Meant to be Seen

Krenzo wrote:For a consumer version, I would try to target the price range of the Rift.

Ok, thanks for the info.

Krenzo wrote:I can't imagine paying for a positional tracking system if you're going to be sitting down. What would that really add to the experience, maybe the ability to lean? Even if you were standing up and moving, you're going to trip over any cords tethering you to a PC.

Yeah, leaning is basically what's missing in the current devkit, and what a lot of people are asking for to limit sickness that some have experienced.

I think most people will be using the Rift in a seated or standing position. In fact that's the only situations that have been shown in all the Rift demos, apart that last video about Immersive Journalism using the PhaseSpace tracking system.

Fredz wrote:Yeah, leaning is basically what's missing in the current devkit, and what a lot of people are asking for to limit sickness that some have experienced.

I feel like the relationship between nausea and positional tracking is a bit over-exaggerated. It was easy for me and many others to become accustomed to positional disparities. Other effects such as violent or uncontrolled camera swings seem much more problematic to me. A lack of positional tracking bothers me more as a missing feature in certain situations. I was recently in a situation (that I am not allowed to describe) where I had the most intense urge to lean forward and it was really frustrating and immersion-breaking that I could not.

PS. Hey Fredz, I just noticed that you and I are neck-and-neck with our post counts.

brantlew wrote:I feel like the relationship between nausea and positional tracking is a bit over-exaggerated. It was easy for me and many others to become accustomed to positional disparities.

Ok, good to know.

brantlew wrote:A lack of positional tracking bothers me more as a missing feature in certain situations. I was recently in a situation (that I am not allowed to describe) where I had the most intense urge to lean forward and it was really frustrating and immersion-breaking that I could not.

I guess there can be quite a lot of situations like this (undisclosed) one. In real life using motion parallax by moving your head is often helpful, I guess it should be the same in VR. Like when you are on a cliff and you look down (Virtual Pit), you sneak around a corner (any military FPS game), in a cockpit situation (Hawken), or when you dodge something, often by reflex (fire balls in Doom 3). I'm not sure a head-neck model would be enough to handle this.

Also motion parallax is an important depth cue, it's more effective than stereopsis for scenes farther than 10m, so it should add to the immersiveness. I've posted something about this here : http://www.mtbs3d.com/phpbb/viewtopic.p ... 686#p71686

brantlew wrote:(that I am not allowed to describe)

Setec Astronomy !

brantlew wrote:PS. Hey Fredz, I just noticed that you and I are neck-and-neck with our post counts.

Heh nice. We are very talkative aren't we?

@Fredz

I think the simplest solution to allow for leaning while seated is an additional 3DOF (or "9DOF") tracker strapped to your chest. Of course, it would also work while standing - it just wouldn't discern translations.

I found this pdf while looking for RF positioning systems.
http://frequenz.schiele-schoen.de/108/1 ... 4_GHz.html
It goes into detail on formulas for position, filtering interference, antenna design... They are designing it for AR use, and claim to have sub mm accuracy.

edit

Looks interesting: http://www.kickstarter.com/projects/108 ... f=category

Might be a good prototyping tool?

...

Krenzo,

I was wondering that given the the level of spacial accuracy you plan to achieve, if it would be possible to use an antenna array and calculate orientation based on the phase difference between them? Initially I was thinking of one plane (yaw) and using accelerometers to determine pitch and roll, but perhaps a unidirectional antenna can be placed on each face of a "cube" so orientation can be deduced in each axis? Also, how do you determine the "Z" axis? Would you place transmitters on different planes perhaps?

Since it doesn't seem that you're using gain levels for your calculations, omni-directional antenna non-linearities shouldn't pose any problems, right?

...

...

That's super awesome Krenzo!! I'm so glad you're finally seeing some tangible results from this project - and it's refreshing to finally get back to some pure VR research talk around here instead of constant Rift-centric chatter. The video looks great. Can you give us a few details like the tracking frequency, accuracy, and range you are able to achieve so far? Also, are you doing any kind of inertial/RF fusion or are those data streams totally separate?

...

Looks amazing!!!!

Krenzo wrote:Thanks, I agree. We need more research discussion.

