Queen’s University
Authors: Samuel Jordan, Michael Greenspan
Abstract: ARPool is an augmented reality system designed to assist shot planning and execution in a game of billiards. Using a projector-camera combination, ARPool is capable of detecting the ID and location of each ball on the table, as well as tracking the position and orientation of the pool cue in real-time. This information is fed through a custom pool physics simulator to obtain a complete table-state timeline of the shot. The shot data is dynamically rendered in real-time directly on the surface of the table using 2D graphics.

A useful, practical, usable (sellable?) tool if you ask me !

I’ll be in NY from 15th to 19th March before going to 3DUI near Boston, and I’d love to meet some VR geeks there !

Also if you have a bed for me, even for one or two nights, I’d happily exchange that with extensive VR discussions

What interesting VR centers are there in the big apple ?

Here’s a game concept by David Arenou, using AR markers to track the player’s movement, but also to register the position of some real life obstacles.

As cool as Time Crisis but with more body immersion!

DIPLOMA : Immersive Rail Shooter from David Arenou on Vimeo.

The basic concept of shuttering eyewear for viewing stereoscopic images isn’t mine.  You can find mentions of it in the literature before my work began.  Missing from the early work were the elegant electro-optical shutters that we now have and a good communications link between the display and the eyewear. (…)

I founded StereoGraphics Corporation in 1980 for the specific purpose of developing stereoscopic technology and offering products to industry. (…)

The initial StereoGraphics products used tethered or wired eyewear for viewing a CRT monitor.  We needed to modify the monitors to run at a high field rate. (…) We wound up making motherboards that were introduced into monitors to goose them into running at 120 fields per second.  We took the sync pulse from the video signal and used that to tell our tethered eyewear when to shutter. (…) The result was a decent stereoscopic image, but the shutters used pretty high voltage (200-300 volts), had low transmission, had parallel electrodes running through them and made an odd buzzing sound that wasn’t exactly comforting close to your eyeballs.

[Insert here *very* technical information about how they created the LCD glass that will allow the light to pass or not]

I knew that the ultimate package would be one that would not involve any cables or wires, or a big controller the size of a hi-fi amplifier.  But how to fit everything into a pair of eyewear?? [Insert very technical information about lowering the power needs of the LCD system] (…)

(…) I became one of the first people to use optical compensation to improve the performance of liquid crystal parts.  This allowed us to leap from a 15:1 dynamic range to one of more than 800:1. (…)

Marv Ackerman designed the infrared link using a pulse width modulation scheme to designate the left from the right fields and to make sure that the shutters were in phase.

The marketing of CrystalEyes was an example of the sale of vaporware. We presold the concept by allowing Silicon Graphics to believe that they invented it. (…) Six months later at SIGGRAPH we showed the first CrystalEyes.   If Jim Clark [one of the founders of Silicon Graphics] had thought about it at all he would have realized that it was impossible to have put anything like CrystalEyes together in six months.

We got them just in time for SIGGRAPH.  I was in my hotel room at the convention assembling them with a little jeweler’s screwdriver, slaving away putting them together.  We got enough of them to work and got orders for hundreds and hundreds of them from Silicon Graphics.  They wanted them in their color with their brand, and we were happy to comply.  In no time at all SGI ate Evan&Sutherland’s lunch with their computer and molecular modeling applications.

We sold over a hundred thousand Crystal Eyes to people in fields like molecular modeling, aerial mapping, oil and gas exploration, and CAD over a period of something less than 20 years.  I believe millions of shuttering eyewear are now going to be sold for 3D TVs.  A great deal of effort has gone into making high-field-rate displays in order to provide sharper images of objects inmotion.  So much time and effort to make LCDs look like CRTs.

That high-field rate technology can be applied to stereoscopic field-sequential television using shuttering eyewear.  It means that the customer can buy after-market shuttering eyewear, which in turn means that the cost of the set can stay low but the stereo function can be enabled by purchasing the eyewear.

I’m happy to be around to see all of this come to pass.  Shuttering eyewear is a good solution for stereoscopic television – providing that people will put up with wearing eyewear at home – because it will become a commodity product, and the end-user at home will be able to buy these eyewear at a low price. And they will get smaller and lighter and eventually be all but indistinguishable from polarizing eyewear.

I didn’t think there was that much research involved. I had imagined that the LCD glass was already existing. So thank you Lenny and your team for giving us my favorite viewing device (after HMDs) !

On a more personnal note, thursday was my last day at Virtools and it was quite moving. But to cheer me up, my colleagues offered me the Fujifilm Finepix Real 3D W1, a stereoscopic camera !! A VR geek come true !! I’ve only played with it for two days but I’m so in love with it.. I’ll try to make a review when I come back from Sri Lanka !

As for me, and as the title of this post suggests, I’m leaving Virtools / Dassault Systèmes.

I won’t discuss  the reasons here. It’s been a tough, personnal and political decision, but sometimes you have to do what you have to do.

I’ll really miss working with David, Geoffrey, the whole Virtools team and some of Dassault Systèmes’ people. I love you, you know who you are! Thanks for the amazing time with you, for all you’ve taught me and all that we shared.

But I won’t miss the changes that DS has forced on us.

Now that I’m regaining freedom, I’ll start by some travelling (is there some VR in Sri Lanka or Costa Rica ??). Oh and I’ve updated my photo gallery.

Then I’ll probably have some time if you need my VR expertise. I’ll remind you of that later but we can already start to discuss it

Here’s to a bright new future !

Marc Lerenard, ESIEA teacher and VR expert has sent me two new videos of the project:

This version is monoscopic, but they’re also working on an anaglyph and active stereoscopy version.

