Lip sync in Sam & max

edited March 2008 in Sam & Max
Hi there,
Now, some people may complain about the older technology used in the episodic Sam & Max games. They may complain that there's not enough polys or fancy textures and effects. I say if there's anything that truly needs improvement, it's the lip sync.

Don't get me wrong, it's not like it's horrible or anything, but with Sam & Max having such heavy focus on dialogue, it makes sense for anything related to dialogue to be the best it can. The acting is great, the scripts are great, the automated lip sync... Not so great. Many times characters completely miss syllables or don't seem to change their mouth shapes quickly enough to follow the dialogue. It kinda spoils the look when other games have such great lip sync. Have a look at the new Chariots of the Dogs trailer, Bosco misses a large amount of his speech.

Perhaps Telltale could spend some time improving their automated lip sync technology? Or maybe they could license Valve's? Their automated lip sync is awesome, not just for realistic characters, but as anyone who's played Team Fortress 2 would know, cartoony characters too.
«1

Comments

  • edited March 2008
    Speaking of audio-relating issues I've noticed the voices have constantly been a bit on the fuzzy side, it's not as clear and precise as it should be. Is this because of mp3 compression or something since you always have to make the complete file around 100mb?

    I'm using the Logitech Z-5500 speakers, perhaps they are too powerful for your games :P (I'd like to see Sam & Max in 5.1, that'd be very interesting..)
  • edited March 2008
    Still a lot better than the syncing in HtR :P
  • edited March 2008
    In their defence, their lip-sync is procedural. The teams are so small, they had to make a program that does all the lip-syncing in the game for them, and it's not going to be as perfect as when the characters are being animated for the cutscenes at the beginning and end of the episodes.

    But like ShaggE said, it's better than what was in HTR.
  • edited March 2008
    You know, I hadn't really noticed. I guess I'm used to lip sinc not being perfect.

    I admit it would be kind of cool if TellTale brought one of those lip-sinc programs that scan the actor's voice and move the lips to exactly match, but it's by no means necessary in my opinion.
  • edited March 2008
    Why buy when they already made their own? :D
  • edited March 2008
    This bothered me a bunch at the beginning of Season One, but I got over it.
  • edited March 2008
    ShaggE wrote: »
    Still a lot better than the syncing in HtR :P
    But wasn't the speaking in HtR just flipping between the 'mouth open' and 'mouth closed' sprites really fast? You can't really call it lip synch of any type if they're not even trying.
  • edited March 2008
    Chris1 wrote: »
    You know, I hadn't really noticed. I guess I'm used to lip sinc not being perfect.
    Yeah, it's the same with me. I'm so used to seeing non-matching lip-synch in pretty much every game I play and every cartoon I watch that it doesn't really bother me.
  • edited March 2008
    ShaggE wrote: »
    Still a lot better than the syncing in HtR :P

    Not necessary : in HtR, the lip sync is completely bad, you don't try to make le link between the voice and the image. In TT's games, its quite good but not enough, so you really get the impression it's not finished. Although, I like the way the lips moves in HtR (it's very old-fashion-cartoon-like), and so is the TT's one. By the way, I don't focus on the lips as I read the subtitles :D
  • edited March 2008
    The fact is this is the program they made and they're sticking with it, plus it's an episodic game. I would only really be complaining if it was a full budget big project. Then, it would annoy me. You should listen to the commentary on episode 101 about how it makes it easier and faster rather then getting animators to work weeks to perfect the lip-sync animation. In short, it's faster this way. :)
  • edited March 2008
    Mwyann wrote: »
    Not necessary : in HtR, the lip sync is completely bad, you don't try to make le link between the voice and the image. In TT's games, its quite good but not enough, so you really get the impression it's not finished. Although, I like the way the lips moves in HtR (it's very old-fashion-cartoon-like), and so is the TT's one. By the way, I don't focus on the lips as I read the subtitles :D
    It's kind of like... um... what's that thing called? Where we scrutinize faces more as they get more realistic. There's a name for it. I just remember that it's why most 'realistic-but-not-real' faces look creepy.
  • edited March 2008
    TrogLlama wrote: »
    It's kind of like... um... what's that thing called? Where we scrutinize faces more as they get more realistic. There's a name for it. I just remember that it's why most 'realistic-but-not-real' faces look creepy.

    The Uncanny Valley.

