3 ms·
> As long as the content is presented to a human for consumption, I think it will be possible to duplicate it, Why do you think so? If you can tell the differ
by patrec 5y ago
> As long as the content is presented to a human for consumption, I think it will be possible to duplicate it,
Why do you think so? If you can tell the difference between a family scene and a screening of a Hollywood film, why can't your camera? It doesn't have to perfect, and you can gradually tighten things down, which is exactly what has been happening for some time.
- nybble41 5y ago> If you can tell the difference between a family scene and a screening of a Hollywood film, … That's a big "if". I wouldn't guarantee to always be able to tell the difference. And there is a lot of non-Hollywood content which looks a lot more like something out of a Hollywood film than just "family scenes". (People do make their own films, you know, not just recordings of family events.) > … why can't your camera? That's asking a lot out of the kind of computing hardware you can reasonably fit into a camera. Even if you assume something like a smartphone with much more computing capacity than a standalone DSLR or mirrorless camera (and a correspondingly higher power budge / shorter battery life), you're not going to be able to train it to recognize every bit of Hollywood content and reliably distinguish it from non-Hollywood content offline. (Online recognition where the camera can upload content to the server for pattern recognition is a different matter, but something like this would presumably be done offline anyway.) The information requirements would be on the order of having all that content downloaded to the phone—never mind the processing involved.