Seems the new d800 36MP camera is causing a lot of debate lately.
It also seems that many people just assume that the extra MP over the 5d2 (and 3) are a linear gain in resolution. It does not seem that can be physically possible, right?
Exactly how much of a gain will one see in image impression by adding those extra MP on a full frame sensor? How much is diminished by the added MP?
The reason I bring this up, is that I used a t1i as a second camera to my 60D, with top-end glass up front. The 60D has more resolution but on a pixel level it is less sharp and contrasty than the t1i. When printed to the same size, I'll be damned to find much difference in the impression of the photo.
In a way the higher MP is nothing more than upscaling a digital image, after a certain point.
The way I look at it, there is only so much light that the sensor can capture to create an image truthful to the detail and color of a scene. At a certain point you've captured the majority of those with a certain MP on a given sensor size. This is evident if you take an 18MP crop sample and 18MP FF sample with lenses with similar resolution/mm of sensor area. The 18MP FF will always show a better representation of the light captured due to increased micro-contrast, color correctness, tonal transitions, etc.
So, what exactly is the gain from a D800 over a 5D? I am guessing the overall photographic impression when printed large is going to be less than a 10% gain. An orange lets loose a lot of juice on the first squeeze, but later you have to really hammer the pulp to get any more drops out of it. I think the d800 is in the territory of pounding pulp. 
Is the d800 going to be better when printed large? Most likely to a degree.
Is it worth dumping all your Canon gear for it over a 5d2 or 3? Most likely not.
D800 still is going to be an incredible camera though! 
