In case anyone hasn't seen this yet I thought I'd share it.
This is an example of motion capture in Unreal 4:
[youtube=420x260]cAxn-koJti8[/youtube]
It's especially amazing when you realize that the one on the left is ALSO being rendered by Unreal 4.
I realize CoT will be played on gaming rigs, not graphics company rigs, and that rendering dozens of people at once with zany special effects zipping about is nothing like detail-rendering one character, and that this was done with some other software tagged onto U4. However, it's still nice to know that U4 can be pushed to be nearly limitless.
This comes back with the computers requirements for CoT ^^
Any infos ?
[hr]
Suivez l'avancement du jeu City of Titans en Français sur https://titanscity.com
http://forum.titanscity.com | www.facebook.com/titanscity | http://twitter.com/TitansCity
[color=red]PR - Europe[/color]
This topic has been discussed several times over the last few years. I don't think the Devs have bothered to give us exact stats yet but it's probably safe to say that practically any PC you've bought say in the last five years and paid say maybe $750 or more for will probably have a decent chance to at least run CoT. Obviously anything newer and/or more expensive/capable the better.
Remember one of the main "selling points" of UE4 was its self proclaimed support for "lower-end hardware". The idea of course is that it's supposed to allow even low end machines to run applications (like CoT) at acceptable performances. Unless you have a very bargain rate PC or are going to try to run CoT on the same ancient machine you ran CoH on I suspect you'll probably be fine.
CoH player from April 25, 2004 to November 30, 2012
[IMG=400x225]https://i.imgur.com/NHUthWM.jpeg[/IMG]
This is the demo that caught my eye, Siren:
[youtube]-vQpc8iYdWc[/youtube]
Ninja Theory used an older version of this system to animate Senua in Hellblade.
Twitter: @SisterSilicon
Yep and the cool part is that it'll probably only be like 5 or 10 years before stuff like this becomes relatively cheap enough for the average consumer to buy. Eventually we'll be able to hook things like this up to MMOs and play games that way. :)
CoH player from April 25, 2004 to November 30, 2012
[IMG=400x225]https://i.imgur.com/NHUthWM.jpeg[/IMG]
More like 5 years [u]ago[/u]
Everquest 2 incorporated facial recognition back in 2012 using any ol' webcam. I used it and it worked pretty well, but sometimes got a bit laggy.
[youtube]9Vg1mHXBfgc[/youtube]
What I thought was really excellent was that it let you calibrate what the 'zero' position is and then your facial movements started from that baseline, so you could start off with any expression or demeanor you wanted as your default.
[hr]I like to take your ideas and supersize them. This isn't criticism, it is flattery. I come with nothing but good will and a spirit of team-building. If you take what I write any other way, that is probably just because I wasn't very clear.
Are you really trying to equate what you're talking about to what SisterSilicon was talking about? I'll certainly grant you that people have been toying around with primitive facial recognition for years now. But being able to [b]render real time photo realistic graphics[/b] based on facial recognition is a bit beyond what anyone or anything could cheaply do in 2012.
Even though the tech that SisterSilicon showed obviously exists today it'll be years before it's cheap and optimized enough to see a bunch of people using it in the average MMO. That was the basis of my "5 to 10 year" prediction.
CoH player from April 25, 2004 to November 30, 2012
[IMG=400x225]https://i.imgur.com/NHUthWM.jpeg[/IMG]
Were you referring to photorealistic graphics, or were you referring to real-time facial recognition? Because those are two separate things.
I wouldn't expect to see photorealistic graphics in an MMO until graphics processing power gets up to those standards.
Until then, the ability to incorporate real-time facial recognition and motion capture into an MMO using [u]contemporary[/u] graphics capabilities has been around more than 5 years. The only difference between contemporary graphics and photorealism would be the number of polygons in the wireframe and the detail of the normal map (texture image) placed upon the wireframe.
[hr]I like to take your ideas and supersize them. This isn't criticism, it is flattery. I come with nothing but good will and a spirit of team-building. If you take what I write any other way, that is probably just because I wasn't very clear.
Should we ask the MWM devs who did the facial animation rigging for CoT's player character models about how much more work would be necessary for Siren's level of fidelity? Or would their answers get moderated for all the profanities? ????
Twitter: @SisterSilicon
Really? You're going to try parse this out to validate your initially blithe assumption that I had absolutely no clue what I was talking about here?
