Look! Up in the Air! It's...Camera Ball?

October 15, 2011 1:09 PM

Take 36 mobile phone camera modules and put them in this and throw it in the air and you get this. Action video of Throwable Panoramic Ball Camera (seen at the top of the page).

Unimpressed with the Ball? Then watch the robot carving styrofoam video instead.

posted by cjorgensen (39 comments total)

Will. Buy. So. Cool.
posted by nevercalm at 1:13 PM on October 15, 2011 [1 favorite]

I love this camera. The concept is fantastic and I have had the pleasure of meeting at least one of the authors. Great stuff.
posted by fake at 1:17 PM on October 15, 2011

Awesome concept! Very cool results! :)
posted by darkstar at 1:18 PM on October 15, 2011

I think this guy does it better.
posted by Chocolate Pickle at 1:21 PM on October 15, 2011 [3 favorites]

I've always wanted something like this, although in my head I picture it capturing video, and also being durable enough to be used in sports, so you could have a filmed video of someone pitching a ball, and someone hitting it with a baseball bat, etc.

Also, patent pending? Really? Isn't it kind of an obvious Idea? Lame.
posted by delmoi at 1:30 PM on October 15, 2011

do not taunt happy fun camera ball. Its sees EVERYTHING.
posted by blue_beetle at 1:41 PM on October 15, 2011 [1 favorite]

Also, patent pending? Really? Isn't it kind of an obvious Idea? Lame.

The USPTO does have a track record of granting frivolous patents (I even hold some of them). The acceleration apex triggering the cameras all at once, and the specific photo stitching algorithms may not be obvious though.
posted by hanoixan at 1:44 PM on October 15, 2011 [2 favorites]

Want.

For an afternoon....then I would get bored with it. But it would be cool for that day.
posted by lampshade at 2:13 PM on October 15, 2011

I feel like you could take some great shots down at Zucotti Park or at with this thing, or at a protest march or the like.
posted by Scientist at 2:27 PM on October 15, 2011

Something in this form factor with Eye-Fi integration would be a ton of fun at parties, protests, funerals, and concerts.
posted by casconed at 2:46 PM on October 15, 2011

> I think this guy does it better.

I'm assuming that's a QT VR camera and that technology is just now becoming available to the average joe. Used to be something that could do that cost in the tens of thousands.

> Also, patent pending? Really? Isn't it kind of an obvious Idea? Lame.

I would think it it was that obvious we'd have seen one by now. The part that I think is amazing is that it's writing 36 distinct digital photos to flash RAM at once. Getting that part done alone is a technical feat worth admiring.
posted by cjorgensen at 2:50 PM on October 15, 2011

I think I could watch the robot cutting Styrofoam all day. Oddly fascinating.
posted by deborah at 3:32 PM on October 15, 2011 [1 favorite]

I would think it it was that obvious we'd have seen one by now.

That's kind of a ridiculous assumption, I know it's obvious because I've thought about it before. Just because no one (that we know of) has done it doesn't mean no one's ever thought about doing it. It may not be practical yet for video, it may just have not been cheap enough to build a prototype for a while, plus the software would take some effort (which I don't think should be patentable in general)

The part that I think is amazing is that it's writing 36 distinct digital photos to flash RAM at once. Getting that part done alone is a technical feat worth admiring.

Just include a buffer for each camera. Also, the flash ram in SSDs is really fast.
posted by delmoi at 3:48 PM on October 15, 2011

I threw my camera up in the air. Where it landed, I know not where.
posted by crunchland at 4:17 PM on October 15, 2011 [2 favorites]

Some might prefer much lower mobile phone trajectories...
posted by cenoxo at 4:18 PM on October 15, 2011

>I feel like you could take some great shots down at Zucotti Park
Here's a not-so-great one.

I like the ball, and imagine it as the embodiment of what would happen if you were using Google Earth to zoom in on your house, and go so close-up that you went SPROING and bounced up a couple feet.
posted by obscurator at 4:20 PM on October 15, 2011

The acceleration apex triggering the cameras all at once...

