Runway Gen-4.5 Builder

Gen-4.5 shots, Aleph 2.0 edits and Act-Two performances · 32 fields · 6 outputs

Library

Filled only

Subject Type

Action Beat

Emotion

Environment

Time and Weather

Shot Size

Camera Movement

Lens and Depth

Motion Pacing

Lighting

Colour Grade

Generation Mode

Reference Input

World Consistency

Multi Angle Plan

Visual Style

Physics and Detail

Performance Line

Performance Mode

Voice Character

Music and Score

Sound Design

Edit Operation

Edit Target

Preservation Rule

Cleanup Priority

Continuity Check

Runway Model

Runway's current generation model for text to video and image to video.

Credit Strategy

Duration

Aspect and Resolution

Negative Prompt

🔒 Preview mode
Try every field. Unlock Runway Gen-4.5 with Pro to copy and save the prompts.

✨ Enhance with AI
Rewrite the main prompt into 3 polished variations. Pro feature.

01Natural Language Prompt

Slow push in medium close up on a 50mm at shallow depth. A solo character turning to face camera with quiet determination, in a rain slicked city street at golden hour. Cinematic rim light, cinematic teal and orange grade, photorealistic cinematic with accurate real world physics, natural real time pacing. Cinematic orchestral with ambient room tone.

02Main Runway Prompt

SUBJECT: Solo Character - Turning to Face Camera - Quiet Determination SCENE: Rain Slicked City Street during Golden Hour SHOT: Medium Close Up on a 50mm at Shallow Depth CAMERA: Slow Push In at Natural Real Time pacing LIGHTING: Cinematic Rim Light GRADE: Cinematic Teal and Orange STYLE: Photorealistic Cinematic with Accurate Real World MODE: Image to Video using Single Character Reference CONSISTENCY: Character and Location Lock PERFORMANCE: Generated Performance VOICE: Warm Female Alto AUDIO: Cinematic Orchestral with Ambient Room Tone MODEL: Gen-4.5 for 10 Seconds at 16:9 at 1080p AVOID: Character Drift

03Direction Brief

DIRECTION BRIEF - RUNWAY REWARDS DIRECTING OVER DESCRIBING LEAD WITH THE SHOT: Medium Close Up is the first thing in the prompt CAMERA: Slow Push In LENS: 50mm at Shallow Depth PACING: Natural Real Time LIGHT: Cinematic Rim Light described as a physical source, not a mood word GRADE: Cinematic Teal and Orange BLOCKING: the Solo Character performs Turning to Face Camera inside the frame PHYSICS: Accurate Real World ONE MOVE ONLY - a second camera move in the same clip degrades the result COVERAGE PLAN: Same Scene Two Angles AVOID: Character Drift

04Reference and Consistency

REFERENCE AND WORLD CONSISTENCY BRIEF MODE: Image to Video REFERENCE INPUT: Single Character Reference LOCK: Character and Location Lock must hold across every generation in this scene IDENTITY: keep the Solo Character face, wardrobe and proportions identical to the reference LOCATION: keep Rain Slicked City Street geometry and object positions fixed between angles COVERAGE: Same Scene Two Angles generated from the same reference set STYLE LOCK: repeat Photorealistic Cinematic and Cinematic Teal and Orange verbatim in every shot LIGHT LOCK: keep Cinematic Rim Light coming from the same direction in all coverage CHAIN SHOTS: use the last frame of an approved clip as the first frame of the next DRIFT CHECK: Check the Final Second AVOID: Character Drift

05Performance and Audio Brief

MODE: Generated Performance VOICE: Warm Female Alto EMOTION: carry Quiet Determination through the face and the body, not just the voice DRIVING VIDEO: film yourself performing the beat, then apply it to the character reference FRAMING FOR CAPTURE: keep the driving performance head on and evenly lit LIP SYNC: mouth shapes locked to the line with no drift at the tail SCORE: Cinematic Orchestral sitting under the voice SOUND DESIGN: Ambient Room Tone ROOM ACOUSTIC: match reverb to a Rain Slicked City Street space AVOID: Character Drift

06Aleph Edit Pass

ALEPH EDIT PASS - APPLY TO FOOTAGE YOU ALREADY HAVE OPERATION: No Edit Pass TARGET: Whole Frame PRESERVE: Keep Subject Untouched MATCH: the edit must match Cinematic Rim Light and Cinematic Teal and Orange in the original plate TEMPORAL: the change must hold steady across every frame, not flicker CLEANUP: Fix Edge Artifacts CONTINUITY: Check the Final Second EDIT BEFORE REGENERATING - an edit pass costs far less than a new generation CREDIT PLAN: Draft in Turbo then Final OUTPUT: 16:9 at 1080p STILL AVOID: Character Drift

📘 Pro Guide

Model Guide

Gen-4.5Current flagship. Best fidelity, physics and prompt adherenceDefault choice
Gen-4.5 TurboFaster and cheaper, slight quality trade for iterationDrafting
Gen-4 and TurboPrevious generation, still strong on consistencyValue option
Gen-4 ImageImage to video specialist, animates a still cleanlyStill to motion
AlephVideo to video editing. Relight, remove, restyle existing footagePost work
Act-TwoPerformance capture. Drive a character with your own actingCharacter work
Third party modelsRunway also hosts other major video models in one dashboardOne subscription
Media RouterAuto selects a model based on quality, speed or cost priorityHands off

Directing Rules

Lead with the shotStart the prompt with the shot size, not the subjectOrder matters
One move per clipTwo camera moves in one generation degrades the resultSplit the shot
Describe light physicallyName the source and direction, not just a mood wordRealism
Runway rewards directionIt is built to be directed rather than describedCore idea
Chain the framesUse the last frame of an approved clip as the next first frameContinuity
Watch the tailDrift usually appears in the final second of a clipQuality control

Aleph Edit Workflow

Edit before regeneratingAn edit pass costs far less than a fresh generationSave credits
Name the targetSay exactly which element changes and which stay untouchedPrecision
RelightingChange the lighting of a finished shot without reshootingCommon use
Object removalRemove one or several elements from existing footageCleanup
Match the plateGrain, grade and motion blur must match the originalInvisible edits
Watch for flickerFrame to frame instability is the usual edit failureCheck first

Act-two Performance

Film yourselfA driving video of you performing beats any text descriptionCore method
Head on and even lightThe driving performance should be clearly lit and frontalCapture tips
Character referenceSupply the character image separately from the performanceTwo inputs
Voice controlA separate audio file can drive matching mouth shapesAdd on
Face body and handsTracking covers head, face, body and hand movementCoverage
Subtle beats read bestSmall expressions transfer more reliably than large onesDirection

Credits and Workflow

Token based pricingRunway moved from unlimited plans to credit consumptionBudget aware
Turbo for exploringFind the shot cheaply, render the keeper at full qualityStandard practice
Cost varies by modelEach model in the dashboard consumes credits differentlyCheck before batch
Batch then selectGenerate several takes and keep one rather than fixing a bad oneEfficiency
Extend beats regenerateExtending an approved clip preserves the look and costs lessLong form

Model status (checked Nov 2026)

Gen-4.5Runway's current generation model for text to video and image to video.
Aleph 2.0Released May 2026 for editing existing footage (swap objects, relight, change backgrounds). Replaced Gen-4 Aleph, retired 30 July 2026.Input clips 2 to 30 seconds, up to 1080p.