This is the whole Runway Gen-4.5 prompt vocabulary on one page — no scrolling through paragraphs. Start with the formula, then copy shot sizes, camera moves, and lighting terms straight out of the tables into your own prompts.
For the full walkthrough behind these tables, see how to prompt Runway for realistic video. For ready-made prompts, jump to the 35 best Runway prompts.
The prompt formula
Runway rewards concrete visual detail in a fixed order. Lead with the subject, end with the film look, and keep it to one clear action plus one clear camera move per clip.
[subject + appearance] → [action / motion] → [setting] → [camera move] → [lighting / mood] → [style / film look]
Here is that formula filled in as a complete text-to-video prompt:
A weathered fisherman in a yellow oilskin coat, silver stubble, hauls a dripping net over the side of a wooden boat, at dawn on a choppy grey sea, slow dolly-in from a low angle, cold golden-hour light breaking through storm clouds with sea spray in the air, shot on 35mm film with shallow depth of field. 16:9, 10s.Why it works: one subject, one action, one camera move, and every clause maps to a slot in the formula — nothing contradicts.
Shot sizes
Shot size sets how much of the subject and scene fills the frame. Name it explicitly so Runway frames correctly instead of guessing.
| Term | When to use | Example phrase |
|---|---|---|
| Extreme wide shot | Establish a location and scale; subject tiny in frame | "extreme wide shot of a lone hiker on a vast salt flat" |
| Wide shot | Show the full subject plus surroundings | "wide shot of a runner crossing an empty bridge at dawn" |
| Medium shot | Subject from the waist up; dialogue and gesture | "medium shot of a barista steaming milk at the counter" |
| Close-up | Face or object detail; emotion and texture | "close-up of rain running down a woman's cheek" |
| Extreme close-up | A single detail fills the frame | "extreme close-up of an eye reflecting neon signs" |
| Over-the-shoulder | Two-person scenes; perspective and depth | "over-the-shoulder shot of a chef plating a dish" |
Camera moves
Gen-4.5 understands real camera terminology. Pick one move per short clip — stacking moves in a 5s shot muddies the motion.
| Term | Effect | Example phrase |
|---|---|---|
| Locked-off / static | No camera motion; let the subject move | "locked-off static shot, camera does not move" |
| Dolly in (push-in) | Slow move toward subject; builds tension | "slow dolly-in toward the subject's face" |
| Dolly out (pull-out) | Reveal context by pulling back | "smooth pull-out revealing the full room" |
| Tracking / follow | Camera travels alongside a moving subject | "tracking shot following the cyclist from the side" |
| Orbit | Circles the subject; showcases a product or hero | "slow 180-degree orbit around the sneaker" |
| Crane up / down | Vertical sweep; grand reveals | "crane up from street level to the rooftops" |
| Handheld | Subtle shake; documentary energy | "handheld camera with natural micro-movement" |
| Whip pan | Fast horizontal blur; transitions | "quick whip pan to the left" |
| Tilt up / down | Camera pivots vertically in place | "slow tilt up from boots to face" |
| Aerial / drone | High overhead sweep of a landscape | "aerial drone shot gliding over the coastline" |
| Dolly zoom | Vertigo effect; dolly and zoom oppose | "dolly zoom as the background stretches away" |
Lighting & mood
Runway understands lighting physics, so lighting terms change the whole feel of a shot. Name the source and quality of the light.
| Term | Look |
|---|---|
| Golden hour | Warm, low, soft sun near sunrise or sunset; long shadows |
| Soft diffused | Even, shadow-light glow like an overcast window; flattering |
| Hard rim light | Sharp edge of light outlining the subject against dark |
| Neon | Saturated magenta and cyan glow; nightlife and cyberpunk |
| Volumetric / god rays | Visible beams through haze, dust, or fog; atmospheric depth |
| High-key | Bright, low-contrast, minimal shadow; clean and upbeat |
| Low-key / chiaroscuro | Deep shadow, small pools of light; moody and dramatic |
| Backlit | Light behind the subject; silhouettes and glowing edges |
| Overcast | Flat, soft, colorless daylight; natural and understated |
Aspect ratios & duration
Set the frame and length up front. Match the ratio to the platform and reach for 10s when a prompt has several actions.
