Throwable 36-Camera Ball Takes Spherical Panoramas 140
MrSeb writes "Jonas Pfeil, a student from the Technical University of Berlin, has created a rugged, grapefruit-sized ball that has 36 fixed-focus, 2-megapixel digital camera sensors built in. The user simply throws the ball into the air and photos are simultaneously taken with all 36 cameras to create a full, spherical panorama of the surrounding scene. The ball itself is made with a 3D printer, and the innards (which includes 36 STM VS6724 CMOS camera sensors, an accelerometer, and two microcontrollers to control the cameras) are adequately padded, so presumably it doesn't matter if you're bad at throwing and catching."
Not the kind... (Score:2)
... of Pokemon Snap I was thinking of.
Re: (Score:2)
Want. (Score:3)
Really cool man!
Viewing is going to be kind of lame (Score:5, Interesting)
It would be cool if those cameras could be upgrade/modified to take full motion video though. You get to be the ball, and look in any direction you want. Heck, with a bit of work you could almost certainly program something that could take a few snaps from this ball in the air to instantly recreate any space in a virtual environment. The combination of parallax from the movement and multiple (presumably overlapping) cameras should make it quite possible for a computer to figure out exactly what is where and what shape it is.
You could make spontaneous virtual tours with something like that. A couple of guys go out to a location, one guy throws the ball at the other, uploads the pictures via cell or wifi to some server that then recreates the space and lets people virtually fly around it. You could even do something like that for crime scene photos or anything that needs to document the exact state of a room.
Re:Viewing is going to be kind of lame (Score:4, Interesting)
Think of the military value though. Toss a ball into a bunker, bounce it around the corner, throw it straight up to see what's on the other side of a wall, etc...
This could be quite the tool for urban combat.
-Rick
Re: (Score:3, Funny)
Yea, all you'd have to do is throw it into a bunker, then go into said bunker to retrieve the ball, come back out of the bunker, plug the ball into a computer and look at the pictures. Then you'll know exactly what was in that bunker you were just in. Revolutionary I tell ya.
Re: (Score:1)
Yea, all you'd have to do is throw it into a bunker, then go into said bunker to retrieve the ball, come back out of the bunker, plug the ball into a computer and look at the pictures. Then you'll know exactly what was in that bunker you were just in. Revolutionary I tell ya.
Not to mention, the pictures would be at a measly 2MP resolution...
So, it would be less knowing "exactly what was in the bunker" and more a fuzzy, pixellated version of what was in the bunker.
Re: (Score:3)
Sure, if they take the bright green prototype created by a bunch of graduate students and deploy that directly into combat.
Or, you know, by the time it was made into something usable by combat troops it would be scaled up to having more megapixels.
Re: (Score:1)
Or, you know, by the time it was made into something usable by combat troops it would be scaled up to having more megapixels.
My bad for not ending my post with a /sarc.
Heck, why not fill the little bugger with explosives while we're at it?
Re: (Score:2)
Because you'd like to look and see if the "bunker" aka urban bulding has enemy combatants or school children in it first?
Re: (Score:3, Funny)
Bad idea, it would make retrieving it dangerous.
Re: (Score:1)
Bad idea, it would make retrieving it dangerous.
Why would you want to retrieve an active grenade?
Re: (Score:2)
Re: (Score:2)
Weaksauce!
Re: (Score:2)
Ummm.... with 36 CCD sensors and the CPU to process the stream of images from them.... I seriously doubt the engineer was dumb enough NOT to at least make provisions for placing an 802.11 chipset in there.
Re: (Score:2)
Re: (Score:2)
Not at all. Enemies will know it's a camera and try to destroy it. Booby trap it so that if they try to crush it, it explodes, or fill it with explosives and have a remote trigger.
Throw it in, grab a few pictures, then detonate if there are no friendlies in range.
Enemies run from potatoes when thrown at them (Score:3)
Enemies will know it's a camera and try to destroy it.
Nope. They will assume it is a grenade and act accordingly. During WW2 a US destroyer and a Japanese submarine nearly collided. The sub was so close the destroyer could not lower its guns far enough, of course the sub crew had no such problem with its deck gun. As the sub's deck gun was being manned sailors on the destroyer noticed a bucket of potatoes that had been brought up to be peeled. They grabbed the bucket and tossed potatoes at the deck gun crew. The guys on the sub immediately began chasing the po
Re: (Score:2)
Heck, why not fill the little bugger with explosives while we're at it?
