I soon found some cheap variable density filters on eBay. I paid £7.79 including postage and it arrived a couple of days later.
Here is the picture from the eBay listing.
The Neewer filter seems well built but I will not tempt disaster by getting it wet.
There are graduations along the side with 'min' at one end and 'max' at the other.
I have read poor reviews of even expensive variable density filters. The problems seems to be inherent in having two sheets of polariser stacked one on top of the other. The graduations are non-linear, in other words, rotating the filter from A to B, say, does not increase the density by as much as between B and C They are not, like any polariser, suitable for wide-angle lenses and there are numerous reports of colour casts.
So, I decided to test the density at each of the marks on the scale. To do so I used the Light Meter app in my iPhone 6. I mounted the iPhone on a tripod, aimed at an evenly lit wall and measured the shutter speed at a fixed aperture. Therefore I obtained the difference in Exposure Value (EV) (i.e. stops) between having having no filter there and the filter at all its marks on the scale except 'max'. At 'max' the filter was unusable because of the appearance of a dark mesh.
Here are the results. The effect of the filter at its minimum setting was one and a third stop compared with the absence of the filter. Sure enough, there was a non-linear increase at each mark, with a steep increase in density towards the maximum end. The verdict was that with this information, density can be dialled in for a stop changes (to the nearest whole stop) from 1 to 6.
I also looked at the effect of the filter, using the same set-up, on colour temperature (Light Meter app gives a read out in degrees Kelvin). There was a shift which varied with setting:
How serious is the maximum shift in colour temperature? Well I took a (Raw) photograph on my Nikon D700 and adjusted its colour temperature to the 5476 K (the without filter red line in the graph above). I then made a copy and adjusted its colour temperature to the minimum value the filter produced (i.e. 5150). The result is below. Remember that these photographs were not taken using the filter, they are a simulation of the maximum shift in colour temperature that the ND filter produces:
|Simulating the effect of the ND filter on white balance|
Left WB adjusted to 5150 K; Right to 5476 K
Finally, to look for any deleterious colour casts, I put the filter on the Sony AX33 camcorder and tried it at the various marks. I left the camcorder on automatic so that it would compensate for exposure and white balance. Could I then see any colour cast from using the filter? The short answer was no.
Therefore, I am really rather pleased with my £7.79 variable neutral density filter. What is more I find it fits my Nikon P610 superzoom bridge camera, but more on that later.