Comment by blhack | Hacker News Reader

Comment by blhack | Hacker News Reader

Maybe this is me being a Luddite, but...what the hell is the point of this?

So you had a "real" actress mimic the facial expressions of this scene...then mapped them onto an obviously fake render of the same person?

Why not just skip all of this and have live action actors perform this scene? What is gained by rendering the same thing in a computer other than appealing to some strange nerd fantasy?

I guess I'm also not a video game player, so maybe somebody can enlighten me here: are the graphics here really better than what we were seeing 5-6 years ago?

Yeah...I guess we can render it in real time (AWESOME FEAT!), but if you want to show off the computational ability of Cell, have it calculate pi or simulate a dust cloud or something. This, at least to me, looks identical to every other video game trailer I've seen for the last 5 years.

If you would have said this was a trailer for half life in 2005 (the dates here may be totally wrong, I have no idea when that game came out), I wouldn't have questioned it at all.

---

As for the story here. This is shallow, boring, and lazy. It's like we're being slapped in the face with whatever the storyteller wants us to feel. The obnoxious "honey" or "baby" or whatever the operator says to the robot is just...uhgg, painfully awkward.

/or maybe I just need more coffee this morning

The graphics are much, much better than something from 2005. For reference, here's a screenshot from a well-regarded 2005 console game: http://upload.wikimedia.org/wikipedia/en/8/82/Gun_Showdown.j...

Especially for the PS3 hardware, the polygon count, lighting solution and facial animation are quite impressive. While this isn't a game but a tech demo, it's still impressive for what it is. I don't see how calculating digits of pi is a more worthy way to demonstrate the capabilities of hardware specifically designed to render 3D scenes than actually rendering a 3D scene.

Yes, but this is a cutscene isn't it?

One of the major points was that they had an actress mimic the facial expressions (and used motion capture for it), and voice actors to "act" the scene.

That doesn't translate to gameplay.

Most 3d games use pre-rendered geometry when avatars walk around swing swords etc. Being able to add highly accurate facial movements to those animations increases realism without needing all that much data relative to modern hardware. And unlike simple video clips you can meld these things into the rest of the games geometry so say a grunts cheek would react reasonably as they lifted their AK to fire on you.

Take it a step further - whether people data-mine and generalize the human motions and expressions or, just cleverly used fixed 'patches' of behavior, well, imagine now using them in live robots. /That/ will be an Uncanny Valley :)

This is not a 'cutscene' in the pre-rendered-and-then-compressed-as-video sense. The site clearly states that it was rendered in realtime: it is a very short path to interactivity.

Interactivity implies that the system be able to reproduce realistic facial expressions in response to the ongoing action without doing motion capture of a live actress.

It's not so obvious to me that the path from this scene to that is very short.

Because a live action performance would be a pretty terrible tech demo of real-time rendering on a popular gaming console.

Capturing and mapping real life "acting" to a virtual character is actually very useful.

Games interweave game play and story telling moments. During story telling moments, using this tech allows the virtual characters to express better emotion.

You don't just insert live action actors in the middle of a game in place of the virtual characters because it would be jarring and confusing. You don't use live action actors during interactive game play because all motion combinations and cameras angles need to be synthesized at real-time.

For non-games, if you don't want your actor to look like a human, but still want it to express human emotions this is applicable. Think Avatar.

If this demo is actually rendering in real-time on a PS3 (ie: not just playing a movie on a PS3), then the graphics are amazing.

You're missing the point. Yes, this is a cutscene - but since it's being rendered in realtime, the camera suddenly doesn't have to be fixed. Players can move around, interact, and view other things while this is going on. Think of NPCs in games going through interactions like this, not just the main character during an unplayable cutscene. If you can create convincing facial emotions in realtime, you can vastly improve the realism in a game.

Reply on news.ycombinator.com