Funky Little Boatrace: Activision’s Astonishing Face Tech



Also, wow. NVIDIA aren’t the only ones doing remarkable things with videogame faces – Activision have their own take on a very similar real-time rendering concept – and indeed using the same bald, parfait-obsessed character. This is apparently all done on current hardare – a GeForce 680 can muster it at 120FPS, they say, although obviously that’s without anything else being rendered. More impressively, Activision R&D are demonstrating this face-sim at GDC, using a two-year-old laptop, whereas NVIDIA talked about their own tech needing a GeForce Titan. I don’t own a GeForce Titan. I might have mentioned that once or twice.

If you aren’t at the show, you can watch this surprising video. I should warn you that the phrase ‘frozen fruit’ may take on faintly nightmarish associations for the rest of your natural life.

Now excitingly enboldened for lazy readers: Oh, and while it’s easy to assume that this is simply someone else slapping their name on NVIDIA’s FaceWorks tech, apparently “this is the same source data from ICT used by Nvidia at GTC, but rendered using a completely different tech.”


Amazing, but even so I feel like I just booked a three-month holiday in the uncanny valley. But it might not be an entirely unpleasant vacation, I must say: it’s not quite the naturalism of LA Noire, but it’s got a whole lot more fidelity. Plus it’ll look a million times better when it’s not only being demonstrated with a fruit-obsessed severed head.

This comes from Activision’s R&D dept rather than being taken from or promised for any particular game, so don’t get any hopes up for seeing it on a screen near you anytime soon. Given that Activsion’s games output is now basically just Call of Duty and Blizzard titles, we can probably take some educated guesses about where this stuff might ultimately fetch up, however.

Find out a bit more about it, and see screenshots of other characters, in this post by Activision’s Jorge Jimenez.


  1. wearedevo says:

    I must say… this is climbing the right side of the uncanny valley.

    • Knufinke says:

      This is still deep down the uncanny valley. The animation is not fluent at all, the face moves a lot like a rubber mask and they probably should pay more attention what a mouth looks like when somebody speaks, most of the times it’s not just a dark cave.

      • Harlander says:

        It does look like someone wearing an incredibly lifelike latex mask, yeah.

        To the extent that if you’d told me it was footage of someone in a rubber mask, I might have believed it (except for the NOT GAME FOOTAGE in the corner :p )

        • Shadowcat says:

          What?! Why are you not wearing your “NOT GAME FOOTAGE” notice? This is a gross violation of protocol, and there will be consequences.

      • Smashbox says:

        If only someone on this team had thought to look at a mouth.

      • Shuck says:

        it’s weird, because the speaking animations of the previous demo looked pretty spot-on, but this one looks like a rubber mask, with all the movements strongly damped down. I wonder how that happened.

    • MajorManiac says:

      Pretty soon artificial humans will look more real than us.

      • Gramarye says:

        Then, the cynical Youtube comment: “Come on, so obvious! REAL!”

    • Ravelle says:

      No matter how advanced facetech is, video game characters can’t put up a real looking smile, also the inside of a mouth isn’t a black hole.

      • tstapp1026 says:

        This. The facetech is truly amazing, but it seems the lack of realism is always in the mouth / teeth area. They need to work on realizing and producing something other than a perfectly off-white bar of bone stretching from one side of the mouth to the other with simple etches to identify one tooth from the next.

        • liquidsoap89 says:

          I think a big part of this issue is that the artists try not to exaggerate the animations of faces. I know when I was in school learning to animate we were generally told how cartoon animation is exaggerated and “realistic” animation is more subtle. I think a lot of animators take that concept to the extreme, and don’t really think that they both cross over quite frequently. When you see a person whose genuinely angry and/or yelling, their face faces will get super scrunchy, as with laughing and crying etc.

          Maybe they’re not doing that simply because polygons are crashing together or something, but I’ll speak for myself here when I say that the biggest thing I had to get over when learning to animate was to try to NOT be so subtle.

    • Turquoise Days says:

      The right side of the valley indeed. Is anyone else getting vibes of Lieutenant Commander Data? It’s the same slightly jerky switches between expressions.

      That’s the only way I can describe these demos – we’ve reached a point where technology can replicate a human pretending to be a robot.

    • jaypettitt says:

      “I must say… this is climbing the right side of the uncanny valley.”

      But what was wrong with the left side of the uncanny valley? Was the grass not as green.

      It’s impressive and all. But an awful lot of outlay for very little return I reckon. Especially once you start thinking about how you’d try and make good use of this in a game setting.

