2 ms·

If I can mechanically perform a musical piece written hundreds of years ago and get a copyright on that, why isn't touching up the output of a neutral network a

• Riverheart 4y ago
  Edited If the neural network gets good enough that no significant work is required for the final product then I imagine you could change some pixels in the output and call it a "touch up". We're not there yet of course but ultimately how much effort does a touch up require to warrant a copyright? Is the touch up creative? Fixing blurry faces/hands seems like replacing a rusty pipe with a new one. For the musical piece, copyright probably expired or has no one to defend it today. At least with performance art, the effort of your derived work is easier to gauge. Your musical abilities and voice are unique to you and what, I think, is being copyrighted. You don't normally claim total ownership of the covered piece. Sam Smith got sued a couple times for his work (notes, not lyrics) being too close to others in what he claimed were original pieces.
• freejazz 4y ago
  Because what you get there is only a copyright in the sound recording, not the underlying musical work. So you can only control uses of your audio recording, you can't stop people from using the same work which is as you noted, in the public domain. By your own logic, the only copyright the edits would entitle one to would be what the edits consisted of and none of the underlying work. This is obviously meaningless in the context of images.