Become a fan of Slashdot on Facebook

 



Forgot your password?
typodupeerror
×
Cloud Graphics Technology

3D Aerial Photos For the Common Man 78

An anonymous reader writes "So you have a RC model aircraft snapping digital photos from the air, but how do you organize them all? This cheap cloud service from a European research giant will upload your photos and automatically convert them into 3D models you can navigate like a video game. And if you don't have a model aircraft, they got those on-the-cheap too. Let the overhead droning begin!"
This discussion has been archived. No new comments can be posted.

3D Aerial Photos For the Common Man

Comments Filter:
  • by Anonymous Coward on Sunday May 22, 2011 @05:05PM (#36212004)

    This is so going to be made illegal when more people start taking high resolution pictures of their neighbors' swimming pools.

    • Re: (Score:3, Insightful)

      It's gonna be illegal anyway, because only a terrorist or pedophile would use the technology, probably to plan out the shortest paths to bomb government buildings or rape kids at the parks.

      Even worse, the so-called "hobbyist" could use the data to make a 3-D video game about bombing government buildings or raping kids and release it under that Marxist GPL license, which would inflame tensions in the Middle East and provoke Iran into destroying Israel.
    • This is so going to be made illegal when more people start taking high resolution pictures of police/DHS/ATF//border activities.

      FTFY

      Strat

  • If you have a 3d world, and can identify the objects in the world, then software can navigate its way around the world and do tasks.
    I'd reason that wiring up Natural Language when having a large database of objects(nouns) would still be rather difficult, but not as difficult as changing camera feeds into 3d world representation.
    Finally you need to build a body for the robot, so it can do things in the world. By understanding Natural Language, anyone can tell a robot what to do in their native tongue. Al
    • The real reason (Score:3, Insightful)

      by Anonymous Coward

      >"I kinda put off actually making Artificial Intelligence in 2002 until someone makes a nice piece of software that you can walk around buildings and turn them into Quake levels."

      No you didn't. Hey, I put off flapping my arms and flying to the moon in 2002, because there wasn't a decent hotel there. Or maybe I'm talking shit. Certainly that's what I assume you are doing.

      You "put it off" because neither you nor anyone else actually knows how to do it.

      • Re: (Score:2, Interesting)

        by CrazyJim1 ( 809850 )
        If you want to check out my webpage on AI from 2002, you can do so here [goodnewsjim.com]. There is a lot of information there that isn't as concise as I have here. AI is easy to understand conceptually when you think of it as simply a computer program taking in input from the outside world, and interacting with it. The AI I'm talking about has little to no machine learning involved. It is all hard coded AI for robots to take commands and follow them. It really isn't as hard as everyone thinks. I think the trap peopl
    • by EdZ ( 755139 ) on Sunday May 22, 2011 @05:33PM (#36212136)

      I'd reason that wiring up Natural Language when having a large database of objects(nouns) would still be rather difficult, but not as difficult as changing camera feeds into 3d world representation.

      With two cameras with a known separation, it's not that hard a problem at all. With one camera and a depth camera (Kinect), it's not hard at all. With one regular camera and a known motion, it's not that hard. With one camera and an unknown motion, it's a bit tricky. Mostof this is covered by Structure From Motion, or occasionally Structured Lighting.

      3D data from 2D data is a bit tricky, but nowhere even close to being as hard as software that can understand natural language. Even PARSING natural language is hard enough for most systems. There are far more systems that can map a 3D environment than there are that can correctly recognise and respond to a simple sentence, e.g: "Computer, will it be rainy today?"

      • Easier might be the wrong word, maybe more cerebral. To do natural language interpretation, I would envision you start with a "dictionary database" full of objects. These objects would be nouns, and then you could program in properties on them known as adjectives. Without a "dictionary database" of known objects, I could not figure out how to code natural language. So you're right that natural language is more difficult than digitizing things because digitized things are a component of natural language.
        • I confess that I have not read parent post while in full critical analysis mode, but I do not think that is needed here.

