Support NeoGAF

[TCKaos]

So we're one step closer to the future.

While this is a very interesting idea I have no idea how one could properly design a game around this. You'd need a single level area (elevation-wise. No bumps or hills or anything) that you can't move from. Maybe a combination of this and that Battlefield 3 room or something. If player movement is truly 1:1 (and anything else would throw you right in to the uncanny valley) then without one of those weird floor-systems like the one in the Battlefield 3 room you'd be left with a play space the size of your living room.

 

[Grakl]

Very nice. I see very little input lag as well.

I think the glasses or whatever that he wears would be great to play games, not necessarily with the motion sensors, but with a regular controller.

 

[Dictator93]

Catching up to cryengine I see

 

[Liamario]

I'm not interested in swinging my arms around to play a video game. That's called exercise!

 

[DieH@rd]

Motions sensors are very expensive for this setup, but they will get cheaper... eventually.

 

[Sethos]

Touching myself will take on whole new dimensions

 

[Flame Lord]

 

[domlolz]

'So we're one step closer to the future.'

lol

 

[roadkillgrill]

Oh wow, I'm a gif on the internet!

I work at YEI, I made this demo and am the guy dancing around. If you have any questions feel free to post them.

Thanks everyone!

 

[wwm0nkey]

Oh wow, I'm a gif on the internet!

I work at YEI, I made this demo and am the guy dancing around. If you have any questions feel free to post them.

Thanks everyone!

How was it to walk around and move freely in a virtual world.

 

[endresults]

There will come a day when you can make your own pre-recorded celebratory gestures or taunts to use in FPS's and sports games using this sort of technology. I eagerly await the endless variations of crotch thrusting, teabagging, and simulated masturbation.

 

[ChosenPredator]

or the day you enter tournaments in your room and your heart is connected via wifi and if you die your heart explodes

 

[Kinyou]

One step closer...

 

[Haunted]

Cool stuff. Combine this with the Oculus Rift and we're partying like it's 1992.

Oh wow, I'm a gif on the internet!

I work at YEI, I made this demo and am the guy dancing around. If you have any questions feel free to post them.

Thanks everyone!

Good job!

 

[Platy]

Hard to think of a good utilization of this for players on their rooms .... you need AT LEAST a controller in one hand to move, or the levels would be as big as the smallest room avaliable (and we are talking about an industry with huge sales on Japan ...)

 

[chertipros]

Oh wow, I'm a gif on the internet!

I work at YEI, I made this demo and am the guy dancing around. If you have any questions feel free to post them.

Thanks everyone!

Good Job!

 

[onQ123]

This is the same idea that I always had for motion controlled games but I know people wouldn't want to wear the bands. but I'm sure we can sell this we just have to make a dancing game that come with the bands , & have a ad with Juicy J 'Bands will make her dance' playing in the background lol.

Bandz A Make Her Dance (Explicit)


someone edit this video to play 'Bandz A Make Her Dance' in the background lol

 

[jet1911]

Dancing around -> Cool
Moving in the game world -> Do want

 

[wabo]

Oh wow, I'm a gif on the internet!

I work at YEI, I made this demo and am the guy dancing around. If you have any questions feel free to post them.

Thanks everyone!

I'm curious to know about the pricing. Old but reliable tech based on "balls" is to a certain extent cheaper* and offers better accuracy. In the other hand, this seems more flexible but output form gyros, as always, looks noisy. So my question is: in real world mocap, how could this be useful. Is it inexpensive or this compensates with countless hours of fixing markers. How easy is to fix de noise? Cheap-ass gyros like the ones in DS3 or Wii could be used?

I'm looking to it as a cheap method of improving animation workflows for indie devs.

* If you don't use many cameras... but then you have to fix many points so it becomes pointless (pun intended).

 

[roadkillgrill]

How was it to walk around and move freely in a virtual world.

Freaking awesome!

On the first take I accidentally pulled out the HMD trying to walk too far.

I funded the Occulus Rift Kickstarter and can't wait to try that one out on this setup. The FOV and response rate on the HMZ-T1 is not great. I can see motions on a secondary monitor a before seeing it in the HMD.

 

[subversus]

Catching up to cryengine I see

Haha, this.

This feature has been implemented in CE3 a year ago or so.

 

[Pillville]

Anyone else instantly think of this?

 

[wwm0nkey]

Freaking awesome!

On the first take I accidentally pulled out the HMD trying to walk too far.

