Runway Gen-4 Prompts
Runway's generation model for filmmakers — the same character, the same world, shot after shot.
About Runway Gen-4
Runway Gen-4 is the generation model from Runway, a company that has spent years building AI tools specifically for filmmakers and video professionals. It produces both images and video, and its defining idea is reference-driven generation: you give it images of a character, a location, or an object, and it keeps them recognizable across new shots.
That makes its core strength continuity — the thing that has separated AI clips from actual filmmaking. Where most models reinvent your protagonist every generation, Gen-4 holds a character's identity steady across angles, lighting setups, and scenes, letting you build a consistent world rather than a pile of disconnected shots. For storyboards, short films, and episodic content, that changes what is possible.
Working this way is a different craft: the reference carries the look, and the prompt directs the scene. The gallery below collects tested Gen-4 prompts with their real outputs, so you can study how working prompts split those responsibilities before shooting your own sequence.
How to write Runway Gen-4 prompts
- 1Start from references, not adjectives. Give Gen-4 clear images of your character or location, then keep those references fixed across every shot in the sequence — that is where the consistency comes from.
- 2When a reference carries the look, spend the prompt on direction instead: describe the action, the emotion, and the camera, not the character's appearance the reference already establishes.
- 3Name one camera move per shot — "slow push-in on her face", "handheld follow through the market", "static wide as the storm rolls in". One instruction keeps the motion readable.
- 4Write a mini world bible and reuse its exact phrases: the same character name, the same location wording, the same lighting description across prompts reinforces continuity between shots.
- 5Keep each generation to a single story beat. Chain several short shots into a sequence instead of cramming a scene's worth of events into one prompt.
- 6Carry the grade across cuts with a repeated mood clause — "overcast steel-blue light, muted grade" — so consecutive shots feel like they belong to the same film.
Frequently asked questions
What is Runway Gen-4 best at?
Continuity. It keeps characters, locations, and objects consistent across shots, which makes it a strong choice for filmmakers building sequences — storyboards, short films, music videos — rather than isolated clips.
How does Gen-4 keep characters consistent across shots?
Through reference images. You provide visuals of your character or world, and Gen-4 preserves their identity in new angles, lighting, and scenes. Fix the references for a whole sequence and let the text prompt vary the action and camera.
Can Runway Gen-4 generate images as well as video?
Yes — Gen-4 covers both. A common filmmaker workflow is to explore a scene as still frames first, pick the frame that works, then move to video with the same references and prompt language to keep the world intact.
Where can I try these Runway Gen-4 prompts?
Gen-4 is available in Runway's web platform, alongside the rest of its filmmaking toolset. Copy a prompt from this page, bring your own character or location references, and adapt the camera and action language to your scene.
Related models
MiniMax H3
MiniMax H3 is a powerful multimodal video generation model that creates up to 15-second 2K videos with native stereo sound from text, image, video, and audio inputs.
Seedance 2.0
ByteDance's #1-ranked video model — multi-shot direction with native lip-sync and spatial audio.
Seedream 5
ByteDance's high-fidelity bilingual image model — the best quality-to-cost ratio for high-volume pipelines.
GPT Image 2
OpenAI's top-ranked image model — it plans the layout, reasons, and self-checks before it draws, nailing complex multi-part instructions.