      The left side of the valley can deliver similar levels of acceptance with far fewer, err, overheads. Avoiding the uncanny valley is what matters.

      • DXN says:

        Gotta cross it some time. How else will we ever have holosuites?

        Personally, I found this footage much more convincing than the other one. The lighting is softer and less ‘Unreal-ish’, the face, well, writhes less, and I thought the eyes worked very well. The mouth did seem to be the weakpoint, but I’m not going to fault them when the task of fooling the brain’s incredibly sensitive facial-recognition flesh-circuits is so incredibly tough. Bravo to them!

  2. Novotny says:

    Staring eyes or what.

  3. Cross says:

    I wonder how tough they are at the baggage check at Uncanny Valley airport.

  4. thegooseking says:

    Given that Activsion’s games output is now basically just Call of Duty and Blizzard titles, we can probably take some educated guesses about where this stuff might ultimately fetch up, however.

    So… VtM:B2?

    It’s not impossible

    • heretic says:


      • SuperNashwanPower says:

        I would prefer they applied it to HG:Zt3 personally.

      • thegooseking says:

        Actually, I just remembered that White Wolf is now owned by CCP, so I guess it is impossible. Ah, dammit.

        Still, VtM:B’s faces were pretty impressive for the time, too.

        • Dowr says:

          “Still, VtM:B’s faces were pretty impressive for the time, too.”

          And they still are; as well as HL2.

          Ah, Source, you’re obsession with the human face is delightful.

      • eclipse mattaru says:

        Bloodlines 2 by Activision? Serioulsy?

        A modern, AAA version of Bloodlines would surely have to go through the darkir/gritty treatment and would have anything resembling a sense of humor removed right away. Plus, who would write it?

        No, thanks. The original is great as it is, and nothing like it is gonna happen again.

  5. Dzamir says:

    You made my horizontal scrollbar small, then I will make your smaller:


  6. Chalky says:

    Very impressive – is there an issue with hair rendering that makes it look better if the character is bald? I imagine that might be the case, but also there’s been a whole load of stuff for “hair rendering” recently which made me think it was pretty advanced as well.

    The mouth is a bit iffy, the edgers of the lips are too sharp I think. I wonder how long it’ll be before it’s impossible to tell it’s rendered.

    • Kreeth says:

      Holy cremola foam.

      Yes, I think the slight weirdness round the mouth is because you’re actually seeing the edge of the polygons: everywhere else the various shader techniques/textures etc are softening everything, but not there.

      Can we please have this in RPGs? Would be a total waste of time if this only got used in Modern Warface or whatever, but human characters who looked like this in Witcher/Dragon Age/Whatever would massively improve the game oi reckons. I’d love to know how much extra effort it would take to create the characters individually using this rather than whatever people use now.

      • Kitsunin says:

        A thorough “Meh” from me. If…Geralt, for instance, actually looked like a real person, I think he would feel way less unique, likewise for most cases. I can se why you would want to use this sort of technology, but I feel that you get a much better experience overall when the creators of a game have complete control over their artstyle.

        But then, perhaps you should take what I say with a grain of salt, seeing as I almost never enjoy watching live action TV shows, preferring anime by a huge amount if I want a serious story. Give me CG or give me stylization, I say.

        • Kreeth says:

          I know what you mean, and I agree with you about stylisation in general, but to my mind if a dev is going for “realism”, then they should really go for it, and something like this could really help. There are stylised games, but I’ve never seen Witcher/Dragon Age/Mass Effect etc as going for that – they seem to be trying to be as “realistic” as possible. Same with TES 6, once they (hopefully) get over their MMO phase.

          It’d look downright bizarre in Monkey Island 5 though.

          Putting this kind of face on a character who was also fully dressed, with lush backgrounds like in Witcher 2, could mean that the slightly odd bits like the mouth get a bit lost, so the whole looks better.

  7. LeMonde says:

    Now there’s a proper song reference!

  8. G-Lord says:

    That tech would be perfect for Warface…

  9. Paul C says:

    There’s still something a little bit weird about the animation – it seems to move in separate sections, like an animatronic head rather than a real person – but that’s still very impressive tech.

  10. Njordsk says:

    Okay I want LA Noire with that.

  11. Chris D says:

    I spent too long trying to work out whether that was just a hoax and they’d just pointed a camera at a bald dude. I’m not exactly an Activision fanboy but damn, that’s impressive.

