Anti Aliasing Filter Simulation
One of the most publicized and discussed features of the K-3 is its Anti-Aliasing Filter Simulation mode. Conventional anti-aliasing filters (also known as optical low-pass filters), work by blurring fine detail across neighboring pixels so that information about it is captured by red, green and blue pixels (at least to some extent). This helps to reduce image artefacts such as aliasing and false colour, but at the expense of overall image sharpness. What AA filters don't do is prevent the artefacts that can stem from trying to capture fine repetitive patterns that clash with the camera's sampling frequency. These are unavoidable without using incredibly strong blur filters.
In recent years, we've seen cameras such as Nikon's D7100 and Sony's Alpha 7R forego their AA filters. In these cameras, the pixel counts are so high that the frequencies that an AA filter would need to blur are also very high. Most of the time there's already sufficient image blurring from either lens aberrations, focus errors, camera/subject motion, or diffraction (at smaller apertures) that an AA filter is unnecessary. It's only really still needed when using a sharp lens, perfectly focused, at optimal apertures.
The K-3 offers a clever approach in these circumstances: it uses its in-body image stabilisation mechanism to deliberately move the sensor during the exposure, so that detail is slightly blurred across more than one pixel. The advantage of this approach is that it's optional, so doesn't have to be applied to images that don't need it.
Filter mode demonstration
The K-3 has two AA Filter Simulation modes, called simply Mode 1 and Mode 2. Ricoh says that Mode 1 aims to strike a balance between moiré reduction and resolution preservation, while Mode 2 pushes the balance further towards moiré reduction. It's given us no further technical details of how this is achieved. You can either manually select one of these modes (or 'Off'), or engage the K-3's AA Filter Simulation Bracketing mode. This takes shots in quick succession at each of the three settings, so you can select the most appropriate one afterwards.
Below you can see a demonstration of the anti-aliasing filter modes in action. This widget contains, as well as . Both Raw and JPEG images are included - Raw with finely detailed sharpening applied to show the differences, and JPEGs (which have much less detail) to show the magnitude of the real-world impact.
As you can probably see, engaging the AA Filter Simulator does result in a visibly softer image, but there's not a huge difference between the two modes. The key thing to understand is that AA filters represent a trade-off: achieving the highest resolution is balanced against preventing artefacts from the Bayer pattern. Compare the Off mode toand you can see the difference in detail.
It's been suggested the difference between the modes is that one mode works in one dimension, whereas the other uses a circular movement; however our test shots suggest this isn't the case. The Off mode shows a distinct interference pattern at the center of the star target. This is greatly reduced inbut there are still slight hints of it, and it's symmetrical - the pattern is occurring at the same point horizontally and vertically. further blurs the pattern (and the fine detail) away, suggesting the difference between modes is one of magnitude (how far the sensor moves during the exposure), not a change in the shape of the movement.
As we've already said, interference patterns can still occur: look closely at the centre ofand it appears that there are diagonal waves running from the top right to the lower left. The same is true, even with Mode 2 engaged. The same can be true with false color: it's much less noticeable in , but it's still visible.
Another nice aspect of the filter bracketing mode is that shots are taken in close-enough succession that there's generally little camera movement between the exposures, even when shooting hand-held. This means that in most circumstances where you'd see any benefits to the leaving the AA filter off, you can probably blend-in cleaner sections from Mode 1 or Mode 2 images to replace any parts of the main 'Off' image that show prominent false color.
Overall, we didn't feel that the AA Filter Simulation is a tremendously significant feature, although it's cleverly enough implemented that its usefulness is maximized. Either the lens quality, aperture-induced diffraction or focus precision is likely to play an anti-aliasing role anyway, so often the AA filter simulator isn't necessary. That said, when you're shooting on a tripod with a lens near its sharpest aperture and you've fine-tuned your focus, it's nice to be able to shoot without an AA filter, but also have the option of engaging AA Filter Simulation bracketing so that you've got a fall-back shot if any artefacts have crept into the image.
The K-3 is one of the few cameras that lets you adjust not only the intensity of its sharpening, but also the radius over which it's applied. We tried all three settings and, while the Extra Fine setting pulls out the most detail, it can cause maze artefacting in high detail areas.
|Standard Sharpening||Fine Sharpening||Extra Fine Sharpening|
This is a 100% crop of the AA Filter Simulation Off image from the Real World example above. We found the Fine setting did the best job of making the most of the camera's detail without adding too much in the way of artefacting or false detail.
We've already published an article about the K-3's behavior with the optional, dedicated FluCard wireless SD card. As detailed in that article, the combination is pretty impressive - providing functions, such as extensive remote control of the camera, that aren't always available from integrated systems.
The implementation isn't perfect - the FluCard's beeping, every time you turn the camera on is distinctly annoying (and can't be disengaged, even if the card is set to 'Off' by the camera). It's a particular problem if you regularly turn the camera off to preserve batteries, when shooting intermittently. The other quirk we didn't enjoy about the FluCard stems from its peripheral nature - because it's not fully integrated into the camera, there's no option to send reduced size JPEGs across Wi-Fi (a feature that most other systems offer). Instead you either have to set the camera to save small JPEGs or you have to use in-camera processing to re-process all the images you want to Wi-Fi, before you send them.
Generally, though, we're impressed with how much Ricoh has managed to wring out of an accessory SD card, especially as it makes Wi-Fi optional, but doesn't depend on an external unit awkwardly dangling off the camera.