4 ms·

> Note that the idea of “true color” for these sorts of astronomical photos is itself a bit ill-defined. Is it? I always figured “true colour” would be whateve

> Note that the idea of “true color” for these sorts of astronomical photos is itself a bit ill-defined. Is it? I always figured “true colour” would be whatever you’d see with your naked eye from an approaching or orbiting spacecraft, or the nearest approximation that (say) a consumer digital camera would give you. Can we get anything like that from our current compositing techniques?

privong 10y ago
> Is it? I always figured “true colour” would be whatever you’d see with your naked eye from an approaching or orbiting spacecraft, or the nearest approximation that (say) a consumer digital camera would give you. Can we get anything like that from our current compositing techniques? For many astronomical observations, the wavelengths of light observed are ones our eyes are not sensitive to. In this particular example of aurora, it's the ultraviolet. For others, it's infrared, millimeter/radio, or X-rays. For those, there's no way to way to make an image that our eye would see, because it wouldn't see anything. Some of the color images from those wavebands are described as "representative color" because the images are created to show certain features linked to the underlying physics or properties of the objects that are being shown. For images that purely in the visible portion of the spectrum, in principle one could construct an image "as the eye would see it". However, as @dakr noted[0], color astronomical images are composites of several black and white images taken through different filters. to create an "as the eye would see it" image requires combining those black and white images in a way that considers the relative sensitivity of the different types of color-sensing cones in our eyes. But, if you really want to get it right, it's more complicated than that. The shapes of the frequency-response of the cones probably differ from the transmission of the filters used in the astronomical observations (not to mention the quantum efficiency of the CCDs used to take the observations). So one would also need to correct for the difference in filterobject versus eye_coneobject (both of those as a function of wavelength). This is typically referred to as a "color correction" (meaning the "color" of the light emitted by the source), and requires some knowledge of the intrinsic spectrum of the source. And there's even another layer after that. If emission (or absorption lines) from atoms lie within the observation filters, those can contribute a non-negligible amount to the signal in an observation with a given filter. Those wouldn't be properly corrected by a color term (for a broadband filter) because they only emit within a relatively narrow wavelength range. So that's another, second order, "color term" to consider. [0] https://news.ycombinator.com/item?id=12013377 https://news.ycombinator.com/item?id=12013377

privong 10y ago

> Is it? I always figured “true colour” would be whatever you’d see with your naked eye from an approaching or orbiting spacecraft, or the nearest approximation that (say) a consumer digital camera would give you. Can we get anything like that from our current compositing techniques? For many astronomical observations, the wavelengths of light observed are ones our eyes are not sensitive to. In this particular example of aurora, it's the ultraviolet. For others, it's infrared, millimeter/radio, or X-rays. For those, there's no way to way to make an image that our eye would see, because it wouldn't see anything. Some of the color images from those wavebands are described as "representative color" because the images are created to show certain features linked to the underlying physics or properties of the objects that are being shown. For images that purely in the visible portion of the spectrum, in principle one could construct an image "as the eye would see it". However, as @dakr noted[0], color astronomical images are composites of several black and white images taken through different filters. to create an "as the eye would see it" image requires combining those black and white images in a way that considers the relative sensitivity of the different types of color-sensing cones in our eyes. But, if you really want to get it right, it's more complicated than that. The shapes of the frequency-response of the cones probably differ from the transmission of the filters used in the astronomical observations (not to mention the quantum efficiency of the CCDs used to take the observations). So one would also need to correct for the difference in filterobject versus eye_coneobject (both of those as a function of wavelength). This is typically referred to as a "color correction" (meaning the "color" of the light emitted by the source), and requires some knowledge of the intrinsic spectrum of the source. And there's even another layer after that. If emission (or absorption lines) from atoms lie within the observation filters, those can contribute a non-negligible amount to the signal in an observation with a given filter. Those wouldn't be properly corrected by a color term (for a broadband filter) because they only emit within a relatively narrow wavelength range. So that's another, second order, "color term" to consider. [0] https://news.ycombinator.com/item?id=12013377 https://news.ycombinator.com/item?id=12013377