High Res Photo Mode

Probably the E-M5 II's most significant trick is its multi-shot High Res photo mode. This uses the camera's stabilization system to shift the sensor slightly between each exposure, meaning you effectively 'scan' the scene at higher resolution. It's something we've seen medium format makers do in the past, but this is the first time we've seen a mass-market product offer such a mode.

Because the camera builds up the final image from eight separate exposures, it can't really be used for moving subjects. Instead, Olympus suggests its use for product photography and, at a push, landscape work. The mode offers two delay options: one that adds a pause before the first exposure to let the camera settle, and a second that adds a delay between each shot, to give your strobes time to recycle.

The E-M5 II's high res feature builds up an image from eight consecutive frames:

The first four frames move the sensor so that all colors are captured at each of the existing pixel positions (canceling-out the effect of the camera's color filter array).

The sensor then shifts 1/2 a pixel width up and to the left, then captures full color information about this new, offset position.

Both of these steps have a role to play.

Unlike the Hasselblad system, which uses four or six shots, the Olympus method uses eight. This is because it's essentially doing two things: the first four movements shift the sensor in whole-pixel jumps in a square pattern. This is designed to overcome the limitations of the Bayer color filter pattern: moving each red and blue pixel into all the gaps that would normally exist between them and their nearest same-color neighbor, meaning you sample every color at every pixel. The sensor then moves up and left by 1/2 a pixel's width and repeats its square pattern - theoretically doubling the vertical and horizontal resolution.

However, because there's some overlap between the original set of four pixels and the set of offset pixels, the system doesn't quite yield four times the original resolution (64MP). Instead, the camera compensates for this overlap and outputs a 40MP file. These files are captured both as 40MP JPEGs and as vast, 100MB Raw files.

40MP capture, viewed at 100%

Standard 16MP capture, viewed at 158% (matched magnification)

The disadvantage of taking so many shots to produce the composite image is, of course, that it takes longer to complete all the exposures and is, consequently, more sensitive to movement. However, the plus side is that it means the camera captures full color information about more of its captured area, giving a truly sharp image, if properly stabilized.

As you might expect, the High Res mode has difficulty with subjects that move during the exposure.

A distinctive cross-hatched pattern is clearly visible in this brightened crop from the foreground of image above.

Hi Res mode is limited to a maximum of ISO 1600 and a minimum aperture of F8, presumably to stop the image becoming too blurred for the extra resolution to be particularly beneficial.

How does it compare with the Nikon D810?

Obviously the multi-shot nature of the E-M5 II makes it much less versatile than the natively 36MP Nikon D810. And, even though you could argue that you need to use a tripod to get the very best out of the Nikon, the results don't suddenly drop to the equivalent of 16MP if you don't. However, we thought it'd be interesting to see how the two compared if you do shoot a static subject.

Olympus E-M5 II, High Res

Nikon D810

As you can see, the higher resolution of the E-M5 II's image doesn't give much of an advantage in terms of detail compared to the Nikon, perhaps in part because these are JPEG results that may not have optimal sharpening.

However, the benefit of sampling all colors at all locations is immediately apparent. The Nikon's Bayer color filter array means that it is only sampling one color (red, green or blue) at each pixel position. As soon as detail gets fine enough that it all falls within a single pixel, the camera doesn't have enough color information about it, resulting in false color (as seen in the crop of the black-and-white etching). It's particularly noticeable on the D810, since it doesn't have a low-pass filter to blur this super-fine detail across into the neighboring pixels.

Here we show the same first four steps as illustrated in the animation at the top of this page. This time, we show all three colors, not just red, and focus in on a quartet of pixels.

The first four steps of the E-M5 II's sensor movement ensure that each red, green and blue-sensitive pixel is moved into the gaps between it and its nearest similarly-colored neighbor - building up full color information about each pixel.

Because the first four steps ensure the Olympus is sampling all colors at all pixel positions (in addition to then repeating the process for an offset position), it means super-fine detail doesn't result in false color.

In addition, it also ensures that the camera's color resolution is as high as its luminance resolution (something we've only seen before with Sigma's Foveon technology). This becomes clearly visible in the final crops, where despite the two image's similar pixel counts, the Olympus image is able to much more accurately render the color resolution targets.