Slashdot Mirror

← Back to Stories (view on slashdot.org)

Matrix-Like VR Coming in the Near Future?

Posted by Zonk on from the can't-see-anything-bad-happening-there. dept.

Anonymongoose writes "A researcher at Brookhaven National Lab reckons it could be just a few years before computers can pass through the uncanny valley. The article refers to this as a 'Graphics Turing Test': 'a computer can be considered intelligent if it can create an artificial world capable of fooling a person into believing it is the real thing.' Michael McGuigan has been performing some interesting experiments using Brookhaven's Blue Gene/L supercomputer and has shown that it can produce realistic lighting effects in real time. McGuigan's original research paper (pdf) is available online."

2 of 249 comments (clear)

  1. Future of Video Games by jfroot · · Score: 5, Interesting

    This is something that has always made me wonder. When computer graphics reach the point where you cannot readily tell if the image you are seeing is real or synthetic, how will this affect video game violence?

    Can you imaging Grant Theft Auto X with full realistic imaging? How would that affect someone when they go beat a whore to death with a baseball bat and the mind cannot as easily dismiss the disturbing imagery as virtual.

  2. Re:Fast Computers aren't enough for realistic imag by Eth1csGrad1ent · · Score: 5, Interesting

    It's going to take a long, long time before you have the algorithms in place that can simulate, animate, and render a realistic person. Animate? Yes... that is still a ways off, though even The Matrix had some pretty convincing versions of Neo.

    But as for still frames and modelling, we're getting there:

    Sexy Girl - http://forums.cgsociety.org/showthread.php?f=121&t=532817
    Tattoo Guy - http://forums.cgsociety.org/showthread.php?f=121&t=550192
    The Artist - http://forums.cgsociety.org/showthread.php?f=121&t=472843

    As for realtime photorealistic animation though, we're a long, long way from there. Lighting is one hurdle, the bigger hurdle is content. Models, textures, rigs... forget rendering, all of this takes a lot of time to BUILD for a photoreal environment.

    Its one thing to come up with a realistic model and scene for a photo-realistic still frame. Its another, to rig all of those models so that they can interact with each other in a pre-determined way. Its something altogether entirely different to do this in real time without predetermined paths and choreographed actions, and modelling all viewable elements based upon the degree of movement that a user has within the space.

    This is very much highlighted in the differences between high-poly count models (for detailed stills) and low-poly models (used for 3D games). The "art" for immersive environments like simulated 3D gaming (fps, racing sims etc) is to come up with a convincing representation of a real world object with the lowest poly count possible.

    Currently the difference between these polycounts is massive.