Light simulation isn't the issue, motion simulation is. Movement and model and texture complexity all have a long way to go.
FWIW I completely agree, and this is a key observation. DaVinci devoted several chapters to the problem of motion in art, and it will always be with us.
That said,
You're still ignoring the fact that realism in CG in increasing. There's a clear trend, and we are closing the gap on the details that make CG look unrealistic. The standards are higher now than last year, which was higher than the year before. Realism has always increased every year, and yet at no time was the increase due to a fundamental change in multiplying colors together.
This is the same argument that's always trotted out. Graphics are improving, but if you diff 2018 to 2013, it's nothing like 2013 to 2008. The fundamental leaps we've been accustomed to seeing are simply not happening anymore. The rate of progress is very clearly slowing down.
It depends which axis you measure, of course. We're able to render more things each year, which is nice. But the visual quality from a fundamental perspective is more or less the same as it was a few years ago.
The quality issues stem largely from light transport -- the colors are all wrong! If you compare them to a photo, you'll see that we don't end up remotely close. You can see this vividly in the YT link above (the Unreal engine walkthrough). If you try to picture yourself in the video, you'll get a strange feeling of being in a candy world, or a shrinkwrapped house.
That's certainly a promising axis to explore, and there are hundreds of papers published each year solely about light simulation.
>> Light simulation isn't the issue, motion simulation is. Movement and model and texture complexity all have a long way to go.
> FWIW I completely agree, and this is a key observation. DaVinci devoted several chapters to the problem of motion in art, and it will always be with us.
Now we are getting somewhere! Let's talk about those things instead of rendering! Which parts of motion are killing realism? Fluids and rigid body dynamics are pretty good these days. Facial animation is still in the uncanny valley. Why?
> Graphics are improving, but if you diff 2018 to 2013, it's nothing like 2013 to 2008. The fundamental leaps we've been accustomed to seeing are simply not happening anymore. The rate of progress is very clearly slowing down.
The rate of progress of rendering is slowing down, that goes right to my point that rendering isn't the main problem anymore, it's approaching good enough for realism. Movement and model and texture complexity progress is increasing. Papers on fluid simulation and facial animation and foliage and texture synthesis and multi-dimensional textures are on the rise.
> The quality issues stem largely from light transport -- the colors are all wrong!
No. The way we handle colors and light transport is fine, those pieces of human knowledge are ready for the jump to realism. You've already acknowledged that by refusing to consider still photos, because they already look realistic. One bad example from a game engine doesn't prove anything. I can show you lots of bad examples from game engines.
You would have a stronger argument if you gathered the very best examples on earth and we talked about those. By pointing at known bad examples and claiming that they say something about the state of the art, it makes me feel like you either don't know what the state of the art is, or you're taking cheap shots.
FWIW I completely agree, and this is a key observation. DaVinci devoted several chapters to the problem of motion in art, and it will always be with us.
That said,
You're still ignoring the fact that realism in CG in increasing. There's a clear trend, and we are closing the gap on the details that make CG look unrealistic. The standards are higher now than last year, which was higher than the year before. Realism has always increased every year, and yet at no time was the increase due to a fundamental change in multiplying colors together.
This is the same argument that's always trotted out. Graphics are improving, but if you diff 2018 to 2013, it's nothing like 2013 to 2008. The fundamental leaps we've been accustomed to seeing are simply not happening anymore. The rate of progress is very clearly slowing down.
It depends which axis you measure, of course. We're able to render more things each year, which is nice. But the visual quality from a fundamental perspective is more or less the same as it was a few years ago.
The quality issues stem largely from light transport -- the colors are all wrong! If you compare them to a photo, you'll see that we don't end up remotely close. You can see this vividly in the YT link above (the Unreal engine walkthrough). If you try to picture yourself in the video, you'll get a strange feeling of being in a candy world, or a shrinkwrapped house.
That's certainly a promising axis to explore, and there are hundreds of papers published each year solely about light simulation.