Urban combat is a nasty process with exceptionally high casualty rates. The last MOUT training I did as a defender (ie: local irregular militia vs significantly larger organized military assault) we inflicted over 70% casualties.
That was a 10+ years ago though, and maybe we've started learning from the Israelis.
But even with flash bangs and grenades, the advantage is in the defender's hands. Especially when you have to be concerned about collateral damage.
Given the assaulter's position, I'd much rather know
Re: (Score:2)
I don't think there's any non-lethal alternative that can't cause long-term injury. E.g. you see flashbangs used all the time in video games as a universal weapon that lets you clear the room full of both bad guys and hostages with not a scratch to the latter, but it can cause permanent blindness/deafness in practice, and if exploded close enough, some real (and even potentially lethal) injuries. Sure, it beats being accidentally shot, but you're not going to win hearts and minds by throwing one in every ap
Re: (Score:2)
Prolonged tear gas exposure can similarly have long-lasting or permanent effect.
Even when it doesn't, it's still quite painful. Suppose you have an apartment building in a block that you need to clear. Somewhere inside are a few combatants, and lots of terrified civilians. Are you suggesting throwing a tear gas grenade in every apartment before entering? That's an easy way of getting a lot of people hating you for a very good reason.
Re: (Score:2)
Protip: "Non-lethal" does NOT mean safe to use indiscriminately.
Alcohol (at low enough doses) is non-lethal to consume. Long term non-lethal consumption will still kill you though.
Tear-gas and similar have the same problem. These are not zero-impact systems, and they CAN be lethal.
Re: (Score:2)
Police could do that. But the military is prohibited from deploying chemical weapons, including tear gas, believe it or not.
Re: (Score:2)
2*36 if it's done right...
Re:Viewing is going to be kind of lame (Score:5, Funny)
Re: (Score:3)
The ideas just go on - what about fitting it with sensors from a kinect too, so you get a 3D model with your photo. IR cameras too, motion sensors - hook a bunch of them into a network, scatter them around, and you could have an "x-ray vision" HUD that shows you what's going on through walls and ceilings.
Very, very clever indeed, I love this idea for the sheer simplicity, and I'll bet you can make the basic hardware for under $US500.
Re: (Score:2)
Re: (Score:2)
I'm confused, he wanted to check if the milk was expired before he shot his way in to an overly thick but poorly insulated refrigerator? Is this because the door weighted too much?
Re: (Score:1)
Re: (Score:2)
(Although on that note, my dad was stopped with a present in his bag at an airport in the 80s. The security guy was able to read the Victorinox logo. It's quite impressive.)
Re: (Score:2)
don't discredit it yet - we just have to teach the bad guys how to play tetherball
Re: (Score:3)
Re: (Score:2)
Actually, a better solution might be a bit lower tech - a long wire. It doesn't matter if it's a single-use thing. Just fire it out, unspool the wire, and stream back (and record) videos until the wire snaps.
That's how many missiles are guided already [wikipedia.org], so doing the same thing for a maybe-recoverable camera seems reasonable.
Re: (Score:2)
It would be awesome if there were some way to send data without wires. :-/
Re: (Score:3, Funny)
Put a countdown led on it and the bad guys might feel like throwing it back out
Military already doing this via wireless (Score:2)
Yea, all you'd have to do is throw it into a bunker, then go into said bunker to retrieve the ball, come back out of the bunker, plug the ball into a computer and look at the pictures. Then you'll know exactly what was in that bunker you were just in. Revolutionary I tell ya.
Years ago a cable tv documentary on robotics was showing a softball sized spherical device with multiple cameras (far fewer than the student's device though) that the military was developing. The idea was to just throw it over an obstacle, into a window or door, etc in an area of interest. The imagery was wirelessly transmitted to a laptop the troops were carrying. Of course this ball was not maneuverable. There were other small robots being testing that could be thrown, tossed onto a roof, through a window
Re: (Score:2)
It's a good thing no engineer has ever been able to improve on a hacked-together prototype built in some guy's garage with improved/new components, otherwise you'd look really stupid.
Re: (Score:2)
High number of video streams, megapixels, battery life, re-projection, data management....
It's not a easy problem.
Re: (Score:2)
High number of video streams, megapixels, battery life, re-projection, data management.... It's not a easy problem.