  12. Blue_Lemming says:

    Did they have to put a menacing string durge along with this?

  13. Lord Custard Smingleigh says:

    I say, that’s not on. You just had me all aflutter thinking that finally Activision has been awarded the prodigous merchandising power of the Boat Race license. I trust I need not elucidate to which Boat Race to which I refer?

    Can’t you see it now? Two eights of digitally recreated oarsmen, gasping and sweating, pulling mightily on their wood, the cox keeping the blues heaving and pulsing and beating as one organism… The soft susurration of the Thames beneath their jolly blue boats, the staccato timekeeping of the coxwains, the splash and creak and groan of the boats, grunting and gurning past the Harrods Furniture Depository… And finally the climax, as the Stag Brewery hoves into view, the sweat flying, droplets of water flicked, glistening, from the oars to hang seemingly weightless in the sunlight before returning to the river as every taut muscle strains to bring their boat ahead of the competition. The muted rumble of approval from the crowd as one or the other nudges ahead to thrust across the finish line by as many lengths as possible at Chiswick Bridge, and the occasional “whoop” from some confused American tourist trying to join in the spirit of things.

    Now that, sirs, would be a game.

    • LeMonde says:

      The Game of The Boat Race?!

      What a ghastly proposition and, might I add, one not even remotely becoming of the race’s ‘Corinthian’ spirit.

      Whatever next!? Are we to expect Activision logos to suddenly besmirch those hitherto pristine craft that have, for generations, been proud, noble champions of the glorious amateur class?

      I hold you and your ilk, who foster such disgraceful, base, corporate and PROFESSIONAL sentiment for this once proud land’s steady decent towards ruination. I wager you ‘hurrahed’ triumphantly when ITV ‘bought’ (the very thought is anathema to me) broadcasting rights to The Race.

      I suppose one might be able to countenance the notion, if and only if the game is so ludicrously expensive as to be playable only by the most deserving. Otherwise, NO.

    • Gap Gen says:

      And how about a required internet connection so that you can only play at 5am, but which enables social features where you can compare erg times with other human triangles and use rowing jargon that no sane person could comprehend.

    • Shadowcat says:

      Smingleigh: Stephen Fry would have been proud of that description :)

      • Gap Gen says:

        Alternatively, it could be where you get two rows of public school boys, have them drink in turn, and then destroy an Indian restaurant.

    • edwardoka says:

      RPS Hivemind, kindly make Lord Custard Smingleigh an offer to enable him to write officially for the site.

  14. GallonOfAlan says:

    Just like the nVidia demo, again with the bald guy. Hair must be too much to ask for at the minute.

    • SuperNashwanPower says:

      You ninja’d me :) Yeah, its exactly the same as NVidia, even with the parfait spiel. Bit odd, maybe this is Activision one-upping NVIDIA?
      EDIT: You did ninja me, but for some reason my comment disappeared. I think the server is having a wobbler again.

      • Alec Meer says:

        I deleted your last comment because you didn’t read the post. Writing a comment is more effort than reading, you know.

        • SuperNashwanPower says:

          [do that again and you get banned]

          • elmo.dudd says:

            Do what again? Unless his original post was really nasty, it sounds like he did the most common thing on the internet, commenting without fully reading a post. Which is even more understanding in this case because it almost looks like a double post, and the difference here is one that would only marginally warrant an update to the original post, not a new one. Frankly I skimmed the text because it opened with a barrage of “A” and since the title involved “Astonishing Face Tech”, I went to the video directly as it obviously had the content of concern. This looked to be one of those posts where RPS was acting as a mirror, rather than a content creator.

            It is a little alarming to see that such common behavior warrants a silent deletion, and his response to you about your rather snarky reply was completely censored with a threat. Looking at his other comments, I have a hard time imagining he really said anything that terrible.

  15. Edawan says:

    Why ? Because animating a whole body (especially the hands, motion-capture doesn’t get that) is hard. And so is simulating cloth and hair in a convincing manner.

  16. MrPo0py says:

    Amazing. The only criticism I have is that they need to model the inside of his mouth. If they can somehow reproduce tongue movements and then show it off instead of have a dark mouth then they have pretty much cracked it. Can you mo-cap a tongue?

  17. herschel says:

    Impressive. But still, the closer we get to perfection of CGI the more likely our brain is to detect “something weird” in a face. Which makes us very uncomfortable and more likely to reject establishing an emotional connection to this person. So more perfect does not equal more emotional and better.

