I know I may be stating the obvious to some of you here, but something to keep in mind is that the original photo used has A LOT to do with how the mirrored images turn out. While it looks like the original picture of Natalie is straight on, it’s not. Her is head is ever so slightly turned to her left (the viewer’s right), which reveals more of the side of her face with the mole.
This means that when you use the left half of the original picture (the side with the mole), more of her face will be used, creating a slightly wider forehead, wider nose, etc. And the opposite will be true when using the right side of the image (slightly thinner face, nose, etc).
And since, like someone has already pointed out, Natalie is very familiar to us all, the slightest deviation from the norm is noticeable…and somewhat off-putting.
i posted something similar to this in a response to one comment here, but i’m sure it’s not going to be seen, so i’ll say it again. there’s a great deal of empirical evidence demonstrating that humans prefer symmetrical faces over non-symmetrical ones (a sampling of articles that deal with this). i would guess that the reason most people think these faces look creepy (and i agree, the left one looks creepy) is because we are familiar with natalie portman being unsymmetrical, and familiarity is also linked to attractiveness (and here’s some reading on that). natalie portman is just a bad subject because we have all been heavily exposed to her face, even just on buzzfeed alone. if you were given a set of faces that you had never seen before and had to choose between the mirrored version and the unmirrored version, provided that you weren’t aware of what the experiment was, then, on average, you would probably favor the symmetrical faces over the others. if you had seen one of the faces before, that would be a confound and make the data provided by you, as a participant in an experiment, less meaningful. i think something that gets lost when popular media tries to play scientist is that there are obviously going to be exceptions to the findings of any given study, but in the corpus of knowledge existing in a given field, those exceptions may already be explained. scientific findings should be understood as an account of statistical averages and tendencies, rather than an exact description of every instance of a given phenomena. in conclusion, tl;dr
Natalie as an Afro-Caribean (and more!)