And its something the military has already built and is evaluating. Fewer camera though. The device only needs to be on for a limited time, when its thrown somewhere of interest and for a few minutes thereafter. A sensor can determine which was is up and turn off cameras pointing into the immediate ground, the laptop with the troops could give priority to the cameras pointing horizontally or slightly upwards, etc.
Re:Viewing is going to be kind of lame (Score:4, Funny)
Think of the military value though. Toss a ball into a bunker, bounce it around the corner, throw it straight up to see what's on the other side of a wall, etc...
Think of the high school teenager value. Toss it into the locker room, bounce it around into the shower, instant 36 counts of manufacturing child porn.
Comment removed (Score:5, Funny)
Re: (Score:2)
Except that in much of the world, there are more females than males, and the US Women's team is pretty damn solid.
So much for "everybody else" huh?
Re: (Score:2)
beating even England
You say that like it's an achievement. I was dragged to the pub for that game. I sat with my back to the screen, and managed to look around for the only exciting bits of the match. It contained a hilarious goal. The English keeper stopped the ball, put it on the ground, and then watched as it rolled slowly across the line...
Baseball/softball sized (Score:2)
This would be a bad idea for the US military as everybody else in the world is better at soccer.
Humor aside, I think the US military has considered this line of thought and their prototype is baseball/softball sized. As a special bonus this size can be easily mistaken for a grenade so its likely to generate some activity in the area it lands.
Re: (Score:2)
Re: (Score:2)
And what, you're going to go building to building and fill each one with impact grenades? I'm sure that will do tons to motivate the locals to help the US.
Not to mention that the M32 is a defensive weapon, full gear and ammo load requires 3 Marines to carry. It is an impressive weapon, but it's use in offensive urban combat and room by room clearing is near nil.
Although, I did have a buddy take an M240-G, clip the carrying strap to the heat shield, forward hand on the barrel swap handle and rear hand on the
Re: (Score:2)
Re: (Score:2)
And what, you're going to go building to building and fill each one with impact grenades? I'm sure that will do tons to motivate the locals to help the US.
...no. I think you are missing the point. The idea is not to motivate the locals to help the US -- that was the so-called "hearts and minds" doctrine that failed so spectacularly in Vietnam. Rather, you want to motivate the locals not to help the terrorists. If the locals begin to understand that their non-combatant status isn't going to prevent them from being mowed down along side the terrorists, they will stop associating with the terrorists, and will in fact start actively cooperating with the occup
Re: (Score:2)
Because obviously, if some foreign nationals invaded the US and destroyed our military, leaving only "freedom fighters" (aka "terrorist") to attempt to repell them, and in an effort to kill those terrorist, the foreign nationals killed your wife and kids, the first thing you would think is, "huh, I guess I should do exactly what they want me to do so they don't kill me too."
That style of occupation has been working for the Israelis so well for the last 60 years too! I'm sure their constent state of war will
Re: (Score:2)
Gorgon Stare [darkgovernment.com]?
I-Ball [gizmag.com]?
Firefly? [gizmag.com]
Crap.. sure, six years ago (Score:2)
hell, my local county Sherrifs department has one of these
I've played with it at a police function....
Re: (Score:1)
wrong (Score:2)
Re: (Score:2)
The only problem with 360 panoramas like this is that viewing it requires you to use some Quicktime-VR sort of setup that always looks bad with the corner distortion and awkward controls. It's hard to map a full spherical image onto a flat display.
I'm going to have to disagree [panoramas.dk] with [panoramas.dk] that [panoramas.dk]. Plenty of great panoramas on that site and are very clean and clear. Just because the source video was crap doesn't mean they all are.
Re: (Score:2)
There's some serious distortion as you move around that image, too. Reminds me of the original DOS version of Tomb Raider. The 3D just wasn't quite right. Takes nothing away from the ball, but the stitching technique needs work.
Re: (Score:2)
There's some serious distortion as you move around that image, too.
This is induced by projection and can't be avoided. Your screen isn't the curved inward like a sphere, so when a spherical image is projected onto it you'll always get edge distortion. Note that when they crop in on the image, the distortion goes away because the image becomes flatter for smaller crops of the sphere, thus the projection becomes less distorted, in the same way that a mercator projection of the Earth is very distorting, but a mercator projection of Seattle will have relatively high fidelity
Re: (Score:2)
It's hard to map a full spherical image onto a flat display.
