"Kara", PS3 tech demo

[BD2003]

Absolutely not. Just like pre-rendered animation, this kind of animation is done with interpolated keyframes. An "animator" doesn't even need to do the animations anymore. With equipment and software, it's now trivial now to capture an actor and map everything automatically. I'm sure some manual clean-up is done, but there's not even any technical achievement here. Unlike rendering, animation and keyframes are not intensive operations. They're easy to do in real-time.

I'm not impressed by the technical achievement, I'm impressed by the results. If its now trivial to do, all the better, can't wait to see more games utilize it. LA Noire ruined mass effect for me. It'd be incredible if they can digitize faces as effortlessly as they can capture voice.

 

[CZroe]

I'm not impressed by the technical achievement, I'm impressed by the results. If its now trivial to do, all the better, can't wait to see more games utilize it. LA Noire ruined mass effect for me. It'd be incredible if they can digitize faces as effortlessly as they can capture voice.

That's like being impressed that they found a multi-lingual voice actor instead of one that only spoke English. It wasn't trivial or easy. Like I said, it took excessive effort but was, none the less, not "impressive" because we see that same level of effort EVERY DAY in a closely-related industry.

Now, you said "That facial animation is incredible." What is "incredible" about it if it was done the exact same way with the exact same effort as any pre-rendered CG animation? You say you are impressed by results and not technical achievement, but you have yet to say why you think those results are "incredible."

 

[Dankk]

Now, you said "That facial animation is incredible." What is "incredible" about it if it was done the exact same way with the exact same effort as any pre-rendered CG animation? You say you are impressed by results and not technical achievement, but you have yet to say why you think those results are "incredible."

I can't tell if you're joking, but... no, it was not done the exact same way with the exact same effort as any pre-rendered animation.

The difference between this and, say, a pixar movie, is that they're able to make such convincing facial animations run in real-time on very old hardware, which as far as I'm concerned, is far more interesting than movie studios who have all the computing power and time in the world to pre-render something.

Same goes for LA Noire. Yes, we've been able to make convincing pre-rendered facial animations for over a decade now, but LA Noire was still impressive because it was the first game to record real facial movements and render them in real time on weakass console hardware.

 

[Mr. Pedantic]

It looks like something from 3DMark06 or Vantage. Not really that impressive, to be honest. The Good Samaritan and the AMD Puppetmaster are much more impressive. Even Unigine Heaven looks better.

The best-looking part of that demo is the text at the beginning.

 

[CZroe]

I can't tell if you're joking, but... no, it was not done the exact same way with the exact same effort as any pre-rendered animation.

The difference between this and, say, a pixar movie, is that they're able to make such convincing facial animations run in real-time on very old hardware, which as far as I'm concerned, is far more interesting than movie studios who have all the computing power and time in the world to pre-render something.

Same goes for LA Noire. Yes, we've been able to make convincing pre-rendered facial animations for over a decade now, but LA Noire was still impressive because it was the first game to record real facial movements and render them in real time on weakass console hardware.

*woosh*

Real-time rendering DOES NOT CHANGE the way things are animated. There are not real-time puppeteers controlling the model animations. If you missed that, believe me, I know more about the difference between real-time and pre-rendered than you. My whole point was that animation, SPECIFICALLY, is irrelevant to whether or not it was rendered in real time. In this particular case, it's even more irrelevant because it was cinematic and non-interactive and, thus, subject to the same shortcuts pre-rendered animation has.

Understand?! They often even use the exact same software, importing and exporting animations from Maya and other tools.

Ever see them animating Toy Story with wire frames and then doing the exact same thing with a video game? That's because they are both PRE-FABBED animation sequences. The final render is only relevant to the animator in that the viewpoint can be controlled in an interactive real-time animation and, thus, it must "work" from any angle. That does not apply to a cinematic real-time sequence such as this. Even if this were a Half-life-like interactive story sequence, it simply means that they put more effort into it than a game typically gets which does not impress because we know that the efforts put into a feature-length theatrical animation are an order of magnitude higher than a few minute segment.

If you think it's so different, then answer me this: How is it any more difficult to apply mocap to a 3D model of a hockey player that is then used in a game than it is to apply mocap to a 3D model of a hockey player that is used in a movie? If they have to make it believable in a movie because it is a SFX shot stand-in for a real actor, it adds more pressure to the other guys involved in the render and they use the advantage of not being real-time to increase the LOD. The animation is the same.

When it comes to animation in real-time vs. pre-rendered and effort/money is the only production difference, generated PHYSICS are the only functional differentiator. This is more like, a reaction to an animation, like a water splash or a cape furl that the animator used an algorithmic tool to develop rather than pain-stakingly and unconvincingly hand-animating it. This was not interactive and, thus, all physics calculations would be pre-calculated and incorporated into the keyframe animation EVEN IN THE REAL TIME RENDER.

 

[BD2003]

*woosh*

Real-time rendering DOES NOT CHANGE the way things are animated. There are not real-time puppeteers controlling the model animations. If you missed that, believe me, I know more about the difference between real-time and pre-rendered than you. My whole point was that animation, SPECIFICALLY, is irrelevant to whether or not it was rendered in real time. In this particular case, it's even more irrelevant because it was cinematic and non-interactive and, thus, subject to the same shortcuts pre-rendered animation has.

Understand?! They often even use the exact same software, importing and exporting animations from Maya and other tools.

Ever see them animating Toy Story with wire frames and then doing the exact same thing with a video game? That's because they are both PRE-FABBED animation sequences. The final render is only relevant to the animator in that the viewpoint can be controlled in an interactive real-time animation and, thus, it must "work" from any angle. That does not apply to a cinematic real-time sequence such as this. Even if this were a Half-life-like interactive story sequence, it simply means that they put more effort into it than a game typically gets which does not impress because we know that the efforts put into a feature-length theatrical animation are an order of magnitude higher than a few minute segment.

If you think it's so different, then answer me this: How is it any more difficult to apply mocap to a 3D model of a hockey player that is then used in a game than it is to apply mocap to a 3D model of a hockey player that is used in a movie? If they have to make it believable in a movie because it is a SFX shot stand-in for a real actor, it adds more pressure to the other guys involved in the render and they use the advantage of not being real-time to increase the LOD. The animation is the same.

When it comes to animation in real-time vs. pre-rendered and effort/money is the only production difference, generated PHYSICS are the only functional differentiator. This is more like, a reaction to an animation, like a water splash or a cape furl that the animator used an algorithmic tool to develop rather than pain-stakingly and unconvincingly hand-animating it. This was not interactive and, thus, all physics calculations would be pre-calculated and incorporated into the keyframe animation EVEN IN THE REAL TIME RENDER.

We get it, you know a lot about animation.

That demo looked fantastic running in real time on a ps3. Don't care how they did it, just that they keep doing it. That's better than just about any other ps3 game running in real time. A game that had that level of quality thoughout would be quite impressive.