I funded the Occulus Rift Kickstarter and can't wait to try that one out on this setup. The FOV and response rate on the HMZ-T1 is not great. I can see motions on a secondary monitor a before seeing it in the HMD.

All you need is that Occulus and This or This and you will be good to go!

 

[nickcv]

Hard to think of a good utilization of this for players on their rooms .... you need AT LEAST a controller in one hand to move, or the levels would be as big as the smallest room avaliable (and we are talking about an industry with huge sales on Japan ...)

i get what you are saying but you could just easily have some sort of button (ie: in your palm) or some gesture to trigger a "walk mode".
as long as you are in that mode walking in place would make your character move foreward in the virtual world.

sure, breaks the immersion, but so does all the rest of equipment you are wearing to make this work.

Freaking awesome!

On the first take I accidentally pulled out the HMD trying to walk too far.

I funded the Occulus Rift Kickstarter and can't wait to try that one out on this setup. The FOV and response rate on the HMZ-T1 is not great. I can see motions on a secondary monitor a before seeing it in the HMD.

i'm really jelly

 

[K.Sabot]

*sees the attempt to kick"

Gaming just got a whole lot less cool looking.

 

[nickcv]

can you imagine playing a platformer like super mario with this kind of techonology? that would really be a massive workout.

 

[Panajev2001a]

Freaking awesome!

On the first take I accidentally pulled out the HMD trying to walk too far.

I funded the Occulus Rift Kickstarter and can't wait to try that one out on this setup. The FOV and response rate on the HMZ-T1 is not great. I can see motions on a secondary monitor a before seeing it in the HMD.

Is it tough not to make the model jittery (probably due to sensor data and trying to balance between jitter and lag)? I still see some jitter in the 3D figure which I do not see on you, not much, but I am interested in knowing how you would plan to attack it.

Great job, btw!

 

[Razek]

I'm not going to cry that this should be stopped, because there are people out there I'm sure who are generally interested in this stuff. So cool that this is hitting another engine.

For everything other than elder scrolls, me waving the sword myself in fights, give me a controller please. I'm just not interested.

 

[Blizzard]

Haha, this.

This feature has been implemented in CE3 a year ago or so.

Are you talking about the mocap system that let developers record animations for games, or something else? I didn't think that was for end users but I may very well be wrong. I vaguely remember a video or two.

 

[Black_Stride]

Catching up to cryengine I see

This was my post you bastard!!!

Are you talking about the mocap system that let developers record animations for games, or something else? I didn't think that was for end users but I may very well be wrong. I vaguely remember a video or two.

Nah CE and im sure even UDK has been able to do real time captures for a while.

In fact many people do less accurate Real Time captures using Kinects because of the depth sensor and what not....get two Kinects working in unison and you can do some complex stuff.
I dont think anything quite a the level of this video....but if you have the tools and hardware, real time motion capture has been around in CE for a long time.

This video shows a small example of whats possible with Kinect and CryEngine
http://www.youtube.com/watch?v=b0OroKutj08


Theres a whole community dedicated to kinect based motion capture....its a very very cheap alternative.

On Topic

Looks pretty damn cool...i want to try it nao!

 

[XtremeXpider]

Motions sensors are very expensive for this setup, but they will get cheaper... eventually.

Kinect?

 

[roadkillgrill]

I'm curious to know about the pricing. Old but reliable tech based on "balls" is to a certain extent cheaper* and offers better accuracy. In the other hand, this seems more flexible but output form gyros, as always, looks noisy. So my question is: in real world mocap, how could this be useful. Is it inexpensive or this compensates with countless hours of fixing markers. How easy is to fix de noise? Cheap-ass gyros like the ones in DS3 or Wii could be used?

I'm looking to it as a cheap method of improving animation workflows for indie devs.

* If you don't use many cameras... but then you have to fix many points so it becomes pointless (pun intended).

The sensor kit used in the demo cost $3995. That includes 17 wireless sensors and 3 dongles.

This demo is a very basic implementation, it is directly applying the orientation calculated from each sensor to the bone with no client side filtering or noise reduction currently.

 

[Zaptruder]

Oh wow, I'm a gif on the internet!

I work at YEI, I made this demo and am the guy dancing around. If you have any questions feel free to post them.

Thanks everyone!

Nice work. I saw this on the MTBS3D forums, but decided not to repost it. Kinda cool that you're here too!

