As I understand, ND4 filter should transmit about 25% of the incident light. I have measured transmittance spectrum of my Fujimi ND4 58mm filter, and got the following result:

transmittance spectrum

So, transmittance is about 40%-45% in the visible range of wavelengths. That's quite far from the expected 25%. Is my filter not really ND4 (although I bought it as such, and it has the marking on its rim)? Or is my expectation wrong, and this filter is normal?

EDIT: I've done some more tests:

  • Luxmeter: covered with the filter it shows 39% of its no-filter readout.

  • Canon EOS 1100D manual mode metering: without filter I get the meter at the center with exposure of 0"5, and with filter it goes 4 steps lower, so that to compensate I have to click the wheel 4 times, getting exposure "5" (i.e. 1/5 s). The ratio (1/5)/0.5 is thus also 40%.

  • Canon EOS 1100D raw photo: raw histograms of the photos with the same exposure have corresponding peaks at values having the ratios 38% to 41%, depending on color channel (corroborating the tint visible in the transmittance spectrum).

  • 2
    \$\begingroup\$ you should buy other nd4 filters to get results to compare with. maybe you have just a error in your measuring system. \$\endgroup\$
    – Horitsu
    Nov 12, 2018 at 5:06
  • \$\begingroup\$ How has your measuring instrument been calibrated? \$\endgroup\$
    – Michael C
    Nov 12, 2018 at 7:48
  • \$\begingroup\$ @MichaelClark it's an Amadeus spectrometer with ILX511 CCD. The only calibration I applied was correction for CCD nonlinearity (which was measured by varying integration time and measuring the output) and subtraction of dark-current bias. I also tried lowering the integration time to be in the linear region of CCD sensitivity, but the results are the same. \$\endgroup\$
    – Ruslan
    Nov 12, 2018 at 7:57
  • 3
    \$\begingroup\$ Could you try setting up your camera on a tripod focused on a blank wall. Meter with and without the filter and see how many stops different you get. \$\endgroup\$
    – Eric S
    Nov 12, 2018 at 14:44
  • 2
    \$\begingroup\$ Interesting that a 40% light transmission corresponds to an ND .4 filter, one of 3 ND standard designations. \$\endgroup\$
    – doug
    Nov 12, 2018 at 20:35

2 Answers 2


The filter is almost certainly an ND0.4 (or ND.4) filter.

Filters specified as ND .4 have a transmissivity of 10^(1/0.4) which is .398. Looking at the 4 measurements the OP made, the reflectance at 520nm (approx. peak of visual response) is .40. The other 3, lux, exposure steps of 1/3 f-stop, and raw histogram range from .38 to .41. This corresponds very closely to an ND.4 filter.

There are several ways neutral filters are designated. The most common is in multiples of half stops. Thus an ND2 filter is 1 f-stop, ND4 is 2 f-stops and so on. This yields transmissivity of .50, and .25 which are not even close. Another convention not in play here is to label the filter with a number followed by "x" which shows the transmissivity decrease factor. Thus one stop of attenuation would be labeling the filter 2x, 2 stops, 4x, 3 stops 8x, etc.

Given the very close results of the 4 techniques used to measure transmissivity, one must conclude the filter is best described as a ND0.4 or ND.4 filter. Perhaps the filter was made by Fujimi to target the Tiffen market which uses this designation. Or some error in their translation or even manufacturing.

Whatever the case, there is little doubt it is a ND .4 filter and not an ND4 or ND2 as they are usually labeled.

  • \$\begingroup\$ It seems Fujimi's markings are totally insane. Just bought an ND16 filter of theirs, and it appears to transmit about 1/125 of incident light (measured by Canon 1100D's light meter). \$\endgroup\$
    – Ruslan
    Nov 18, 2018 at 17:13
  • \$\begingroup\$ @Ruslan Sheesh. That's pretty bad. Labeled by the usual convention, that would be an ND14 or ND2.2 (ish). \$\endgroup\$
    – doug
    Nov 18, 2018 at 17:51

The easier way to know empirically how many effective f-stops it has:

  1. It’s to buy a large (A3) mate card any mid grey shade will work.
  2. Then go outdoors on a sunny day, place the card on a wall or similar using pins or tape, make sure it’s evenly light.
  3. Set your camera to AP (aperture priority) and select an aperture of f 5.6, for example.
  4. Without the ND filter. Don’t use the matrix mode. Point your lenses and set yourself at a distance from the gray card so the only thing showed through the viewfinder it is the gray card, and not the pins. Annotate the shutter speed. Tip: Give yourself some leeway, most viewfinders will show less than 100% of picture area and you don’t need to focus.
  5. Mount the ND filter and proceed like explained in step 4 (don’t touch the aperture between steps 4 and 5), annotate the shutter speed.
  6. The shutter speed you have annotated in step 5 should be lower than the one from step 4.

e.g. Let’s say that in step 4 you got a shutter speed 1/1000”, since your ND reduces the light by 4 f-stops, step 5 should give you around 1/62”. Halving the original speed 4 times or mathematically: shutter_speed/2^f-stops in our example (1/1000)/16 = 1/62,5”


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.