There isn't any acceleration apex. The acceleration is constant between when it leaves their hands and when it hits the ground: 9.8 m/s/s

There's a position apex, but you can't detect that with an accelerometer.
posted by Chocolate Pickle at 4:35 PM on October 15, 2011 [1 favorite]

There isn't any acceleration apex. The acceleration is constant between when it leaves their hands and when it hits the ground: 9.8 m/s/s

I'm not going to call bullshit on this, I'm just going to ask for an explanation.

You're basically saying an accelerometer can't detect changes in direction? It is moving in one direction (up), stops (apex), then goes back in the opposite direction (down), and throughout that journey, an accelerometer takes the same reading the whole time? I guess it makes sense mathematically - yes, it is constantly accelerating towards the ground at 9.8 m/s/s...it just seems counter intuitive. If there is a mass inside the ball that moves with acceleration (in my mind, that's how accelerometers work, I might be wrong), then when they throw the ball up, the mass is "pressed" in a downward direction. Once the ball starts falling down, the mass is "pressed" in an upwards direction... like when you go around corners in a car, no?
posted by Jimbob at 5:14 PM on October 15, 2011

Acceleration wouldn't be constant either. It would start fast, then slow down until gravity made it change directions, then it would increase in speed until it reached maximum velocity. Terminal velocity probably wouldn't be that great.
posted by cjorgensen at 5:30 PM on October 15, 2011 [2 favorites]

Chocolate pickle is quite right. It's hard to mentally separate velocity and acceleration, but the ball (in the absence of fluid drag, and in the limit of heights small compared to the Earth's radius) experiences the same magnitude and direction of acceleration the whole way through the trip.

Although air resistance would create a little variation in the acceleration - perhaps that's what this thing uses to detect the apex?
posted by Salvor Hardin at 5:43 PM on October 15, 2011

Crazily, I was at a party last night where one of the attendees described to me this very camera being used at Burning Man.

He mentioned there's a mercury switch inside the ball that fires the cameras, so (IIRC) the picture is taken just as the camera begins its descent.

And even nuttier is I wasn't even supposed to be at said party; I got asked along by co-workers. A housewarming party with a bunch of folk in town for a steampunk convention and a live band that played something like 1920s vaudeville/jazz/blues. Best party I've been to in years, and I had to leave early. Sigh.
posted by dw at 5:48 PM on October 15, 2011

After an extensive conversation on IRC with people who know physics, I see that Chocolate Pickle is quite right.

Fucking physics.
posted by Jimbob at 5:48 PM on October 15, 2011 [3 favorites]

As they say in their blurb...

Our camera contains an accelerometer which we use to measure launch acceleration. Integration lets us predict rise time to the highest point, where we trigger the exposure.

So they're not detecting the apex, they're predicting it based on how hard you threw it. Which they can do because...

> The acceleration is constant between when it leaves their hands and when it hits the ground: 9.8 m/s/s

Heh. I dunno... seems at least a little non-obvious to me. :) Also given they're at the Technical Univesity of Berlin, it's probably not the USPTO?
posted by adamt at 5:53 PM on October 15, 2011 [1 favorite]

Our camera contains an accelerometer which we use to measure launch acceleration. Integration lets us predict rise time to the highest point, where we trigger the exposure.

Ha. I just came back to say "Hey I wonder if it measures launch acceleration..."
posted by Jimbob at 5:55 PM on October 15, 2011

You're basically saying an accelerometer can't detect changes in direction?

It depends on how they're induced. It cannot detect changes in direction induced by gravity.

An accelerometer has a reference weight, which has mass and therefore inertia. If force is applied to the object (say, by your hand), and thus to the case of the accelerometer, then it has to transmit force to the reference weight to overcome the weight's inertia. The accelerometer has strain detectors and they pick up the force being applied to the supports holding the reference weight.

But when acceleration is being induced by gravity, the gravitational force applies to the accelerometer case and to the reference weight too. There isn't any need to overcome the weight's inertia, and the strain detectors don't detect anything.

Strictly speaking, gravity isn't really a "force". That's what General Relativity tells us. In a sense, an object in any kind of orbit (even a short one) is following a geodesic in non-Euclidean space. The apparent curve of movement is an illusion caused by the fact that spacetime is distorted by the presence of mass.