Not much to say other than great proof of concept demo.

The really nice thing about it is how much more personalized and emotive animation is through full motion body capture like this (or whatever you call the accelorometer based version of it).

When big budget multiplayer games get around to including this kinda stuff - we'll really be scratching out heads over why we were so obsessed with standard gamepad controls.

The level of interaction, immersion, and communication goes way beyond the canned animation options available to a standard control system.


Can I make a suggestion about walking though? I recommend a walking-in-place function. Couple this thing up with say... a move navigation controller. Push on the navigation controller to determine the the direction of travel - but speed of travel is determined by how high and fast you raise your knees. i.e. pushing up on the controller and running on the spot will cause your avatar to run forwards.

Ideally, the foot fall rate of the avatar would be the same as yourself to create the illusion of synchronicity - the closer you can marry the movement to the sensation, the better the immersion will be. But on the spot movement is the most practical compromise for navigating large scale virtual environments that exceed the bounds of the physical space you're in, without additional (and often expensive) hardware requirements.

 

[blu]

Oh wow, I'm a gif on the internet!

I work at YEI, I made this demo and am the guy dancing around. If you have any questions feel free to post them.

Thanks everyone!

The one on the left or the one on the right? : ]

 

[wabo]

The sensor kit used in the demo cost $3995. That includes 17 wireless sensors and 3 dongles.

This demo is a very basic implementation, it is directly applying the orientation calculated from each sensor to the bone with no client side filtering or noise reduction currently.

Ok, then its not "that cheap" but cheaper than traditional mocap. Would want to see how much noise can it filter, remember 2 hours long Carmack's conference at Quakecon talking about how hard was to filter the noise in the VR gyros and getting a good response.

 

[roadkillgrill]

Is it tough not to make the model jittery (probably due to sensor data and trying to balance between jitter and lag)? I still see some jitter in the 3D figure which I do not see on you, not much, but I am interested in knowing how you would plan to attack it.

Great job, btw!

There are 3 points that the jitter is coming in to play.

First one is the physical placement of sensors, they are not screwed directly to my bones so they will shift around a small amount while moving. This one is unavoidable in real-time without adding some more constraints on the model.

Second part is the simple floor and translation method, it is currently looking at the lowest point to use as a tag point and once another bone becomes the lowest point it shifts the model so that that is the new pivot. There is no interpolation currently being applied to this so it occasionally snaps to the next point or toggles between two close bones. With a bit more time this should be able to be smoothed out. This was made on a short deadline for a conference originally.

Third part is that the sensors have some noise inherent with their parts. While the error is less than a tenth of a degree the error compounds with additional sensors being used. With 4 sensors to track an arm from shoulder to hand it can be off by about a half degree. We are still working on reducing the noise further and is a very active section of our research.


The unrealscript used in the demo is planed to be released, I want to clean it up a bit first before making it public. Maybe add a bit more to it as well.

 

[shaowebb]

I'd use it to help render out animations for stuff with the engine since its a pretty good engine for rendering, but I wouldn't really think of it as a tool thats going to do a whole lot for gameplay. Generally, simplifying motions required to perform actions in games is the best bet so long as you don't go QTE. Could you imagine trying to play a game like DMC for long periods if it required this? Not me. Its why I always felt Kinect should have just been marketed at animators and software engineers and not gamers for stuff.

 

[wabo]

I'd use it to help render out animations for stuff with the engine since its a pretty good engine for rendering, but I wouldn't really think of it as a tool thats going to do a whole lot for gameplay. Generally, simplifying motions required to perform actions in games is the best bet so long as you don't go QTE. Could you imagine trying to play a game like DMC for long periods if it required this? Not me. Its why I always felt Kinect should have just been marketed at animators and software engineers and not gamers for stuff.

MS has won millions with Kinect so, from a business point of view, it was the right choice.

 

[Spat_triate]

Cue hundreds of Gangnam Style videos starring Marcus and his backup dancers the Gears.

 

[Black_Stride]

Its why I always felt Kinect should have just been marketed at animators and software engineers and not gamers for stuff.

Kinect SDK IS marketed to software engineers.....its heavily heavily marketed to software engineers and UI designers.

Kinect for Xbox 360 is for gaming
Kinect SDK for Windows is for everyone else.

