AI

The shift from physical surfaces to generative video research

Generative video research provides professional tools through a single API to integrate AI into visual effects pipelines. Runway Ruby and Act-Two enable production-ready shots by converting videos to 16-bit EXR sequences and capturing realistic facial expressions.

The shift from physical surfaces to generative video research

Aviation foundations and the digital transition

A runway is an elongated, rectangular surface designed for the landing and takeoff of an aircraft. These surfaces consist of asphalt, concrete, or a mixture of both. Pilots consider the prevailing wind direction when determining runway orientation to mitigate crosswind operations. In 1928, Ford Airfield became the first U.S. airport to have concrete runways. I see how the physical requirements of takeoff and landing differ from the digital needs of media production. The aviation industry once relied on circular groomed and drained turf called all-over fields. This technology became obsolete in the United States during the 1920s. Aviation pioneers like Orville Wright underlined the need for distinctly marked and carefully prepared landing places, though he noted that preparing the surface of reasonably flat ground is an expensive undertaking. Runway lengths are now commonly given in meters worldwide, except in North America where feet are commonly used. Runways are named by a number between 01 and 36, which is generally the magnetic azimuth of the heading in decadegrees.

AI models and API integration

The research in generative video provides specific tools through a single API. The platform provides access to frontier models from Runway and other labs the day they release. The API allows users to select different model identifiers on the same endpoint or use the Model Router to select a model automatically based on quality, latency, or cost. This router picks the best model for every request. Users can set a preference for quality, latency, cost, or speed. The platform provides full visibility into cost per model, per endpoint, and per team. The Runway Dev MCP connects your account to your coding agent so it can build, debug, and maintain the integration without leaving the editor. Users can create their own pipelines by combining multiple models, modalities, and tasks. The platform includes models like Google Gemini Omni 1.2 Flash, which can generate, edit, or extend videos with image and video references, keyframes, and native audio.

Capability Model Type Specific Detail
Text/Image to Video Runway’s state-of-the-art High-quality generation
Motion Capture Runway’s next-gen Captures movement
Image to Video Runway’s fastest High-speed generation
Video Conversion Runway Ruby 16-bit EXR or 10/12-bit ProRes

The API includes models for image, video, audio, and real-time interaction. You can access these models through a single integration. The platform provides prompting and workflow expertise. One model generates video of a character speaking from a text script or an audio file, powered by GWM-1. Kling 4.0 at 720p or 1080p allows users to generate from text, first, middle, and last frames, or edit a video.

Professional visual effects pipelines

The toolset helps teams integrate generative AI into visual effects pipelines. This technology produces everything from high fidelity quick-turnaround previs to "production-ready shots". These tools help teams shrink timelines and stretch budgets. Runway Ruby converts videos into 16-bit EXR sequences or 10-bit and 12-bit ProRes and HEVC files. The training covers AI video creation for movies, games, advertising, or animation. Users can add voice and movement to creations. Act-Two captures facial expressions and body gestures. The tools allow for the generation of speech and the creation of lip-synced videos. The system generates speech, sound effects, or rich audio scenes with cloned voices. Users can master camera movements, angles, and cinematic techniques. Magnific Precision Upscaler V2 provides image upscaling up to 16x with sharpness and grain controls. ElevenLabs models can translate audio to up to 29 other languages. Users can also use prompt expansion for text to video tasks. How will these tools change the speed of traditional film sets?