IMACOMPUTA
Member
Laser quest + AI (bosses?), various scenery/backdrops, visual feedback, who knows what else.
-
Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.
positional tracking: What it is, how its done, and why Valves tracker is a revolution
- Thread starter Krejlooc
- Start date
Laser quest + AI (bosses?), various scenery/backdrops, visual feedback, who knows what else.
ChronotriggerJM
Member
Yeah something like this makes total sense for a moving experience, and it really drives the point for how effortless Sony's setup is for a more robust semi seated approach, lights and a camera lol. Spacial VR is certainly going to be pretty awesome to experience though, definitely don't have the space for it currently however, but someday!
BiGBoSSMk23
A company being excited for their new game is a huge slap in the face to all the fans that liked their old games.
Who said anything about grabbing? Just some rudimentary finger tracking shouldn't be that taxing. And yes I've tried VR, but it's not quite there yet.
Thanks krejlooc for taking the time to explain it. When I was younger I tried to imagine how a holodeck would work, and the conclusion I came to is that the computer would be able to trick the mind into thinking that it is moving straight while the body is actually doing circles. It's kind of crazy to start seeing that actually happen. VR is really going to push us to new places, not just virtually, but in our understanding of how the human mind works and how it can be manipulated.
I could see them use UWB or 60Ghz wireless. Obviously it has range problems (at most a few metres) and problems with obstruction.
If this works as well as I picture it will (eventually), you could have a classroom with 20 people in it all walking in different directions and not hitting each other, as long as it's the same computer controlling what they're all seeing.
THAT spins me out,.
onesvenus
Member
Do you have any link to some of these? I worked on a VR research lab for almost 4 years and then (I left 5 years ago) outside-in tracking devices were the most reliable with difference. In fact the best professional solutions sold today are almost all outside-in tracking systems (take a look at the intersense website for example). Of course they also have some practical problems that make these solutions unfeasible for being used at home, the ultrasonic tracking system we used for example was a 2x2 meters sensor grid hanged from the ceiling that tracked a 9x9 meter volume with millimeter precision.
Even being an outside-in supporter I think inside-out solutions are better suited to home spaces I just want to read why Carmack says those are better.
The Hermit
Member
This is dexmo:
The spider-like appendages are break points. They don't pull back on your fingers (that's a big no-no) but rather prevent your fingers from moving forward. By tracking the position of your fingers in 3D space using forward kinematics, you can watch for collision between them and in-game objects, and cause the breaks to kick in when they intersect. By doing this, you can feel objects that don't exist - spheres and walls and such.
The obvious problem at the moment is durability and that touch with this thing is binary - you don't feel texture, merely only blunt "existence" or "non-existence."
But it's an incredibly cool first step. This stuff is cheap, too - they had planned to sell them at $100 for a pair. They went into hibernation a while back to let their concept cook a bit more.
Astral/H3X
Member
It's actually outlined in the OP why they are better. Regardless, let's put it this way: we have a millimeter precision in a 15x15 area, with angle accuracy that is just as good. Ultra sonic would be good for finding the location of an object, but not necessarily the tilt or rotation of that object. This has both, and is super freaking cheap AND usable for home scenarios.
TheCongressman1
Member
Have you ever tried to walk in a straight line with your eyes closed? Even if you are being very careful, over a distance, you will assuredly veer off course into a weird direction, even without you knowing. This is because our visual information is the single most important tool we use to determine our orientation as we move through space. While, true, we have other senses that help us determine where we are (this is called proprioception), our eyes are the primary thing we use to orient our bodies while we walk.
With this in mind, you can take advantage of this and curve your display (have it rotating slightly as you walk) to direct your person into walking in concentric circles. This is called redirected walking, and it actually works:
The people walking in these curved paths would swear they walked in a straight line. The obvious problem with redirected walking is that your computer has to know where you are in relation to walls to help direct you into the room. Hence it had the same problems inside-out tracking used to have - you'd have to meticulously set up your environment so your computer would know the parameters of your room to get you walking in circles.
Valve's positional tracking lighthouse does this for your computer automatically.
