I recently bought a 10-stop ND filter for my 5D3. It works beautifully, and from normal shooting I'm not detecting any noticeable color cast, which is great. Here's a quick test I did outside:
Without filter:
With filter:
But I'd like to be a bit more thorough, so I did a few controlled indoor test shots with a color-checker, and I need some help to analyze the result.
The first scenario is set on Auto-white-balance (which is what I normally shoot with anyway).
The no-filter test case resulted in a normally exposed shot at a shutter speed of 1/30s, Temp=4250, Tint=+21:
With the filter, I get an shutter speed of 25s, Temp= 4150, Tint=+8
My first impression is that they look very similar, the filtered shot perhaps a tad warmer. A slight tweak will make them look identical. But then I noticed that the Temp and Tint values are not the same as the control shot. The control has a slightly higher Temp (4250 vs 4150), and a significantly higher tint toward magenta (+21 vs +8). So even though the resulting shot looks pretty similar, their Temp and Tint values are different to reach a similar output.
So the 2nd scenario is to use manual white balance so the temp & tint stays the same in both shots, and here's what I got:
Control shot (no filter): WB set to 4300 (and apparently a tint of +8):
With filter: Same Temp =4300, tint=+8
My reaction is that the filtered shot looks better white-balanced, but what really matters is that clearly it doesn't look the same as the no-filter control shop. The temperature doesn't differ much between the shots, but the no-filter shot has a green tint compared to the filtered shot's magenta tint.
So does this mean that my 10-stop filter does have a color-cast toward magenta? Even thought with Auto WB the shots look nearly the same, and sliding the tint slider in post can eliminate any perceived differences.
What conclusions should I draw?
:: My 
