Accurate human rendering in game [2014-2016]
- Thread starter babcat
- Start date
- Status
The facial expressions are massively detailed yet, they have some uncanny valley into them. Like there is no cohesiveness between the different areas of the expression which destroys the believabilityDamn, what an impressive Hollywood cast that will be mocap-ed! Best one in the videogame yet! The only actor from Wing Commander era that I wish they also got is Malcolm McDowell.
And it looks like they have upgraded Ryse mocap/ CryEngine3 rendering quite nicely.
That face needs to blink a lot more. Eyes are easily one of the biggest giveaway when it comes to animated faces. One of the biggest issues I find here is that the blinking almost never looks natural, the speed of the blink is way too slow and the blinks themselves are separated so far apart that whenever you do see them it seems very artificial and carefully planted rather than natural. Real people blink fast, and they do so quite often in a random manner...sometimes even twice or three times in a row !
Apart from that the other issue is iris movement, when a character is looking around and turning his head like Gary is doing in that video there is no way the iris is going to be that still looking, a real person's iris would jump places and twitch around. Some games like Until Dawn manage to achieve this detail though.
Apart from that the other issue is iris movement, when a character is looking around and turning his head like Gary is doing in that video there is no way the iris is going to be that still looking, a real person's iris would jump places and twitch around. Some games like Until Dawn manage to achieve this detail though.
cheapchips
Veteran
Oldman looked good I thought. The in game faces on the AC2 demo were pretty naff looking.
I like the blood flow tech they are using. Too bad we don't know when and if SC will ever release :|
I tried the Gamescom alpha and my impression of it is that it doesn't even look like a game yet.
Edit: Gary Oldman interview + some footage from perf capture
I tried the Gamescom alpha and my impression of it is that it doesn't even look like a game yet.
Edit: Gary Oldman interview + some footage from perf capture
Last edited:
What? Please tell me that you are not believing all this Derek Smat bullshit.
SQ42 will be out in 2016, or at least the 1st half of the campaign that was promised to come out first.
What i played from the game was disappointing to say the least, i haven't followed the latest drama closely so i was basing that off of what i played. Also, SC is a project that is riddled with delays and promises so i can't really trust them to release SQ42 in 2016.
You do realize that blinks are straight from the capture, right?
With that said, this would be one of the first things to be tweaked with manual animation so there is indeed some truth to your point here.
Yeah the eyes have a minimum turn radius of 2-4 degrees or so, basically jumping instead of a constant follow movement. Again, something for the animators.
All in all it's still impressive and I imagine they're going to keep working on it. But the sheer amount of animation they need is going to require some compromises.
#1 I have very rarely seen him in a interview setting, so it is strange to hear him without his American "movie" accent.
#2 does the face camera recording has any issue with him having glasses? He's the first actor that I saw in mocap gear that wear glasses.
Last edited:
Does any current lip-sync technology consider the difference in light-wave vs. sound-wave speed? I have the weird feeling that the lips lag, even though I do believe they are matched exactly. But in reality shouldn't actually the sound lag (to make the brain feel like observing real phenomena), because it takes a while?
If there are any sync issues it's usually more about the tech stuff and not speed of light vs sound.
I actually think that this presentation was a little bit rushed maybe, having a set deadline for this citizencon or whatever; and also perhaps driven by a need to prove that the budget is well spent, as there's been some article on the Escapist recently with some pretty strong claims about management problems at the studio. So they basically decided to show of stuff that was probably not as polished as it could have been.
You see, as far as I'm aware there's no facial mocap solution that can just deliver perfect quality without a lot of manual animation on top. Not even for places like Weta or ILM. It's basically the old 20-80 rule, which means that usually the first 20% of effort will get you to 80% of the quality you seek, but then you have to spend 80% of your effort to get that last 20% of the result.
Coincidentally the same is true for body mocap, but that tech is a lot more mature and more understood by now, so we usually don't get to see the rougher first pass results.