My retinas seem to handle it just fine. Yes, you'll have to use some special software to render the image, but there's no reason you can't render a viewpoint into the image without distortion.
Re: (Score:2)
Your retinas aren't flat, and your display (the image your brain has processed) isn't projected on a flat surface. Try again!
Re: (Score:2)
The only problem with 360 panoramas like this is that viewing it requires you to use some Quicktime-VR sort of setup that always looks bad with the corner distortion and awkward controls.
I imagine the "real" intended use of this is for oppressive police/military units to toss into an area to map the room and identify potential threats easily. Things like where people are in relation to windows, any exits, how many have we wounded so far, etc.... I really doubt this will be a successful consumer product but as a product to make killing consumers easier, safer and more efficient I think it'll sell quite well.
Just toss it in a room and you've got the picture. Toss a grenade first and you get
Re: (Score:2)
I imagine the "real" intended use of this is for oppressive police/military units to toss into an area to map the room and identify potential threats easily.
It's also usable by non-oppressive police/military units.
In fact, this technology will likely cut down on accidental shootings of civilians, because the police/military will be better able to determine what sort of threat they face before exposing themselves. Fewer split-second shoot/don't-shoot decisions means fewer wrong shoot/don't-shoot decisions. (the oppressive police/military, OTOH, won't use this because they don't care who's in the room; they'll just blow up the room immediately and sort out the
Re: (Score:2)
Re: (Score:2)
The only problem with 360 panoramas like this is that viewing it requires you to use some Quicktime-VR sort of setup that always looks bad with the corner distortion and awkward controls. It's hard to map a full spherical image onto a flat display.
That's the "only problem" with *any* photography, with the edge distortion getting more pronounced the wider angle you try to project. A traditional lens makes straight lines curve - most noticeable with fisheye lenses. A pinhole keeps the lines straight, but angles get distorted.
The solution in a scrollable view, is to zoom into the scene a bit more. Try it with Google Streetview -- zoom out and the edges look odd (although they maintain a rectilinear projection); zoom in a couple of steps and it's not a p
Hmmm... (Score:5, Funny)
Hmmm so now I can take photos WITH my balls.
Re: (Score:2)
Re: (Score:2)
Stereoscopic? (Score:1)
Should be stereoscopic. And there should be an immersive stereoscopic viewer.
-Max
Smile! NO DUCK!!!!! (Score:3)
...but seriously, neat idea but hardly for everyday use. The seams are horrible in the resulting panorama. I presume each camera is using it's own auto exposure. What you need to overcome this is for all the cameras to communicate and decide upon a single exposure. Also might be difficult for the photographer to look natural when the shot is taken, but still catch the ball.
Good to see people trying different things.
Re: (Score:2)
Re: (Score:2)
I don't think the problem is exposure, as my understanding is that these are just sensors all controlled by the controller, so I'm fairly sure that's not the issue (as it would be if you had just jammed 36 off-the-shelf digi-cams in there. But even if that were the problem, it would be trivial to solve. Even if you can't force them all to use the same fixed exposure and white balance settings, that can be corrected after the fact by reading the settings out of the exif and compensating in software.
The bigge
Re: (Score:2)
The seams and exposure differences can be fixed with a little automatic processing. I sometimes take panoramas with my cellphone camera, a 4-year-old Sony Ericsson W570i with a 2MP sensor. Even if the source pictures are differently exposed, as in the image in the article, stitching them together with Hygins normalizes the exposure levels and the seams are unnoticeable. I believe Hygins uses enblend internally. (Of course, the panoramas still look somewhat cheap, due to JPEG artifacts and crappy optics, bu
Re: (Score:2)
Yeah, I'd love to get my hands on a set of the source images, since the stitching quality in the samples they showed was *horrible* compared to what Hugin can do. They clearly aren't trying to do seam blending or photometric correction, and I think one of the images is just plain registered wrong.
Re: (Score:2)
You're missing the point. This was a tech demonstration, not an end-user finished product, you can see that in the end credits (VTFV replaces RTFA, I guess). Yes, the stitching is hackish, but that doesn't matter. The proof of concept is brilliant, and I could easily see this kind of thing taking off. Even without stitching, it gives you the ability to take pictures of the surrounding area from a reasonable height, anywhere. I could see this being really useful at concerts and events where you want a pictur
Re: (Score:2)
...but seriously, neat idea but hardly for everyday use.
Or, perfect for everyday, low-end use.
