Full Body Dance Dance Revolution
tasty_beanburger writes "NewScientistTech has a story about a full body version of Dance Dance Revolution. It uses vision recognition to award points after assessing a player's ability to correctly mimic silhouetted dance shapes. Check out the video clip of it being demoed at SIGGRAPH 2006."
The researchers believe the system could have more practical applications in the future. They say it could be used to automatically translate sign language, for example
Yes, if this technology exists, and slips out into mainstream, they better put it into more practical uses than DDD.
Wait, a dance game where you have to actually dance? How revolutionary!
I'm currently pouting at the lack of specs. Is this going to be on my PS2 (*sigh*, PS3) or do I have to shell out for a new standalone application? Granted I didn't download the video...
Seems like a lot of work for something that isn't that novel. I sense that a DDR with additional EyeToy functionality would be just as good, and I already *have* most everything for that.
Does this mean that the fat kid who lives at the arcade and plays 3+ hours of DDR every day yet never loses weight isn't going to be able to play DDD, because his silhouette can't possibly match up with the figures on the screen? I think that's unfair. Give sweaty headband kid a chance!
At first I thought there's no way you can get 3D motion from a single video frame (not a completely off-the-cuff comment, I know a thing or two about computer vision). Then the geometer in me said "but the dimensionality of the configuration space is limited by limb rigidity." Not every point on the body can be in any point in 3D space; e.g. the hand has to be ~12" from the elbow. If you can track the shoulder, elbow, and hand, you can estimate how long they are and deduce (up to some reflections) where they are in 3D space based on the rigid body kinematics (assuming the shoulder to be fixed as an example). Eliminate the reflections by ruling out those that would violate anatomically impossible poses and/or produce occlusions. It has flaws, but I'm curious if anyone's tried it. An old prof of mine was working on something similar involving tracking limb movement for pattern recognition, so it wouldn't surprise me if someone has looked into this.
:)
That's right, I just generated and solved my own argument. The electronic extension of the voices in my head.
m0nstr42.blogspot.com
The game has an "autoplay" mode (press F8 during game-play, if memory serves, or select it in the main options menu). In this mode, the game does not keep score, but merely regards each step as perfectly accurate. I just jump around on the floor in front of it (and wear holes in the carpet -- good thing it was already torn by my old cheap office chair).
My computer is usually set up this way for exercise, because all I want to do is get my heart-rate up to a certain level, not keep score. In any case, failing a song would only interrupt the routine: I use the "endless" mode for exercise, so that there's a more or less constant stream of random songs. A modestly experienced player has a pretty good idea how well they're doing at any given song anyhow.
proof, n. A demonstration that a conclusion is implied by certain premises and axioms.
I've been through the honeymoon stage, and then reached the level of skill where a soft mat no longer responded quickly enough to my feet. Then, like an apprentice Jedi building his own lightsaber (pretentious, much?) I built my own hard dance platform. After much abuse, I eventually broke it. Likewise the MKII platform.
During one of these "no working platform" moments, I decided to try it sans platform entirely. I found it was actually better to work without the platform when exercising: when you're exercising, specifically, you aren't there to play the game as such.
My current exercise programme runs at approximately one hour a day, six days a week. I doubt that I've ever been fitter. I certainly haven't been slimmer in the last fifteen years. Yay for exercise-by-video-game!
proof, n. A demonstration that a conclusion is implied by certain premises and axioms.