I'm designing a light source that emits several wavelengths of physical light (via LEDs). The light emitted will be measured with a CMOS sensor (from a mirrorless camera), and in order to properly calibrate the measurements I want to do, I need to make sure that each wavelength of light is at the same level of brightness. I have a decent handheld photo-spectrometer (Sekonic C-700-U, and I've taken some measurements, but the results I am getting back are a bit confusing.

The wavelengths I want to measure are roughly 440 nm, 535 nm and 660 nm (corresponding to status M densities in film densitometry). The photo-spectrometer I am using can measure the brightness of an ambient light in lux, and also produce a line graph of the power (I think?) of ambient light in Wm2nm-1 vs the color in nm.

From the measurements I have taken, these are the approximate brightness values in both lux and Wm2nm-1 ((watt × meters2 × nanometers-1, which is what the meter documentation claims. I'm not super familiar with this measurement, but I think it reflects power over an area modified by wavelength) of the three colors:

  • 440 5260 lx 8 Wm2nm-1
  • 535 37700 lx 1.8 Wm2nm-1
  • 660 5980 lx 3.5 Wm2nm-1

As you can see, there is very little correlation between the power emitted at each wavelength and the brightness in lux. For my purposes, which measurement is more useful, the power per meter squared per the inverse of the wavelength, or the lux measurement?

  • 1
    $\begingroup$ Depends on application the reading is used for. If you are making a cutting laser, you only care about the energy in it and do not care how the human eye might perceive it and if you are making a flashlight you only care how the human eye would perceive it. $\endgroup$
    – DKNguyen
    Commented Jul 22, 2022 at 15:20
  • $\begingroup$ In this case I only care about how a CMOS sensor would perceive it. Now of course all sensors have their own specific sensitivities to different wavelengths, so in order to get it balanced relative to the sensor I would need to calibrate it to that sensor. But I would like to start that calibration from a known point of equal output. My guess is that from the perspective of the sensor, lux is more important because it is counting photons rather than the energy transmitted by those photons. $\endgroup$
    – flimsy
    Commented Jul 22, 2022 at 16:09
  • $\begingroup$ @flimsy Seems roundabout to use lux for that unless what you care about is what the human eye sees when viewing the video produced by the camera. Because otherwise you might as well skip the photometric middleman and just expose the CMOS sensor to known light sources of a known radiometric intensity and map that to the output intensity. $\endgroup$
    – DKNguyen
    Commented Jul 22, 2022 at 19:30
  • $\begingroup$ I actually don't care about the human eye in this case. This is all about sensors measuring the relative intensity of different frequencies of light. $\endgroup$
    – flimsy
    Commented Jul 22, 2022 at 21:09
  • $\begingroup$ @flimsy But what kind of intensity? Normally the needs you must meet will dictate this right off the bat but it does not sound like you have any. $\endgroup$
    – DKNguyen
    Commented Jul 22, 2022 at 22:41


Your Answer

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

Browse other questions tagged or ask your own question.