I have a 6-stop neutral density filter (uniform density, not graduated) that I have been using with my D800E and AF Nikkor 24-85mm 1:2.8-4 D lens.
Comparing the meter reading (in A mode) with and without the filter, I get a difference of about 6 stops (e.g. 1/125 sec. without, 1/1.6 sec. with, both at f/5.6). But while the 1/125 sec. exposure without the filter appears correct, the 1/1.6 exposure is very dark. Exposing for 15 sec. at f/5.6 (in M mode) gives me a pleasing result, although the light meter thinks this is much too long.
I've heard that a linear polarizer can fool a light meter, but can a ND filter do so as well? Has anyone seen this phenomenon before?
I guess I should experiment with other lenses/bodies. If I do, I'll report the results.
With long exposures using the Vari-N-Duo I often have to dial in exposure compensation of a stop or more. It depends on the degree of ND effect dialed in as well as the scene. While center weighted metering helps, it's not the full story.
I don't worry about it too much - I use the histogram and take multiple exposures to "bracket" the exposure for the scene. After a couple of frames, you can estimate it pretty accurately.
In practice I use Aperture priority much like manual mode. Exposure compensation is my adjustment.
IR photography has a similar issue. The scene can make a big difference in the difference between the meter reading based on visible light and the required exposure based on IR. You just get used to the scenes, exposures, and what kind of adjustment is required.