Others have answered this question pretty well for large public keys, but I wonder about the same argument for shorter elliptic curve keys where the key length

Others have answered this question pretty well for large public keys, but I wonder about the same argument for shorter elliptic curve keys where the key length and hashed fingerprint length may be the same. For example, is comparing 8 bytes of a Curve25519 public key as good as comparing 8 bytes of a SHA256 hash of that key? My gut says no, since generating partial cryptographic hash collisions is completely random while a structured public key of any kind is presumably less random, but I'm not sure how much less random (and thus easier) it would be.