image at 100%:

In this 200% crop, you can more clearly see the moiré artifacts in the AC unit on the left, particularly the lower left of the grate:

Reducing Maze Artifacts

Since the RAW processing software used interets what the sensor records, using different RAW converters can have a big impact on the rendering of moiré, depending on the sophistication of the software. The above sample was rendered using Adobe Lightroom 4.

Processing the same file with another program, Raw Photo Processor, renders an image with almost no color moiré and a drastic reduction in the maze effect.

Here’s a final image of the above file converted with RPP using the program’s sharpening setting of 4.0 and a local contrast adjustment of 5, which makes it closer to the standard sharpening setting of Adobe Lightroom.

There’s still a degree of aliasing due to the linear detail at the limit of the sensor’s resolution, but the maze effect is still greatly diminished.

Short of using different RAW converters, maze artifacts are very hard to remove. The upshot is that since the errors occur at the pixel leve, they generally affect small areas of the image and may not be noticeable at most normal output and viewing distances.

AVOIDING MOIRÉ