Oh dear lord, if this works properly that's going to be huge.
ryanthelion123
Neo Member
Have you ever tried to walk in a straight line with your eyes closed? Even if you are being very careful, over a distance, you will assuredly veer off course into a weird direction, even without you knowing. This is because our visual information is the single most important tool we use to determine our orientation as we move through space. While, true, we have other senses that help us determine where we are (this is called proprioception), our eyes are the primary thing we use to orient our bodies while we walk.
With this in mind, you can take advantage of this and curve your display (have it rotating slightly as you walk) to direct your person into walking in concentric circles. This is called redirected walking, and it actually works:
The people walking in these curved paths would swear they walked in a straight line. The obvious problem with redirected walking is that your computer has to know where you are in relation to walls to help direct you into the room. Hence it had the same problems inside-out tracking used to have - you'd have to meticulously set up your environment so your computer would know the parameters of your room to get you walking in circles.
Valve's positional tracking lighthouse does this for your computer automatically.
Wow, this would be monumental if Valve could do this.
aeroslash
Member
All this talk about "haking" your mind reminded me of airplane simulators. Not the tipical flight simulator in your house but the real big full motion ones used by airlines to train their pilots.
On those, they use gravity to make you feel something entirely different from what you really feel. The simplest example is the simulator pitching up a little when you throttle up. That, which actually pulls you against the seat, paired with your vision of going straight ahead, fucks your mind into thinking you are actually acceleraring in a straight line.
it's really a feeling hard to describe, but very good at tricking your mind.
I'm very excited to see where all this studies takes us on the future. Bright times for VR for sure!
On those, they use gravity to make you feel something entirely different from what you really feel. The simplest example is the simulator pitching up a little when you throttle up. That, which actually pulls you against the seat, paired with your vision of going straight ahead, fucks your mind into thinking you are actually acceleraring in a straight line.
it's really a feeling hard to describe, but very good at tricking your mind.
I'm very excited to see where all this studies takes us on the future. Bright times for VR for sure!
onesvenus
Member
.
The OP is talking about occlusion as the main problem with outside-in tracking but ultrasonic tracking systems don't have that problem.
The tilt and rotation of an object aren't a problem for ultrasonic systems as they are almost always mixed with an inertial system like what valve is doing with its HMD.
Does somewhere talk about resolution or are you guessing about it having mm precision?
Super-accurate positional tracking was never really the problem with swordfighting simulations. The lack of feedback is the biggest issue - when your opponent blocks there's nothing to actually stop your real-world motion, so how can it handle you doing something you shouldn't be allowed to do?
Astral/H3X
Member
Those inertial systems are, more than likely, going to be subject to drift, just as the Wiiremote would be if it did not have the inside out tracking. These aren't using inertial sensors to track angles, they are using cameras point at every direction, comparing the images from every direction.
Also, I imagine those ultrasonic systems are neither cheap nor small, of which (comparably) this is both.
It mentioned the MM accuracy somewhere today, it's in one of the articles. I don't remember which one.
onesvenus
Member
So are you saying ps move has rotational drift for example? the biggest problem is positional drift and its completely gone with an ultrasonic tracker.
That's true. I'm not discussing about if they are more suitable or not, they are not, I'm just questioning the rational behind saying they are better.
Thanks, I'll try to find it.
NumberThree
Member
We don't have actual numbers but they talk about "sub-millimeter precision": http://www.engadget.com/2015/03/04/valve-vr-input/
Katzenjammer
Member
Thanks for the writeup, hadn't read about valve's new tracking system yet. I wonder how the pattern emitted by the base is used by the cameras. Are the cameras only interested in following static points, or is there also information encoded in the pattern? And in that case, how can that pattern be decoded when the projection of the pattern gets distorted by objects on the walls (with the worst example being mirrors)?
DouglasteR
Member
This is a concept Oculus has been high on for a while. I've done work with it, too - it is really cool stuff. This is definitely something big. Here's a cool representation of it:
The path on the left is what the user thinks he is walking. The path on the right is what he actually walks.
Its in the same time frightening and beautifull.
Imagine it rendering the game world in your bedroom/hall.
Its not really up to Valve. This is on the developers. Valve has simply opened up the possibility for it.
