Disney VR


It’s been about 18 1/2 years since I joined Disney for a super-secret VR project in Glendale.

The first experience we built simulated Aladdin’s Magic Carpet Ride in 1st person and 60 FPS, as if you were flying the magic carpet above the streets of Agrabah. The giant system that ran it was a three-headed (think SLI/Crossfire) SGI graphics super computer called the RealityEngine2, big as a refrigerator and producing ten times the heat.

We knew 1/2 million dollars per user wasn’t exactly a scalable for-profit experience. Today, an XBox is about 1-2x as powerful for 2,000 times less money. A Tegra3 chipset is in the same ballpark, but for about 25,000 times less. So the smart idea was to buy our way into the future and get a big head start on solving VR, on only a $20M budget!

  • BTW, apparently, in 1994, we could not take good screenshots of anything, either that or bit rot. But here are some photos I recently recovered. It kills me that I had an actual HD video of the ride but I brought it to a gig and lost it — curse you Zombie! 

The bulky HMD was the best I’d ever seen (note: my Oculus is still on order, too soon to compare) and the experience rivaled what you could get even today, except it was only meant to last 5 minutes vs. 100+ hours of Fallout 2 I’m still finishing. We put 100,000 people through this experience at Epcot Center in Florida.

I was there for the first 2-3 months, tuning the ride on-site. I actually had to work in a glass walled room while thousands of civilians filed by — “a programmer in his natural environment” the sign could have said, “please do not tap glass.”

We learned quickly that people were cognitively overloaded in VR and they tended to simply blow right past the great interactive 3D animated characters we built for them, like drunken drivers. So we invented a system called “weenie wires” that made the ride a cross between something totally free-form and a dark-ride where you’re on rails. You could always choose where to go, but we’d nudge you in the right direction and help you have more fun.

I’m reminded of this because my friend JWalt, who did so much work on the Weenie Wires, went on to win an academy award for his work in previsualizing movies. He now tours the country doing a one-man interactive show. He and I also built what might be the first 60fps full-body motion capture system for pupeteering “skeletally retargeted” 3D characters in real-time. Kinect uses a nice improvement on the idea — and it finally gets rid of those damn wires.

Speaking of a Who’s Who of Disney VR, I have fond memories of the whole crew: JWalt as mentioned above, Scott Watson, who is now CTO of Disney R&D, Jon Snoddy, who I rejoined later in life for a  startup called BigStage (making digital 3D copies of users) and who is now back at Disney along with another cool guy, Eric Rice. The always brilliant Randy Pausch did research in VR using our platform (if you haven’t watched his “last lecture” stop everything and please do so now). I managed Jesse Schell for a brief time, when he was just a junior show programmer fresh out of CMU, and look at what amazing things he’s gone on to do without even asking if it was okay. Aaron Pulkka was in the same boat as Jesse and has also done extremely well.

Many of the other folks from that time went on to become VPs of Disney or other companies. It was quite an amazing bunch of people to work with.

Anyway, after Aladdin VR, I went over to the main Disney R&D building and continued to hack away at new tech ideas.

The first thing I did was take a cool motion base they had built and made it fully interactive. This thing used six air-cells that we programmed via MIDI to simulate an undulating water surface. We put a raft on it, gave everyone an oar (with 6DOF tracking — think Kinect, but with wires) and even modulated the exhaust to blow wind and squirt water in people’s faces on command — yes I got to douse Michael Eisner. This became the Virtual Jungle Cruise ride in DisneyQuest, when it was still around.

I was also fascinated with ways to do VR/AR without an HMD. So we worked hard on projected virtual environments and audio, trying to put pixels and sounds everywhere we looked. The most successful was the 6-sided CAVE environment that showed up in part in the Hercules 3D ride. The least successful, from an “internal politics” POV was the multi-bright projector (shown at right).

True story: our top level VP of R&D had a pet project to build a HD movie projector pumping out so much light that he could use black velvet as a screen, just to get the purest blacks imaginable. It cost $100k or more and harnessed what I think was a singularity for power. I saw it and quickly realized I could tile 6 small off-the-shelf projectors at 1/2 the price and twice the brightness. I showed it to my boss, he liked it, and told me to never show it again. Ever. This may have been the first time I realized that fitting into a large company might take some getting used to.