So what I believe to be the case here is that they didn't have enough time to fully polish the animation and get rid of these tiny little issues, which however are pretty damn important to sell the character. Both the tech and the acquired datasets seem to indicate that their system can produce great results - but no matter how much R&D goes into it, there still has to be animators who can provide that final but important touch to bring the characters to life.
But also, as I've mentioned above, this effort takes a lot of human resources, which are quite expensive; and a game like this would probably require hours and hours of facial animation. So no matter how big the budget is, there's always going to be a limited amount of work that's possible to do on the rough p-cap and so we can't reasonably expect every second of the results to be perfect. Here, it comes down to the proficiency of the supervisors to allocate the resources in the most efficient way and make sure that the important and most visible moments get the greatest effort.
So I guess it's a bit too early to tell just how much of the potential of their tech they will actually be able to realize. What has been shown is certainly promising.
I actually think that this presentation was a little bit rushed maybe, having a set deadline for this citizencon or whatever; and also perhaps driven by a need to prove that the budget is well spent, as there's been some article on the Escapist recently with some pretty strong claims about management problems at the studio. So they basically decided to show of stuff that was probably not as polished as it could have been.
You see, as far as I'm aware there's no facial mocap solution that can just deliver perfect quality without a lot of manual animation on top. Not even for places like Weta or ILM. It's basically the old 20-80 rule, which means that usually the first 20% of effort will get you to 80% of the quality you seek, but then you have to spend 80% of your effort to get that last 20% of the result.
Coincidentally the same is true for body mocap, but that tech is a lot more mature and more understood by now, so we usually don't get to see the rougher first pass results.
So what I believe to be the case here is that they didn't have enough time to fully polish the animation and get rid of these tiny little issues, which however are pretty damn important to sell the character. Both the tech and the acquired datasets seem to indicate that their system can produce great results - but no matter how much R&D goes into it, there still has to be animators who can provide that final but important touch to bring the characters to life.
But also, as I've mentioned above, this effort takes a lot of human resources, which are quite expensive; and a game like this would probably require hours and hours of facial animation. So no matter how big the budget is, there's always going to be a limited amount of work that's possible to do on the rough p-cap and so we can't reasonably expect every second of the results to be perfect. Here, it comes down to the proficiency of the supervisors to allocate the resources in the most efficient way and make sure that the important and most visible moments get the greatest effort.
So I guess it's a bit too early to tell just how much of the potential of their tech they will actually be able to realize. What has been shown is certainly promising.
sound would travel roughly 12 feet in 1 hundredth of a second so the lips would be less than 1 frame behind at 60fps
Higher quality videos at gamersyde:
http://www.gamersyde.com/news_star_citizen_showcases_squadron_42-17197_en.html
I could see the splitte much easier in the gamersyde Videos.
The Squadron 42: Facial Animation Tech Video is interesting too.
http://www.gamersyde.com/news_star_citizen_showcases_squadron_42-17197_en.html
I could see the splitte much easier in the gamersyde Videos.
The Squadron 42: Facial Animation Tech Video is interesting too.
ArcticCircle
Newcomer
I think the main problem here is Chris Roberts lack of director skills and artistic vision. I mean man did direct and write awful Wing Commander movie and was producer for Who's Your Caddy?, one of the worst rated non indie movies in imdb.
Cast is just about as good as you can get, as it was in many Wing Commander games too. It didn't stop everything from being cheesy garbage, despite the A list actor talent. Man that SC senate speech... Oldman should have pulled Harrison Ford quote here - "Chris, you can type this shit, but you can't say it"
Cast is just about as good as you can get, as it was in many Wing Commander games too. It didn't stop everything from being cheesy garbage, despite the A list actor talent. Man that SC senate speech... Oldman should have pulled Harrison Ford quote here - "Chris, you can type this shit, but you can't say it"
This has nothing to do with tech.
There will total 10 hours of performance capture (face and Body) be used in Squadron 42 and this small part is a propaganda speech. Oldman compared Roberts with George Lucas. If all things were so bad much of the famous actors would not take part in there. They earn less in video games than in films. They also probably have seen their part of the script before they have assumed the role.