For reasons of symmetry and esthetic elegance, physicists nowadays really, really want to unite gravity with the other three fundamental forces, all of which are transmitted by particles (photons for the Electric force, gluons for the Strong force, and weak vector bosons for the Weak force) and to that end they've hypothesized a graviton to do the same thing for gravity. But General Relativity (which is the only real theory of gravity we have) says that gravity isn't a force at all. It's a consequence of the way that mass distorts spacetime. There isn't any evidence that there is any such thing as a graviton.

So since an accelerometer detects forces, and gravity isn't a force, an accelerometer doesn't detect it.

...now if an object with an accelerometer is sitting on the ground, the accelerometer will detect a lot of force. But what it's detecting is the force being applied by the ground to the object to prevent it from following the natural geodesic. It is indeed accelerating constantly because of that. It isn't detecting gravity, though.

And I have a headache now.
posted by Chocolate Pickle at 6:23 PM on October 15, 2011

Can I rewrite some of that? I didn't quite get it right.

The apparent curve of movement is an illusion caused by the fact that spacetime is distorted by the presence of mass.

No, not really. The problem here is that geodesics don't follow Euclidean straight lines. We don't live in a Euclidean universe, but all our intuitions are Euclidean.

In non-Euclidean geometry, a "geodesic" is the shortest line between two points. But depending on the nature of the geometry, it may be "straight" or it may follow strange curves (when observed in Euclidean terms).

Inertial movement in General relativity doesn't follow a Euclidean straight line. It follows a geodesic, and it takes force to make it diverge from the geodesic. Accelerometers will detect that force -- but gravity doesn't do that. Gravity, in a sense, is the name we give to the way an object follow a curved geodesic but it isn't a force as such.

And here's the key: in General Relativity, an object which is following a geodesic isn't accelerating.

But it is, in Newtonian physics, which is fundamentally Euclidean in outlook, and when we talk about gravitational acceleration being 9.8 m/s/s we're talking in Newtonian terms. Only problem is that Newtonian physics isn't correct.

And now I really have a headache.
posted by Chocolate Pickle at 6:45 PM on October 15, 2011 [1 favorite]

You might have a headache, but you've actually just taught this poor, pathetic ecologist a hell of a lot about gravity...
posted by Jimbob at 6:50 PM on October 15, 2011

(self propelled version of this) + (wireless data upload) + (machine vision) x 1,000,000 = a city mapped in a few weeks, give or take.
posted by AndrewKemendo at 6:57 PM on October 15, 2011

Well, I think I botched it up. I'm looking forward to a real physicist showing up here and doing a better job. Volunteers?
posted by Chocolate Pickle at 7:24 PM on October 15, 2011

I liked the panorama banana.
posted by underthehat at 8:53 PM on October 15, 2011

Volunteers????? I'm still struggling with the physics of those stick bombs.
posted by taff at 1:37 AM on October 16, 2011

In the days before gps, I could really have used this a few times when I was lost in the scub cedar of west Texas, trying to orient by compass and topographic sheet. Cedars are a pain because they're just tall enough to obscure the view, and impossible to climb. I'd tote one of these with an iPad on occasion for orienteering, instead of a gps, just for fun.
posted by Devils Rancher at 6:50 AM on October 16, 2011

The accelerometer would absolutely be able to detect when it has reached the apex and begun the decent. While it is true that everything on the planet is subject to a constant downward acceleration of approximately 1 g, as you pointed out an accelerometer does not measure gravity, it measures forces acting on the inertial proof mass suspended inside the sensor and in free fall will report 0 g since the mass inside is falling at the same speed as the rest of the sensor. As you also described, while sitting on the ground the sensor is measuring not gravity but the restoring force from the ground pushing upwards.