It's pretty accurate. It's less than a tenth of an inch in accuracy from eye balling the data. If you look at the console output, you can get an idea of the noise of the system. It's not calibrated though so it's not 1 to 1 precision yet. The resolution will only get better as I add more transmitters. The system is actually only using data from 3 of the transmitters right now, and combining data from the fourth will also increase precision. It's not optimized so the actual update rate is only around 20 Hz. I am aiming for 60 Hz in the final system.

I've still only tested the range out to about 10-15 feet. I need to add more transmitters to fully cover my garage and test the coverage.

There's no sensor fusion right now. That will help in the final system. My IMU is really neglected as I didn't even have the rotation data filtered for the video which might be noticeable if you look for it. I had another IMU, the ArduIMU from Sparkfun, in there when I went to record the demo and found it spitting out garbage data. I had to spend a few days throwing together my own IMU to be able to make the recording.

Cool. Correct me if I'm wrong but since time of flight is negligible here, you can retain that accuracy at pretty much any practical distance right? So your range is only limited by your transmit power and you could achieve that 1 or 2 millimeter accuracy at 10 feet or 100 feet or 1000 feet? If so this would be incredible for something like VR laser tag type installments. It has similar properties to optical installments but wouldn't require dozens of cameras for coverage and would also allow the use of props and obstacles without occlusion problems. The implications are amazing!

...

Ahhh. I misunderstood. I thought the antenna unit you showed was only 1 of the 4 transmitters - but that is all four. Good idea about a set configuration to eliminate calibration. But in theory you could provide different configurations optimized for different ranges - or in an advanced scenario a variable antenna that would require calibration.

And then of course you have the FCC issue. In an indoor scenario can you dampen the radio leakage to acceptable regulations and still achieve 100 or 200 feet range?

The quality of your work looks excellent, Krenzo! Out of curiosity, what's holding the system back from being able to update substantially faster than 60Hz? Is it processing power?

I do have some concerns regarding the use of magnetometers for yaw orientation. Some people on the Oculus forums have mentioned severe magnetic field distortions in their apartment. I suppose you could map out the user space, but it would be easy for people to do improperly. Any thoughts on this?

...

Fantastic and exciting work!

You mentioned price around a OR kit.

What's the breakdown on that?
- each transmitter
- each receiver antenna with amplifier
- digitizer
- fpga board
- rotation tracker part

Because I'd be most excited about following crazyhead setup:
4 transmitters
12 receivers with 12 amps
1 multi digitizer for 12 receivers
1 multi fpga board
0 rotation trackers

Receivers on each wrist, ellbow, shoulder, hip, knees, ankles. cables running to digitizer and fpga strapped to the back.
Would make a precise, low latency, orientation and occlusion resistant rt mocap system. 2 rotation trackers on wrists.

Freestanding
Combined with the omni.
Or hanging/halfsitting in a climbing harness controlling aliens/avatar style bots or turrets
with wrist, elbow and shoulder receivers and aim calibration any plastic toy gun could be used without tracking needed built into it....

a lot of receivers, but depending on cost breakdown not THAT expensive for nearly endless possibilities.
If you go KS such a tier might be very popular

Krenzo wrote:I would severely doubt 100 ft is possible but maybe with a different antenna.

What are the limiting factors?

Krenzo wrote:I've been aiming to keep costs as low as possible, but reading replies to my video, it looks like people would only be happy if it were dirt cheap! I'm starting to doubt whether I should continue to aim for a consumer product when I could charge ten times as much and target the professional motion capture market.

Your big competitor obviously is optical. For a few tracked points RF seems to win out on cost and complexity, but for lots of points (full body and facial tracking) reflective optical seems like a win. If your range really is very limited then it seems like your main advantage is occlusion - but only for a relatively small number of points.

