I don't understand why you think that comparing relatively uniform areas on-screen will give you a good idea about "noise characteristics" in a camera
Actually, this the only way to compare noise characteristics, although "comparing" should be done by "measuring" the noise, not "judging" it.
Example: shots with the Canon 40D, once @ ISO 200 and once @ ISO 400,
with the same exposure, i.e. the same amount of light has been captured. An area has been selected and the noise measured on the noisiest channel, which happens to be the red in the illumination of that setup.
Of course, the initial intensity in the ISO 400 shot is twice as high as in the ISO 200 shot. Actually, the red channel is 1.09 EV higher (9.90 vs. 8.81).
The noise in the ISO 200 shot is 48.2%, while in the ISO 400 shot it is only 33.4%.
Don't you think, that this is an
objective measurement of the noise? It does not depend on the scenery (only that the measured areas have to depict smooth, unicolored, evenly illuminated surfaces), nor on the color, nor on the actual illumination, nor on the exposure, nor on the lens, etc.
Even if you're correct that the per-pixel noise is similar, the clue seems to be:
maximum printable size
Is that an objective, comparable measurement? Like measuring something in nanotechnology with a carpenter's measure tape?