Meant to be Seen

niall wrote:

Trilion99 wrote:We recently came up with a demo for this: http://www.lp-research.com/lpms-head-tracking-demo/
The 3D graphics is a bit old-schoolish (+ a few bugs), though. But the main thing is that latency is low and here is no drift (thanks to the magnetometer). Anyway we combined the system with an optical tracker as well: http://www.lp-research.com/simultaneous ... -tracking/

Could make it really cheap, too. So it might be interesting for low cost VR.

Putting those onto gloves like these:

http://benkrasnow.blogspot.com.au/2010/ ... s-for.html

and getting relative positioning going between the HMD and the gloves (and relative finger positions, and pressure sensitve finger/palm sensors down the road)...... I'd buy it!

What really excites me about all of this is that we're getting closer to the point where we'll be able to to "see" our hands through the HMD - with relatively accurate position, orientation, fingers etc. Any kind of gun or device in the game could be overlayed on that virtual hand, using your finger sensors for control e.g. index finger trigger movement fires, thumbs up does something else, ... think multi-touch but in 3 dimensions so 1 and 2 hand gestures, say send a Street Figher II fireball with the appropriate wrists together gesture.

Forget keyboards, mice, joypads, guns... we've got to head in the direction that Apple is going and just use what we have - our hands and body. Maybe Kinect will get there one day, but you'd want to be surrounded by sensors in a cave like environment. If we can just put on a pair of gloves, which talk back to the base station/HMD via Bluetooth, then it's just a matter of time for the HMD to go wireless also and we can use these tools in any environment outdoors or indoors, alone or in groups, sitting or standing, blah blah blah

So what you do is mount depth cameras onto the Oculus Rift, scanning the projection area of the display, like so:

http://www.technologyreview.com/news/42 ... he-future/

and then when your hands are in view we can render a hand, a gun, etc. and trigger actions by hand and finger gestures *pew pew*

Then add positional and orientation tracking gloves for when your hands are not 'visible' in the virtual world. Down the road full sensor gloves could replace the depth cameras, giving you gestures regardless of where your hands are.

Is it just me, or is this not the long-term direction that most people are dreaming of here...?

Oculus got a new recruit?

http://kotaku.com/5948741/cliff-bleszin ... epic-games

seems cliff blesinski an avid fan of the rift has left epic, quite surprising imo

urban wrote:Oculus got a new recruit?

http://kotaku.com/5948741/cliff-bleszin ... epic-games

seems cliff blesinski an avid fan of the rift has left epic, quite surprising imo

I thought about that too!

In the last Bonus Round, he did say: "Mark my words: VR is coming back".

He posted on NeoGAF saying he had no plans for now besides going on vacation with his wife... Which of course, is probably a half-truth.

The guy is a game designer, what would he do with Oculus though?

My guess is that he'll start his own studio, which of course will focus on VR. Believe!

http://www.youtube.com/watch?feature=pl ... STge5IDxF4#!

an awesome Oculus accessory!

Damo3D wrote:http://www.youtube.com/watch?feature=pl ... STge5IDxF4#!

an awesome Oculus accessory!

Exactly!

But losing the need for a controller should be the final piece of the puzzle - full finger movement, orientation and two-hand gestures can then take the place of any controller and we can play any style of game with any style of UI with our fingers, hands and arms representing in-game objects like weapons, picked up objects that can be manipulated (being able to use a Rubik's Cube in the game would be a nice test case for sensor sensitivity and intuitiveness), do any style of one or two-handed gesture to represent in-game commands and actions etc.

Have to keep an eye on this, after the HMD and the first round of game content kickstarts this industry, it'll all get down to how we interact and control objects within the game.

Personally, I want to use this to work in not just for games - typing on a virtual keyboard, swiping between virtual desktops with my hands, looking around a spherical workspace around me filled with GUI applications, terminals and browser windows etc. 1080p and eventually 4K would be needed for this, but the ability to focus like this and not let a noisy work environment disrupt your productivity with endless distractions... that's what really excites me, turning work into a game and say lying back in a nice leather recliner, jacked in...