By the way, Lucky Number Slevin, Lord of War, The Punisher and The Jacket were good. Wing Commander is a bad movie but there were several problems on both sides. Roberts and the publisher.
There will total 10 hours of performance capture (face and Body) be used in Squadron 42 and this small part is a propaganda speech. Oldman compared Roberts with George Lucas. If all things were so bad much of the famous actors would not take part in there. They earn less in video games than in films. They also probably have seen their part of the script before they have assumed the role.
By the way, Lucky Number Slevin, Lord of War, The Punisher and The Jacket were good. Wing Commander is a bad movie but there were several problems on both sides. Roberts and the publisher.
Silent_Buddha
Legend
In other words, SC is proceeding just like any other AAA title.
I think the problem people have is that unfortunately for CIS, they have let people see development progress from Day 1.
I have a feeling if anyone had been around for Uncharted's development from day 1, people would have been crying out about how the project is riddled with delays and promises. Or Bioshock. Or Gears of War. Or Halo. Or Killzone. Or StarCraft. Or just about any other AAA title that was the first in a franchise (or stand alone).
But for those the public didn't even get a hint that the title was in development until it had been in development already for 2-4 years. And then another 2-3 years until they actually released.
If Star Citizen had followed your regular AAA development schedule, they'd only just now (being sometime this year) be releasing information to the public about an upcoming game they've been working on. Or if they wanted to release information super early, they might have released info. late last year about the game to the public.
It's a double edged sword for crowd funded games that want to let people be involved with development from day 1. And a mistake for most developers, IMO, to do that. Because then you have all this unnecessary wild speculation by people who don't know any better.
Regards,
SB
^ I completley agree with that. That's the trouble with crowdfunded games, especially those who have longer development time. I don't think we ever saw any other AAA game with this kind of commuinity outreach [the only other example I saw was Project CARS which had fantastic but also very long develpment], so almost everyone is surprised to see that games don't just "appear working great" from the very start and they are crafted in a year.
Not only that, malicious trools can start dishing out actual full blown online hostile attacks, and since there is large pool of people who have "gave their hard earned money" to this unfinished project, even if small percentage of those people get swayed, all of them will get VERY LOUD... and even small amount very loud people can influence general online population easily. In the case of SC, one insane troll managed to stir quite a lot of unnecessary trouble. Shame that happened.
Not only that, malicious trools can start dishing out actual full blown online hostile attacks, and since there is large pool of people who have "gave their hard earned money" to this unfinished project, even if small percentage of those people get swayed, all of them will get VERY LOUD... and even small amount very loud people can influence general online population easily. In the case of SC, one insane troll managed to stir quite a lot of unnecessary trouble. Shame that happened.
Can only agree SB and DieH@rd, for a game of this scope a 5-7 year cycle is not that surprising given that the engine has yet to prove itself for anything other than shooters (I really feel for the tech team). It's not as if the end of your KS is day 1 on the project, it's more like day -180 (at best) as you still have to rent a building(s), hire all the department heads, then the team leads, then the engineers/artists/designers/actors/writers/sound designers/etc/etc.
CIG haven't helped themselves though both by shooting for the moon and by shying away from being upfront about how long a moonshot might take but to go from that to 'it's never happening' when the timelines are this young is a bit much (especially the 'he's living the high life off ill gotten KS funds' nonsense that's been floating around of late). 'They only show ships' is the kind of thing you say when you don't understand that models can and should be created independent of the engine that makes them do anything, if art had to wait for an alpha engine to work you'd be paying an awful lot of artists to freshen up their portfolios fulltime.
CIG haven't helped themselves though both by shooting for the moon and by shying away from being upfront about how long a moonshot might take but to go from that to 'it's never happening' when the timelines are this young is a bit much (especially the 'he's living the high life off ill gotten KS funds' nonsense that's been floating around of late). 'They only show ships' is the kind of thing you say when you don't understand that models can and should be created independent of the engine that makes them do anything, if art had to wait for an alpha engine to work you'd be paying an awful lot of artists to freshen up their portfolios fulltime.
- Status