 

[majik13]

I think the best implementation of this to work in standard games would to have sort of single or multi handed controller ergonomically tied to the hand. You probably only need a thumbstick and a few buttons, to help solve the walking/traversing issue, and to let you do other game interaction stuff. I wonder how that would work out in practice? May be a little disorienting. you should test that roadkillgrill. Easiest thing to use I would guess could be a wii nunchuck or the psMove nav or whatever.

Also, something like this doesn't need to be made for all or most games, and not everyone wants to play something that requires so much physicality. But it could be really cool for some games and new types of games

 

[Parakeetman]

Damned impressive stuff! Thanks for sharing the vid!

 

[Azure Dream]

or This

At first I thought "LOL", but it grew on me as it went.

And then I imagined several Kinect-style sensors embedded along the sphere, mapping you out all the time in full 360°.

 

[Desty]

Very nice. I'm starting to believe.

 

[red731]

I want that Virtusphere. That would be like a dream come true.
I was so jealous about the guys that could play with that Battlefield 3 VR setup.
This is smaller and we could put it in our homes.

 

[Panajev2001a]

There are 3 points that the jitter is coming in to play.

First one is the physical placement of sensors, they are not screwed directly to my bones so they will shift around a small amount while moving. This one is unavoidable in real-time without adding some more constraints on the model.

Second part is the simple floor and translation method, it is currently looking at the lowest point to use as a tag point and once another bone becomes the lowest point it shifts the model so that that is the new pivot. There is no interpolation currently being applied to this so it occasionally snaps to the next point or toggles between two close bones. With a bit more time this should be able to be smoothed out. This was made on a short deadline for a conference originally.

Third part is that the sensors have some noise inherent with their parts. While the error is less than a tenth of a degree the error compounds with additional sensors being used. With 4 sensors to track an arm from shoulder to hand it can be off by about a half degree. We are still working on reducing the noise further and is a very active section of our research.


The unrealscript used in the demo is planed to be released, I want to clean it up a bit first before making it public. Maybe add a bit more to it as well.

Thank you, very detailed answer .

 

[whitehawk]

Thanks roadkillgrill!

The video looks very exciting. I can't wait until motion and VR are used in a consumer environment. It seems to be like the next big innovation in gaming. It's been done in the 90s, but we're almost at the point where we have the proper technology to actually to it right.

 

[Zaptruder]

All you need is that Occulus and This or This and you will be good to go!

I have a feeling that this kinda tech is going to make more headway once VR starts taking off.

The problem has always been that VR tech hasn't been able to reach the mass market, and as a result, never been able to generate the kind of content that makes it justifiable for people to take it to the next level.

I'm not saying that this kinda stuff is going to become a defacto component of VR - but once the money starts pouring in, a lot of VR tech will move very quickly.

Haptic feedback suits, omni-directional treadmills, exoskeleton arms/legs, even the pictured roller ball... will all increase in quality much faster, will be more desirable, will lower in cost somewhat (but they'll always remain premium items thanks to their bulk and associated entry barriers), and be accessed by a larger community of enthusiasts.

But all that relies largely on the bottom end of the tech been suitable for the greatest number of people.

i.e. Rift+ quality headset @ $500 or below, kinect or leap style motion system to control everything else.

Above that, you can have split fish motion controllers (i.e. something like the razer hydra) for a bit more input flexibility...

and then at the enthusaist level, you can have all those other haptic simulation technologies.

With that been said, the hamster ball isn't without its set of flaws. The curvature of the ball is sufficient for your foot to sense that it's not walking naturally on flat ground. And the weight of the ball means that it has a lot of inertia - you can't stop on a dime; you kinda have to slow down and brake the ball.

Those factors combined means it's a partial immersion solution, rather than a full immersion solution.

Similarly, walking on the spot is a partial immersion solution - cheap and easy, but allows you to use your limbs to operate your character, and allows you to feel some degree of proprioception based haptic feedback (i.e. the sense of movement of your own body).


The end game for mechanically based haptic feedback tech (i.e. not brain interfacing) is an exoskeleton rig - something that puts an exoskeleton in a gimble mount. Even then, you might need to combine it with a haptic feedback suit to get the most out of it.

That tech is almost doable today. Just at a very high cost. We simply don't have the content to justify that for anyone but researchers and the military right now.

 

[Dion Blaster]

Wow this stuff is amazing.

Would this work for non first person perspective games? Like could you run around on that mat or in that ball and a game could work in 3rd person? I'm imagining playing a platformer or something like RE4 with this.