One more time I was referring to the vid SisterSilicon provided which very clearly involved [b]a [color=red]COMBINATION[/color] of real time rendering of photorealistic graphics [color=red]WITH[/color] facial recognition[/b]. For some reason you seemed to gloss over (or maybe didn't even notice?) the exciting real time photo-realistic rendering part of this vid and focused mainly on the relatively boring facial recognition part of this. I was never remotely attempting to unlink the two technologies - instead what's obviously important here is that they are [b]marrying[/b] the two techs together into a SINGLE product.
Do you really think I might not be aware of general state of facial recognition or the fact that it's been around for years? I know games, smartphones, etc. have toyed around with primitive forms facial recognition in the past and the fact that up until fairly recently it's been relatively limited [b]precisely because[/b] of the lack of cheaper processing power to make it more worthwhile. Why on earth would I "predict" that facial recognition (as a standalone tech) wouldn't be a common thing by itself for 5 or 10 years when it's been around in various experimental capacities for decades now?
Clearly the [b]real time rendering of photorealistic graphics[/b] was the "cool part" of this COMBINED TECH and that's what's going to prevent this COMBINED TECH from being be a ubiquitous consumer thing for many years due to the graphics processing power required.
A mountain made out of a molehill once again...
CoH player from April 25, 2004 to November 30, 2012
[IMG=400x225]https://i.imgur.com/NHUthWM.jpeg[/IMG]
Now who's being pedantic?
My God, can you ever acknowledge a point without being so disturbingly difficult?
Look at your original post. You state that some day soon we will be able to use facial recognition and motion capture to roleplay in MMOs. At least that's what I understood you as saying.
So, when I say that it's been done in Everquest since 2012, your response is, "not to that extent, it hasn't"
Seriously. If your only point is that what Sony Online Entertainment provided in 2012 wasn't as visually impressive as that technology demonstration SisterSilicon referenced. Then Yes. I agree wholeheartedly. But to take the position you have is just downright argumentative for the sake of being argumentative. Give it a break.
[hr]I like to take your ideas and supersize them. This isn't criticism, it is flattery. I come with nothing but good will and a spirit of team-building. If you take what I write any other way, that is probably just because I wasn't very clear.
You'll notice I highlighted (in bold red) the point you've totally tried to GLOSS OVER yet again because again I suspect you didn't even originally notice it in SisterSilicon's vid. I didn't have to use the literal words "WITH REAL TIME PHOTO REALISTIC GRAPHICS RENDERING FOR MMOs" in my initial post because that should have been PAINFULLY OBVIOUS given the topic of this thread.
On more time I was referring to the vid SisterSilicon provided. YOU were the one that had to jump off into the weeds to discuss something that as far as I'm concerned was compeltely "Apples and Oranges" different. Who gives a Flying F#@k about something that dealt with relatively ancient graphics technology that was at least 5 or 6 years old in reference to this thread? Why do you even think SisterSilicon posted that vid in this thread in the first place? Because it ACTUALLY HAD SOMETHING to do with the GROUNDBREAKING [b]real time photo realistic rendering tech[/b] referenced in the original post of this thread.
I'm sorry you apparently wanted to try to make me look foolish and got called out on it. Better luck next time...
CoH player from April 25, 2004 to November 30, 2012
[IMG=400x225]https://i.imgur.com/NHUthWM.jpeg[/IMG]
Chill, Lothic. Chill.
I see absolutely no reason whatsoever why you should have responded so vehemently. I made no attempt make you look foolish. Just because you feel foolish doesn't mean it was my attempt.
Again. Go back and look at your post.
I'll quote it for you here:
No where in there is a mention of the photorealism. But there is a mention of using "things like this" in an MMO.
So maybe. Just maybe, I did not pick up on your [u]unstated[/u] reference to photorealism, and instead thought that "things like this" referred to real-time facial recognition with motion capture. I don't think that is an unrealistic interpretation of what you wrote.
Surely a simple lack of clarity on your part can lead to a misunderstanding on my part without it being perceived as a personal attack upon you.
[hr]I like to take your ideas and supersize them. This isn't criticism, it is flattery. I come with nothing but good will and a spirit of team-building. If you take what I write any other way, that is probably just because I wasn't very clear.
[img]https://i.imgur.com/lFOj8oz.jpg[/img]
[center][img=44x100]https://i.imgur.com/sMUQ928.gif[/img]
[i]Verbogeny is one of many pleasurettes afforded a creatific thinkerizer.[/i][/center]
That is a classic meme. How many layers of dust was on it? :P
"Let the past die. Kill it if you have to."
That demo'd combo of realism in the character model, near-real-time responses to input, and past-the-uncanny-valley quality of facial expression recognition was quite impressive. I'll have to investigate more, but if the character model was also generated rapidly from the human demonstrator using the same gear that was doing the motion capture, then I'll be even more impressed.