          All natural languages that I am aware of are based on core objects that have no physical representation. Such as "friendship", a multitude of different kinds of "love", "dishonesty", "truth", and so on. It is nearly impossible to describe anything in any human language without referencing these purely conceptual objects. The references are often buried in tacit assumptions or postulates

    • by Anonymous Coward

      Wrong. The limiting factor is people want to be animals and don't want to learn. They need incentives and enforcement, not just any form of cheap education.

    • I kinda put off actually making Artificial Intelligence in 2002 until someone makes a nice piece of software ...

      Nice try, Marvin.

  • Would this work with aerial photographs that can be seen on google maps? That would cover a large area already.
    • by caseih ( 160668 )

      No because it relies on two photos of the same thing, but taken at slightly different angles. In other words, it would work with a series of photos with some movement between them, but not with google maps since they are static.

      • Maybe not with the Google maps we have access to, but I'm sure Google has more than one image of the same buildings. (which I assume they are using for Google Earth 3D)

  • I'm in a place with no good aerial photography and have considered doing it myself and adding the streets to openstreetmap. What's a good known system to get started with this?

    • I'm in a place with no good aerial photography and have considered doing it myself and adding the streets to openstreetmap. What's a good known system to get started with this?

      You might check out TFA. They have a system to sell you (no prices).Draganfly [draganfly.com] will sell you one for about 10 grand. A quick consultation with Google gives prices from $400 up. I looked into this a while back for pretty much the same reason. A decent system that could get high enough to get over trees / hills / etc. and hold a big enough camera to get reasonable resolution would be closer to the 10K than $400. If you wanted to make a significant fraction of it yourself (like the camera mounts) you could

      • You might check out TFA. They have a system to sell you (no prices).

        CHF 10'000 [sensefly.com] ~= USD 12'000. Cool little thingy, though. [sensefly.com]

        • That just feels expensive. I have an electric RC helicopter that cost me less than $2k for the heli, radio and all the tools needed to get it to fly. I could easily attach a camera to it (even video) and stream 10+ minutes (battery life) of photo/video for a nominal added cost.

          I'm assuming that cost is quite a bit on the software/R&D side of things.

          • I agree it feels expensive. It's less than the USD 15'500 the similar Boomerang [attopiloti...tional.com] costs, though.
            I guess most of the cost is in R&D and software as you say, and most of the value in it working out-of-the box, autonomously. It's amazing how it looks in that video: Seems the guy just throws a piece of black cardboard in the air and the thing starts flying along a defined route. The waypoints can even be changed in mid-air.

            If I had more free time, I would seriously consider this as a hobby. Open source softwa

      • I imagine that an anchored weather balloon or five acting as your lift with a basket underneath would work well. Seems cheap enough, esp since you don't need to go superultralight on your hardware (just very light)

    • RTFA: [smartertechnology.com]
      "Don't have a ready UAV? EPFL has spun off another startup that makes an inexpensive drone. Called the senseFly, "

      senseFly.com [sensefly.com] is selling something they call the swinglet CAM, a cheap looking Toys-R-Us RC plane [sensefly.com] that the want an astounding $10,000 for [searchthenetnow.com] (so much for "on the cheap" like the article said!).

      I recommend this guy and his $100 camera. [rcplanecams.com] The results speak for themselves [rcplanecams.com]

      If you really want to go cheap pick up a 808 keychain camera [youtube.com]. Takes excellent video, it's as small as a car keyless e
      • that guy is no longer in business.

        I've seen video from a gopro on top of some planes, lots of examples on youtube. A couple friends have rigs.

  • This would be great to have for flight simulators.
  • by Jane Q. Public ( 1010737 ) on Sunday May 22, 2011 @05:50PM (#36212230)
    Several times I have seriously considered getting into doing this kind of business: aerial photos and surveys. I thought it would be both profitable and fun. And it probably is. But their "cheap" aerial reconnaissance drone costs almost $12,000.00 U.S. How much aerial surveying would it take to pay that back? A lot, I think.
    • Have you researched market opportunities? I can't imagine aerial surveys coming cheap, there must be at least several hundred dollars per deployment in there I should think

      There seem to be interesting things going on in the DIY/hobbyist sector as well. I could think of a whole range of services one could offer and for which these would be perfectly adequate.