Personally, I wouldn't get overexcited about this aside from the geek tech aspect. You wont be doing this redirected walking stuff in your average VR game/experience anytime soon. I think it has more commercial application than consumer application for the time being. Places where you can have a large, open area, with a well design overhead tethering system, and specifically designed software that takes advantage of it.
For gaming applications, what a lot of people here are going to be most interested in, I see Valve's tracking solution mostly as a way for us to get rid of the occlusion problem of external camera tracking. I don't think this necessarily means we're going to be wandering around more, but it helps with seated experiences by allowing people to be facing whatever direction. Now, even the Crescent Bay allows full 360 degree head tracking, and you'd be able to face whatever direction and use a gamepad or whatever, but what it doesn't have the capability for is 360 degree positional tracking of motion controllers. You have implementations like the STEM system, which would offer that capability, but from what I understand, it would not be quite as accurate as the form of positional tracking used with Valve's 'Lighthouse' setup. It also quite expensive at the moment....
All in all, its really cool that Valve has tackled this tracking issue(more or less), but I think its benefits for our normal gaming and viewing experiences will still be limited to seated, or at least stationary experiences. You cant just expect that everybody will have a large enough room, you wont be able to move quickly in-game(walking only) and you are still tethered to the PC so cabling is an obvious concern. It also raises some safety concerns. One journalist hurt himself in one of the 'moving around' Morpheus demos. And that was a controlled environment(though granted, he still said he loved it!).
EventHorizon
Member
That's why I propose the compromise of a hybrid solution. Use a run button for forward movement and tracking for everything else. All the turning, dodging, sidestepping... would be done 1-to-1 by tracking. You wouldn't move very far in the real world and the game could use this redirecting trick to make sure that little movement stayed in the center of your movable area.
If you inadvertently get next to the real world boundary then a wall appears before you in game notifying you that you need to back up a couple of steps. This setup would allow the use of a variable amount of space for the movable area. At the minimum amount you'd constantly see the barrier walls and would have to consciously constrain your movements. At the full 15' x15' area you'd rarely see the barrier.
I think you'd very quickly start to run into walls, unless its a very relaxed and leisurely experience that allows you to constantly remind yourself of reality. If you're engaged in the world, and honestly, in VR, you often will be(especially as it gets better and better), you're going to very quickly lose track of reality. Even if all you're doing is side stepping or dodging or turning, you'll end up moving around quite a bit. I think even in an open 15x15 space, which I would suspect only a small minority of people will actually have(especially to play a game in), you'll be hitting the boundaries quite a bit. We can easily cover a 2ft distance in a single step, and that's without being hyperactive due to adrenaline pumping or whatever! For me, in about a 5x9 useable area, it would be completely unfeasible. And I suspect many people's PC setups are not in big, open spaces. This is all speaking nothing of the cabling problems.
It would be cool to see stuff like this implemented as perhaps alternative controls, for those who do have the space, or maybe for things like a VR arcade or something, but I think most consumers will be better served by developing regular old seated or mostly stationary experiences for primary controls.
Great write-up! I knew about most of the stuff, just not that Valve was already that far along with it. Amazing. Potential game changer. Not just for VR either.
Like Seanspeed I can't see people walking around their living room, I'd be too afraid for damage or injury no matter how safe people say it is, but it mitigates a number of problems with using it in real life settings, and perhaps above all means people don't have to install cameras in their living room.
Like Seanspeed I can't see people walking around their living room, I'd be too afraid for damage or injury no matter how safe people say it is, but it mitigates a number of problems with using it in real life settings, and perhaps above all means people don't have to install cameras in their living room.
does inside out tracking allow for absolute positioning information, or just accurate movement tracking? i.e does it know that controller A is 2ft to the left and 4inches above controller B, or does it just know that controller A moved 2 inches forward?
Just that one of the impressions was that the positional tracking of the HTC controllers wasn't perfect when trying to pick something up.
Just that one of the impressions was that the positional tracking of the HTC controllers wasn't perfect when trying to pick something up.
