Not being a security expert, I have a rather theoratical question that I have been thinking about for quite a while now:
Let's imagine an encryption software that takes a user-provided password, passes it through a key derivation function and then uses its output as a key for AES-256 in order to encrypt a file. Assuming a key space of 100 characters and further assuming that an attacker knows the setup and is aware of the length of the password, is it correct to say that adding more characters to a 39-character password does not increase the user's security because
2^256 < 100^39 and, thus, the attacker would rather try to brute-force the actual derived key instead of the user-entered password?
(I'm aware that brute-forcing a random password of that length is completely unfeasible now and in the foreseeable future.)