  • with all the talk of aerial drones and such im wondering how long it will be before you can buy an mini antiaircraft battery to put in your backyard.

    • I'm hoping for one small enough that I can further postpone cleaning my garage, via better mapping of the rakes, bicycles, and recycling bin placement.
    • And I always thought a small sized AA battery was a AAA battery! I know: miniaturization and everything, but I think you might need more juice than that to power your spy drone.
      • Anti Anti Aircraft Gun?

      • by adolf ( 21054 )

        There are also AAAA batteries. A common 9V alkaline battery is full of them.

        For whatever that's worth.

        (On the end of the spectrum, F-sized batteries are also reasonably common, and are one size bigger than D.)

        • You'd need a pretty big cup to fit an F sized... oh, battery, whoops!
          • You'd need a pretty big cup to fit an F sized...

            In the days before LED lamps, the Fx2 (two 'F' cells, packaged) was very popular in the UK. Let the sniggering begin.

            They're still available [hitchnhike.co.uk]. Then someone thought to try making a triple pack [hitchnhike.co.uk] (inspired by Eccentrica Gallumbits) and later a pentuple [hitchnhike.co.uk] pack. (Actually, they've stopped selling those now, due to the discontinued consumer supply of NiCad cells, not because the design is "wrong" per se. They use D-cells instead. BFD.)

    • It's called a "shotgun".
  • It's hard to tell for sure because of the depth-of-field effect applied to the rendering (which I imagine was the reason they used that effect), but it seems like the quality of the model drops off dramatically the further you get from looking straight down. In the few unblurred street-level frames I caught of the high-resolution video, it's almost as though I'm looking at a clay model of the city which has had really high-quality texturemaps applied to it.

    It's still pretty cool, but I don't think anyone is

    • On their website, the sample photos from their drone are basically straight down. That would severely limit the quality of the imagery on vertical walls. I'm curious why they don't angle the camera at, say, 45*. That should allow good coverage of roofs and walls.

  • It will all boil down to what render quality costing factors are involved.
  • This could be used for:
    - games (both FPS and tactical)
    - better civil engineering planning (seeing how a project will affect its surroundings)
    - archaelogy (understanding old battlefields)
    - current military (understanding new battlefields)
    - terrorism (see "current military")
    - tourism (find hotels/other with good location)

    What other uses can we think of?
    (I think it's pretty nifty. :)
  • The OBJ format sample model on the Pix4D site doesn't play nice with nice with Alias [iu.edu]. Having to tweak either the OBJ data or the viewer to get a firsthand view of the model tells me this service is not quite ready for prime time, and likely in need of significant funding to get to production quality.
  • In Capitalist west smart RC fans thank you.
    In Tribal Region smart freedom fighters read up on 3d thanks to you.
    In the past you needed the CIA, sat images of Soviet bases, now you need a toy plane, camera, a laptop and code.
  • If you live in the US just go to government servers and download high quality 3d terrain models in several different formats. Used them frequently in engineering. We did have to resort to flying our own photogrammetry when tolerances were tight, but that was rare.
  • So now we can make a 3D model of our neighbour in her bathing suit? Sweet!
  • How low can you go and still get a decent 3-D representation? And what kind of resolution? Could this be used for surveying difficult terrain?

  • The price is nearly $10,000 US.

    • It's nearly ten thousand CHF, which is about 11,500 USD. I almost never deal in either currency, so I had no idea the franc had surpassed the dollar.

  • I have overlaping air photos of about 200 square miles of BC mountains. A simple spectrometer yields an exagerated 3D image. How could I emulate this with software?

Per buck you get more computing action with the small computer. -- R.W. Hamming

Working...