Originally Posted by pooby
I wouldnt say 'we're there' yet shading wise. We can fool people in certain types of cinematic shots with certain lighting, but I dont feel a blanket 'we're there', which implies full mastery, is an accurate statement.
For example, even if you just take a still from this, I have never seen any shading that is as convincing on this level, and certainly not in motion. you look at this and you just know that it isn't CGI.
I spend much time studying this area, so I'd love to be shown a CGI example which has a similar level of actual realism.
Exactly, that's what where my line of thinking is as well.
There's so much subtlety to everything in that video
And I mean, that video is only 1080p30....what if there was an 8k high-fps version of it? What if the camera pulled in even closer and we were just staring at a patch of his skin below the eyelid?
I've spent my career studying how biology structures are actually made like this close up of skin:
and how those bulges are then made up of these:
they all have micro-translucent properties that collectively form a complex macro appearance of SSS. That macro SSS is what CG artists currently dial in, but that doesn't hold up when you zoom in further also with higher resolutions.
All those translucent shapes are layered together as dead flaking skin, alive skin, and fat:
Then there's also muscles, blood/pulse, and other things that change the macro shading
If you could see that last image in motion, it'd blow people's mind how complicated it is with the skin stretching and bulging while gliding and squishing over the fat layer with tension points where it moves more in some areas and not as much in others.
not to mention a lot of skin is moist with refractive oil in all those cracks further complicating the shading
I mean, how many artists bother to put hairs on lips to get the shading that much more accurate?
Instead we fake it with a blurry reflection slider instead of using normal sharp reflections reflecting a complex and dynamic micro surface to create the reflection blur.
It's the same thing as the old argument of using specularity vs reflections. Reflections are better, but more expensive. Before blurry reflection algorithms became the norm, people couldn't fathom using reflections on every object in their scene - same with SSS/translucency. People were often content using straight lamberts for a lot of the objects and ignored the subtle broad blurry reflections that were actually happening and the small details they can pic up when you crop in close to them.
We simulate everything right now with pixel maps instead of geometry - which works fine for today's current film standards. People have pulled off fantastic shots that are perfectly good enough to fool people in that specific instance, but resolutions are getting higher. Frame rates are getting higher. Stereo 3D is getting better. The magnification across the board is getting more extreme.
I've been dealing with this at my job for awhile now with doctors wanting to show the person's head so people are quickly oriented, and then zoom into a craniotomy to show what they see under their surgical microscope - and have that transition seamlessly hold up rendered at (as of a 6 months ago) 4k stereo at 60fps.
Techniques and the amount of detail we have down right now won't hold up as well in the future under the scrutiny of newer technology and newer cinematic shots that'll demand realism beyond what's currently capable.
Chances are, whatever CG we think looks uncanny valley right now probably would have looked fine on low-res VHS. I mean, just look at this VHS/DVD/bluray image: