Runway Unveils Characters: Real-Time Conversational Avatars from a Single Image
tlakomy · x · 2026-07-23
At SIGGRAPH, Runway showcased its new Characters feature. The technology can generate a real-time conversational digital avatar from just a single reference image in about 15 seconds.
The generated characters support real-time interactive conversations at 24fps, featuring natural expressions and lip-sync. The feature is compatible with various image styles, transforming photorealistic selfies, cartoon brand mascots, or abstract creatures into living conversational agents. It is now accessible via Runway's web app and API.
Related event: Runway Unveils Real-Time Interactive Characters from a Single Image(2 posts)→
More from Multimodal
- User says image models ignore reference photos unless outfits are spelled out — Ok-Star-6755 · 2026-07-23
- Seedance 2.0 gets Japanese dance videos right, with a funny ending — anthara_ai · 2026-07-23
- Storyboard-first workflow makes AI dance videos far more consistent — anthara_ai · 2026-07-23
- Forge Neo gets an Image2Prompt tab for turning images into prompts locally — adeliogentile · 2026-07-23
- Google Flow adds direct image annotations to steer video motion with Omni Flash — Mr_AllenT · 2026-07-23
- A multimodal “task” prompt for agents bundles voice, screen, marks, and text — omarsar0 · 2026-07-23