I saw this question, which was too broad, and I thought of a more specific question that can hopefully be answered.

There are "normal" (visible light spectrum) cameras, infrared cameras, and UV cameras, and there are also specialized VNIR cameras that capture the visible spectrum plus some of the near-infrared range. As I understand it, the typical application of VNIR cameras use filters to shift the NIR frequencies into the visible spectrum, to be captured by red, green, and blue sensors, producing false-color images that are essentially a composite of the NIR color information and the visible.

But what if, instead of concerning ourselves with the ability to compress the range of frequencies down into the visible spectrum for the sake of human eyesight, we designed the camera with the goal of recording the widest possible range of frequencies (for example, inventing a new image format with more than 3 channels per pixel - say, R, G, B, IR-A, IR-B, IR-C, UV-A, UV-B, UV-C, designing a sensor matrix of subpixels sensitive to those channels, etc.)? Accepting that we would have to substantially improve the manufacturing of the sensor array to achieve the same resolution as an ordinary RGB matrix, let's prioritize the minimizing of chromatic aberration over maximizing detail.

How far can we extend the frequency range of a single camera? If all the light (or electromagnetic radiation) we capture still has to pass through the same aperture and lens(es) at the same time, could we get all of the infrared, visible, and UV spectra at once? Could we go further than that? What factors impose the hardest limitations on such an endeavor? If possible, please differentiate between something that's definitely, completely impossible, vs. just prohibitively expensive.

  • 1
    $\begingroup$ Assuming we could do that, how would you know if you captured the results without converting it to some sort of false-color alternate? Even things like the Hubble and Chandra have to convert the information into something a human can interpret. $\endgroup$
    – grfrazee
    Mar 17 '16 at 22:12
  • 1
    $\begingroup$ This is a duplicate of a (closed) question somewhere on physics.SE . TL:DR: it's a meaningless question since various methods can capture almost any wavelength. $\endgroup$ Mar 18 '16 at 11:38
  • $\begingroup$ BTW, you're dead wrong that IR cameras "shift" wavelengths. That's done when the output display is created, by assigning various visible colors to the signal strengths recorded for various IR bands. $\endgroup$ Mar 18 '16 at 11:39

A raw microbolometer array1 is completely insensitive to the wavelength (energy per photon) of the incoming radiation — it simply measures the temperature rise that's due to the total incoming energy. We generally use them only for longwave IR, because they're the best technology for doing that, and we have better technologies for shorter wavelengths anyway.

But you could conceivably create a sensor that had multiple bolometers per pixel, each one looking through a suitable filter for a range of wavelengths.

1Note that complete cameras have intentionally limited response to short wavelengths because of the materials used in the lenses.

  • $\begingroup$ No, they don't. The response to shortwave is either due to the limited QE of the sensing layer or to make the output image more closely match what our eyeballs' response is. $\endgroup$ Mar 18 '16 at 11:41

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