How does a camera sensor capture colour?
It does not, directly — sensor pixels only measure brightness. Colour is produced by placing tiny coloured filters over them and then reconstructing the missing information computationally.
The Bayer filter. Almost every camera sensor has a mosaic of red, green and blue filters, one per pixel, arranged so that half are green and a quarter each red and blue. Green is doubled because human vision is most sensitive to green wavelengths and derives most of its detail information from them.
Demosaicing. Each pixel therefore records only one colour. The full-colour image is reconstructed by interpolating the missing two values at every location from the neighbours — a computational step, not a measurement. This is why a raw file needs processing before it looks like a photograph, and why different processing software produces visibly different results from identical data.
What follows from this:
Effective colour resolution is lower than the megapixel count implies, which is one reason megapixels alone describe image quality poorly.
Demosaicing artefacts appear on fine repeating detail — moiré patterns and false colour on fabrics, brickwork and distant railings. Some cameras historically fitted an optical low-pass filter to blur slightly and prevent it, trading sharpness for freedom from artefacts.
Alternative designs exist: sensors with stacked layers capturing all three colours at every location, and monochrome sensors with no filter at all, which are substantially more sensitive and sharper because no light is filtered away and no interpolation occurs — used in astronomy and specialist photography.
Where the rest of image quality comes from: sensor size, which determines how much light each pixel can gather and therefore noise performance; the lens; the analogue-to-digital conversion; and increasingly computational processing, which combines multiple exposures and applies learned models — now the dominant factor in phone photography.
Why phone photographs look the way they do. Small sensors gather little light, so the processing pipeline does enormous work — noise reduction, multi-frame stacking, sharpening and tone mapping — which produces the characteristic look and is why two phones photograph the same scene differently.