Runway’s New AI Generates Worlds as You Play
GWM Worlds 2 makes AI-generated video interactive, creating each moment in response to your actions.
Runway unveiled GWM Worlds 2 (opens in a new tab) on September 3, showing AI-generated worlds that respond as people move through them. The model creates the next moments of video and sound from the actions it receives.
One demo starts on a wet city sidewalk. A woman in a green coat approaches a street sweeper, and the two talk as the camera follows her. Instructions describe their movements and dialogue; the model turns them into the scene below.
Playing inside the scene
The next demo puts those instructions on a keyboard and mouse. A player guides a mage through the snow, casts a spell, and opens a portal. Each input gives the model a new action to generate while the scene keeps moving.
Runway calls the instruction format WorldPrompt. It sets out the characters, actions, and camera movements. A separate demo splits control between a player and a director, so one person can move the character while the other changes the surroundings.
According to Runway, GWM Worlds 2 generates 720p video at 24 frames per second with sound. That continuous response lets a person play through the scene as it is being created.
An agent at the controls
Those controls also work with an AI agent. Runway gave one a generated robot in a warehouse and a goal of reaching the red flag, then the blue one. The agent chose the movements, and the model generated the robot following that route.
Access
GWM Worlds 2 remains a research preview. Runway notes that fast camera turns can lose detail and the model can forget earlier parts of a scene. Public pricing and a general release date have not been announced.
Businesses can contact Runway through its research page (opens in a new tab) to discuss access. The demos offer an early look at creating a scene you can keep directing and exploring as it unfolds.
More from Capture & Create.
Browse all stories
Blockworks · 2 min read
Blockworks Turns Your Words Into Builds with AI on Apple Vision Pro
A new AI demo makes a boat, a bicycle picture, and a statue from virtual blocks you can pick up and take apart.
Tripo AI · 4 min read
Tripo Adds 3D Animation From Text Prompts
Creators can generate short actions from prompts and combine up to five motion segments, with skeleton options for Unreal Engine, Unity, and avatar workflows.
Robot simulation · 1 min read
NVIDIA’s Robo Olympics sends simulated robots sprinting, skating, and kickboxing
Sports videos and written prompts guided OpenAI’s Astra in building the events, with NVIDIA’s simulation tools turning motion references into tested robot movements.