5
\$\begingroup\$

I cannot find any resources that detail a standardized method of measuring the spectral response of a camera. Does such a standard exist? Does anyone know how to do this?

To give some context, I'm trying to measure the spectral response of a camera module for which the manufacturer has not provided the spectral response curves (I'm not sure why...). I imagine this would involve a monochromator, and maybe an integrating sphere, but again I'm not sure.

I hope this is the correct place to ask this! I think this probably falls within the realm of computational photography.

Improve this question
\$\endgroup\$

2 Answers 2

Reset to default
3
\$\begingroup\$

Here are some of the ways: http://www.image-engineering.de/iq-products/iq-tools/measurement-devices/camspec http://www.image-engineering.de/iq-products/iq-tools/measurement-devices/camspecs-express I use a monochromator, light sphere, and a photodiode. You can find most of those supplies at Edmund Optics http://www.edmundoptics.com/ and similar shops. You can also opt for simpler methods using diffraction grids.

Improve this answer
\$\endgroup\$
2
  • \$\begingroup\$ Thanks for this! Is there any particular reason you used a photodiode? This could be swapped out with a spectrometer, correct? \$\endgroup\$
    – Clyde W
    Commented May 1, 2015 at 23:18
  • 1
    \$\begingroup\$ @Clyde W : photodiode is in fact a more precise and simpler way, and something I have a habit to use for the particular purpose of real-time brightness control; but of course you may use a spectrometer. \$\endgroup\$
    – Iliah Borg
    Commented May 1, 2015 at 23:28
1
\$\begingroup\$

Yes, color sensitivity is often measured and specified in bits. DxOMark provides one example and incorporates this into their camera ratings. From the description of their "Color Depth" test:

Color sensitivity indicates to what degree of subtlety color nuances can be distinguished from one another, often meaning a hit or a miss on a Pantone palette. Maximum color sensitivity reports, in bits, the number of colors that the sensor is able to distinguish. The higher the color sensitivity, the more color nuances that can be distinguished.

As with dynamic range, color sensitivity is greatest when ISO speed is minimal, and falls rapidly with rising ISO settings. DxO Labs has focused on measuring only maximum color sensitivity. A color sensitivity of 22bits is excellent, and differences below 1 bit are barely noticeable.

Presumably spectral accuracy, which it sounds like you may also be concerned with, is just a matter of calibration.

Improve this answer
\$\endgroup\$
5
  • \$\begingroup\$ Actually, I'm looking to measure the spectral responsivity of each colour channel (R, G, B) over a range of wavelengths. I want to produce plots like this: maxmax.com/images/Cameras/Technical/… Sorry, my question isn't terribly clear. \$\endgroup\$
    – Clyde W
    Commented May 1, 2015 at 21:03
  • \$\begingroup\$ Ah, yes, that's not something that would concern photographers, but is obviously important for some lab and other scientific applications of camera sensors. As you note it would be straightforward to sample such a response curve using a monochromator. \$\endgroup\$
    – feetwet
    Commented May 1, 2015 at 21:20
  • \$\begingroup\$ @Clyde W Those plots are from JPEGs and have little to do with colour channels responses. \$\endgroup\$
    – Iliah Borg
    Commented May 1, 2015 at 22:38
  • \$\begingroup\$ @IliahBorg I'm not quite sure I understand what you mean. That plot is from the following link: maxmax.com/spectral_response.htm It seems like they are looking at each R, G, or B pixel (from the Bayer array) in raw images, not JPEGs. Could you clarify what you are saying? \$\endgroup\$
    – Clyde W
    Commented May 1, 2015 at 23:16
  • 1
    \$\begingroup\$ @Clyde W : I asked them directly at that time because their curves were far off from sensor specs and my own measurements; and here is the quote from the answer from Dan Llewellyn himself: "Since measuring the response requires taking a picture every 20nm from 300nm to 1000nm, and you need to take both UV-VIS and VIS-IR from separate monochromators, you end up with a lot of pictures. So far, we have just been taking JPG pictures to keep the file sizes and processing time down." \$\endgroup\$
    – Iliah Borg
    Commented May 1, 2015 at 23:26

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.