Runway Gen-4

规格参数

定价

Runway Gen-4 is the fourth-generation image-to-video model from Runway, the company that pioneered AI video tools in professional post-production workflows. Where earlier Gen models introduced the world to AI video, Gen-4 refines Runway's approach with a landmark feature: multi-frame reference (MFR) technology that keeps characters, props, and environments visually consistent across the entire duration of a clip — solving one of the most persistent challenges in AI video generation.

能力说明

Gen-4's multi-frame reference system allows you to provide multiple reference images — for example, front and side portraits of a character — and the model will maintain that character's appearance throughout all 16 seconds of video, regardless of pose changes or camera angles. This enables a new class of use cases: consistent character animation for short films, branded mascot videos, and advertising content where visual identity must remain locked. Gen-4 also supports the anamorphic 2.35:1 aspect ratio, the cinematic standard for theatrical widescreen, positioning it as the preferred tool for filmmakers and commercial directors.

最佳实践

To maximise consistency in Gen-4, provide clean reference images with the subject clearly isolated from a neutral background — ideally shot under even, diffuse lighting. For character references, include both a front-facing and a three-quarter view to give the model enough 3D information to maintain appearance across camera angles. In your motion prompt, reference the visual style of specific film genres or directors ("slow dolly push, warm tungsten interior lighting reminiscent of late 1970s American cinema") to activate Gen-4's understanding of cinematographic conventions.