Unfortunately, you're wrong. Even if they can get that working perfectly, it can't do the things you want to do. Because there's no haptic feedback. You can't replace a controller with hand and finger tracking and expect people to still be able to do controller-like things, because those rely on haptic feedback (and on precision). With that system, people would still want a physical gun prop on their body somewhere that they can draw and use if they want to shoot things.

What about two Novint Falcons? I have one and love it.

Well, there is the Novint XIO (http://www.novint.com/index.php/novintxio), but it's been listed as "coming soon" for two years or so by now. I'm guessing that they have moved away from the consumer space for now and are concentrating more on the commercial / military market. Perhaps something like the Oculus can bring them back though.

I've seen other companies do similar devices (but cheaper looking) for the entertainment industry. I'm not sure how feasible it would be for someone to make something like that at home. And as long as the sensors don't rely on optical line of sight it would be possible for someone to hold a plastic toy gun at the same time and use the real object for haptic feedback.

Sidetracking a bit I can mention that the 2006 SF book "Rainbows End" by Vernor Vinge has quite a few interesting VR/AR ideas that I've though about lately. In that world haptic feedback is provided by machines that basically move themselves to be where you are touching an object. Eg a wall may have an AR surface, and it would then have parts that move to allow you to touch it. (The book is set 10-15 years into the future.)

Carmack mentioned this in his Quakecon talk, and he raised the very practical issue of the dangers of this if the devices are not precisely controlled. (We are only now having industrial robots which are designed to work together with humans and not closed off in their own security zone. It's not completely unlikely that the tech can be possible in another 10 or so years time.)

2EyeGuy wrote:Unfortunately, you're wrong. Even if they can get that working perfectly, it can't do the things you want to do. Because there's no haptic feedback. You can't replace a controller with hand and finger tracking and expect people to still be able to do controller-like things, because those rely on haptic feedback (and on precision). With that system, people would still want a physical gun prop on their body somewhere that they can draw and use if they want to shoot things.

So you add haptic feedback, and let precision evolve as this technology gets better, faster and cheaper..?

Sorry it just doesn't seem like that big of a hurdle to me, certainly not insurmountable or so challenging that this direction shouldn't be pursued. Lots of things to do first, but that's where VR will inevitably head - using our bodies as input.

I agree with niall, haptic feedback will obviously improve the concept, but it needs to start somewhere, and that will be tracking our hands and/or body without any feedback. That real-time Mocap video looks awesome. I have been waiting a long time to see this sort of thing become a reality.

Damo3D wrote:http://www.youtube.com/watch?feature=pl ... STge5IDxF4#!

an awesome Oculus accessory!

It's easy to see 10 years into the future, kids running around in a park, playing Quake outdoors.

Oculus Rift with a tiny 3d cam in front (and a battery)
Real time tracking of their body and motion with wearable sensors,
A gun in their hand (trigger, buttons, aiming),
One stationary pc in the middle of the park to gather all this data wirelessly and send it,
Steam Cloud to compute how everything will look (the rocket launcher in the kids hand, his armor with damage on it, the explosion of the rocket) and send it to the stationary pc, rendered on top of the cam feed with some post processing.
(again, the pc in the middle that sends every feed to every corresponding HMD)

I know there are some things in AR that still need some figuring out but I'm sure that will happen. I'll be sitting on a banch, eating some special brownies, watching my kid blow some other kid up with a grenade.

Google will have their own thing of course. Wearing those glasses and saying "google, I'll walk to my crappy job, I want to see some female Na'vis on the sidewalk, throw in a few Autobots on the road and a few Nazgûls with those flying beasts in the air."

Next:
[youtube]http://www.youtube.com/watch?v=q54rFJ0vWQw[/youtube]

donkaradiablo wrote:...
It's easy to see 10 years into the future, kids running around in a park, playing Quake outdoors.
...

I think you're partially right, but it won't be out doors. It will be indoors with special skate shoes that simulate the walking and running, but keep you in the same spot. Your entire body will be tracked with a very high speed kinect type setup.

Future parent:
"OMG, your brain is going to rot if you sit and read those books all day. Go and play video games with your friends!"

I think you're partially right, but it won't be out doors. It will be indoors with special skate shoes that simulate the walking and running, but keep you in the same spot. Your entire body will be tracked with a very high speed kinect type setup.