I read about fear of running into walls, but also what about furniture? A 2ft high glass coffee table in the area is an instant 'no way in hell should you try this' dealbreaker alone. Even a wooden coffee table would be bad enough(can a system warn you of knee-high objects at all?). Plus couches, endtables, lamps, entertainment centers(wanna run into your $1000 TV?), desks, beds, stairs, blah blah blah.
I think the question quickly goes from 'can you do it?' to 'should you do it?'. Especially for developers.
Its positional tracking in the literal sense, so it will pick up the positions regardless of movement.
the pictures of the basestation had 'class 1 laser' notices on them. So if they're shining laser dots on the walls, won't they be pretty much the same size? They won't vary much in a 15ft space.
Fine Ham Abounds
Member
@redirected walking, cool in concept. But you're not going to be able to get it working well in many usage cases. Even though I have a 2m diameter clear space in the space I operate my computer in... I doubt I can get it working sufficiently well in this area.
An alternative is required.
But the tracking tech does provide the ability to move around space in a 1:1 manner that is highly awesome - using the controllers to move you around in large spaces, then actual movement to explore your space. The grid should light up as you approach physical elements in your physical space.
It's a pretty decent compromise and allows for a heightened degree of immersion for more users in more scenarios than could otherwise be accomplished.
An alternative is required.
But the tracking tech does provide the ability to move around space in a 1:1 manner that is highly awesome - using the controllers to move you around in large spaces, then actual movement to explore your space. The grid should light up as you approach physical elements in your physical space.
It's a pretty decent compromise and allows for a heightened degree of immersion for more users in more scenarios than could otherwise be accomplished.
The most important news to come out of GDC is this invention.
The idea is so simple and elegant it makes you scratch your head wondering why it wasn't done before. Between the Wiimote, PSMove & Kinect you would think someone would stumble over this combo earlier. Or maybe it was just tech holding us back from realizing the concept earlier?
The idea is so simple and elegant it makes you scratch your head wondering why it wasn't done before. Between the Wiimote, PSMove & Kinect you would think someone would stumble over this combo earlier. Or maybe it was just tech holding us back from realizing the concept earlier?
Awesome write-up, cheers for that.
There's just one thing that you could change in your first paragraph or so to make it more accurate. An IMU != an accelerometer. It's a device that contains several sensors to determine linear and rotational accelerations. This means it needs at least one 3 degrees of freedom (DOF) accelerometer and one 3DOF gyro. Some IMUs also have a 3DOF magnetometer as well (which gives you your absolute orientation - which sounds awesome on paper but it doesn't work well in buildings because of all the steel in the walls and stuff). Higher end IMUs tend to have several accelerometers and several gyros to get more accurate measurements.
There's just one thing that you could change in your first paragraph or so to make it more accurate. An IMU != an accelerometer. It's a device that contains several sensors to determine linear and rotational accelerations. This means it needs at least one 3 degrees of freedom (DOF) accelerometer and one 3DOF gyro. Some IMUs also have a 3DOF magnetometer as well (which gives you your absolute orientation - which sounds awesome on paper but it doesn't work well in buildings because of all the steel in the walls and stuff). Higher end IMUs tend to have several accelerometers and several gyros to get more accurate measurements.
In regards to the effect of tricking your brain into thinking you're walking in a straight line when in actual fact you are walking in a curve, I wonder what other clever brain hacks could be used.
The rubber hand illusion shows how the brain can quickly be fooled into thinking that a fake rubber hand is actually your own. In VR it might be possible to induce the same effect as the video below by using vibration in the controller rather than a brush tickling your skin. Maybe at the start of a game, designers could have a virtual cell phone vibrate along wth the physical controller you are holding to induce the effect. Although I suppose the rumble of a machine gun would do the same thing.
Then when you see other objects like foliage brushing your virtual arm, it might feel real to you.
http://youtu.be/sxwn1w7MJvk
The rubber hand illusion shows how the brain can quickly be fooled into thinking that a fake rubber hand is actually your own. In VR it might be possible to induce the same effect as the video below by using vibration in the controller rather than a brush tickling your skin. Maybe at the start of a game, designers could have a virtual cell phone vibrate along wth the physical controller you are holding to induce the effect. Although I suppose the rumble of a machine gun would do the same thing.
