Text to Video Prompt Examples: 15 Copy-Paste Starters

Text to video prompts work when you paste a complete scene: who or what is on screen, what changes, how the camera moves, and how many seconds the clip runs. This page gives you fifteen copy-paste starters built for Kling AI, MiniMax Hailuo, Luma Dream Machine, and Runway Gen-4 as of August 2026. Each line is blank-canvas text-to-video language meant for an empty upload slot. You leave with ready pastes across product, people, and place jobs, a one-change adapt rule, mid-2026 host notes, and a soft path to PromptMake /image when you want a hero still before you open a motion UI.

What these text to video prompts cover

A text to video prompt describes a clip from words alone. You type the scene into Kling, Hailuo, Luma, or Runway. The host invents every frame. That path differs from image-to-video, where an upload locks the subject and your text only adds motion. It also differs from a structure lesson that teaches pillar labels without a paste library. Here the deliverable is fifteen starters you can run today.

Searchers who type "text to video prompts" want lines that survive a first credit spend. Forum strings full of "cinematic 8K masterpiece" burn tokens and leave motion vague. These starters name a concrete subject, one timed action, one camera path, and a duration that matches the beat list. Swap nouns for your product after the first clean render.

Strong fit for marketers cutting five-second product teasers, creators testing social B-roll from scratch, and designers storyboarding motion before a longer edit. Skip this page if your source is already a locked still and you only need motion language on top of pixels. Skip it if you want a four-pillar craft drill with no paste bank. Use this library when the canvas starts empty and you need a first working sentence.

How to use the starters on Kling, Hailuo, Luma, and Runway

Paste discipline beats prompt length. Pick one starter that matches your job type. Set duration in the host UI to the seconds written in the line. Generate once. Score subject clarity, motion finish, and camera obedience. Change one variable on the next pass. That loop works on free tiers where credits rotate daily as of mid-2026.

Read the starter aloud before you spend. If you cannot picture the action inside the stated seconds, shorten the beat or pick a different line. Do not stack three camera moves onto a four-second clip. Do not paste image-only adjectives (beautiful lighting, hyper-detailed) without a verb that changes over time. The subsections cover host settings and the edit rule that keeps retries cheap.

Host settings before you paste

Confirm the mode is text-to-video with an empty upload slot. Set aspect to match the job: 9:16 for vertical social, 16:9 for wide hero, 1:1 for feed squares when the host offers it. Set duration to the number in the starter (four to six seconds on most consumer lanes).

Kling AI and Hailuo often read clause-separated prose. Luma Dream Machine likes short sentences on five-second clips. Runway Gen-4 accepts film vocabulary when you keep one camera path. Put resolution and frame rate in UI controls when they exist. Leave those words out of the prompt unless the vendor docs for your route say otherwise.

The one-change rule after the first render

Score three checks: Did the subject stay readable? Did the motion finish inside the window? Did the camera follow one path? Fix the failing check only. Subject drift: tighten material and color nouns. Motion mush: cut a beat or add one second in the UI. Camera chaos: delete extra move words until one verb remains.

Save the winning paste next to the host name and date. Next week you swap the product noun and keep the motion grammar. That habit turns fifteen starters into a personal kit list without rewriting from scratch each time.

15 copy-paste text to video prompts

Fifteen starters below. Each line is paste-ready. Suggested duration sits in the text so you can match the UI slider. Hosts named in the intro all accept these as blank-canvas prose; dialect tweaks come after you lock a clean first render. Swap bracket-style nouns in your head (bottle brand, jacket color, city name) after the motion holds. Do not invent celebrity likenesses. Keep subjects generic or fictional.

Group one covers products and objects. Group two covers people and lifestyle action. Group three covers places and weather. Run one starter from your group before you rewrite adjectives. A clean five-second product clip teaches more than ten half-finished experiments.

Starters 1–5: product and object

1. Medium studio shot of a matte black wireless earbud case on brushed aluminum, soft key from upper left. Case lid closed for two seconds, then lid opens once to reveal white earbuds. Slow dolly in only. Five seconds.
2. Overhead flat-lay of a ceramic pour-over dripper on oak, hot water kettle spout at frame edge. Thin steam rises for one second, then a slow pour fills the dripper for three seconds. Camera locked. Four seconds.
3. Close-up of a brushed steel wristwatch on a linen cuff, daylight from a window on the right. Second hand ticks once per second for the full clip. Camera locked, shallow depth. Four seconds.
4. Three-quarter shot of a white leather sneaker on a slow turntable against seamless gray. Shoe rotates a quarter turn over five seconds. Soft overhead key, static camera. Five seconds.
5. Extreme close-up of a glass perfume bottle on black acrylic. One condensation bead slides down the front face from top to mid glass. Slow push-in only. Soft rim light from behind. Five seconds.