Wow! Bravo, Krenzo! You still have a lot of work to do on this thing (aggressive miniaturization, FCC testing, 1:1 configuration, finding a level of smoothing that doesn't kill the responsiveness, etc.), but what I'm looking at in that video is the best homebrew positional tracking solution I've ever seen. I may proceed with my IR optical experiments for seated use regardless, but it's pretty clear what you've got here is going to be better in every way for nearly every usage case. I have a number of gameplay ideas that are utterly dependent on low-latency positional tracking absent the kinds of occlusion problems that hamper optical solutions, so consider me an enthusiastic supporter. I'm not sure how I might be able to help but if you're pursuing a patent on this I've been mired in that process for weeks and would be happy to share my experiences so far.

You're definitely on the right track here, keep it up!

Krenzo wrote:I would severely doubt 100 ft is possible but maybe with a different antenna.

Thanks, MSat. That level of quality is absolutely required when working at GHz frequencies. Physics doesn't play around! The limit to update rate is cost. You could achieve 1000 Hz if cost was no factor. I've been aiming to keep costs as low as possible, but reading replies to my video, it looks like people would only be happy if it were dirt cheap! I'm starting to doubt whether I should continue to aim for a consumer product when I could charge ten times as much and target the professional motion capture market.

I don't think mapping out the user space would be hard to do, but you can never be 100% fool proof.

Wouldn't a lower frequency give you better range for a given power output, and also reduce attenuation from obstructions? Is it a matter of trying to keep antenna size down?

There will always be people who complain something is too expensive. Reading comments in a Kotaku article regarding the Rift, there were some people that said they wouldn't buy it unless it was less than $100. Go figure. Yet cellphones routinely cost $500-700 with a mark-up (at least strictly over the cost of the hardware) of more than 100%. There's no reason why you couldn't offer both commercial and consumer versions. Maybe going with a commercial version first might be a good idea as you can have a substantial mark-up and have less customers to deal with. Then as the necessary hardware gets better and cheaper, a consumer version might be more feasible. This is great tech though, and I would love to eventually see it in the consumer market.

So it looks like your receiver consists of an antenna->amplifier->FPGA (which does a lot of heavy lifting). As I understand it, one of the biggest reasons you need an FPGA capable of working at such high frequencies is primarily for high-resolution time stamping of the incoming (unmodulated?) RF signal, correct? I'm not at all familiar with FPGA market and associated costs, so I might be making a moot point, but is it possible to go with a cheaper high-performance device that has less gates and offload some of the functions to specialized components on the market?

...

Kenzo, I'm throwing money at the screen and nothing happens. How come?

I just read through the thread again as I apparently forgot your earlier explanations. Yet the more I read it, the less I understand and realize just how ridiculously impressive this is. You are doing magic on the FPGA! In order to prevent people from knowing just how dumb I really am, I'll keep my mouth shut from now on except to cheer you on

If transmit power is the limiting factor, then in fact it is an artificial (or regulatory) limit, true? Is it possible to transmit at high power indoors and dampen the signal leakage to acceptable legal limits?

Krenzo for the love of mercy please don't listen to folks on youtube in my experience people there are usually too negative , at the very least please consider a small production run of developer kits.

I'm a developer on a tight budget and even I'd happily scrape together $400-$500 for a dev kit to use in conjunction with my rift.Compared to the cheapest mocap rig ( around $6K-$7K as you know) this is already dirt cheap , sure people will have to deal with a few cables here and there but that's what 1st gen is all about right ? A bit clunky but functional.

Hah, yeah, I don't plan on listening to those people. I will probably put it up on Kickstarter as soon as I can, but right now, I'm having to focus on job searching which is distracting me from finishing the system.

Forgive my presumptuousness and intrusion Krenzo, but apply here! http://www.oculusvr.com/careers/

What you are doing is amazing, but you know that, wish I had the background to understand the details of it, but I am excited none the less of what kind of possibilities you are opening up for the future of VR.

Aabel wrote:Forgive my presumptuousness and intrusion Krenzo, but apply here! http://www.oculusvr.com/careers/

I will just say "no comment" on that.

Aabel wrote: What you are doing is amazing, but you know that, wish I had the background to understand the details of it, but I am excited none the less of what kind of possibilities you are opening up for the future of VR.

Thanks

SOLD - You've got my attention.... I'd buy a kit straight up.
I'm sure there are loads of forum members reading over this thread who'd scoop one up without even blinking an eye...

Place a survey & actually get a feel for how many members would be interested...

