Stories
Slash Boxes
Comments

SoylentNews is people

posted by martyb on Sunday November 15 2015, @08:49AM   Printer-friendly
from the image-up-jpeg-back? dept.

As smartphones become people's primary computers and their primary cameras, there is growing demand for mobile versions of image-processing applications.

Image processing, however, can be computationally intensive and could quickly drain a cellphone's battery. Some mobile applications try to solve this problem by sending image files to a central server, which processes the images and sends them back. But with large images, this introduces significant delays and could incur costs for increased data usage.

At the Siggraph Asia conference last week, researchers from MIT, Stanford University, and Adobe Systems presented a system that, in experiments, reduced the bandwidth consumed by server-based image processing by as much as 98.5 percent, and the power consumption by as much as 85 percent.

The system sends the server a highly compressed version of an image, and the server sends back an even smaller file, which contains simple instructions for modifying the original image.


Original Submission

 
This discussion has been archived. No new comments can be posted.
Display Options Threshold/Breakthrough Mark All as Read Mark All as Unread
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
  • (Score: 2, Interesting) by anubi on Sunday November 15 2015, @09:50AM

    by anubi (2828) on Sunday November 15 2015, @09:50AM (#263606) Journal

    It seems like it ought to be possible to analyze a video clip, thousands of frames shot from varying perspectives, and from that derive models for all objects in that space, filled in as perspective brings those surfaces into the view plane.

    In the event that some surfaces never come into view, extrapolate from symmetry or known similar objects.

    Can you imagine sending the cloud a videoclip of whatever, and having it send back extremely detailed 3D models of each object detected in the clip, including joint vertices, for re-insertion into modelling software? One would have no trouble making fantasy movies with any models one can submit footage of.

    I could see a computer modeling a scene like we do, creating the 3D image in our "mind" from the 2D image stream presented. The more frames one can analyze, the sharper the image can become as we have more data to integrate the noise ( thermal, quantization, and optical focus ) out.

    With computer power getting to be what it is, I do not think there are many things that are beyond feasibility to do anymore. Its gonna drive us all nuts, I suppose. Can you imagine someone take some family footage and one day see a movie of you doing something you know good and well you have never done?

    The ultimate 3D photoshop.

    --
    "Prove all things; hold fast that which is good." [KJV: I Thessalonians 5:21]
    Starting Score:    1  point
    Moderation   +1  
       Interesting=1, Total=1
    Extra 'Interesting' Modifier   0  

    Total Score:   2  
  • (Score: 0) by Anonymous Coward on Sunday November 15 2015, @02:12PM

    by Anonymous Coward on Sunday November 15 2015, @02:12PM (#263646)

    Wow we could like have a Matrix of online connectivity! Couple that with VR and occular implants and the politicians will show us whatever it is they want us to believe and it'll be very real! I assume the modeling software for the cloud version 1.1a will be to generate funding to work on the second a version works on more than just registered voters that use facebook.

    Anyway google has a 2d version of "know good and well you have never done". There are some articles about it. They "improve" pictures, add arms or point where eyes look, or remove people to make an image more pleasing according to some algorythm. One person, I believe, became upset when a number of pictures from a vacation of one year that were taken in the same spot the next year ended up with a dead family member inserted due to that person missing from the later photos (for reasons not clear to the google robots that opted to not leave that idea on the cutting room floor) but in the prior year's photos taken at the same location.

  • (Score: 2) by Zz9zZ on Sunday November 15 2015, @07:59PM

    by Zz9zZ (1348) on Sunday November 15 2015, @07:59PM (#263738)

    I forget where, but some researchers have already done this using a drone. From a stereo camera (I believe, could've been single) they were able to create a 3D model of a few city blocks.

    --
    ~Tilting at windmills~
    • (Score: 1) by anubi on Monday November 16 2015, @05:36AM

      by anubi (2828) on Monday November 16 2015, @05:36AM (#263862) Journal

      Could that have been a Google Earth thing?

      One does not need two cameras to get a stereo image of a static object, as one only has to wait an interval of time on a moving camera to get another perspective view from the same camera.

      Now, things get tricky if there are things moving in respect to each other in the image... think taking pictures of a chorus line of gorgeous ladies twirling every which-a-way... what would it take to analyze the images and provide a 3-D model of each dancer - given the camera has seen where all the joints and muscles are - and be able to extrapolate any pose? To me, it almost seems impossible, but then many things that seemed impossible to me a few years ago are now commonplace.

      I believe the world of visual reality as we knew it will soon change radically. We are already at the point of 2D photorealistic imaging - where computer-assisted art can be ( for me, anyway ) indistinguishable from a proof taken from a camera.

      I believe the day is soon coming that anything we see on a screen could easily be nothing more than a figment of an artist's imagination. They are already damned good at it. I get the idea that in ten years, even the newscaster on TV will be an entity existing only in the memory core of a computer somewhere - yet visually indistinguishable from a living breathing human. Not to even mention ads - where the pitchman delivers the exact presentation the client wants ( and since that person never existed - he does not have to risk anyone meeting him on the street which bought one of his pitched products and was disappointed ).

      --
      "Prove all things; hold fast that which is good." [KJV: I Thessalonians 5:21]