As a thought experiment, imagine the riders on the vomit comet are the proof masses and connected via a spring scale to the top and bottom of fuselage. On the ground they will exert a downward force on the spring. In level flight they will exert about the same downward force. While accelerating into a climb or through an aerobatic maneuver, they will exert more force. And when the airplane noses over to begin its unpowered dive, they will gradually exert less and less force since they are falling at the same speed as the scales until the top and bottom scale both read 0 m/s^2, just like the falling camera ball. If the airplane accelerates downwards faster than gravity by using its engines, the riders will be pulled towards the ceiling, leading to a "negative G" reading. I didn't read about any thrusters on the camera, so we can ignore this case.

This is exactly how the drop-sensors in your laptop know to park the hard drive heads when the laptop falls off the table, or how the ball can know when to snap the picture at just past apogee.
posted by autopilot at 8:42 AM on October 16, 2011

I've always wanted something like this, although in my head I picture it capturing video, and also being durable enough to be used in sports

I'm actually editing some stuff sort of like this now, cell phone company gave a panorama photographer a bunch of phones and he 3d printed a 360 rig that shoots video. They put it on a mountain bike. Video's not out yet but I've been playing around with action footage in the 360 player and it's pretty awesome. Fun challenge to edit something when you can't be sure what the audience will be looking at.
posted by nathancaswell at 9:14 AM on October 16, 2011

The accelerometer would absolutely be able to detect when it has reached the apex and begun the decent.

No, it absolutely won't. The ball can't know from instantaneous measurements when it is at apogee: as soon as it leaves the thrower's hands, there are no external forces acting on it for the accelerometer to detect.

Its acceleration is the constant 9/8 m/s^2 of Earth g. Its velocity will change linearly, steadily decreasing from the launch velocity, crossing zero at apogee. Its height will describe a parabola.

The trick is: the ball uses its accelerometer to measure how hard and for how long the thrower accelerates the ball upwards. That gives it enough information to calculate its launch velocity; and given the launch velocity, calculating the time to apogee is trivial.

(I'm not sure the vomit comet thought experiment helps; as far as the riders are concerned they experience free-fall for the duration that the plane is in its parabola.)
posted by We had a deal, Kyle at 4:03 PM on October 16, 2011

Upon reading the NASA reduced gravity research - Flight Trajectory page, I see that I had a misunderstanding as to the portion of the flight during which zero-g was felt by the occupants. It begins half-way up the arc when the engines are throttled back to only overcome drag, not just on the way down. This makes the riders strapped to their spring scales on the way up the path into the equivalent of the camera-ball once it has left the thrower's hand.
posted by autopilot at 4:53 PM on October 16, 2011

Heh. I dunno... seems at least a little non-obvious to me. :) Also given they're at the Technical Univesity of Berlin, it's probably not the USPTO?

Sure, I suppose the 'take a picture at the apex' thing was somewhat non-obvious. I was thinking of doing video, from which you could pick any frame. Also, you could get around the patent by simply having a timer.

If the patent is simply for "cameras on a ball" then I think that's kind of bogus.

So since an accelerometer detects forces, and gravity isn't a force, an accelerometer doesn't detect it.

Well, technically it detects a change in momentum. The 'force' that actually moves it is the electroweak force pushing on the atoms and molecules in the object that's being moved. But it's not measuring the electromagnetic force itself.

No, it absolutely won't. The ball can't know from instantaneous measurements when it is at apogee: as soon as it leaves the thrower's hands, there are no external forces acting on it for the accelerometer to detect.

Its acceleration is the constant 9/8 m/s^2 of Earth g. Its velocity will change linearly, steadily decreasing from the launch velocity, crossing zero at apogee. Its height will describe a parabola

Well, you're splitting hairs between "Know" and "Detect". It will "know" when it's apogee, but it won't "detect" it, at least if you're in a vaccum.

If the accelerometer is precise enough to detect (de)acceleration due to air resistance, it will go from seeing a slight negative (towards the earth) acceleration to seeing a slight positive (away from earth) acceleration, so it should be able to 'detect' the change.
posted by delmoi at 8:54 PM on October 16, 2011

I love how I see a green ball with cameras in it and say, "Hey, metafilter, look!" and then the thread gets so complex that I can't follow it.
posted by cjorgensen at 4:37 PM on October 17, 2011

So basically the ball is a self-aware n-dimensional timecube?
posted by blue_beetle at 5:37 AM on October 18, 2011