Skip to main content
Runway Gen-4.5 Prompting Guide

Runway Gen-4.5 video.
Cinematic motion, six aspect ratios.

Runway Gen-4.5 is the latest from Runway's flagship video line — cinematic motion, strong camera language understanding, six aspect ratios from cinematic 21:9 to vertical 9:16, and 5 or 10 second clips at 3 credits per second.

Overview

Runway Gen-4.5 is the latest in Runway's flagship video model line. It produces consistently cinematic output with strong understanding of camera moves, lighting language, and atmospheric scenes. Think of it as the model to reach for when you want the result to look directed.

Two clip durations (5 or 10 seconds), six aspect ratios spanning cinematic 21:9 through vertical 9:16, and both text-to-video and image-to-video modes. Pricing is 3 credits per second — 15 credits for 5s, 30 credits for 10s — same across modes and aspects.

5 · 10s

Clip durations

6

Aspect ratios

3 cr/s

Flat pricing

Key Features

Looks directed out of the box

Cinematic by Default

Runway Gen-4.5 produces cinematic results from sparse prompts — natural composition, smooth motion arcs, atmospheric lighting. Other models often need explicit cinematography language to hit this look; Gen-4.5 hits it by default.

Drone push, tracking shot, dolly in

Reads Camera Language Literally

Name the camera move — "slow drone push", "low tracking shot", "handheld follow", "static locked-off wide" — and Gen-4.5 executes it. Picks up specific filmmaking vocabulary (anamorphic, shallow DOF, golden hour) more cleanly than most consumer video models.

Animate any starting frame

Image-to-Video Mode

Pass a starting image and Gen-4.5 writes motion onto it. The source frame anchors composition, character identity, and color palette while your prompt describes the action and camera move. Great for animating illustrations, product stills, and storyboard frames.

21:9 · 16:9 · 4:3 · 1:1 · 3:4 · 9:16

Six Aspect Ratios

From cinematic 21:9 widescreen through square 1:1 social spot to vertical 9:16 reels — every common output frame covered in a single model. Pick the ratio for your downstream channel without upscale/crop loss.

Example Videos

Each example shows the exact prompt that produced the result. Copy any prompt with one click.

Aerial Drone Push

16:9 · 5s · 15 credits

A slow drone push over a misty Pacific Northwest forest at dawn, dense Douglas firs piercing low fog, soft golden sunlight breaking through the canopy, painterly cinematic, no camera shake

"Slow drone push" is one of Runway's strongest camera moves — Gen-4.5 holds steady altitude and smooth forward motion. Add "no camera shake" or "stabilized" to lock down handheld jitter if the model defaults to it.

Cinematic Character Walk

16:9 · 5s · 15 credits

A young woman in a wool coat walks slowly down a rain-soaked Tokyo backstreet at night, neon reflections in puddles, anamorphic lens flare, shallow depth of field, the camera tracks alongside her at hip height

Tracking shots at "hip height" or "shoulder height" land beautifully. Pair the subject's motion (walks slowly) with the camera's motion (tracks alongside) so Gen-4.5 knows how the two relate. Anamorphic flare is a particular strength.

Product Macro

16:9 · 5s · 15 credits

Macro slow-motion of a single ice cube splashing into a crystal tumbler of amber whiskey on a polished bar, droplets suspended mid-air, warm spotlight catching the glass facets, deep blacks, premium beverage commercial aesthetic

Macro slow-motion product shots reward Runway's strong physics — droplet motion, glass refraction, suspended particles. "Premium beverage commercial" is a useful aesthetic anchor that nudges the model toward studio lighting and clean negative space.

Vertical Reel

9:16 · 5s · 15 credits

A surfer rides a glassy turquoise wave at sunrise, vertical handheld follow shot, water spray catching golden light, fluid camera motion, cinematic color grade

9:16 vertical reads naturally on Runway. Name the camera type explicitly ("handheld follow shot") — vertical handheld is what social-native footage looks like. "Cinematic color grade" elevates raw documentary to polished content.

Prompting Tips

Name one dominant camera move

"Slow drone push", "low tracking shot", "static wide", "handheld follow at hip height" — pick ONE and commit. Runway executes a single move beautifully; combining two or three usually produces compromised motion.

5 seconds is plenty for one beat

5-second clips ($15) are ideal for single-beat moments — one action, one camera move, one mood. Pay for 10 seconds only when you genuinely need a multi-beat sequence or a longer atmospheric hold.

Anchor with cinematography language

"Anamorphic lens flare", "shallow depth of field", "golden hour", "Kodak Portra", "ARRI Alexa". Runway reads these as literal aesthetic targets. Most useful in shorter prompts where vibe matters more than action.

Use 1:1 only with image-to-video

Runway's text-to-video mode rejects 1:1 (square) — only 16:9 and 9:16 are supported for T2V. For square output, attach a starting image and use image-to-video mode. The error is unambiguous if you forget.

Pick aspect by downstream channel

21:9 for cinematic, hero, widescreen edits. 16:9 for standard YouTube / horizontal social. 9:16 for Reels / TikTok / Shorts. 1:1 for square social (I2V only). 4:3 and 3:4 for traditional/retro feel.

Image-to-video for character continuity

When you need a specific character or product to stay consistent, attach a starting image. Gen-4.5 anchors composition and identity to the source frame, then writes the prompted motion onto it.

Settings Reference

SettingValuesNotes
ModeText-to-video · Image-to-videoI2V requires a starting frame. T2V composes from text alone but excludes the 1:1 aspect.
Duration5s · 10sFlat 3 credits per second — 15 cr for 5s, 30 cr for 10s.
Aspect ratio21:9 · 16:9 · 4:3 · 1:1 · 3:4 · 9:161:1 requires image-to-video mode.
SeedInteger (optional)Pin for reproducibility. Omit for varied output across re-runs.
Pricing3 credits per secondSame across modes and aspects. No premium tier for I2V.
Source for I2VPass image as starting frameAnchors composition, identity, and palette. Prompt then describes motion.

FAQ

Runway leads on cinematic feel out of the box and reads cinematography vocabulary literally — best when you want a 'directed' look from sparse prompts. Veo 3 is stronger on native audio and dialogue. Kling 2.6 Pro has audio + more aspect range. WAN 2.7 is the most cost-efficient. Pick by what matters most for your shot.