I guess you're right, it might be not clear enough, or too far-fetched for other reasons.
I intended the person to be a morphian...their visual perception is actually independent from their "eyes" since their whole surface is sensitive to light, so he would have "seen" the girl approaching and standing there.
Of course I assume that morphian imposters are more than smart enough to generally pretend eyesight, by narrowing their eyes when looking against the sun, widen their eyes or maybe close one eye when looking at something really small and close, even faking a concentrated look. They would cover their eyes with their hands or turn away their heads when such a behavior would be appropriate for humanoids (shock, decency, embarrassment...), and so on.
All of this would be rather redundant and pointless, but morphians would have to do adopt such behavior. This very special individual, however, was not prepared for this special case of social interaction, i.e. he did not realize he was expected NOT to know who was standing there.
So..after all, it's some kind of twist that there is no such thing as a perfect disguise or flawless magic, and that even the most sophisticated deception might be revealed for the most simple reason.
Of course, there is an important distinction between ingame/character knowledge, and GM ressources - even if the characters know that morphians exist (which might be an obscure rumor or even completely unknown in some regions), they would not naturally know enough about ther physiology and perception to instantly make such conclusions. So yes, a story should definitively not depend on the players seeing through whats's happening.
But than again...in some longer campaign, I guess the scene might happen and pass rather innocently, with either the GM waiting for the moment of dawning comprehension, or as a way to introduce morphians in the first place - at least I hope so ;-)