Why aren't more developers living up to L.A. Noire?

Title Why aren't more developers living up to L.A. Noire?
Author Tom Chick
Posted in Games
When February 4, 2013

One of the clearest indications of L.A. Noire's ability to express human performances better than almost any other game is the above gag reel, released by Depth Analysis, the folks responsible for its facial motion capture technology. L.A..

Read the full article

The expressiveness is look pretty amazing, but there is still a pretty big risk of spending a ton of extra money on a more humanized character and still being in the so called 'uncanny valley.' So until this type of technology becomes a whole lot cheaper, I don't any companies budgeting for this level of reality.

I'm not sure how expensive the technology actually is, but Depth Analysis' website implies that it's inexpensive. Of course, they would say that. :)

But maybe you're right that it's cheaper to just have an in-house artist do facial expressions.

I was under the impression that it was a big part of LA Noire's cost.

Probably because A) It costs money B) It has an uncanny valley look C) Not all games want to be films and D) LA Noire didn't do too well saleswise.

It's rare even to see human expression in movies and TV shows. And we're all sat watching/playing with vacant expressions or just kinda grimacing.

Their faces look nice and everything, but the rest of the animations are horrible. The heads aren't the right sizes for any of the bodies, and the movements of the models is very stiff.

You aren't seeing more of it because it only allows capturing facial expressions of people who actually exist. There's no real life Dante, no real life Nathan Drake and no real life Isaac Clarke. The tech is impressive, but it's also a one trick pony. Another problem (which becomes very evident when looking at L.A. Noire) is that body motion capture has to be done separately. The floating head sensation I got from L.A. Noire reminded me of the terrible Green Lantern movie.

The tech is also a tremendous resource hog and extremely rigid. It's essentially fmv mapped onto the surface of a low poly head. Every frame of animation has its own awfully low-res texture and normal map. No such things as specular channels or other snazzy texture effects either, because that would have exceeded the rendering budget. Oh, and don't even think about blending animations. Won't work.

The tech is a creativity-stifling, utter dead end.

I think you're wanting to open pandora's box here. You just know that a lot of studios would go cheap on the acting talent, or halfass the direction, and then you'd have bad (physical) acting to contend with as well.

As long as it's just voice acting you can have guys like Martin Sheen come in on a weekend, on a project he doesn't much care about, and still deliver a passable performance.

LA Noire faces weren't bad at all in cut scenes and so on. Though if those characters were in a period movie, they'd all have much smoother faces due to the layers of makeup.

But my god, those facial animation tic cycles in the game interviews! A complete failure, despite the evident power of the system that implemented it. That dreadful chunk of combined gameplay and tech art was worse than giving the characters lego faces. Those tic cycles were some of the worst things I've seen in a long time. Uncanny valley indeed.

Perhaps the tech will evolve and become more developer friendly. It's not like the tools used to create the above reel are frozen in time. It's in Depth Analysis' interest to make their performance capture technology as easy and generous to use as possible.

Maybe, but the tech that's widely being used already is getting more and more capable and developer friendly as well, all while giving the artists a lot more freedom. Just compare the wooden facial animations in Heavy Rain to what Quantic Dream manages to pull of in their new game Beyond. Heck, the tech isn't even being used for movies. Even the faces in Avatar were done using simple mocap with a lot of tweaks by some highly skilled artists.

Not what I want game makers to spend their resources on for most games.

Also, was more impressed with Silent Hill 3 considering it was on a freaking PS2.

They have this thing called "actors" now, that allows real life people to portray non-existent, fictional characters. I think its going to be a hot new trend.

I've at least seen some evolving of face tech over the past two years, but most of the advances I've seen have occurred outside the Unreal engine or whatever modified Gamebryo thing Bethesda uses. To me, both of those engines create the most awful looking humans I've ever seen. I'm still trying to steel myself to play Dishonored and dealing with everyone running around with Unreal Engine lobster claw hands.

That said, even with older and more stylized tech, I thought Rockstar games like Bully, RDR, and maybe GTAV did ok with facial tech. The humans in Sleeping Dogs look great, too. The Witcher 2 and the Red Engine do a nice job.

I think the real takeaway from this tech (which still looks amazing to me) is that it's a goalpost, but not a necessary process. If I'm CDPR, I'm trying to make my facial tech in the just-announced Witcher 3 come as close to what was achieved in LA Noir without perhaps actually using that specific tech. Instead of saying "I want to use Depth Analysis's tech" it seems like a few of the more ambitious studios are saying "I want our engine to use that as a target, without actually using that third-party tech."

We used p-cap on Crysis 3. I was really pleased with the results.

https://www.youtube.com/watch?...

You can see my ear in one of the shots :).

The point was that not every game maker wants their characters to look like the people who act in the mo-cap suits. They have character designers for a reason.

I'm not particularly in love with the thought of Nathan Filion on the cover of Uncharted 4 or Lara Croft: Tomb Raider starring Angelina Jolie.

The interviews were both the best and worst part of the game. The best because it gave you the illusion of being detective. The worst because the makers seemed SO impressed with their tech that they seemed to expect you to base every correct answer on the facial expressions alone.

The problem was every suspect's facial expression seemed to be trying not to notice a monster sneaking up behind you.