Unwanted Effects of ND Filters

2012.10.01

The goal of a Neutral Density filter is to uniformly reduce the amount of light reaching the sensor. This allows a longer exposure, wider aperture and even higher ISO than a normal exposure. Unless photographing something extremely bright like the sun, ND filters are mostly used for getting creative.

Neutral Density filters come in different sizes and critically, different strengths. There are more than one way to measure strength but the most intuitive is the fraction of light it lets through. Most ND filter manufacturers label their filters this way: An ND8 lets 1/8th of light through, an ND16 lets 1/16th of light through, etc.

A perfect ND filter would filter exactly the right amount of light across all wavelengths and uniformly across its surface. Having quite a few ND filters at hand, I decided analyze how the filter and camera behave when an ND filter is mounted on the lens. Naturally, the camera's behavior will vary according to the camera. The one used for the experiment is a Pentax K-5 DSLR with a DA* 55mm F/1.4 lens. As an additional point of comparison, I included an HD Polarizer which also cuts light by 1/2 or 1-stop but not uniformly. As for the ND filters, they are the Hoya HMC ND8, Pro 1D ND16 and HMC ND400.

The camera was mounted with each filter in turn and several shots of both a white-balance card and color-calibration chart were taken. The first observation is that the camera produces highly varied results when left to its own, particularly with exposure.

Using Multi-Segment metering:

• The HD polarizer gives a +1 EV delta which is perfect.
• The ND8 gave a +2 EV delta which is one stop short.
• The ND16 also gave a +2 EV delta which is two stop short.
• The ND400 also gave a +2 EV delta which is way too short. This suggest the metering limit of the camera was reached.

Under the same conditions with Spot metering:

• The HD polarizer gives a +1 EV delta which is perfect.
• The ND8 gave a +2.5 EV delta which is half a stop short.
• The ND16 gave a +2.5 EV delta which is 1.5 stops short.
• The ND400 gave a +3.5 EV delta which is way too short again. This shows that Spot metering has a lower than Multi-Segment.

Manually setting exposures to the correct amount according to the strength of the ND filter gives perfectly consistent results in terms of exposure with all filters but reveals variations in white-balance.

Given a filterless exposure of 1s at an aperture of F/4 and ISO 200:

• The HD polarizer needs 2s and has virtually no effect on white-balance. This I actually knew already since it was the reason I sold my B+W filters which gave terrible color-shifts and replaced them with Hoya HD Polarizers.
• The ND8 needed 8s for the same exposure. White-balance however became noticeably greenish. The grey patch on the color chart with RGB values (199, 199, 199) unfiltered became (189, 202, 179).
• The ND16 needed 16s for the same exposure and had absolutely zero effect on white-balance, keeping the grey patches all perfectly neutral.
• The ND400 was exposed for 401s which gave the same brightness but gave results a slight bluish tint. The RGB (199, 199, 199) patch became (199, 213, 216), so technically less red, rather than more blue.

It appears that ND filters are well calibrated in terms of exposure but the camera struggles with lower light levels. Other than the ND16 Pro 1D and HD Polarizer, these filters have an effect on color. Not having multiple copies of the same one, one cannot say if this is a sample variation or simply inherent properties of the filter.

Clearly the best filter is the Hoya Pro 1D ND16 and it certainly deserves its Pro label and higher price tag. This one comes in strength between 1 and 5 stops. There is no ND400 version or even close. To get perfect WB with such a dense filter, it is probably better to shoot RAW since reduced light-levels do not allow a Custom WB reading to be made with the filters on. The HMC ND8 filters should be avoided as much as possible.

• Digital Photography