Runway's path to Gen-2 began with a video problem that good individual images do not solve: frames can flicker or change content instead of forming a coherent action. In its development retrospective, the company explains that rather than attempting end-to-end video generation from the start, it first solved a simpler version with Gen-1, where an input video determines the output's structure. Gen-2 then tackled text-guided video generation without that structural input, followed by an image-to-video update. The progression changed what a creator had to supply, while keeping consistent motion across frames as a central technical challenge.
