Quick answer
Plan your video as a series of short shots. Use image-to-video with a strong first frame for more control, add reference images for consistency, write prompts that describe subject, action and camera movement, and use editing tools like Aleph to change existing footage. Expect to generate several variations and finish the edit in a video editor.
What Runway offers
According to Runway's public changelog, features and models have included:
- Gen-4.5: introduced in December 2025 for paid plans as its flagship video model, with first-frame image input added later that month.
- Gen-4 image and References: generate images using reference pictures for more consistent characters, objects and styles.
- Aleph: a video-to-video model available on paid plans since July 2025 for editing and transforming clips.
- Act-Two: performance-capture style animation, transferring a recorded performance onto a character.
- Chat mode and Workflows: a conversational interface and node-based workflows that can be published as apps for your workspace.
- Audio tools: text to speech, sound effects and speech to speech.
- Third-party models: in January 2026 Runway announced additional partner models available inside the platform.
- API: access for developers to some models.
Model availability differs by plan, and credits are consumed differently by different models.
Prompting for video
Describe shot, subject, action, camera and mood
Slow dolly-in on a ceramic mug of tea on a wooden windowsill, morning light, steam rising, soft rain outside the window, shallow depth of field, calm and cozy mood.
Image-to-video for control
Start from a still image that already has the look you want, then prompt only the motion:
The camera slowly pushes in. The steam curls upward. Raindrops slide down the glass. No other movement.
Tips
- One main action per clip. Complex choreography tends to fail.
- Use camera language: push in, pan left, static shot, low angle.
- Say what should not move if you need stability.
- Iterate: change one thing at a time so you learn what affects the result.
Workflows
Make a 30-second concept video
- Write a shot list of six to eight shots, each a few seconds.
- Create key frames as images first, using references to keep characters and style consistent.
- Animate each frame with image-to-video.
- Generate several takes per shot and pick the best.
- Add voice, music and sound effects, then assemble in an editor.
Fix or restyle existing footage
- Upload a short clip.
- Use a video-to-video editing tool such as Aleph with a specific instruction: "Remove the parked car on the left" or "Change the lighting to dusk."
- Compare against the original frame by frame for artifacts.
- Keep the original safe and use the edited version only if it holds up.
Turn a still image into a social clip
- Choose a strong image with clear subject and space.
- Prompt a subtle camera move or ambient motion.
- Crop to the target aspect ratio and add captions in your editor.
Quality checks
- Consistency: check faces, clothing, logos and objects across shots; AI video can drift.
- Physics and anatomy: watch hands, limbs, reflections, text and moving liquids.
- Text: add titles and captions in an editor instead of generating them.
- Audio sync: confirm that voice and effects match action.
- Accuracy: do not use generated footage to depict real events or products in a misleading way.
Rights, consent and disclosure
- Get permission before using anyone's likeness or voice.
- Avoid prompts that imitate copyrighted characters, brands or living artists' distinctive work.
- Read Runway's current terms on commercial use for your plan.
- Disclose AI use where a platform, client, or law expects it.
Common mistakes
- Trying to generate a whole story in a single prompt.
- Writing vague prompts with no camera or motion direction.
- Using up credits on exploratory clips with premium models before testing cheaper options.
- Skipping final editing, sound design and color matching.
- Using realistic fake footage in ways that could deceive.
Limitations and alternatives
- Clips are short and multi-shot consistency is still difficult.
- Results vary run to run, so budget for several attempts.
- Precise, frame-accurate control is easier in traditional editing and VFX software.
- For simple trimming, captions and cuts, a standard video editor is faster and cheaper.
FAQ
What is Runway used for?
Generating video and images from text or images and editing footage with AI.
What is Aleph?
Runway's video-to-video model for adding, removing or replacing elements, changing lighting or style, and generating new angles.
Can Runway make a long film in one go?
Not reliably. Most projects are built from many short clips.
Do I need permission to use people's faces or voices?
Yes. Get consent and follow legal and platform rules.