One way to make CG images more real is to pass them through a neural network. Some GANs can do that. Apple had a paper about synthesizing eyes using such a process.
Thank you. I'd almost given up on this comment chain, but yours made wading through them worthwhile.
Turns out you can't point out that no one has achieved fully-simulated videorealism without being labeled ignorant. You'd think HN would be the one place that people would be intrigued.
> Turns out you can't point out that no one has achieved fully-simulated videorealism without being labeled ignorant.
I guess that's a minor dig at me since I'm the only other person who used that word.
I agreed multiple times that no-one's made your idea of CG realism yet, so you're distorting the truth.
I'm sorry if you felt it was personal, I'm not using ignorance as an insult. I'm using it to help explain why your proposed "scientific test" can't work.
That said, you did demonstrate ignorance of basic math & rendering issues by claiming that color multiplication is meaningless. You demonstrated ignorance of the state CG production, and of basic logic, by claiming that the lack of realism that meets your criteria somehow proves it's not possible and that current techniques won't get us there. You might turn out to be right, but claiming it's a scientific absolute like you did is wrong. It's either maliciously wrong, or wrong due to ignorance, and I choose to assume the best about you.
> You'd think HN would be the one place that people would be intrigued
I'd love to have a productive discussion on how to achieve video realism. I have yet to see any attempt from you to start one. I tried to respond positively regarding motion simulation, and you didn't reply. I've posted a realistic video, twice, to which you haven't commented.
Here: "Improving the Realism of Synthetic Images" https://machinelearning.apple.com/2017/07/07/GAN.html
Also: "Visual Attribute Transfer through Deep Image Analogy" https://arxiv.org/abs/1705.01088 (this paper even shows a refined CG Mona Lisa)