Wow. Pretty nerdy sounding, but this technology is quite impressive, and could be incredibly useful for aspect ratio conversion, object removal, rescaling, etc.

The object removal is very very impressive. I can think of a bunch of interesting uses for this, especially if you artificially mess with the weights for effect only. Id love to see this on a sequence of frames.

via reddit

Update : here is Dr Ariel Shamir’s homepage, with PDF and higher quality Quicktime.

3D goodness means getting cozy with your local graphics API — and getting ready to nerd out in a big way. OpenGL continues to progress with a major overhaul. It’s a way off, but you’ve still got lots of eye candy with OpenGL 2.1. So … if you’re not Electronic Arts or Bungie, is there really any reason to use DirectX?

With the release of Windows Vista, we’ve been hearing a lot about DirectX, Microsoft’s Windows-only API for accessing graphics hardware. Of course, most of what you’ve been hearing is Windows gaming lovers complaining because they have to upgrade to Vista just to get DirectX 10 — and they take a compatibility and performance hit for many existing games as a result. (The latter isn’t DirectX 10’s fault; it’s a side effect of a new driver and display model in Vista itself, which impacts OpenGL and DX9, as well.) So what’s going on in the OpenGL camp? At SIGGRAPH, OpenGL 3 was announced. The full spec isn’t available yet, and actual OpenGL 3 hardware will be some ways off, but the future looks bright. In a presentation on the new OpenGL, NVIDIA’s Michael Gold pointed to these major hallmarks:

• Getting “back to the bare metal” for performance. This includes cutting back on overhead, streamlining the API, and actually revamping the object model in a way that should boost raw speed.
• Simpler, more efficient application development.
• Simpler driver development.

So that all sounds good. The object model appears to be the major change, with new object meta-classes that make it easier and more efficient to, well, make stuff. Good luck deciphering this at this point (I expect it’ll be easier once the real spec is out), but here’s more on the announcement, with slides:

OpenGL 3.0 Birds of a Feather at SIGGRAPH
PDF with slides, via NVIDIA’s Michael Gold

Us visualists, of course, can leave most of this to developers and hardware makers. What’s nice is that when we do want to make things look slick, we have access to a cross-platform 3D API in tools like Processing/Java, Pure Data (via GEM, etc.), and Max/MSP/Jitter.

As it happens, I’ve been looking at both OpenGL and DirectX solutions while putting together tools and frameworks to do new 3D work.

Most of the emphasis on working with multi-touch and alternative controllers has been on our sister site, Create Digital Music. But in a way, visuals are even more demanding of new hardware. After all, musicians have all kinds of hardware that work perfectly for performance (keyboards, knobs, drums, violins, sousaphones, kazoos, and whatnot). But new visual performance media demand something different if they’re to evolve.

Oh yeah, that, and most pro visual apps are kind of a b**** to use with a mouse and aren’t all that much better with a tablet. (Unless you’re somehow discovered the secret and find a Wacom as easy to use as a ballpoint. Please, tell me how.)

That makes this tidbit all the more interesting:

Jazzmutant is proud to have been selected by the Siggraph Emerging Technologies Committee in San Diego to demo a new prototype device for digital imaging involving multi-touch control. This solution will go beyond mere finger-drawing and clearly illustrate a new way to interact and improve productivity with drawing and video editing software. Furthermore, the solution presented will be the very first multi-touch enabled Tablet PC shown to the public.

JazzMutant news

What’s that now? Visual editing on a multi-touch surface? JazzMutant is best known for the creation of the Lemur multi-touch hardware. It wasn’t specifically intended for music, but that’s where it got most attention; you can, incidentally, route its native OSC control to Processing, Max/MSP/Jitter, Pd/GEM, Flash, and so on. But it was pricey (US$2500), and while you could design your own interfaces for it, it wasn’t quite the same as having a computer.

Now we get a one-two punch of tantalizing possibilities: a controller specific to visuals, whatever that may mean, and the possibility of using an actual computer with multi-touch input. I’d love to have that with some of what I’m building with Processing these days for performance. I’m a little more skeptical on the visual hardware side, only because so far that has tended to mean a selection of templates for Lemur-like hardware. But either way, this is promising — we’ll be watching the news out of SIGGRAPH very closely indeed.