Olympus took a quantum leap forward last year with the release of the OM-D E-M5. Its image quality was far above anything seen from a Micro Four Thirds camera before it, and we hoped that Panasonic's newest flagship camera would succeed, or at least match its quality. And it seems that while its JPEGs might lag behind (sort of), the GH3's raw files stand up easily to the E-M5's.
The test images below were taken with the same Panasonic 12-35mm f/2.8 lens, opened to f/5.6 on both cameras. Images were recorded in JPEG at the highest quality setting, with noise reduction turned off. Raw files were captured with the same settings and exported from Adobe Lightroom 4 using the highest quality JPEG settings, using the General - Zeroed preset without any additional post-processing involved.
Surprisingly, from the lower ISO settings the Olympus E-M5 seemed to be recording more detail, with sharper-looking images. This is probably a result of in-camera sharpening, but it also could lend credence to the rumors that the latest Olympus MFT cameras have weak or no AA filters.
At ISO 1600, the GH3's images start to show some fine noise over the entire image. While the E-M5's image remains noise-free overall, some fine details are starting to show wear and tear – for example, straight lines are beginning to look a little ragged. At ISO 3200, the GH3 has more noise, but with a finer grain, while the noise from the E-M5 looks more disruptive.
ISO 6400 is worse on the GH3, with a clear watercolor effect, and the return of the jaggies we first saw on our review of the Panasonic GF5. ISO 6400 is probably the best balance between high sensitivity and image noise that can be struck with the E-M5. ISO 12800 should be avoided if possible on the GH3, while the E-M5 does slightly better. ISO 25,600 is unusable on both cameras.
In contrast to JPEG, the raw files from the GH3 and E-M5 look very similar. In fact, we'd be hard pressed to distinguish between the two if they were put together in a blind test, leading us to conclude that whatever differences arose in JPEG as a result of each camera's unique processing engine.
Again, we see that the Olympus E-M5 seems to capture slightly more detail using the same lens, but the difference is not as apparent as with the JPEGs. Also, the E-M5's noise patterns are less destructive in raw, leading us to think that the E-M5's JPEG sharpening and noise reduction is quite aggressive.
ISO performance at every level is nearly identical. Up to ISO 3200 both cameras perform very well with minimal noise and is a good upper usable limit. Like their JPEGs, ISO 6400 is the ultimate limit and above that images are just too noisy.
In a nutshell, when shooting JPEGs the E-M5 seems to capture more clarity in its images, while the GH3's details are softer, even when shot with the same lens. The GH3's JPEG noise patterns look more artificial at higher settings, with blocky jaggies sometimes visible at ISO 6400. The E-M5's JPEG noise patterns look more 'organic', and while less noise appears overall, what does appear is less forgiving to fine details.
When shooting raw however, differences become slight, and the image quality between the two is virtually identical. The Panasonic's JPEG jaggies do not appear when shooting raw (the same happened with the GF5) and the resulting noise quality is much more organic, similar to the E-M5's. It seems that the GH3's sensor, the first in Panasonic's next-generation G series camera after the E-M5, has stepped up to match the best of what Olympus has to offer.
Of course, this is only a simple JPEG versus raw performance test between the two cameras. In totality, they are very different tools with unique pros and cons. The Olympus E-M5 has an advanced built-in 5-axis image stabilization for example, but cramped controls. The GH3 has no built-in image stabilization, but has more advanced video shooting features and more luxurious controls in a bigger body.