  18. Loque says:

    The more you try to simulate reality, the more you look fake. I am sorry, I understand this is an amazing technology… but it looks terrifying and fake as hell.

    • SuperNashwanPower says:

      Isn’t that the uncanny valley though? You go from cartoon, to more real, then into freaky, then out the valley into “looks like a human”? A few tweaks to this and we’re there, I think. Then maybe we will find ourselves just going “this is a video, surely?”. I think its all pretty awesome to be honest.

    • Juan Carlo says:

      I agree. This is amazing, but so, so, creepy. It looks and moves like a robot with really realistic skin. And the animations suck. When it smiles or opens its mouth wide it looks like Garry’s Mod’s facial animation feature.

      It’s passable as real as long as it doesn’t move, though.

  19. Premium User Badge

    Ninja Dodo says:

    Pretty nice, but lip-sync is still terrible. Also this tech is kind of useless until they can produce believable animation without facial capture. As long as they’re relying on 100% captured performances it’s really only good for cutscenes, or scripted sequences.

    They need to break the facial movement into part captured / part hand-animated / part procedural components that can be recombined dynamically into a character that reacts to the player in real-time with emotion and perceivable thought process (which is as much an AI/design challenge as it is tech).

    If they do that, THEN I’ll be impressed.

  20. Inzimus says:

    don’t care.
    8-bit demos on a stock C-64 will always be more amazing than this crap.

  21. Loque says:

    I would hardly get any emotional connection to this kind of CGI. Id’ rather play with a 10 years old 3D model and have fun, instead of foucsing on skin details and zombie eyes.

  22. Panda Powered says:

    Also add the Moby-Dick-rendered-fake-Norwegian-director-Kojima-Productions-Metal-Gear-my-brain-is-melting mindfuck to the list of recent uncanny valley face tech (at least they used every trick in the book to get maximum fakery).

  23. CaLe says:

    I hope they put as much effort into body deformation due to bullet impact because I want to shoot his stupid face.

    • TechnicalBen says:

      I hope you never meat him in real life. Or anyone for that fact, as faces are rather common.

  24. 11temporal says:

    More realistic then my own face.

    • Llewyn says:

      Yes, but judging by your photo next to your name that wouldn’t be incredibly difficult.

  25. Moni says:


  26. Corporate Dog says:

    I feel like everybody’s repeating the phrase “Uncanny Valley” just to hear themselves say, “Uncanny Valley”.

    For me, this blasts right through the Uncanny Valley to the other side. If not for the severed neck area that you occasionally catch a glimpse of (and the looping facial expressions I saw in the Nvidia video), I’m not sure if I’d be able to ascertain that he was 3D rendered, if I hadn’t been told ahead of time that he was.

    The technical and artistic aspects of it are all very, very good. It’s just the showmanship that needs a little work.

    • Lanfranc says:

      I feel like everybody’s repeating the phrase “Uncanny Valley” just to hear themselves say, “Uncanny Valley”.

      So, an uncanny volley of “Uncanny Valleys”?

      • Corporate Dog says:

        Something like that, yes.

        Some of us think the tech looks good!

        In fact I can’t wait until they use it in Mass Effect 4, so we can see Uncanny Volus.

    • Gryz says:

      I think the whole concept of Uncanny Valley is that the rendered images look so good, that suddenly the smallest imperfections will stand out. And give you the creeps.

      My mum told me (when I was a kid) that she once sat in the train when she was young. There was another traveler in the same compartment. The guy looked weird. But she couldn’t see why. After 10 minutes, she realized the guy had no ears. It wasn’t noticeable at once. But because you see a “normal” picture, with only a slightly unusual element, your brain becomes alert.

      I think Uncanny Valley works like that. Look at Doom, you know it’s a game. But when you look at a head like in this demo, your first impression might be “that’s real”. And then the little suspicious details will start to alert your brain. And give you that uncanny feeling.

      The head looks awesome. But the mouth does not. And the teeth do not. So that might be what gives us that uncanny feeling. And I bet when they try to render characters with hair, we’ll do two steps back. But we’re slowly getting there. Give it 10 more years.

    • TechnicalBen says:

      Technically it’s not “3D rendered”, it is “3D video captured and then replayed in 3D projection in realtime”. While a small difference, it is at least partially why this looks so real. It looks real because photographs look real. It’s captured directly from the actor.

