This website: generated.photos has 1,971,820 million free generated faces and has an API to generate more faces – maybe memory league can incorporate these photos in the Names and Faces database?
It would be a nice change of pace haha
Kind of scary that an artificial intelligence can generate such realistic looking folk! Cool find!
Is it? Why?
I don’t know, brings to my mind dystopian visions of redacted/corrected photographs being used as evidence against person in court. Like, “Mr. Bjoern, we have evidence that you WERE at that place on the night in question”, and some doctored CCTV footage gets you sent to the clink for a long one.
Get Orwellian vibes just seeing the things, as fascinating as they are
I hope I can find myself there. Joke.
It seems they use template images of people to generate faces, so you will find similar faces. That is sketchy.
If that is the same site where the faces for the first day of the ML Tournament were taken from, I would be against using those. Those faces were much too similar, roughly the same age and look, almost like clones/close relatives (that is why on day 2, the normal faces were used again).
In context, not surprised if the page used certain templates as basis for the AI faces.
If you want to be even more scared, you have this website
Right, that makes sense. What if we used their API and changes the parameters to either have a) less training data (a bit risky), or b) more diverse training data? That would mean different faces and so a better memorization experience.
This is funny, but sometimes when I’m doing memory league names I either chance upon someone I know or a face I’ve just seen…