Starters 6–10: people and lifestyle

1. Medium shot of a courier in a red jacket locking a bike on a wet sidewalk at dusk. Hands turn the key once, then step back half a pace. Soft neon storefront glow behind. Camera locked. Five seconds.
2. Chest-up of a barista in a navy apron at an espresso machine. Steam wand hisses once, then a thin milk pour starts into a ceramic cup. Soft kitchen practicals. Slow dolly in only. Five seconds.
3. Wide shot of a runner in a gray tank on a gravel path under overcast sky. Runner approaches from mid-ground for three seconds, then passes camera left. Static tripod, 35mm feel. Five seconds.
4. Medium shot of a chef plating a single seared scallop on white ceramic. Tongs place the scallop, then a green oil drop lands once beside it. Soft overhead kitchen light. Camera locked. Four seconds.
5. Street-level shot of a person in an olive trench coat walking toward camera on a crosswalk. Coat hem shifts once in a light wind. Soft morning sun from camera left. Slow lateral truck right only. Six seconds.

Starters 11–15: places and atmosphere

1. Low wide shot of a wet asphalt alley at night, single neon bar sign soft in the background. Rain streaks fall for the full clip. One taxi light passes mid-frame at three seconds. Static tripod, 24mm feel. Six seconds.
2. Cliff-edge wide shot of ocean waves under late afternoon sun. One wave breaks against rock for three seconds, spray lifts once. Soft haze on horizon. Slow tilt up only. Five seconds.
3. Forest path medium shot under morning mist. Single shaft of light moves across moss as fog thins. Leaves still. Slow dolly forward only. Five seconds.
4. Rooftop dusk shot of a city skyline, two window lights blink on in distant towers. Sky grade shifts warmer for four seconds. Camera locked, 50mm feel. Four seconds.
5. Desert highway wide shot at noon, heat shimmer on asphalt. Single car approaches from far distance for five seconds, then soft brake lights flash once. Static tripod, 28mm feel. Six seconds.

How to customize a starter without breaking motion

Customization fails when you rewrite everything at once. Keep the motion spine and the camera path from the starter you chose. Change subject nouns, materials, and place names first. Change camera only after a clean subject swap. That order protects the timed beats that make text to video prompts readable to Kling and Luma.

Write your edit as a short checklist before you paste: new subject noun, same motion verbs, same camera verb, same seconds. If the new subject is larger or smaller than the original (sneaker to sofa), shorten motion or add one second so the action still fits. The subsections show noun swaps and camera swaps as separate passes.

Swap nouns and materials first

Replace the hero object or person with your brief. Keep count language: single bottle, one runner, two window lights. Keep material words concrete: matte ceramic, brushed steel, wet asphalt. Vague swaps like "premium product" invite invent clutter.

Example: starter 1 becomes "matte sage green earbud case on walnut" while lid open timing and dolly stay identical. Generate once. If color drifts, name the hex-adjacent word again in a second pass rather than adding style praise.

Swap camera second, duration last

Change camera only when framing fails the job. Product detail often wants a push-in. Environment establish often wants a locked wide. Replace one move with another move. Never add a second move in the same paste.

Adjust duration in the UI after camera is stable. If the new subject needs more time to travel across frame (runner, car, wave), add one or two seconds and keep the same beat order. If the clip ends with empty frames, cut a second or end the action earlier in the text.

Common mistakes with text to video prompts

Mistake 1: Image caption pasted into a video box. Still language freezes pose and light. Add a timed verb and one camera path or the host invents chaos.

Mistake 2: Three camera moves in one four-second clip. Orbit plus dolly plus tilt collapses on Kling and Hailuo. One path per generation.

Mistake 3: Duration longer than the beats. Ten seconds of UI time with one steam curl wastes credits on empty tail frames.

Mistake 4: Subject motion and camera motion fused into "dynamic reveal." Split them: "Subject stays fixed; camera pushes in slow."

Mistake 5: Celebrity or private likeness hacks. Use generic wardrobe and age-band language. Skip real names meant to clone a face.

Mistake 6: Rewriting the full starter after one miss. Change one axis: subject, motion, camera, or seconds.

