Slashdot Mirror
3D TV For The Masses?
scubacuda writes: "Technology Review has an article on new software that could make 3D television a reality. Previously encumbered by an expensive process that takes up to nine cameras per scene, a company called DDD now takes existing 2D film and creates a "depth map" for each frame. A TV that can handle this sort of software rendering currently costs $25K, but DDD estimates that in a few years, a 3D TV could only cost only 20% more than its 2D counterpart."
Having spent the last 8 years developing content preparation technology for 3D presentation systems, I'd like to add my 2 cents.
The problem with 3D TV, apart from all the viewing paraphenalia, is that it's not wide enough. Even in Imax 3D, with a field of view approaching 80 degrees, directors have problems composing shots that fit in the "viewable pyramid" formed by the viewer's nose and the four corners of the screen. (It's fairly well established that anything in 3D projection that clips this pyramid destroys the illusion of 3D, because one eye view clips before the other, causing the audience to be subconsiously disturbed in their viewing). In any case, the 3D effect only operates within this pyramid.
This company has been pumped on and off for some time on various message boards that cover 3D - especially Imax boards. AFAIC, maybe their technology will do well on good 3D presentation systems, but TV-sized screens just won't cut it - all the tests I've seen of 3D on a TV are pretty much limited to novelty value.
/stillconfused
No. Well...maybe. Actually, yes. It really just depends.
I saw 3D lenticular displays exhibiting DDD's work at SIGGRAPH 2001 and SID (Society for Information Displays) 2002.
I was not that impressed. Basically you see a large moving stereogram but the stereo separation is no where near as good as your average ViewMaster. Most of the time it doesn't look 3D at all. And your depth of field is very limited when it is 3D.
I talked to them for a while about how their technology works. Basically, they attempt to interpolate around the edges of objects in the foreground. Sometimes they can't and that limits how close to the viewer an object can appear. For example, imagine a camera moving by a tree only a few feet in front of it and with a complex landscape in the background. Around the edges of the tree there should be new landscape that one eye would see and the other would not. There is no way to recover this image data because in the original film it was blocked by the tree. Thin objects, like wires or poles or window frames are also especially difficult. Most of the time it requires an artist to do some hand tuning of the images.
I don't see what is special about their technique. Even if they do have some novel ideas for getting 3D out of 2D, I don't see how the data would be useful considering how bad 3D lenticular displays look - eg. limited depth of field, incorrect focal length for objects at different depths, very limited viewing angle.
Video game manufacturers have yet to realize how difficult it is for those with even slight red/green colorblindness to tell the difference between saturated yellow and saturated green. Super Puzzle Fighter is a perfect example.
Web designers are also bad. You know how often I've seen red text in front of a greenish background? Or cyan in front of white?
Even people like me who didn't realize that they had any color defiency until they started using computer and playing video games run into trouble when color is used to convey important information in the digital world.
3D is the same way. There are a sizable percentage of people who are "monocular" and thus are unable to use goggle or lenticular based 3D solutions.
Thats why we should all use:
www.actuality-systems.com
The idea that so many people have missed is that 3D for its own sake is just a gee-whiz widget.
:)
What they need to be aiming for is immersion.
Immersion is the experience of being inside an environment rather than being an external viewer.
You can get the immersive effect in 2D, and in fact this has been done with great effect by Disney on some of their rides.
What factors influence the experience of immersion? Foremost is a wide viewing angle; this is where most 3D simulators fail. You must see 'reality' everywhere, not floating in box in front of your face. Also very important is proper audio cueing, as much of a human's experience of spatial orientation is from subtle echoes and pitch changes. Other things that I think add more to the experience than 3D is view tracking, an engaging plot line and breeze control. Also, odor control, as in it cant be stinky
Porn tends to be the one of the first industries to pick up on new technology (video, multiple camera angles in DVDs etc). They are looking for anything that gives them an edge in a very competitive market. Admittedly at $1M/feature it's probably too high, but if the price does drop, expect it to appear in porn well before it arrives in a holywood blockbuster.