Besides the obvious "make it affordable / reliable / comfortable" challenges in the coming years, it emphasized to me the need for a flexible and user-friendly expression calibration system - one that can handle mapping a human's expressions to faces not so close to their own. Popularizing the tech and making it usable for MMOs will require a trainable system that guides a regular user as they map their expressions to a wide range of human and non-human faces, going well beyond what the Everquest video showed. Better user-calibrated control over the frog's mouth movement while speaking is one example. Another would be ensuring that the model can show all common expressions (smile / snarl / etc), handle varying degrees of each expression, and map the human version to the model version of that expression - even when the model is so drastically different that it may not have a mouth with which to smile.
Lothic - I seem to have no trouble understanding your points. If I failed to connect a direct quote (w/ video) and "like this", I'd put myself in a forum time-out. I've begun to expect a kind of... fanatical devotion to "reply" at you.
Suirprisingly that was also what SOE was able to do in 2012. Just look at the example in the video I posted in which the character's face was a frog.
And like I mentioned, it did have a way to calibrate it. I don't recall how helpful the tutorial was, however. I do remember that if I set my face a certain way when I calibrated it, and then let my face relax to it's normal expression, I could in effect create a default expression upon the character's face without having to constantly maintain that expression on my own.
[hr]I like to take your ideas and supersize them. This isn't criticism, it is flattery. I come with nothing but good will and a spirit of team-building. If you take what I write any other way, that is probably just because I wasn't very clear.
I must have been in the midst of editing when you replied. My edit partly addressed the frog expressions, which seemed to be struggling to match human expressions (odd eye movements and mouth shapes not well matched to either the speech or expression, grimacing instead of smiling), though it improved for the brief time when the calibration screen was open. Perhaps the difficulties I saw were just for that video and user, while your own testing had more success. Until we can see demo video showing a solid mapping of human user to human character expressions in a realistic gamer setting - that doesn't end up looking like those uncanny valley Mass Effect expressions - I think widespread popularity is years away. Getting that calibration system to be good enough to duplicate natural expressions onto a human character (as the CES video did achieve) but without having a team of experts, bright lighting, a clean shave, recalibration every few minutes, and so forth, seems to be a big hurdle.
I suppose what I'm thinking of as an interim solution is a second pathway, more limited in some ways but more flexible in others. It wouldn't be what we saw in both videos - a kind of direct, dev-assumed mapping of human expressions to character expressions regardless of how unusual the character may be (alien, robot, frog, etc). It's better described as an option to use my expressions as macros to trigger emotes (prerecorded by the devs) that are designed to match the type of head the character is using. To do so, it needs a way for each user to teach the recognition system to perform [Emote 1] when the user makes [expression A]. The user would select the emote they want to train the system to recognize, then perform a distinct expression; the character emote would then be performed whenever the user's face closely matches that stored [expression A].
Given the broad range of possible characters - who might have an eye count not equal to two, might be able to control ear movement far more than a human, or may have no face at all yet still have ways of expressing emotion through physical appearance, I don't see how they could all be handled with the sort of direct "user's face movement" --> "character's face movement" mapping that both videos seemed to enforce. We may also want to account for people who have trouble getting the system to recognize their expressions (due to facial hair, injury, glasses, etc) by letting them substitute a different input expression that the system can recognize.
An example of this might be with a Chameleon-human character - unlike a human, its "confused" emote might be designed to involve sticking out its tongue and changing the color of its skin to purple, while its smile involves rolling its eyes in opposite directions and sticking out its tongue. I can't properly emote either of those for the recognition system to duplicate, but if the system had a macro feature as I described, I could program it to perform the respective emotes when I show my human equivalent of confusion and a smile.
I like your second pathway idea for all the reasons you stated. And I think it would be nice for the player to divorce his or her own facial expressions from the characters at those certain times of maximum eye rollage. No one can stay in character [u]all[/u] the time.
Then there is the added benefit that using a webcam to recognize certain key expressions as triggers to activate an emote doesn't have to be limited to just emotes. Once this is in place, users could theoretically be able to trigger any in-game action with facial expressions. In other words, using facial expressions as an actual input device for game actions. Why not?
Can you imagine the accessibility advantages that would give to people with motor control difficulties?
[hr]I like to take your ideas and supersize them. This isn't criticism, it is flattery. I come with nothing but good will and a spirit of team-building. If you take what I write any other way, that is probably just because I wasn't very clear.