That's exactly what happened. People complain about the way high-frame-rate cameras look cheap.
The language of cinema is always unreal. Even the most "natural" performances are always highly stylized. Camera motion, lighting, makeup, etc are all very different from real life. You can see it even in behind-the-scenes footage; everything looks so fake. It's designed to feel right in the captured image, and that's the only goal.
They'll use whatever tools they have for it. This might be a good tool, but it's hard to predict. They never really worked out a good use for 3D in film, even though it's also more "natural" than a moving 2D image. It's a different kind of uncanny valley: it didn't resemble real life enough, but you were content with the really unrealistic language of 2D cinema.
Newer generations growing up with 30+ FPS and better animation will develop different tastes.
Yep, and it’s also mistakenly assumed to be a necessary ingredient of all the great films you remember, rather than a nearly unavoidable ingredient. After all, those great films were all 24 FPS! And they used that medium in really creative and technically proficient ways!
And that’s true and great. But all the bad films were also 24 FPS, and it would be weird to say “imagine how much worse that terrible low budget film from 1978 would have been if they had modern professional digital cameras and shot in 48 FPS.”
Everyone is entitled to their opinion, especially regarding aesthetics. But this is pretty much the same sort of complaint that I was criticizing. You can have a genuine steadfast aesthetic preference for black and white films, or silent films, or SD digital video, but I disagree with extending that to anything of the form “technology X is undesirable because it makes films more realistic and films are not supposed to be realistic.”
People liked some things, like talkies. And they rejected other things, like 3D. I could come up with an after-the-fact rationalization for why some things worked and some didn't, but it's pretty much impossible to tell beforehand.
AAA isn't a value marker. It's a shame every time I see discussions of gamedev it's always "but can you do AAA with this library?" Who cares!
A previous title in the same franchise (Wind Waker) was also lambasted on release for it's alternative art style & today has aged considerably better than most games of that era.
If anyone is interested in an example, I found this: https://m.youtube.com/watch?v=IP0B45hUPgo
We also happen to be very good judges of what looks real.
And at some point I want to play realistic open world VR D&D, with multi-model intelligent NPC’s, enemies and beasts! Oh, how I have been waiting…
I want game characters that feel alive in all the ways my brain expects creatures to act alive, which I have yet to see done with most NPCs. Even key NPCs usually fall very short.
Cartoons have done this for decades, but it's (relatively) easy when everything can be preplanned.
This makes that easy whether you want to go realistic or not.
Exceptions exist of course, japanese developers are able to get very cartoony visuals but that means they won't be able to use this metahuman 3d model and animation technology.
Still, this is impressive tech, don't get me wrong. But we are still in the Uncanny Valley. Getting better though!
>Yes, and also: the actual actress felt a tiny bit uncomfortable, maybe the performance in front of a live audience, maybe because she could have a slip of the tongue or whatever
To be honest, the way her mouth moves in real life felt very off. The "(over-)acting" during the capture was very unnatural, likely down to pressure of being on-stage as well as that just being a common problem for untrained actors doing anything scripted.
Would be interested to see how it handles a capture of someone just engaging in natural conversation or something else less dramatic & exaggerated.
I'm not sure that explains it all, but there are expected differences compared to how someone would move while talking normally.
I agree. It's very impressive though, a clear leap forward.
Hellblade 2's demo is even worse though. It looks like the lips are completely disconnected from the gums and flap around loosely as they move.
Very video-gamey, not super convincing. But then again this is running a model in just a few seconds, with input from consumer hardware! Makes me wonder what you could do with more capable hardware (or if the iPhone hardware is really THAT good).
It does make me think of one excellent application for this software: scanning your own face to make an in-game character who looks exactly like you. I assume I'm not the only one who's labored over character generators for hours to produce a decent-looking character. I think there's a neat opportunity here to get video game players to scan their own face as input instead.
Generating a high-quality model from a camera phone is a whole other can of worms.
My significant other works for Epic on the cinematics team so we've played around with Metahuman a bit. Scans we've done weren't immediately perfect (like the video on that linked page mentions, hair and ears don't scan well) but cleaning it up was basically just deleting those portions of the mesh, letting RealityCapture fill in the rest of the head for the model, and then adding similar looking hair from Metahuman's database.
So there's probably a little bit of work needed before we get "scan your head with your phone to create a video game character automatically" but if you're cool requiring hair selection as opposed to auto generating it from the scan we're pretty much there.
So yeah.
Maybe what you're getting at is "uncanny valley" - which I did get a little bit. But it's impressive how little I felt the valley for such a low-effort demo. I am sure with good actors and professional grade capture hardware this would be significantly less.
What's really exciting to me is that this tech lowers the bar. Indie game developers are going to soon be have this sort of photorealistic quality to their characters.
But outside of that, there are the limits of rotoscoping coming into play. Human motion has a lot going on and our 3d captures aren’t anywhere near that kind of fidelity yet. Therefore we still need to rely on what the old masters of animation discovered when they were getting past the rotoscoped era. To make a performance more believable there has to be a little more acting and exaggeration brought in. The eyes can’t just get wider. They need to have anticipation, movement, secondary movement through other pieces of the face, etc. Same with the mouth. It’s not that the mouth can’t stretch in an abnormal way. It’s that it requires a little bit extra like a neck movement creating anticipation of an angry snarl.
As far as we have gotten along with mocap tech, when it comes to film, there’s a loooot of hard work that goes into fine tuning done by professional animators to make the performance of each actor look good when it’s translated to the 3d character.
Seems promising if we allow for iterative improvements.
In my case I just try not to focus too hard on any one thing, and that helps me get immersed.
No, you're not the only one. You may still be crazy.
I was expecting something cutesy like Apple's animated animal avatar emoji things, but instead, it's near-live generation of more creepy frozen-faced fizzogs I've seen in various videogame demos -- and that rather strange Final Fantasy movie, 20Y or so back.