Mistake 7: Treating text-to-video starters as image-to-video preserve briefs. Preserve language belongs on upload routes. Blank-canvas lines need full subject prose.

Mistake 8: Chasing host folklore before a clean first paste. Master one starter on the model you can access today, then read release notes for dialect tweaks.

Model notes for Kling, Hailuo, Luma, Runway (2026)

Public model names shift. Treat the notes below as an August 2026 snapshot. Confirm pickers inside Kling AI, MiniMax Hailuo, Luma Dream Machine, and Runway before you paste old forum strings. Shared rule: one subject, one motion spine, one camera path, seconds matched in the UI.

Free and low-credit tiers favor four-to-six-second tests. Spend those credits on starters that already name timed beats. Save twelve-second first drafts for paid lanes on Runway Gen-4 or Kling when your account exposes them.

Kling AI reads separated clauses well. Paste subject, motion, and camera as clear sentences. Hailuo rewards timed verbs ("0–2s static, 2–4s lid opens once"). Luma Dream Machine prefers short readable prose on five-second clips. Runway Gen-4 accepts shot-type lead-ins (medium studio shot, low wide shot) when you keep energy consistent across the line.

Audio fields appear on some 2026 builds. If your host generates sound, add a short ambient cue or write "no music" after the visual line so a score does not land on a quiet product clip. Keep audio optional. Visual beats still decide whether the clip works.

When PromptMake /image helps before text-to-video

Text-to-video starts from words. Many teams still want a hero still first so they can approve color, material, and crop before they spend motion credits. PromptMake /image turns a reference photo into Midjourney, FLUX, DALL·E, Stable Diffusion, or Leonardo formatted prompt text when you need that still. Soft path: https://promptmake.net/image.

Workflow: upload a product photo or mood still, pick Recreate Exactly to hold composition or Create Variation for alternate hero frames, export a clean PNG from your image model, then decide. If you keep text-to-video, write subject nouns from the still into one of the fifteen starters. If you switch to image-to-video later, the same still becomes the upload anchor and your text shrinks to motion and camera only.

Guest accounts get about three /image runs per day. Free registration raises the image quota. Quotas for /image and /text stay separate. Spend image runs on subject lock, not on hunting synonyms for motion you have not chosen yet. PromptMake does not render video clips. It shortens the still step so your text to video prompts attach to a frame you already trust.

FAQ

What are text to video prompts?

Text to video prompts are written briefs that tell a motion model what to generate without an upload. Strong lines name the subject, the timed change, the camera path, and the clip length. Weak lines stack still-image praise words and leave motion vague. The fifteen starters on this page are blank-canvas pastes for Kling, Hailuo, Luma, and Runway Gen-4.

How do I write text to video prompts that work on the first try?

Start from a starter in this library that matches your job. Set the host duration to the seconds in the line. Generate once. Change one axis if something fails. Keep one camera path and one primary motion beat per four to six seconds on consumer tiers as of August 2026.

Can I use these text to video prompts on Kling, Hailuo, Luma, and Runway?

Yes. The starters use portable prose those four hosts accept in mid-2026 text-to-video modes. Stay out of the image-to-video upload lane. Adjust clause style after a clean render: Kling and Hailuo like clear sentence breaks; Luma likes short sentences; Runway likes shot-type lead-ins.

How are these different from image-to-video prompts?

Text-to-video lines describe the full scene from scratch. Image-to-video lines assume an upload already locked the subject and focus on motion and camera. Do not paste preserve-upload language into a blank text box. Do not paste a full scene rebuild on top of a still you want to keep.

How long should a text to video prompt be?

Length matters less than beat clarity. Two to five dense sentences beat a paragraph of empty adjectives. Match motion scope to seconds in the UI. One primary action per two to four seconds is a safe starting ratio on free tiers.

Which model should I start with in 2026?

Start with the host you can open today and a four-to-five-second clip. Luma and Kling fit quick blank-canvas tests. Hailuo fits short timed beats. Runway Gen-4 fits longer text-to-video when your plan exposes ten-second lanes. Paste the same starter across hosts only after it works on one.

When should I use PromptMake /image with text to video prompts?

Use /image when you want a cleaner hero still or Midjourney or FLUX text from a reference before you invent motion. Soft start: https://promptmake.net/image. After the still holds, copy subject nouns into a starter from this page, or promote the still into an image-to-video workflow if you prefer upload lock.