Space photography

A space telescope's image sensor is a CCD or CMOS chip — a grid of photon-counters. Each pixel sits on the focal plane for an exposure (anywhere from milliseconds to multiple hours), accumulating electron counts. The output is a number per pixel: how many photons of which narrow-band filter hit there. Multiple exposures are stacked to beat down read noise, registered to sub-pixel alignment, and median-combined to reject cosmic-ray hits.

Colour is the most-misunderstood part. Telescope filters are narrow-band — they pass a specific 10–100 nm slice of the spectrum: hydrogen-alpha at 656.3 nm, doubly-ionised oxygen at 500.7 nm, infrared bands at 2 µm or 12 µm. The Hubble Palette assigns those filters to RGB channels in non-physical ways (S-II red, H-α green, O-III blue) to maximise contrast between physical processes. JWST's mid-infrared images map 21 µm, 11 µm, and 7 µm to RGB — colours your eye couldn't see if you were there.

The processing pipeline is open-source for most NASA / ESA observatories. Raw exposures land in MAST (Mikulski Archive for Space Telescopes) or ESA Datalabs the moment they hit the ground; image-pipeline tutorials show exactly how a final composite is assembled. The pictures aren't manufactured — they're documented to the byte.

For ground-based imaging the additional challenge is the atmosphere. [Adaptive optics](observation/adaptive-optics) corrects it in real time by sensing the turbulence with a laser guide star and counter-bending a deformable mirror. JWST and Hubble are in space partly to skip this problem entirely; the atmosphere is also opaque at most infrared and UV bands, which is the other reason.