Someone in the general chatter thread linked to this topic recently, which I found very intriguing. I’m gonna bump it up and revive it: I thus posit…
ImageBrainz.
That’s all I’m gonna say. There’s a lot of visual media out there, in ever-increasing amounts, and often transient in nature… but it’d be pretty fantastic to have a way to permanently track metadata and relational data for them, especially illustrations and artwork. Some integrated feature to hash and uniquely identify roughly identical images, ala AcoustID, would of course be swell as well.