Runway News | Bringing Runway Characters to SIGGRAPH's Biggest Stage

Runway presented its Characters technology at SIGGRAPH's Real-Time Live! - one of the most closely watched showcases in computer graphics - marking a notable moment for generative AI in production-adjacent contexts. The demo put a real-time generative character in front of an audience that skews heavily toward researchers, engineers, and technical artists, offering a different kind of pressure test than a product launch or online reveal.

Real-Time Live! has historically been a venue for rendering breakthroughs, game engine demos, and simulation techniques, so the appearance of a generative AI character system in that lineup reflects how quickly the field is shifting. For Runway, it was also an opportunity to get direct feedback from an audience that tends to ask precise, technical questions rather than general ones.

In a writeup from the team, members described what surprised them on the day - both in terms of researcher reactions and the performance of the system itself under live conditions. Real-time generative characters sit at a difficult intersection of latency, visual coherence, and controllability, and demoing that live leaves little room for the kinds of post-processing or retakes that accompany most AI video showcases.

Runway Characters, as a product direction, is focused on creating consistent, controllable digital figures that can be used across generated video - addressing one of the more persistent pain points in AI filmmaking, where maintaining a character's appearance across shots has been technically difficult. The SIGGRAPH appearance suggests Runway is pushing that work further into real-time territory, which would have significant implications for interactive applications, virtual production, and live content creation. The team's notes on what comes next hint at continued development on both the technical and usability sides of the system.