Leon Benzakein wrote:It looks like the VND introduces IR contamination.
It now makes sense to have the IR cut the closest to the lens because you need to remove the contamination introduced by the VND.
I know this is reviving the thread a little, but a few weeks isn't too bad and I thought this was worth a clarification.
Neutral Density filters do not introduce IR contamination. The IR contamination is already there. What happens is that they remove other wavelengths of light, leaving the IR contamination more dominant. It's a subtle difference, but I think an important one.
When you put a ND filter on what it is doing is blocking a certain amount of certain wavelengths of light. ND filters are tuned to specially block visible light. Different manufacturers and types use different processes, with VNDs typically leaving behind the most untouched IR (and UV, but that's usually not as big a deal). Uneven blocking of different wavelengths is why different NDs will give a slight colour cast.
For example, if you go on Hoya's site they provide transmission curves for most of their filters. If you look at their ND8 filter
http://www.hoyafilter.com/hoya/products/ndfilters/ndx8/you can see that the transmission curve is pretty good, but not perfectly flat. It passes slightly more red that blue light with some minor wiggle in-between. Their curve cuts off at the 700 mark right where IR light starts, but I suspect if it continued further there would be a noticeable curve upwards.
This is also why combinations of different ND filters can produce quite different results in when used with different IR filters on different cameras. It's a combination of what wavelengths the ND is passing, what wavelengths the IR is blocking and what wavelengths the camera sensor is particularly sensitive to. Different manufacturers of all three are going to have different cut-off points.