I think both setups will be reality. Probably an indoors laser tag type one first IMHO, probably easier to solve the tracking issue than the walking issue.

That said, the tracking issue is the biggie currently.

android78 wrote:

donkaradiablo wrote:...
It's easy to see 10 years into the future, kids running around in a park, playing Quake outdoors.
...

I think you're partially right, but it won't be out doors. It will be indoors with special skate shoes that simulate the walking and running, but keep you in the same spot. Your entire body will be tracked with a very high speed kinect type setup.

I said 10 years not 5

Everyone get the KickStarter E-mail for shipping Address..Feels like things are moving along..come on November/December..Get here already...

3dvison wrote:Everyone get the KickStarter E-mail for shipping Address..Feels like things are moving along..come on November/December..Get here already...

I received my shipping confirmation email - giddy up November!

Just filled in the kickstarter shipping info. Can't wait!

What's taking so long Oculus?

I've filled my shipping info like an hour ago and I still haven't received my order!

haha yea i was so happy to get that email

I hope we can see some updated concepts or early assemblies before november is up.

I just had an idea to eliminate (or reduce) the issues with regards to the end-to-end latency from the tracking sensors to the image render being displayed. I would think this would have most effect for the turning left and right, and thinking that you could have a pixel shift at the screen end, to compensate for the latency. If the screen used regular analog av signals, this would probably be easier to achieve by creating an artificial h-sync that would move the image a few pixels either way, but there must be a way to do this with digital.
I do wonder if maybe it would be possible to add a post render effect that would get the difference in the head rotation between when the scene was started rendering and when it was complete, and shift it to compensate. I know that it wouldn't be 100% correct since the actual viewpoint wouldn't be adjusted, but I think it would be close enough... or at least something worth considering. Maybe Cyber could comment on if this would be possible to implement?

received the mail, yeah!!
Too bad I will probably see my rift in January because here in Italy it takes a month for a US order to arrive...

@Android

To clarify, your idea is to shift the buffer horizontally immediately after receiving sensor data, but before the current frame has been completely rendered? That sounds like it could be an interesting idea. I don't have the skill set to do implement it quickly, but I would love to see the results.

FingerFlinger wrote:@Android

To clarify, your idea is to shift the buffer horizontally immediately after receiving sensor data, but before the current frame has been completely rendered? That sounds like it could be an interesting idea. I don't have the skill set to do implement it quickly, but I would love to see the results.

What I mean is that, before rendering starts, it will have the head position at that time. The renderer will take some time to actually render the scene, by which time, your head rotation may have changed. To compensate for the change in rotation, you could shift the image post-rendering, by a few pixels, to compensate for the change since the rendering started. I think this could be incorporated in the rift driver that cyber is creating, since that is already altering the post-renderer to re-map the pixels.

@android78: Something like that may be technically possible, but I don't think it would help much. If anything it may look jarring and add nausea.

First snow falling already. An indication to me that Christmas is not too far away.

cybereality wrote:@android78: Something like that may be technically possible, but I don't think it would help much. If anything it may look jarring and add nausea.

Do you think so? I don't imagine that it would change the resultant image very much (assuming the game is rendering at 60fps), it only has to account for the movement in the last 16ms, so the viewpoint shouldn't have changed much. I came up with this idea when thinking about why it's not a problem moving your head while playing a game on the TV, so I imagine it would be no more nauseating then that.
I think the main problem would be determining the 'average distance' of the image, so you know how far to move it. ie. if you have a wall that is rendered close to you, you wouldn't want the picture to move so much for a 1 degree as if the wall was in the distance. But it would be great if the driver could have an option to turn it on and you could have a very basic value for a multiplier (or divider) of the offset from the previous gyro value to the rendered image gyro value.
Given the work, I would think this pretty easy to test out.

