When it comes to photographic filters (like Neutral Density – ND or Graduated Neutral Density – GND, it’s not possible to avoid talking about colour cast. Although the characteristics of the filters are many and all important, colour neutrality is the one that is immediately understandable to everyone. If we take a photo without a ND / GND filter, and after mounting it we find a strong magenta or blue or yellow dominant (with the same white balance and tint), surely there is something wrong. You invest a great deal of money and time in your photography. Camera, lens, tripod, travelling, time, the list goes on. And all of this culminates in that final shot. So you need to be assured that, to coin a phrase, what you see is what you get. You deserve a perfect colour accuracy.
White Balance and tint
Obviously, you can blame white balance and sometimes, white balance is the culprit. But if you are using a filter where you have been assured neutrality, such as NiSi filters, and the colour is off, you know it’s the white balance.
This can happen especially with strong ND filters, such as 6-8-10-15 Stop. In fact, setting the auto white balance, the camera tries to guess the right temperature for the scene, unfortunately through very dark ND filters this operation is very difficult, and in the end the temperature (in kelvin) and tint (green/magenta shift) values are not accurately chosen.
Obviously this inconvenience is not a problem itself, the important thing is to shoot in RAW and then set the white balance and tint in post-production, just take the value of the shot without filter and set it as WB and Tint. Each evaluation of the filter colour accuracy must be done with the same temperature and tint between before and after. For NiSi filters, with practically zero colour cast, you will get perfect results. If you are doing as per the above with other brands filters and the problem persists. It’s time to upgrade to NiSi.
A final thought, always make sure to mount your filters correctly, properly using the gasket provided in NiSi filters, mounting the ND as first element close to the lens (if using a square filter system such as NiSi V6). Don’t forget to close the viewfinder if you use a DSLR (no need to do that with mirrorless cameras).
The colour cast in ND filters
The purpose of ND or Neutral Density is to be, well, neutral. What we mean by this is that there is no preference in transmission to any wavelength. The filter absorbs all wavelengths uniformly. If NDs do not do their job properly and permit more of one wavelength than another through, you will get a colour shift. Most common colour shifts are down to excess IR (Infrared light) at one end of the spectrum which “warms” the shot giving a magenta colour cast. At the other end of the spectrum, UV light, “cools” the shot giving a blue cast. Achieving the perfect balance of neutrality is an extremely precise art.
An ND64 (6 stops) filter reduces the light reaching the sensor by a factor of 64. So in a perfectly made ND64 at each frequency we must have a light reduction of 64 times. At 450nm (blue), at 500nm (green) or 590 (orange) we must detect the same reduction of light on the other side of the filter. What happens if we detect twice the light for example at 450nm? The photo will most likely have a strong blue tint because the light of that wavelength has not been effectively reduced as in the other colours.
Measuring colour casts with a graph
From the above logic, we derive a graph to indicate the casts of an ND filter. We can put the wavelengths on a graph (in nm) and the corresponding % transmission. We’ll check if the % transmission remains uniform throughout the spectrum. We do not care about the absolute transmission value, this depends on the density (stop) of the filter. If the transmission is 25% it means that 1/4 of the light passes through the filter, (an ND4 2 Stop, 4 times reduction), but this is not the purpose of the test. The test below shows 4 curves of 4 ND filters of different brands.
All NiSi filters (square or circular) have the same glass and optical treatments, so this is a valid test for any NiSi filter on the market today (excluding the old AR series which is no longer in production). The remaining curves are from three popular brands.
Let’s observe the curve of the NiSi filter (top left), the trend is very regular over the whole spectrum of the visible light and even beyond. This indicates that the filter absorbs light very uniformly over the whole spectrum, without being unbalanced in any wavelength. Thus, the NiSi filter is virtually without dominance.
Looking at the others, e.g. “other filter 2“, we see that there is a very uneven transmission from 460nm to 500nm (over 10% drop) which results in a higher absorption around 500nm (blue-green) than 460nm (blue), so most likely a blue dominant. We get close to infrared (around 700nm and above) and light transmission spikes, leading to problems in long exposures above one minute (magenta cast).
Looking at the two graphs on the right: “other filter 1” and “other filter 3“, it is clear how these two filters are basically transparent at red and infrared frequencies. Filter 1 is quite constant at first, but ends up off the scale from 670nm onwards. For long exposures (above 30/60 seconds) the resulting photo will be very prone to magenta. Filter number 3 is the worst among those tested, in addition to being inconsistent from the beginning already at 600nm (yellow-orange) begins a rapid rise up to 100%, already at 690nm. The 100% transmission at 690nm indicates that if we are using a 6 stop ND64 filter this will have no effect on the light from 690nm onwards. So the deeper reds and all the infrared will reach the sensor, creating a reddish /magenta cast. Already at 660-670nm, we are at 50% transmission increase from previous values, so we expect a virtually unusable shot with this filter.
Please note that we used filters that are in a price range of ±£20 from NiSi filters, so in the same category. Also, these are top range products, no better versions of each filter/brand was available at time of the test.
The absence of colour cast in practice
The measurements we have made do not remain a purely academic exercise but have strong relevance in practice. Let’s look at the picture above, taken without any ND filter. What happens when using a non-NiSi filter compared to a NiSi filter?
The filter on the left is the one we tested and called “other filter 2“, even just reading the curve you could have noticed the strong blue dominant. We would like to note that all the photos were taken with the same WB and tint and then exported to JPG from the RAW file without any modification.
Also, the colour cast on the left it’s not constant between shadows and lights (we know looking at the results of the test). An uneven colour cast is the nightmare of every photographer. In comparison, the NiSi filter on the right is virtually free of colour cast, both looking at the graph and the final image.