We’re fast moving into the age of realtime, and those of us working with video, graphics and CGI are finally being invited to the party. Over the last year we’ve seen the web move increasingly towards a realtime medium, with the growth of live and mobile streaming video, realtime blogging (see, for instance the P2 theme for WordPress), and the persistent, ambient messaging culture of Twitter & co. stepping center stage.
NVIDIA's Tesla 2070, a supercomputer disguised as a graphics card
Video post-production on the other hand, and 2D & 3D computer graphics production even more so, have long involved a horribly long-winded process. In video editing, titling and post-effects, we spend an unbelievable amount of time bringing in (or generating) footage, pre-rendering, making changes, running RAM previews, encoding, transcoding, the list goes on. Each step along the way breaks the flow between conceiving of an idea, applying it, and seeing the results play back in realtime, as they’ll eventually end up.
In 3D this breakdown of immediate feedback and constant iterative process punctuated by long waits is even more pronounced as the complexity of 3D models, lighting, shading, texturing, physics simulations, and animation all take their toll on a system’s CPU, RAM and to some extent graphics card. This often means that to get any sort of usable results in a remotely realistic time frame, users are heavily reliant on multiple CPUs, render farms, and any number of software licenses. Even then, production and post-production timelines can be challenging and more often than not drawn out by the limitations of the tools at hand.
All of this is changing, and while there are other players starting to make moves, NVIDIA technology seems to be leading the way at the moment with their next-generation Quadro FX and Tesla graphics cards.
The key to the shift is an emphasis on incredibly powerful GPUs taking pressure away from the CPU, allowing giant leaps in render times, and visual feedback of complex systems in near and actual realtime. NVIDIA’s CUDA architecture is already being taken advantage of, and looks like it will increasingly more as 2010 unfolds. Adobe’s CS5 software, for instance, looks to make incredible use of the combination of 64-bit operating systems and CUDA graphics acceleration to make working with video and graphics an entirely different experience to anything we’ve seen before.
Then there’s the next generation of CUDA-powered cards set to make use of the power of their (codename) FERMI architecture. It’s hard to even fathom the impact of what amounts to a supercomputer (with equally impressive RAM capabilities) working alongside the CPU to deliver previously unthinkable graphics processing speed from a single workstation.
As these technologies work their way through the ecosystem, more and more software will take advantage of what they have to offer. We’re already seeing a huge trend emerging towards realtime lighting, rendering, texturing and animation in products like Mental Ray/iray, Vray RT, Arion, 3DS Max Mental Mill integration, Softimage Gigacore, Motionbuilder and several others, and many of these solutions are still largely CPU-dependent. As the next generation graphics cards enter the market and gain support, we should expect to see these early prototypes taken to another level.
So what does all this amount to?
From my point of view, the move towards accelerated and realtime graphics processing means:
- A creative production and post-production workflow more akin to traditional art media than we’ve ever seen. Just as a sculptor, pianist or cinematographer adjusts and interacts with their work instantaneously, so those of us working with visual effects, 3D animation and video post-production will be able to finely tune our projects without endless delays and waits between the smallest changes. That means greater chance for experimentation and greater opportunities to land at the best possible results
- The ability to spend more of our time and computing resources creating than waiting on our systems to see how our attempts to do so have panned out
- The ability for small shops and teams to efficiently work to tighter deadlines, without recourse to gigantic, expensive, hard to maintain render farms; and for larger shops and teams to take on increasingly ambitious projects that budgets might not have previously stretched to
For me personally, as someone producing the multimedia content at WordPress.com, this is all very exciting news. WordPress develops new features and offerings at an incredible pace, which makes for short production turnaround times when it comes to announcement and learning content. As a consequence, the more power available for realtime design decisions, and heavy duty, accelerated rendering, the higher the overall quality of our media output can be.
We’re on the verge of some incredible changes, and I’m looking forward to seeing how they’ll impact on our workflows and content in the months and years ahead.
43.062096
141.354376
