Runway has just introduced a groundbreaking update with its Gen-4 model—References.
This powerful new feature allows creators to generate consistent, high-quality AI-generated images using up to three reference images.
Whether you're a filmmaker, digital artist, or content creator, this update changes the game for storytelling, character development, and visual continuity.
In this post, you’ll learn exactly how to use the References feature, step-by-step.
We’ll also explore real-world use cases and pro tips to help you get the most out of it—even if you’re just getting started with Runway AI.
References lets you upload up to three images—like a person, a location, or an object—and generate new visuals that maintain consistency with those inputs.
This solves one of the biggest challenges in AI content creation: preserving identity and style across multiple outputs.
Want the same character to appear in different scenes?
Want to create a cinematic sequence where your protagonist travels from a moody alley to a vibrant marketplace?
Runway References makes this possible.
From concept art and previsualization to social content and full-scale production, References brings an entirely new layer of realism and control to AI visuals.
Log in to your Runway account and open a new image generation session.
Click on the Image tab to start working with visual inputs.
You can upload up to three reference images. These could be:
Drag and drop them directly into the reference panel.
If you're starting with just one reference (like a character), that's perfectly fine too—you can add more as your scene evolves.
Hover over the uploaded reference and click the option to give it a name. For example, name your character “DetectiveJohn” or your location “RooftopAtNight.”
This step allows you to recall the reference later using the @ symbol (e.g., @DetectiveJohn) in any prompt.
Now you can get creative with your prompt. The References feature allows for natural, conversational language. For example:
Show me @DetectiveJohn standing on @RooftopAtNight with a neon sign glowing in the background.
Or try something more specific:
Show a tight close-up of @DetectiveJohn leaning against the back wall of @RooftopAtNight with warm light illuminating the side of his face.
If you like the result, you can reuse it by dragging the output image back into the references panel.
This lets you keep evolving the visual story with continuity.
Want a different camera angle? Just prompt it:
Show me a bird's-eye view of @DetectiveJohn walking through @RooftopAtNight.
If you didn’t name your references, you can refer to them as “image one,” “image two,” or “image three,” based on the order you uploaded them.
You can also reference specific parts of the image, such as:
Plan scenes in advance with consistent characters and settings. Great for pre-visualization and pitching ideas.
Design characters and environments that stay coherent across multiple drafts or story beats.
Create series-based content with a consistent look. Perfect for storytelling reels, character intros, and brand visuals.
Produce cohesive image collections that share a defined aesthetic and thematic foundation.
Currently, the References feature is optimized for characters and locations.
However, Runway has confirmed that support for object preservation, styles, and more is in development.
This will open up even more advanced workflows for creators.
The References feature in Runway Gen-4 is one of the most exciting advancements in AI-generated media.
It bridges the gap between raw creativity and real-world usability by giving creators powerful tools to maintain continuity, style, and narrative structure.
Whether you're an absolute beginner or an experienced digital creator, this tool offers a major upgrade to how you approach visual storytelling.
It's time to experiment, build worlds, and bring your vision to life—one consistent image at a time.
Ready to try it yourself?
Head over to runwayml.com and start creating with References today.