      When we were young, and 8 bit was all we had, the impression was that it was impossible to render such things on a computer. While today we have been proven wrong, I take comfort in the knowledge that our interpretation was that the workflow would stay the same. Individual polygon modelling and hand painted textures with limited resources. The ability to record in full 3D seems to have opened up a possibility that was not there before.

    • yhancik says:

      Thanks for that post :D I wanted to say something about people who can’t help to spurt out “Uncanny Valley” every time there’s a 3D face on a blog post, but you nailed it

  27. golem09 says:

    Looks a tiny bit better, but it’s still deep in the valley.
    L.A. Noire’s faces still animate a lot better. They are not even close to this detaillevel, but the amount of expressions and the movement of the mouth is a lot better.
    Sry Nvidia and Activision, you got beaton by 2 year old tech

  28. Artista says:

    Tobey Maguire is looking creepy these days.

  29. androidpk says:

    In the grim, dark future there is only frozen fruit.

  30. PopeRatzo says:

    A nice cucumber facial mask would clear up those uncanny pores.

  31. Shooop says:

    This is really impressive, even though a few of the speech animations are still off. They’ve got the static model and blinking perfect. The subtle expressions are really good too.

    I’m more shocked that it’s Activision of all companies who’s doing this actually. I’ve never thought of them to be interested at all in pushing new tech for obvious reasons. This is a welcome surprise.

  32. norfolk says:

    For the record, an entirely fruit parfait beats an entirely yogurt parfait any day of the week.

  33. Hoaxfish says:

    Nobody ever does a tech-demo of just a person’s arm, or just their toes. Always the head.

    • MacTheGeek says:

      Knuckle hair is still beyond the ability of modern hardware to render accurately.

  34. welverin says:

    “Given that Activsion’s games output is now basically just Call of Duty and Blizzard titles”

    This made me realize have no idea what Activision has released than CoD, is Prototype 2 by Activision, because that would be one thing.

  35. almostDead says:

    At about 0.43 there is the beginnings of a gurn. Just a sneaky little one, but it was forming. Perhaps the guy making the video knew he’d get in trouble for going full gurn, but he left a little clue to know it was possible.

    Is the monologue about a parfait from a movie or TV or something? Where on earth would you get the idea to put that in your tech video?

  36. baozi says:

    i only realized i loved my face after it had been burned with acid. before, it was just my face. i only started loving it again when it had partially disappeared.

    …is what this reminds me of. that sentence, i mean.

  37. phenom_x8 says:

    Yeah, jorge jimenenz, I remember downloading his real time SSS program almost a year ago.
    Its already great so there’s no surprise here for me (Just like Nvidia Real Face Rendering tech demoed recently )
    Our PC already do this since a year ago (even my old 6850 caught between 90 – 120 FPS), its just feels like a new tech due to “next gen’ console hype recently.
    Update this is the link for the real time demo :
    link to

  38. TechnicalBen says:

    I do not wish to be in the same room as the people who want this tech so they can shoot virtual people more “realistically”. In fact, being on the same planet as them gives me the creeps.

  39. bj says:

    As much as I admire the work of Jorge Jimenez, both this and Nvidia’s demo rely heavily on the work of Paul Debevec and his team. Take that away and all you have is a mannequin. I wonder if ICT would even get a mention if both demos weren’t based on the same data.

    I really don’t think all these people on a quest for emotion are approaching it in the right way. More polys, new shaders and bigger textures can’t save a character from bad writing. Telltale’s characters in The Walking Dead had far more of an emotional impact on me than anything David Cage has ever done. They weren’t held back by their appearance.

    Basically what I’m saying is that you’re all cyber-racists.

  40. zeekthegeek says:

    The next generation is all games about disembodied heads. Warface was just the beginning.

  41. voidburn says:

    The only concern I have with this tech is that to be used it will require actors, good actors. This will further impact the budgets of AAA games, and I’m not entirely sure if the costs of delivering GOOD content with this will actually improve games in a significant way.

  42. cervor says:

    Now we know sexism in games was just a self-preservative reflex against the traumatising effects of the uncanny valley. So we can all agree on its moral necessity. Whatever. (I actually know there is a difference between sexist styles and those which are not apart from simulated realism.)

  43. KenTWOu says:

    I don’t get this head/face obsession. Real games could use this technology only in few short cut-scenes, so why do we need to use such high fidelity?

    • El_Emmental says:

      Because you don’t convey much emotions with your toe (unless you’re into foot fetish).

      Also, because it’s clearly not for real-time gameplay, and what you see in cutscenes is mostly faces talking/expressing emotions.