What does encapsulation give you over C's struct-plus-functions style?

It moves an invariant from "documented and hoped-for" to "compiler-enforced." In C, a struct's fields are public, so any code can poke them, nothing stops a caller from desynchronizing a ring buffer's head and tail by writing the fields directly, or calling uart_send before uart_init. A class marks that state private, so the only way to change it is through the public methods, which maintain the invariant. The compiler rejects external access at build time. You get a small, deliberate interface and a guarantee that the internal consistency rules can't be violated from outside, which dramatically narrows where a bug could originate. It's the same data and functions as the C version, just with an enforced boundary.