If that were true, it should be easy to point to some simulated video and say "Look! This one looks identical to nature. It's so perfect it looks like you were

If that were true, it should be easy to point to some simulated video and say "Look! This one looks identical to nature. It's so perfect it looks like you were really there, filming it with your own camera." But no one has ever made a video like that, so we can only show videos of trivial scenes (like a room composed solely of cubes with only one or two lights). There'd be no contest with any real-world comparison -- say, a messy desk with a laptop on it, with the sun shining through a window. The moment you give the viewer enough visual cues, they'll pick out the simulated video nearly 100% of the time, let alone getting anywhere close to fooling them half the time. More precisely, if you show a playlist containing a mix of simulated videos and real-life footage, viewers shouldn't be able to select the simulated footage any better than random chance. That's the proof no one has ever done this.