    See the "Heavy Rain" tech demo for a good example of it. Gave me the chills the first time I watched it.
  • edited March 2008
    The fact is this is the program they made and they're sticking with it, plus it's an episodic game. I would only really be complaining if it was a full budget big project. Then, it would annoy me. You should listen to the commentary on episode 101 about how it makes it easier and faster rather then getting animators to work weeks to perfect the lip-sync animation. In short, it's faster this way. :)
    I think that BECAUSE it's episodic they have the opportunity to improve it. Why? Each episode they can slowly make improvements, until they have a really good automatic lip sync program. Yes, I had listened the commentary, and indeed having the process automated does make it faster than hand animating the mouth movements, but that doesn't mean the program that does it can't be improved!
  • edited March 2008
    ShaggE wrote: »
    The Uncanny Valley.

    See the "Heavy Rain" tech demo for a good example of it. Gave me the chills the first time I watched it.

    I heard that the developers for that tech demo were under high pressure to get it finished before E3, so they didnt have time to polish it some more. Which is too bad, because it was already pretty good.

    On-topic: I agree, lipsynching is probably one of the things that need improvement most. Mostly beacuse people talking are always staring at the screen, so you notice it more that in games where they might as well be facing away from you.

    Perhaps for season 3?
  • edited March 2008
    Yeah, pretty good idea. Concur.

    But it would render the idea of episodic Sam & Max practically impossible.
  • jmmjmm
    edited March 2008
    Come on guys. There are lots of more critical things to be added, enhanced or fixed than this.
    I'd rather have the engine development team to work on better memory management for instance or to fix some of the outstanding problems on Vista.
  • JakeJake Telltale Alumni
    edited March 2008
    Better lip sync is something we'd love to do, and have some ideas for. Don't hold your breath because it's something that's just been sort of talked about as a thing to do that would be cool, but it's also something we're definitely aware of.
  • edited March 2008
    Any thoughts on licensing the lip sync technology Valve uses, Jake?
  • JakeJake Telltale Alumni
    edited March 2008
    Is their lip synch technology good? I remember very much enjoying their facial animation, especially the eyes, but I don't remember the lip sync. I am about a far disconnected from lip sync technology as you can get in the company, but I imagine that if we did another pass at lip synching, we would write our own software. (The current lip sync stuff is based off of existing software but has been fiddled with a bit to up its quality... at least to my knowledge! Not entirely sure.)
  • edited March 2008
    Just as long as you guys don't end up making the lip sync look like the "open, close, form an O with the 'ooo' sounds" method in Guitar Hero 3. It works for most songs, but in others, it looks like a really bad 3D anime character c.Speed Racer.
  • edited March 2008
    Jake: Is Valve's lip sync technology good? Have you even played Team Fortress 2? :P It's excellent! Particularly the new version they used for Team Fortress 2. Not only does all the mouth movements almost perfectly match what the character is saying (as Half-Life 2 did well) but with TF2 having a more cartoony style, it has a little more exaggeration to make it a bit more "animated'. Perfect for Sam & Max I say.
  • edited March 2008
    I would guess that TF2's lip sync is done by hand, given how little spoken dialogue there is.
  • edited March 2008
    Jake wrote: »
    Is their lip synch technology good?

    See for yourself.

    Yeah, I'd say it's pretty good.
  • edited March 2008
    tabacco wrote: »
    I would guess that TF2's lip sync is done by hand, given how little spoken dialogue there is.

    With the Source SDK it's as simple as typing in a phrase, the program extracts the phonemes, and automatically does the syncing near-perfectly. Might take some tweaking sometimes.. as in some cases it might be glitchy.

    It's actually cool because it makes its best guess on where the particular words go based on the soundwaves in the actual audio recorded.
  • JakeJake Telltale Alumni
    edited March 2008
    That's what the off-the-shelf software we use does, though it's fighting an uphill battle with things like dog mouthes, and a gigantic rabbit maw which takes up 1/3 of an oval face.
  • WillWill Telltale Alumni
    edited March 2008
    Also, another thing to bear in mind is that our characters don't exactly have standard facial structure. It's a wonder that Max can produce anything vaguely resembling words with his gaping maw.
  • edited March 2008
    Will wrote: »
    Also, another thing to bear in mind is that our characters don't exactly have standard facial structure. It's a wonder that Max can produce anything vaguely resembling words with his gaping maw.

    It's a wonder you guys were able to completely close his mouth, too, at the end of 101.:D
  • edited March 2008
    tabacco: No, they're definitely not hand animated. The trailers are though, and although those look much better than the actual game, what you seen in the game is still very very good.
    Jake wrote: »
    That's what the off-the-shelf software we use does, though it's fighting an uphill battle with things like dog mouthes, and a gigantic rabbit maw which takes up 1/3 of an oval face.
    I was under the impression the software used pre-set mouth poses chosen by the animators. i.e if sound is like this, use this pose.
  • edited March 2008
    billyboy12 wrote: »
    I was under the impression the software used pre-set moth poses chosen by the animators. i.e if sound is like this, use this pose.

