sRGB

What it defines

A colour space needs three things: primaries, a white point and a transfer function. sRGB fixes all three — a specific red, green and blue, a D65 white point, and a transfer curve close to a gamma of 2.2 that determines how stored numbers map to light output.

Standardised in 1996 and modelled on the CRT monitors of the day, it has outlasted the hardware that motivated it because everything settled on it. It is the assumption a browser makes when an image has no profile attached.

Why it is the safe default

An untagged image is interpreted as sRGB, so an image that genuinely is sRGB renders correctly everywhere, tagged or not. That makes it the format-independent equivalent of a lowest common denominator that actually works.

For anything shipped to a browser without a controlled colour pipeline, converting to sRGB before export is the reliable move. It costs a little saturation on images that used a wider gamut and removes an entire class of "why does this look different on my phone" problems.

What it cannot represent

Its gamut is comparatively small. Deep saturated cyans, greens and oranges that a modern phone camera captures and a modern phone screen can display simply do not exist in sRGB, and converting into it clips them to the nearest representable value.

For photography destined for print or for wide-gamut display, working in a larger space and converting to sRGB only at web export preserves those colours in the master while keeping the delivered file safe.

The mistake to avoid is the reverse: exporting a wide-gamut file untagged. It is then interpreted as sRGB, and every saturated colour renders as though it were a less saturated sRGB colour — which looks flat and dull, the opposite of the intent.