Well that is some peculiar cheating.... I have used the cam extesnively for landscape shooting (a trip in the nearby hills) and compared to my GH4 which scores as almost the best mFT sensor out there the difference is really notable. Noise a bit but it is dynamic range which is related to noise but somehow not reflected in DxO scores that is a real difference. A visible one. SOmeone showed me a pic of an E-M1 shots of mountains yesterday and I was disappointed. Thought I fooled myself because my first reaction was..."Well that is no better than my GH4, it looks identical" only after a few seconds I noted this was shot with the E-M1. A sensor identical to my GH4.
I can assure you em1.2 shots look visibly better in every aspect of the IQ and it is a notable but surely not a huge step up. GH4 and GX8 and PenF already have very solid still IQ. But this is very similar to A6500 shots.
I did not expect anything like it, I too thought this is just that 20 Mp sensor and well it is an excellent cam with fantastic Hires.
Which is a second point: the HiRes mode is thought to be rtestricted to 50MP at best in PenF and here we do see some noise coming in whereas the 40Mp shot of the EM5.2 does not. However: on the Em1.2 I can take the RAW now and with the cheapish 20 mm 1.7 Panny the IQ is brilliant at 65 Mp. It is sharp and noisefree. This has got to be due to the better 20MPO sensor. Combining 7 files with better IQ jsut gives a better overall performance.
So some trickery or not: I cannot say it is or is not so but I can say that compared to my GH4 every aspect of the IQ out of this cam is visibly better. Not just noise..