This is sort of a weird place to put my little opinion on this issue but I think Vitaliy thought the whole idea was "flame" so I think this is the only appropriate location to state it.
I would love to see a scientifically controlled comparison between all these cameras (arri alexa, red 1, epic, scarlet, gh2, gh1 etc etc.) but this isn't quite it since they compared a package of goods against another package of goods as in the package of red epic and a red lens against the package of the hacked gh2 with the Olmypus zoom lens. The clear problem with this is you can't tell if its the lens or the camera body that is falling short. In this case I have to suspect the red lens just isn't as good as the Olympus zoom lens.
I thought the people at Zacuto did a fantastic series of tests with the 2010 and 2011 SHOOTOUT. They put top end 10-30k lenses (same lens on all the camera bodies). With that test the lens quality is taken out of hte equation as were environmental conditions which were also held perfectly equal. At that point the only difference was coming from individual camera bodies.
A test like that with the epic is the only way to compare it to the gh2 or any other camera body for that matter.
This is just my suspicion but I believe the Red lenses aren't on the same level of as say the top end Zeiss, voightlander or the top end Olympus lenses. I've seen tests with the same camera body where zeiss lens came out with more clarity and color than either sony F2 lenses and red lenses. My own experience is that zeiss lenses are about the same as top end olympus lenses on quality of clarity and color coming through the lenses. So it would not be a surprise if the olympus lens is indeed better than the red lens used in that test.
At that point its not an accurate test of the red epic against the gh2. One has to use the exact same lens on both cameras first. They should have used the red lens with a pl-m43 adapter on the gh2 and then compare that footage and see how they both look.
I personally think both are great but its a very high probability that the epic is better overall, not just on abilities but clear out footage quality as well. At the very least its easier to grade in post production. But I have to admit I like the idea of a gh23 barely being able to keep up with a red epic, and to such a degree that if you are really careful you can smoothly cut gh2 footage with red epic footage.
On the 2K versus 4K issue, it's always an advantage to record at a higher capture resolution than your target distribution resolution. It gives you the ability to work more wonders in post-production while minimizing the risk of degrading the final results to something less than 2K. When your final rendering step is downsampling from 4K to 2k, you have many options for selectively smoothing and/or sharpening the print as desired.
On the GH2 versus Red issue, one thing I noticed about the Scarlet is that its crop factor at 3K is almost identical to the GH2. This means that with interchangeable lenses, you will get the same field of view on the 3K Scarlet as on the 1080p GH2. With well-matched film modes between the two cameras, intercutting GH2 into Scarlet footage should be very practical, with the Scarlet's 3K resolution insuring excellent quality in the distributed 2K prints.