The only remaining problem, as for any VR application, is that the required head tracking takes some space (you don’t see that but there are at least six Optitrack cameras dispatched around) and that for the stereoscopic version you would need glasses. You wouldn’t be able to do that with four autostereoscopic screens since their viewing range is too small and they only have horizontal parallax.

Other solutions like the Holopub by french company BeInRelief exist :

Although it’s totally not holographic (I hate that when ‘holography’ is misused as it is 99% of the time!!), it’s a pretty cool marketing solution : a screen, a semi-transparent mirror, and you’re done !  No glasses, no tracking, illusion of depth (no stereoscopy, no parallax), integration of 3d images on top of a physical object.

We’re in the future baby!

Here’s a summary of these links :

- Barco about the light efficiency of their stereo systems http://bit.ly/6rTKjP

- Namco Bandai will use RealD stereoscopic API for their games : http://bit.ly/7BLR5d

- Google Goggles, incredible visual search : http://www.google.com/mobile/goggles

- Great AR algorithm demo by Oxford @ Ismar 2007 http://bit.ly/phGam

- Haptic Ring http://bit.ly/20XMgj

- Autostereoscopic shell for the iPhone, 49$ http://www.spatialview.com/en/node/489

- Firefighters in a CAVE : http://www.youtube.com/watch?v=qk1VdSKRa4U

- Youtube in 3D Stereo (old) : http://googlesystem.blogspot.com/2009/07/youtube-3d.html

- AlloSphere, A 360-Degree Virtual Reality Chamber http://www.scientificamerican.com/article.cfm?id=allosphere-ucsb

- A french 3D Webcam : http://www.gizmodo.fr/2009/11/22/la-webcam-3d-crazy-cam-vous-donne-du-volume.html

- Oxford AR group publishes all their vision paper : http://www.robots.ox.ac.uk/ActiveVision/Papers/index.html#tag2009

- VR setup at CNRS : http://www.limsi.fr/venise/index.en.html

- Asus has a laptop compatible with nVidia 3D Vision ! http://bit.ly/5MnIO

- Softkinetic, Optrima Partner On 3D Gesture Recognition Tech (http://www.gamasutra.com/php-bin/news_index.php?story=26095)

- Nice iPhone app to take stereo pix : http://www.juicybitssoftware.com/3dcamera/

I’m also going to Lyon for the big VR orgy, so see you there !

I’ve seen such a system in Weimar at IPT/EGVE 2007 and I hear several systems are starting to use this.

So here are three methods, demonstrated thanks to my Advanced Sketching System (c) :

Method 1 : Users are separated through a polarized filter

This can be seen as a regular active stereoscopy system.

The system has two active projectors, one polarization filter in front of each projector, regular shutter glasses plus matching polarization filters.

Thanks to the polarization, each user sees only one (active) projector. Each projector displays left and right image as is regularly done with active projectors.

In theory, you have it, each user sees in stereo ! This is the “easy” way which doesn’t need too much modification apart from adding polarized filters on shutter glasses.

The active stereo provides a good separation between left and right eyes, but the polarization separation is not perfect, so each user sees a ghost image of the other user.

This unperfect separation of polarized filters is less problematic when it is used for only one user since the left and right images aren’t that much different. But when you see the picture of another user, it is potentially very different and can also move while you are static. That’s a real presence breaker !

So let’s try another way :

Method 2: Users are separated through a shutter mechanism

Now this can be seen as a regular passive stereoscopy system. For each user, the left/right images are separated thanks to a polarized filter.

The tricky part : The first (active) projector displays the left image for user 1, then the left image for user 2. At the same time, the second (active) projector display the right image for user 1, then the right image for user 2.

This means that half the time you can see the left and right images for user 1, then half the time you see the left and right images for user 2.

But how can you hide the left/right images of the other user ? With modified shutter glasses that shut both eyes at the same time rather than each eye alternatively !

This method also requires some tricks on the image generator. Often times the generation of left/right images in quad buffers is hardwired in the renderer, so you need a system that is flexible enough to allow the generation of two left images instead of left/right.

Method 3 : multiple time splices

Another way is to take the same approach as the active stereoscopy that also requires hardware modifications : draw each image sequentially : User 1 Left, User 1 Right, User 2 Left, User 2 Right, User 3 …

Shutter glasses would then open only for User 1 left eye, all other eyes/users being shut, then User 1 right eye, etc.

This has several implications, ranging from flickering if the projector frequency is not enough, to decreased luminosity. You also need special shutter glasses and software to support all this.

Conclusion

This is not an easy topic! But an important one if we want real cooperation. I’ve left the option for multiple HMDs aside because you don’t see the other user, and some people just don’t want to wear HMDs.

I’m sure others methods exist, such as this one presented at IEEE VR 2009: Image Blending and View Clustering for Multi-Viewer Immersive Projection Environments, Jonathan Marbach

Do you know other methods ?

I’m finally starting using Twitter, codename VRGeek (what else?). I don’t know yet how I’m going to use it, but probably just post links to interesting VR stuff when I don’t have time or not enough resources to speak about it.

I’ve also added a newsletter to the blog, which will simply send you the latest articles by mail. It seems not everyone is fond of RSS !

You can subscribe to the newsletter thanks to the widget on the right column of the blog.

There’s also a fancy new tag cloud !

Now you can’t miss anything I’m saying !!

Maybe next week I could start a TV show ? =)

Just a small word to say that 21 teams have registered to the 3DUI Grand Prize !

We’re very happy about that, we didn’t expect that much participation =)

All the big 3D user interfaces labs are participating with at least one team, and we also have one private company (you know who you are!)

We’re sure we’re going to get very interesting solutions and we hope that this contest will give a small boost to the 3dui field !!

Again thank you all for participating, and have fun while working on it !