Very interesting idea indeed, however I'm having a bit of trouble believing the gradients on both wheels add up uniformly. I don't doubt my geometry can be seriously flawed, but my intuition tells me there is going to be some "vignetting". :(

Ah, forget it. I wasn't thinking straight. The 3D representation fooled me because I wasn't imagining the stacking of the gradients correctly with their centers on the same axis (light path axis).

The yellow arrow is on the right spot, you can only use the top half of the filters because the way they are represented (perfectly aligned) they will be uniform on all the filter (top & bottom) but as soon as they start rotating in the black arrow direction it will become darker on top (yellow arrow zone) but it will become lighter on the bottom and the discrete transition from black to white would immediately show a pattern.

It makes perfect sense now. It's a great invention, the only drawback I see is that the diameter of the filter has to be at least twice the size of a similar VND. Well worth the advantages (light transmission up to 90%).

@nomad Thanks for the extra input.

I didn't knew about those other issues, I rarely have gone into trouble even with cheaps VND (perhaps because I rarely use them). The cross effect only shows when I'm using it at the maximum. Now, 25% light transmission against 90% that's decisive. :)