Remember when digital cameras were expensive? Remember when those really cheap ones started showing up -- tiny, toy lens, 640x480 resolution -- yet they were fun.
TFA suggests these balls will sell for $100. I'd buy one, and I wouldn't care about the odd stitching artefact.
HDR? (Score:5, Interesting)
Re: (Score:2)
Re: (Score:2)
Looks like the camera images have a fair bit of overlap - they could probably assign at least a low/high exposure patterning that would give additional exposure data beyond whatever the sensors can provide (I presume they used consumer sensors at 10bit at best but probably already coming out processed to 8bit).
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
"True HDR" requires a sensor that can distinguish between a high range of brightnesses. That's all.
Film is one example. There's a huge amount of detail in a negative, that you throw away when you print, because neither paper nor eyes have the same range.
There are certainly electronic sensors that are equally able to capture a wide dynamic range.
Achieving it with multiple exposures is a hack (quite a cool hack, of course).
Is it canine-tested? (Score:3)
Except for small, hard rubber spheres, they haven't made a ball that my dog can't tear to shreds.
Nothing to really add.. (Score:1)
hey! (Score:2)
But does it come with a GPS? (Score:2)
Re: (Score:2)
Very Useful Little Gadget (Score:1)
I can see the military and police going ape over this. Toss a ball, get a quick survey of hostile territory BEFORE going in. Even with distortion, it will be very useful without needing VR glasses.
It's also cheap(comparatively speaking) and light, so several can be carried.
For civilians, just think what it will do for paintball! Just make sure the lenses are easy to clean. :)
Re: (Score:2)
They have camera-carrying UAVs for that already. And satellites. This would be better for if you're in a space you can't see from the air.
Throw one of these into a cave or building, see if there are any bad guys around the corner, etc.
And as for what I've seen of high-zoot paintball matches, it consists primarily of hiding behind something and lobbing thousands of paintballs into the air trying to get lucky when the enemy peeks out from the thing it's hiding behind lobbing thousands of paintballs at you.
literally (Score:2)
The Japanese ball is much better. (Score:2)
Much better surveillance and pursuit ball [youtube.com] from Japan. It would be easy enough to add more cameras to that thing.
Next, do this: (Score:2)
Use 36 video cameras and an attitude sensor, and combine the images and stabilize them to a particular attitude in realtime.
Now you can toss this thing at random and always see a 4-pi steradial view of the area no matter how it's tumbling.
This I would buy (Score:2)
I'd expect different types of balls. For instance one that's nearly unbreakable. Or a bigger one which you can roll down a slope.
Imagine tossing the ball from one cabrio to another. Or taking a birds eye picture of a sport your buddy is playing. Or a ball fixed on a helmet.
Great for hobby/sport rocketry or kites (Score:2)
Beautiful Red (Score:2)
Man... (Score:2)
Re: (Score:3)
I can see parallax in the panoramas and I think it's a design flaw. To eliminate parallax all pictures of the panorama must be taken from the same focal point. Since each camera on this ball has its focal point in a different location, all panorama's taken with it will have parallax and the images won't line up perfectly.
However, with parallax data it is possible to extract depth information, enabling 3D images.
Re: (Score:2)
And the data should be there - they just need a much more advanced image stitching system.
Re: (Score:2)
Yeah, that's the impression i got. With better processing this could look way better than it does in the video. It's probably not hard to infer the ball's angular velocity from the motion blur present in all the source images, which *ought* to enable some sort of deconvolution to make it all sharper.
Also, where camera areas overlap, you have "luminance comparisons" from one camera to the next. If the same surface looks darker from one camera than from another, that probably means that a light source has for
Re: (Score:2)
For all the cameras to have the same focal point, the focal point would need to be in the centre of the ball. I don't think that's impossible to achieve, but it boggles my mind a bit trying to visualise how it would fit together.
Re: (Score:2)
Why not actually put a decent camera in there?
No reason at all -- you can buy sensors and lenses of nearly any quality off the shelf.
Looking forward to seeing your new and improved pictures! Let us know when you're done.
Re: (Score:1)
RTFA. They measure acceleration and compute the apex from that. Easy peasy and it works really well. Unless the planet you're on has wildly varying gravity, of course.
Re: (Score:2)
"I hereby put the above thoughts into the public domain: screw you patent system."
You are assuming that you own this idea. You would have to patent it before anyone else does for that to be the case.