This is the first assignment for the Brown CS course CS143, Introduction to Computer Vision. The aim of the project is to create 'Hybrid Images' of the sort made by Aude Olivia, Antonio Torralba, and Phillippe Schyns in their 2006 SIGGRAPH paper by the same name.
Hybrid images are neat fusings of the low frequencies of one image and the high frequencies of another in such a way that the human visual system interprets the resulting image differently depending on the distance from which it is viewed.
A laplacian pyramid constructed from one picture is merged with a corresponding gaussian pyramid from another. The number of levels from the laplacian pyramid collapsed into the final image dictates the frequency cutoff.
To whatever extent this illusion can be considered 'convincing' that factor is still greatly context dependent. The merging of the rhino and the car is as such significantly more jarring than the merging of the cat and the dog.
If you're low on personal space, or just feeling like a bit of a scallywag, you might want to try cheating the distance requirement of the illusion by squinting or making the pictures smaller.
NEXTWhen set as a profile image on one of those new-fangled 'social networking' sites, this effect could be used to send different messages to those who can see your full photos and those who can only see thumbnails.