so, I had this thought...
what if the caricature could pull up similar images per the google custom search api? you see an audience of doppelgangers.... They are all watching you watching them... the infinite selfie recursion...
what if I used the Kyle's face tracker to match the orientation of the faces - could it be fast enough to have the google images faces "turn" as our participant's face turns? I know arturo did something related - he mapped Paris Hilton's face onto his...
I just grabbed a Google API key and instantiated a CSE. Will do a quick prototype and see how responsive the api is, then see how I could filter the faces...The only thing about faceTracker is it doesn't do great with profiles or bad lighting...
I made a Smiley Face app a couple years ago that mimicked the users expression. It did a lot of fakey-mc-fake-fake interpretation due to various issues.
--EITHER THAT OR--
if I go with the comic book idea, then the last created caricature is playing to the left of the main screen. The new participant can react to the previous animation and extend the story, which she will see on the right.
I'd better limit the shot sequence to like 5 seconds or somethign in order to asap upload and display the video.... Should I use youtube or store it locally?