3D Visualization Moves Forward

Chris writes "Showing for the first time at the Society for Information Display (SID) conference in Boston was a three-dimensional display with 100 million volume pixels or "voxels". The Perspecta is a hardware and software combination that projects 3D images inside a 500 mm transparent spherical dome. Images 250 mm in diameter can be seen from a full 360 degrees without goggles, allowing the viewer to walk around the image. It can be used to visualize protein structures and to plan surgical and radiation treatment by locating the exact position of a tumour on an x-ray or mammogram. It could also be used in air traffic control, prototype designing and security scanning of luggage. Perspecta uses Texas Instruments' digital light processor technology and a spinning projection screen, which sweeps the sphere." We've done some previous stories about this globe from Actuality Systems. The trend seems to be toward simulating 3D with high-resolution flat screens, though.

Comments

Do Your Own 3D Modelling

[Anonymous Coward]

http://www.opendx.org

Data Explorer. Formerly from IBM. Runs on Linux, and Windows.

3D protein modelling for the rest of us with only flat-screen monitors.

Aaaaaand, why bother.

[Bowie J. Poag]

Ok, its cool and all.. yeah, being able to project something volumetrically, but is it really _useful_ ? I fail to see how paying $20,000 for a bleeding edge "display sphere" makes more sense than rendering something in stereo, and crossing your eyes, which most visualization packages are capable of doing nowadays anyway.

Where I used to work, we had a number of visualization packages that allowed researchers to view molecules/proteins/DNA sequences in stereo. It was routine, and required no specialized hardware.. You just render two views of the same object, side by side on the screen, with one view taken slightly from the left or to the right of the other. You can manipulate them in realtime, in stereo. Doesnt require glasses.. Just have to cross your eyes. Hell, go visit my site, i've got a couple stereoscopic wallpapers up, and theres nothing stopping me from producing stereoscopic 3D animation in Blender.

:)
Cheers,

Re:Aaaaaand, why bother.

[Easy2RememberNick]

You know that was the first thing I thought when I read the article, was someone who lost vision in an eye, but then as I read it I realized that it's better than most 3D displays because you don't need special glasses. Wasn't the big point, the excitement, the breakthrough of the thing that you don't need special 3D glasses to view the image? So even with one eye you could see an object in the sphere, not depth but still...objects. It's still a pretty cool display.

100,000,000 voxels: not as impressive as it sounds

[segfaultdot]

that's a cube of 464*464*464 pixels. It's a great start, but i'd rather have a Radeon 7000 :)

Re:Glasses suck

[kevinmik]

Displays that require the user to wear glasses aren't what I'd call and adequtes solution

I believe what he was refering to were the screens that actually create a 3d image without glasses. They're really impressive. They do this by having two hi res lcd's, one on top of the other, that have a slight offset in the image between the two. The top one is translucent and the difference in the two pictures creates the illusion of 3d without having to use glasses or anything. It's like you've got your own holographic monitor!

I just can't wait until they come into my college budget price range (yeah, like that's gonna happen :)

Did this about a year ago...

[Uller-RM]

I did a contract coding job similar to this about two years ago - for an exhibit at a tech expo, we rigged up a pair of curved mirrors and a plexiglass semisphere with a hinged hatch. A projector shot a 1024x768 image through the pair of mirrors, producing an image that gave you roughly 270deg FOV horiz and 90deg vertical. Add a joystick and a rudimentary tunnel shooter... :)

My part of it was hacking up the game engine (Virtools' kit) to render from two in-game viewpoints each frame and distorting the image in a third rendering pass so you'd get a correct image on the screen - lots of optimization, since we were rendering at that resolution two years ago when the Geforce2GTS and 1GHz P3s were the height of consumer technology. That, and some other blocks for the scripting language for level transfers and whatnot.

(The engine used to be marketed under the name Nemo, now called just Virtools Dev. Not too impressive graphically by today's standards, but it has the most artist-friendly scripting system I've EVER seen. If they strapped a decent rendering tech onto it and some network code, they'd have an absolutely outstanding project on their hands.)

Re:Did this about a year ago...

[Uller-RM]

After compositing the two camera images in the framebuffer, it did a read from the back buffer into main memory (specifically, into a block in the heap with buffer space above and below to get a power of 2), was uploaded back to the card as a OpenGL texture, and rendered bilinear across a precalculated set of vertices and UV coords (stored in vertex arrays and display lists). If we were running it on a Voodoo, I could have used Glide to do it all on the card, but this was right when they were ditching Glide, and the rendering engine didn't have a Glide plugin :-\

I would have liked to work out an entirely 2D routine for doing it instead of having to do the texturing, but I was under a rather nasty timeframe to do it, and the example code we got from the inventor of the technique took some time to decipher. Why people insist on doing horrible pointer arithmetic instead of [] (which expands to *(x+i) in the preprocessor anyways), I'll never know.

On a Geforce2GTS the fill rate and AGP2x limited us to about 20fps, but on an SGI NT workstation it absolutely FLEW since main memory and video memory were shared. Unfortunately, we had to do 10 kiosks with this, and the budget didn't allow getting an SGI for each one :-\