Im not sure if dycus and palmer haven't already started looking into this... but the latest generation of MEMs are very exciting.... even if we dont get it in the devkit, i definetly see these maybe making it to consumer (with some absolute tracking) :O. Ofcourse, this depends on just how many hardware/software engineers oculus has or will get

But in particular MEMS that looks promising ("9 in 1" MEMS)
http://www.invensense.com/mems/gyro/mpu9150.html

some highlights:

Reduced settling effects and sensor drift by elimination of board-level cross-axis alignment errors between accelerometer, gyroscope, and compass

-essentially going for 3+ separate chips to one chip! Eliminating more room for error

Embedded algorithms for run-time bias and compass calibration. No user intervention required

-Dynamic calibration, No need for the user to calibrate the kit (or atleast not they don't have to to obtain good results/accuracy)

-Each of the sensor seems to have better resolution, response time, better processing, which is leading to even better accuracy/ less drift.

Another neat aspect - Motion Fusion / Digital Motion Processing engine
http://www.invensense.com/mems/devices.html

Not ground breaking but could be useful

Digital input on FSYNC pin to support video Electronic Image Stabilization and GPS

-not so much the GPS but if oculus ever mounts 1-2 cell phone camera for tracking/looking outside the unit for basic maneuvering.


Didn't mean for a long post put either way, we are starting to get even better acell/gyro/mag. Sensor. This is like 2-3 generations newer MEMS and most importantly is from the same chip manufacture as what hillcrest tracking is using (1-2 of the chips). Exciting time/future!

Namielus wrote:First snow falling already. An indication to me that Christmas is not too far away.

It snowed where I am too today (in Adelaide, Australia, in the middle of Spring):
Once-in-century October snow across SA
That's not normal. It never even snows in Winter here.

But if you meant that metaphorically, I didn't receive any shipping confirmation email. I preordered. I may have to change my shipping details though, since I haven't updated my address on PayPal and it will get sent to my parents' house, and I may or may not be house-sitting there depending on the shipping date.

I meant it literally, tho I did receive the confirmation e-mail just a few days a go.
We get crazy amounts of snow here, btw. So much that I have my own tractor just to handle the snow on my own property.

The confirmation email is for pre-orders only, not the kickstarter ones, yeah?

android78 wrote:The confirmation email is for pre-orders only, not the kickstarter ones, yeah?

I got one for the kickstarter

That 9 axis chip looks really nifty. Only $80 for an eval board. I'm not sure how to read that error/noise spec though, could someone translate that for me? It could be a promising tracker for all kinds of projects!

I got one for kickstarter too. Thanks for the response, Krisper.

android78 wrote:The confirmation email is for pre-orders only, not the kickstarter ones, yeah?

I think it's only for kickstarters, not pre-orders. I didn't get one for my pre-order. Pre-orders aren't supposed to ship until January.

There are three types of orders at this point, right?

1. Pre-Kickstarter "paypal" based orders
2. Kickstarter orders
3. Post-Kickstarter "pre-orders".

It's probably a bit confusing if you're not paying close attention.

mysticeti wrote:There are three types of orders at this point, right?

1. Pre-Kickstarter "paypal" based orders
2. Kickstarter orders
3. Post-Kickstarter "pre-orders".

It's probably a bit confusing if you're not paying close attention.

Right. Although 100 of the Kickstarter orders you have to assemble yourself and they ship in November, instead of December, and they shouldn't be confused with the DIY Rifts that people build themselves from scratch. And 1000 of the Post-Kickstarter "pre-orders" come with Doom 3 BFG, and the rest don't. And some lucky post-kickstarter pre-order customers will get theirs earlier than normal because Palmer is a fan. And some VIPs like John Carmack already got the hand-made prototypes that have a lower FOV and a different tracker, that we see in all the videos.

As we are almost half way through October, I am hoping that Palmer will give a status update on the intital 100 units shipping in November (estimated ship) as well as the December estimated ship date.

space123321 wrote:As we are almost half way through October, I am hoping that Palmer will give a status update on the intital 100 units shipping in November (estimated ship) as well as the December estimated ship date.

Think in another post he or somone else, said somthing about being in Korea. Could that have somthing to do with parts/production ?
In order to produce 1000's of Rift units, to be ship out in November/December, I wonder when production would have to start ? I wonder if it already has ?

I have been wondering "why korea?". I know he was there for the Korean Gamers Conference, but they are focusing a lot on Korea.