    Technology, as illustrated in this thread, has moved past the Nightmare Before Christmas lip sync process. :)

    And in case you didn't get that, that's how they did the lip sync in the film. They had a pre-set collection of mouths (or in Jack's case, heads) that the animators picked through a computer program based on the phonetics of the script. Interesting program, but I'm glad we are far from that method.
  • edited March 2008
    No no, I mean they had a series of poses for different sounds, and the programs tweens between them as it analyzes the audio.
  • edited March 2008
    Which was the case in Nightmare before Christmas, and to some extent what is going on in HalfLife2 and in the TTTool. It's just that these days, the lip sync is more complicated, as Jake said, because of the unusual shape of Sam & Max's mouths.

    Now, you would think that characters like Superball or Abe would be easier for the lip sync to read properly, but I guess their rigging isn't as complicated as a HalfLife2 character's face.
  • DanHDanH Telltale Alumni
    edited March 2008
    Will wrote: »
    Also, another thing to bear in mind is that our characters don't exactly have standard facial structure. It's a wonder that Max can produce anything vaguely resembling words with his gaping maw.

    ...a wonder indeed. :D
  • JakeJake Telltale Alumni
    edited March 2008
    billyboy12 wrote: »
    No no, I mean they had a series of poses for different sounds, and the programs tweens between them as it analyzes the audio.

    That is true, but it blends less than optimally. As I said, we have plans to improve on it. :)
  • edited March 2008
    Well, if you're planning on improving what you have now, here's my suggestions, feel free to use/not use them:
    1) Faster reaction time - Some characters completely miss phonemes because they're currently in the process of forming a different phoneme.
    2) Faster mouths - The character's mouthes seem to move unnaturally slow. For someone like Max, who talks reasonably fast, it looks a little odd.

    Basically, characters need to hit their phonemes faster, and more accurately ;)
  • edited March 2008
    billyboy12 wrote: »
    Basically, characters need to hit their phonemes faster, and more accurately ;)

    That's a give-and-take situation. I can run HalfLife2 on my computer, even though it is a POS, because it meets the minimum video requirements to run. The unfortunate thing is it isn't powerful enough to where the lip sync won't mess up if there is too much going on at the screen at once, mostly because the game has an insane amount of detail.

    With Sam & Max, if you look at their opening and ending cut scenes for any episode this season or last, the lips tend to hit their mark rather nicely. At least they do on my end. For someone who is running on bare minimum, the lip sync in those cut scenes are probably just as "bad" as the lip sync in-game.

    In other words, they can improve it so that it hits the phonics faster, but it all depends on if YOUR computer can do the same thing.
  • edited March 2008
    Zeek wrote: »
    That's a give-and-take situation. I can run HalfLife2 on my computer, even though it is a POS, because it meets the minimum video requirements to run. The unfortunate thing is it isn't powerful enough to where the lip sync won't mess up if there is too much going on at the screen at once, mostly because the game has an insane amount of detail.

    With Sam & Max, if you look at their opening and ending cut scenes for any episode this season or last, the lips tend to hit their mark rather nicely. At least they do on my end. For someone who is running on bare minimum, the lip sync in those cut scenes are probably just as "bad" as the lip sync in-game.

    In other words, they can improve it so that it hits the phonics faster, but it all depends on if YOUR computer can do the same thing.
    Correct me if I'm wrong, but aren't the intro/outro sequences animated in Maya?
  • edited March 2008
    TrogLlama wrote: »
    Correct me if I'm wrong, but aren't the intro/outro sequences animated in Maya?

    I didn't know Ms. Angelou was an animator! :p
  • edited March 2008
    TrogLlama wrote: »
    Correct me if I'm wrong, but aren't the intro/outro sequences animated in Maya?

    Yeah, but they still have to export the animation sequence from Maya and into the game, because not everyone that buys the episode will have Maya. Furthermore, they said in last seasons audio commentary that lip syncing in Maya is a pain in the ass, as the dialouge can get so long and complicated that it is better to just use the automated program to save them time. That's why if you know what to look for, you can tell when the Maya animated sequence is done and when the TTTool takes over.

    They've gotten better at that blend this season from my end, but there were several parts in the opening and closing of 203 where the cut scenes would skip in the animation because of processing issues. And that's DURING the parts I could tell that were animated in Maya.
  • JakeJake Telltale Alumni
    edited March 2008
    I think you may be misidentifying the issue, sadly.
  • edited March 2008
    Probably, but that's my logic for you. :p
Sign in to comment in this discussion.