Runway Research

Runway Research is the AI research arm of Runway, the creative AI company, dedicated to advancing the state of the art in generative AI for video, image, and multimodal applications. The research division has published significant work on diffusion models, video generation, and creative AI tools that has influenced the broader AI research community and powers Runway's creative platform products.

The division's most significant contribution has been its work on latent diffusion models, which formed the technical foundation for Stable Diffusion — one of the most widely deployed open-source image generation models. Runway researchers collaborated with academic partners to develop and publish this influential architecture that democratized high-quality image generation.

Runway Research's video generation work has produced the Gen-1 and Gen-2 models that power Runway's commercial video generation products. These models have consistently represented the frontier of commercial video generation quality, enabling realistic video synthesis from text descriptions and enabling style transfer and editing capabilities not available elsewhere.

The research team's work on motion estimation, temporal consistency in video generation, and multi-frame prediction addresses fundamental challenges in video generation that distinguish it from image generation. Maintaining consistent appearance of objects and characters across frames while enabling natural motion is technically challenging, and Runway's research has made significant advances in this area.

As AI video generation continues to advance rapidly, Runway Research represents one of the most active and impactful research programs focused specifically on creative AI applications for video. Their work shapes both the capabilities available to creators and the broader trajectory of generative AI development.