Olympus E-620 Review / Test Report - Resolution
DSLR Reviews -
Sunday, 09 August 2009 04:44
Page 3 of 7
Let's start with some basic explanations to understand the following analysis. Upon naive observation you may think that all cameras with the same number of mega-pixels provide the same output resolution. However, things are not as simple. The output quality depends on a variety of factors: lens quality + AA filter + sensor resolution + in-camera processing + (optionally) external (PC-based) post-processing.
Our benchmark lens for the following analysis was the Olympus 12-60mm f/2.8 @ 14mm f/4.5. Originally we used a Zeiss ZS 50mm f/1.4 but there were some unresolvable problems with the setup for whatever reason. However, the center resolution of this wide-angle zoom lens should be as good as the prime here.
Olympus DSLRs are rather complicated beings with respect to their effective sensor resolution. The Olympus engineers have obviously a primary design objective: avoid jaggies/moirees/false colors. This is achieved by implementing a very strong AA filter ("Anti-Aliasing" or low-pass filter). The other manufacturers are also using AA filters (except Leica and Sigma) but the current trend is to use rather weak ones. A strong AA filter works out fine for eliminating moiree effects but it has a negative impact on the resolution characteristic. As you can conclude from the MTF chart below the E-620 performs mediocre here although it's not really worlds apart from the rest. The delta in resolution is lost in the AA filter (reads: blurred). The old Panasonic L10 has a vastly higher RAW output resolution (when using ACR and not Silkypix) despite of its older 10mp sensor for instance - it relies on a very/extremely weak AA filter (just like the Panasonic G1) but consequently it is also more vulnerable to moiree problems.
JPEG images straight from the camera tend to be softer on most cameras but Olympus managed to surpass Adobe Camera RAW with direct JPEGs (set to "Noise Reduction OFF"). This may be surprising but then camera JPEGs are nothing else but RAW data converted within the camera. The "sharp" JPEG output is one reason why you won't see any resolution differences compared to JPEGs from other cameras.
I guess many Oly users will complain that using an Adobe product (here Adobe Camera Raw) for RAW conversions yields in less than optimal results. The chart below compares ACR vs Olympus Studio (RAW converter) conversions and the straight JPEG quality. As you can see the MTF results are _roughly_ the same. Now it is certainly valid to state that ACR-processed images look softer. Why is that ? Olympus images are (very) "oversharpened" whereas ACR images are "undersharpened". Or in other words: Olympus applies a heavy amount of sharpening during the post-processing whereas ACR does not. Technically (less so visually) ACR images contain at least as much scene information which can be emphasized by applying more sharpening (which would also yield in slightly higher MTFs again).
Here's a (400% magnified) comparison between the ACR output and Olympus Studio. The rather extreme oversharpening should be fairly obvious here (the edge halos are an indicator for oversharpening):
Adobe Camera RAW:
Olympus Studio (NR off, Sharpen=Normal, applies to straight JPEGs as well):
When looking closely at the Olympus Studio output above you'll notice a typical side effect of strong sharpening so
let's investigate the impact on the sensor noise characteristic now ...