Mostly, I was very interested in what we’d today call NUI. The image at left above depicts a user holding a simple stick or baton that we tracked in 6 degrees of freedom to navigate/fly through an environment. Any object worked fine, even a hat. The CAVE above was meant to allow us to point, using a wand, at any 3D object even in stereo 3D. The oars in the rowing ride were another example of multi-participant 3D spatial controls. Just think how hard it is for 4 people to drive one car. That’s the key problem to solve. Oars work very well, since it’s just like in the real world. You have to communicate or you go in circles.

One interesting outcome from that particular line of research was a game called “Crazy Taxi.” Strangely enough, my research into the idea of a whole family simultaneously shouting directions (“go left! look out!”) at a virtual taxi driver was the initial seed idea for a “Wild Taxi” ride (think: “Mr. Toad’s Wild Ride” meets the famous Blues Brothers car chase) that we showed to Sega. Then Sega (taking one of our creative directors) went off and built the arcade version for a single player using a joystick and produced billions of dollars of profit and several lawsuits, the latter of which I’ve previously ridiculed.

I’d still love to see the full-size car mock-up of the original Disney Ride. I’m sure someone will do this for Kinect at some point.

All in all, my 4 years at Disney was extremely fun and rewarding. A+, would do it again.

  1. #1 by Nir B.D. on October 18, 2012 - 11:00 am

    Hi,

    As I was 12 years old, I was one of the kids who had experienced the Aladdin VR. I was picked from the “Aladdin VR Tour” line and it was one of the most incredible experiences I have ever had.
    From a young age I was a fond of 3D graphics and it enchanted me. I even remember the “programmer in his natural environment” and believe it or not, I even knew, being 12 years old, what Silicon Graphics was.
    After a decade later, I went again to Epcot, and couldn’t find any VR of any kind there. That was really disappointing.
    Now I am an engineer and Co-Founder of an AR company, specializing in 3D graphics.
    I was wondering why this project was stopped and why VR had not yet made it to the mainstream? In addition, if you had such good HMD two decades ago, why this hadn’t gone to mass market yet?
    I heard some answers to that are that people starting being delusional after being immersed in VR, have you experienced any of these back then?

    Would appreciate your reply,

    Thank you,
    Nir

    • #2 by Avi on October 18, 2012 - 4:34 pm

      Hi Nir,

      Thanks for your comments! I don’t think anyone was ever delusional, although I have a funny memory of getting the (then) US Senate Majority Leader extremely nauseated. This was back in our labs in Glendale, before we deployed anything, on a day when the controls weren’t working. Short-notice demo — you should always have a backup plan, which we didn’t. So I had to improvise and use a mouse to copy what he was trying to do with the handlebars as quickly and accurately as I could. I don’t think he knew, but it was very laggy, and so he turned green. A few people did get sick on the actual ride, but not more than for a good rollercoaster, afaik. Some people get sick on park benches.

      More people had to deal with the cognitive overloading in the world and just had a hard time adapting.

      Anyway, the main reason it wasn’t commercialized was the cost and logistics. As far as Disney rides go, they tend to be very expensive. But you have to look at the throughput to make it up. For VR, it took 5 minutes to on-board each guest plus a 5 minute ride @ 500k/seat (not including the HMD and motion base, which we left off). That’s less than a hundred people a day per seat, vs. many thousands for a rollercoaster. DisneyQuest tried it on a cheaper and smaller basis, with smaller PCs, etc.. but they had some issues to contend with.

      Also, keep in mind, while the HMD worked great, it was so heavy it required weight relief cables from the ceiling, which makes it not good for use in your home, for example. And the weight still added a lot of inertia to head motions, even under full relief. The Disney*Vision HMD also had very low effective latency, since it used CRTs and wired sensors and SGI added some custom extensions to fix latency in their pipeline.

      I’m definitely looking forward to the Oculus Rift to see something cheap, light, bright, and wide FOV finally.

      Avi

(will not be published)