| Setting | Value | Use it for |
|---|---|---|
| 16:9 | Landscape | YouTube, standard video, most cinematic b-roll |
| 9:16 | Vertical | Reels, TikTok, Shorts, Stories |
| 4:3 / 3:4 | Classic / tall | Retro looks and portrait framing |
| 1:1 | Square | Feed posts and grid tiles |
| 21:9 | Cinematic widescreen | Filmic, letterboxed hero shots |
| Duration | 2–10s (5s & 10s presets) | 10s for multi-action or camera choreography |
| Gen-4 Turbo | ≈ 5 credits/sec | Fast, cheap drafting and iteration |
| Gen-4 | ≈ 12 credits/sec | Highest-fidelity final renders |
| Resolution | Up to 4K | Delivery and large-screen work |
Features quick-ref
The core Runway features and the one rule that matters for each.
| Feature | What it does / the rule |
|---|---|
| Text-to-Video | Generate from text alone; follow the six-part formula above |
| Image-to-Video | The image sets subject, color, lighting, and style — describe motion only |
| Keyframes | Set first, middle, and last frame images to control a shot's arc |
| Act-Two | Maps face, hands, and full-body gestures from a driving video onto your character |
| Lip Sync | Up to 45s, text-to-speech or uploaded audio, up to 4 faces |
For motion-only prompts in depth, see the image-to-video prompt pack.
Things to avoid
Most weak Runway clips come from asking for too much at once. Cut these and quality jumps.
| Avoid | Instead |
|---|---|
| Multiple simultaneous camera moves | One clear camera move per clip |
| Contradictory motions (walking + sitting) | One primary action, one subject |
| Long readable on-screen text | Add text later in an editor |
| Multi-line dialogue inside a generation | Use Lip Sync or keep speech short |
| Run-on, comma-stuffed prompts | Clean clauses in formula order |
| Re-describing the image in image-to-video | Describe only the motion |
Full example prompts
Four complete prompts assembled from the tables above — copy, swap the nouns, and generate.
Extreme wide shot of a lone red kayak crossing a mirror-still alpine lake, paddler stroking in a steady rhythm, snow-capped peaks and pine forest behind, slow aerial drone shot gliding forward and up, soft golden-hour light with mist rising off the water, cinematic film look with muted greens. 21:9, 10s.Best for: establishing b-roll and travel intros.
Medium shot of a young woman in a mustard raincoat laughing as she turns to look over her shoulder, rain-soaked city street glowing behind her, tracking shot following her from the side, hard neon rim light in magenta and cyan reflecting off wet pavement, moody cinematic style. 16:9, 5s.Best for: character-driven, night-city moods.
Extreme close-up of a wristwatch as a hand tilts it into the light, second hand sweeping, brushed steel and sapphire glass catching reflections, slow 180-degree orbit around the watch face, soft diffused studio light with a single hard highlight, glossy premium product look on a matte black surface. 1:1, 5s.Best for: product hero loops. For more like this, see the product & ad prompt pack.
Wide shot of an empty subway platform at night, a single train arriving and slowing to a stop as steam drifts across the tracks, locked-off static shot, low-key chiaroscuro lighting with flickering practical lights and volumetric haze, gritty 35mm film grain. 16:9, 10s.Best for: atmospheric, slow-burn cinematic scenes.
Frequently Asked Questions
What is the Runway prompt formula?
Order your text-to-video prompt as subject and appearance, then action or motion, then setting, then camera move, then lighting and mood, then style or film look. Lead with concrete visual detail and skip conversational filler.
How is an image-to-video prompt different?
With image-to-video the input still already sets subject, composition, color, lighting, and style, so your text prompt should describe motion only — the camera move and how the subject or scene moves. Do not re-describe what is already visible in the frame.
How long can a Runway clip be?
Gen-4.5 generates 2 to 10 seconds per clip, with 5s and 10s the common presets. Build longer sequences by extending clips. Use 10s when a prompt has multiple actions or a camera choreography that needs room to play out.
Which aspect ratios does Runway support?
Gen-4.5 supports 16:9, 9:16, 4:3, 3:4, 1:1, and 21:9 cinematic widescreen, at up to 4K resolution. Pick 9:16 for social, 16:9 for standard video, and 21:9 for a filmic look.
Turbo or Gen-4 — which should I use?
Gen-4 Turbo is faster and cheaper at roughly 5 credits per second and is ideal for drafting and iterating. Full Gen-4 costs around 12 credits per second and delivers the highest fidelity for final renders.
Can Runway do dialogue and lip sync?
Yes. Act-Two maps face, hands, and full-body gestures from a driving video onto your character, and Lip Sync handles up to 45 seconds of text-to-speech or uploaded audio across up to 4 faces. On-screen readable text and long multi-line dialogue inside a generation stay unreliable, so add those in an editor.