Then when you see other objects like foliage brushing your virtual arm, it might feel real to you.
http://youtu.be/sxwn1w7MJvk
Shuggananas
Banned
Sounds cool but I'd need a new place for this, I don't really have much space for that haha
EventHorizon
Member
The only movement you'd be doing the vast majority of the time would be turning in place. Remember there is a run button. You are never going to be actually running. You could run all the way across the entire map of Dragon Age: Inquisition and not move an inch.
The most movement you'd do would be to sidestep or dodge. That's a one time act. You are not going to be side stepping in a straight line for 5 ft. and in all likelihood if you sidestepped one way to avoid something, you'll sidestep the other way the next time you need to avoid that thing. In fact game AI could be turned to make you do exactly that.
In addition, the length of the sidestep in the virtual environment can be expanded vs the real environment. In the real world you'd move 2 ft to the right but in the virtual world you move 2 ft 5. This slight expansion of distance makes you even less likely to step out of bounds. The same can be done with any turns you make expanding or shrinking the amount to make sure you are oriented so that you always have some space to your sides.
Is it just a one-time act, though? I mean, if you only ever do it once in a play session, its a one-time act. If its a regular mechanic, then you'll likely be doing it many times over the course of your session. There's no reason to think that it will necessarily balance out. It will all be dependent on the situation. Maybe you side step 6 times to the right and only 2 times to the left over the course of 10 minutes. That still leaves you 8ft to the right of where you were, which, if you started from the center, means you'd have reached the boundary at some point. Stepping forward, backwards, even turning involves a step and a bit of movement often enough, its all going to lead you all over the place as you play. Not immediately, but soon enough. And this all assumes you've got a full 15x15 area to work with, which most people probably wont.
And what happens when you do get to the boundary? You take a step away, but you're still only one step away from that boundary again. So you take multiple steps away. Well, by the same nature that allows for redirectional walking, are you really going straight the other way? Or are you also moving slightly forwards/backwards while doing it, in a slightly horizontal(seen from above) direction?
Your solution sounds interesting and I'd love to try it. I'm just not convinced it will be all that viable as a sort of standard movement scheme for people's home VR solutions. Especially with PC's still commonly the domain of bedrooms and offices and not the larger rooms of the house.
Not when it's being projected by a laser, it won't be.
(EDIT: although hitting a surface at an oblique angle will make a difference)
EventHorizon
Member
That is where the redirected movement comes into play. Just because you turned 45 degrees or move 2 ft in the game doesn't mean that's how far you really moved in real life. The game can stretch or shorten those movements a little bit to change your position and orientation. You'll never be going in a straight line. You'll be circling around the center even though you won't realize it.
Redirection only works if you don't have a frame of reference. If you can see the barrier then that gives you that reference so you can walk directly away from it.
Hey I'd love to try it too. I've never been in VR so all of this is theory to me. I'm just going off of what others have said is possible and putting it all together. For example, one test I'd love to try is to see how well such a system would work fighting a huge ogre swinging a club at you. You'd have to sidestep his vertical strikes and duck his horizontal ones while trying to strafe around him to strike with your sword. I wonder just how much space that would take.
Another piece of the puzzle is complete. Hopefully after reading the OP people can understand what the Wii brought and how that tech has evolved to what Valve has today. Motion control is finally 1:1.
This is what Tactical Haptics reactive grip was trying to do. Their KS failed, but they are still working on it.
https://www.youtube.com/watch?v=XBB_OFMJ-Go
Haven't heard anything about them from GDC though.
This is what Tactical Haptics reactive grip was trying to do. Their KS failed, but they are still working on it.
https://www.youtube.com/watch?v=XBB_OFMJ-Go
Haven't heard anything about them from GDC though.
Yes. This happened to me with DK1 and Lunar Flight. Seeing my body in the screen totally hijacked my perception of where my body was. Surreal and Im pretty sure it has medical implications to be developed in the future.
That's a really cool controller. But it doesn't have much to do with the rubber hand illusion.
SquiddyCracker
Banned
What if the lasers hits your eyes?
They're class 1, so safe.