Image Quality

Resolution

One of our concerns about the Olympus E-M5 II was that, in an age where 24MP is commonplace, its 16MP sensor seemed a bit behind the times. The PEN-F addresses this with a jump to 20MP. In Raw, detail is properly resolved without aliasing to roughly Nyquist / √2, which is normal for a Bayer type sensor and gives the 12% increase in linear resolution that we'd expect to see from a 16 to 20MP jump. We see a similar performance from the GX8, but it's a tiny bit softer than the PEN-F, probably because of shutter shock.

It's an issue that is familiar to Olympus, and can be seen on previous models such as the E-P5. Thankfully, not only does the PEN-F offer the 'anti-shock' electronic first curtain mode (which is the 'standard' shutter mode for shooting our test scene). Even with the mechanical shutter, there's no significant shutter shock on the PEN-F.

JPEG Performance

There are some key differences in the JPEG engines between the GX8 and the PEN-F, starting with sharpening. The text only tells us a little bit; the Olympus shows mild sharpening halos and might have a slight bit more contrast, while the Panasonic seems to leave behind more fine detail. In fact, the Panasonic's JPEG can end up with similar amount of detail to the Olympus, even though it starts out with slightly less detail in the Raw file. Conversely, other parts of the scene show that the Panasonic's sharpening can be too much with certain details, leaving behind an unsightly 'stair stepping' pattern along sharp edges.

Color at base ISO is fairly similar, with the Panasonic showing more saturation in the blues and the Olympus showing a bit more kick in the reds. Compared to the E-M5 II, we see red has actually been turned down a touch, which is most evident in skintones, both dark and light. The results aren't exactly better for every skintone.

As the ISO is increased, the Olympus does a good job of maintaining saturation. By comparison, the Panasonic's end up a little bit dull and unsaturated. By default, the PEN's auto white balance, the doesn't try to neutralize our tungsten light as much as the Panasonic, however, this is a setting that can be adjusted in the menus.

The two also handle noise reduction differently, with the Panasonic employing context sensitive NR, and the Olympus applying a single NR level that strikes a balance between noise and detail retention. This allows the Olympus to preserve more fine details at high ISOs. 

Raw Performance

The new sensor for the PEN-F does offer a slight resolution increase over the E-M5 II without impacting low light performance. It's a bit disappointing to see that while there might be a tiny bit less noise on the PEN-F, there's no noteworthy increase in performance that doesn't fall outside our margin of error. Then again, it's not a huge surprise, as this new sensor does not feature any noise decreasing innovations like backside illumination. We do see a bit better performance over the E-M1, which could be due to the light cost of the E-M1's PDAF system. 

The great news is with this new sensor there is no noise cost to using the silent electronic shutter. This means low light street shooters need not worry about the added stealth impacting noise performance. 

Compared to the GX8 at both base ISO and high ISO, the Panasonic appears to have slightly more noise. This is most likely due to the slightly stronger saturation of the ACR profile for the GX8, and can most likely be resolved through processing. 

All in all, the PEN-F offers more features over the GX8 that allow it to make the most of the new 20MP sensor's resolution, but otherwise have near-identical performance in Raw and are mainly differentiated by how their processors handle JPEG files. For low-light, macro, and telephoto shooters (or any combination of the three) our test chart leans towards favoring the PEN-F.

High Resolution Mode

Like the E-M5 II, the PEN-F offers Olympus's unique 8-step high resolution pixel-shift mode, where the camera uses its IBIS system to capture 8 shots, two sets of four, each canceling out the Bayer pattern. The second set of four images is offset by 1/2 pixel vertically and 1/2 pixel horizontally, quadrupling the resolution.

The issue with this mode is it reduces the effective pixel pitch by a half, makes each pixel effectively 1/4 of its former area. This stresses the more affordable Olympus 45mm F1.8, our standard Micro Four-Thirds studio scene lens, so for this mode we switched to the top of the line Panasonic-Leica 42.5mm F1.2 Nocticron.

It is apparent Olympus knows the stress this mode puts on lenses, because they really crank up the sharpening for the 50MP JPEG files that come out of the camera alongside the 80MP Raw files. That said, there still is plenty of detail in the resulting JPEG images. In fact, the detail retained is comparable to an OOC Jpeg from the Canon 5DS, although the Canon still shows more detail when viewed in Raw mode.

The advantage of using two sets of four shots to build up the high res image (rather than just using two, offset images) is that the camera captures every color at every pixel. This means that you get greater color resolution than a single Bayer image and much less risk of false color and aliasing in high-frequency detail. The downside is that your scene needs to remain essentially unchanged for the duration of the eight shots.

The other advantage towards this pixel-shift mode is it also reduces noise. At ISO 1600, the Olympus shows slightly less noise when compared to the a7R II (despite the Sony's much larger sensor), however that advantage will be lost once sharpening is applied, since it will exaggerate noise is the file. In conclusion, the High Res mode does have the ability to produce some detailed images, but can't compete with a higher native resolution and the appropriate glass to go with it.