If you do a small production run... I'm definitely in

Awesome idea! I would however not buy any tracking system now that was not officially endorsed by the Oculus VR team.

Would be nice to get the best technology but I would compromise for something that will be supported in the SDK for Oculus Rift.


@Krenzo, I posted this on the Oculus VR development forum in their existing discussion forum about this concept, hope that you will post there too:

https://developer.oculusvr.com/forums/v ... f=25&t=787


This 'GPS-like' technology concept really looks to be the most promising Position Tracking System for VR that I read about so far!

Achieving unobstructed line of sight non-magnetic position tracking support should be a high priority for Oculus VR?

Assume you could have multiple Oculus Rift users each with own RF receivers using the same RF senders in one room?


Does anyone know if this guy have even been in contact with the Oculus VR team about collaboration around this idea?

Perhaps Oculus VR can employ him, buy the whole project, or at least license the concept for it if he own the patents?


I only wonder how cheaply the parts for this could be built or sourced if it was mass produced by the Oculus VR team?

If the first-generation consumer Oculus Rift will retail for $299 US-dollar when released, then I think that a Position Tracking System sold as an official Addon Peripheral for it could not go for more than $199 US-dollar at retail, and preferably less, more like $99 US-dollar.

Maybe more business-like would be to sell the Oculus Rift without a Position Tracking System for $299(US), then sell Oculus Rift bundled with a Position Tracking System for $399(US), and sell the Position Tracking System separately as an Addon Peripheral for $199(US).

That is kind-of how Microsoft initially sold the Kinect for Xbox 360 game console, as you can not expect people to pay more for controller addons than for the main product.

Harley C wrote: Does anyone know if this guy have even been in contact with the Oculus VR team about collaboration around this idea?

Perhaps Oculus VR can employ him, buy the whole project, or at least license the concept for it if he own the patents?

I'd suggest (if you read between the lines) his response in the post above http://www.mtbs3d.com/phpBB/viewtopic.p ... 40#p123931 answers that question.

V8Griff wrote:

Harley C wrote: Does anyone know if this guy have even been in contact with the Oculus VR team about collaboration around this idea?

Perhaps Oculus VR can employ him, buy the whole project, or at least license the concept for it if he own the patents?

I'd suggest (if you read between the lines) his response in the post above http://www.mtbs3d.com/phpBB/viewtopic.p ... 40#p123931 answers that question.

I read between the lines but unfortunately a comment like that can be interpreted in so many different ways.

Just a few examples that I could think of right away:

1. No I want to first do more research to see if I can do this on my own before I think of selling it to Oculus VR.

2. No I am in contact with a other company or individuals about collaboration or employment and can not talk about it yet.

3. Yes I am in already contact with Oculus VR team but signed an NDA so are not allowed talk about it right now.

4. Yes I am in already contact with Oculus VR team and talks are progressing but do not want to talk about it right now.

5. Yes I have been in contact with Oculus VR team but it did not go well so do not want to talk about it right now.

Just saw the video and I must say great work! This looks very promising and I hope you can take it places In terms of ignoring the consumer market, why not do both professional and consumer? Obviously you'd have to choose one to start, but if it is a matter if better or more expensive hardware for the pro market but essentially the same software and platform otherwise but slower update rate for consumers, why not? I am a dreamer

This is looking fantastic. It's starting to look like all of the pieces required for a feasible holodeck are almost within our grasp. I don't know that this will be as useful as a consumer product as it would for a commercial venture though. The problem is one of space, and I don't see a huge demand when you need a reasonably large area set aside with no obsticles to get much use out of this.
As a commercial product, I think the combination of this with rift and individual tracking for each hand and each foot within a reasonably large hall (you would probably have to stagger transmitters within the space since range will still be an issue) would provide almost limitless experiences.
I look forward to seeing where this goes.

BRB, going to Seattle

Great work Krenzo!

I'm working on putting together a VR arcade franchise kit (subspacevr.com), which is similar to your original vision of VR laser tag. Your tracking technology would allow me to expand the play space and to add additional tracked objects.

I'm interested in purchasing a system, even if it is still in the prototyping phase. Send me a PM with a price.