Wan 3.0 Text to Video Generator

Describe one scene: who moves, where they are, and how the camera follows. Start with your own prompt or try an idea below.

Inspiration
0 / 5000
  • Preparing your generation…

Workflow

Text to video

Workflow

Model

Wan 3.0

Paid WAN3 generation requires an active paid Trial Pack or a Standard or higher subscription.

No matching models

16:9 / 5s / 720p / 1
Estimated cost—credits
Try an idea

A stop in the desert

A lone rider approaches a faded gas station in the desert. Hold a wide shot as the horse crosses the orange dust.

Prompt workshop · a watch repair in one shot

First draft
A watchmaker works in a beautiful workshop. Make the scene cinematic and detailed.
Revised prompt
Close view of a watchmaker lowering a brass gear into an open pocket watch with tweezers. The watch rests flat on a folded green cloth. Warm light from a desk lamp catches the gear teeth. Hold the camera level with the bench until the gear settles into place.

The original prompt leaves the model to invent the task, choose the object and decide when the shot is over. The revision names the tool, the point of contact and the final frame. You can now judge the result by watching whether the gear reaches its seat. If it vanishes just before contact, try framing the watch from slightly above so the hands leave the mechanism visible. Keep the green cloth and desk lamp in the next attempt. Those details give you a stable background against which to compare motion.

From a written scene to a moving shot

What can I make from text alone?

An AI text to video generator takes a description of an event and imagines how it unfolds between the first and last frame. Try it when you know what should happen but have no source image. A bakery opening for the morning is a subject. A baker lifting a metal shutter until a strip of sunlight reaches the flour on the counter is a shot. The second idea gives you something concrete to look for when you play the clip. It also leaves the model room to invent the shop. If the counter must match a real location, take a picture of it and consider starting with the image to video tool.

How do I turn a script into a prompt?

Read a line of your script and underline the part a camera could record. A character realizing she has missed the train might look at an empty platform, then loosen her grip on a paper ticket. Write that gesture in the prompt. Pick the viewpoint that makes the gesture legible. A wide view shows the empty tracks, while a close view shows the ticket creasing between her fingers. If both details matter, plan two clips and join them in an editor. This approach also works for voiceover scripts. Underline the image each sentence calls to mind, then decide which of those images actually needs motion.

How long should my first shot be?

Act out the movement once and time it. Opening an envelope takes a different amount of screen time from walking across a courtyard to deliver it. Choose a duration in the workspace that leaves a moment to see the subject before the action starts and another to read the result. If the action needs several locations or a change of costume, break it into separate shots. A longer generation gives an action more time, but the extra seconds also need a visual purpose. For a looping background, describe a repeating movement such as a curtain breathing in a breeze and inspect the join in your editor.

Free online text to video prompt examples

Product reveal · control the reflection

Try this prompt for a tabletop product video. A dark green glass bottle stands on wet slate beside a window. A narrow band of morning light travels across its shoulder as the camera slides slowly to the right. The bottle stays still. The shot ends when the reflection reaches the edge of the glass. To adapt it for pottery, replace the traveling reflection with light catching the ridges in the glaze. Watch where your eye goes on playback. If the water attracts more attention than the product, remove the wet surface for the next version. For packaging with an exact label, start with an approved product photograph.

Travel scene · show the direction of movement

Try a street scene with one clear path. From beneath the awning of a closed flower shop, the camera watches a cyclist enter from the left and ride toward the bridge on the right. Puddles catch the yellow light from the shop across the road. The camera holds its position as the cyclist leaves the frame. This prompt gives the rider an entrance, a destination and a reason to disappear. To turn it into a tracking shot, change the camera sentence so that the camera rides alongside the cyclist. Keep the same road and lighting when you compare the two versions. Look for the moment the bridge becomes visible.

Character beat · make an expression readable

A young stagehand waits behind a red curtain holding a paper star. She hears her cue, looks down at the star and smiles. Frame her from the chest up with the curtain in soft focus behind her. End with her eyes on the star. For a silent clip, the glance can carry the story even if nobody hears the cue. To make the gesture easier to read on a phone, bring the paper star closer to her face in the prompt. Review the relationship between her hands, the prop and her gaze. If a handoff or a second character distracts from the expression, give that action its own shot.

Choose settings for the screen and the budget

Vertical, square or widescreen?

Choose the aspect ratio before describing the composition. For a vertical video of a dancer, picture the top of the raised hand and the bottom of the moving foot. Give both some room. In a landscape version of the same dance, a sideways step can carry the viewer across the frame. A square composition suits a compact action such as two hands tying a parcel. Use the ratios offered by the selected model. If you need two delivery formats, review the action in each crop before approving the shot. An attractive landscape preview can lose the object that explains the story when cropped for a phone.

Which model and resolution should I choose?

Start with the requirement that is hardest to change later. A storyboard may need a particular frame shape. A cut synchronized to music may need a particular duration. Select a Wan model in the workspace and check which combinations it actually offers. Resolution affects how much detail you can inspect, while the prompt determines which detail you ask the viewer to notice. A higher resolution cannot repair a camera pointed at the wrong end of a workbench. For a product shot, first verify that the camera reveals the object in the right order. Then evaluate the final version at the size where your audience will see it.

How much does a generation cost?

You can explore the prompt examples online for free. Eligible new Google accounts receive 30 welcome credits. Wan 3.0 generation requires a supported paid plan or Trial Pack and enough credits. Check the credit estimate for your settings before submitting a job. If you are comparing two configurations, record the model, duration, resolution and displayed estimate together. A price seen with one set of settings can be a poor guide to another. Budget separately for tests and for the version you intend to deliver. A short rehearsal of a coffee pour can tell you whether the stream lands in the cup. It cannot tell you whether a longer take finishes cleanly when the pot returns to the counter. Reserve enough room in your plan to inspect that ending before committing to a whole sequence.

Diagnose a clip and choose the next edit

Why does the camera move when I want a still frame?

Read your prompt once for movement words. In a shot of a train pulling away, a request to follow its windows asks for camera movement even if another sentence calls the view static. Pick a fixed point of observation such as a bench on the platform. Describe the train leaving that viewpoint while the edge of the bench remains in the foreground. On playback, compare the bench at the beginning and end. If it appears to drift, revise the camera instruction before changing the lighting or style. This gives you a practical way to judge camera control across two otherwise similar versions of the scene.

Why does the subject change halfway through?

Pause at the first frame where you notice the change. Did the actor disappear behind a door, turn away or pass into a much darker area? Move the action into an uninterrupted view and describe one distinguishing feature that stays visible. A mustard scarf against a gray coat can be easier to track than a small badge on a moving sleeve. When a person or an object must match supplied artwork, compare the reference to video and image to video options in the tools directory. Decide whether you need a visual reference for identity or a particular opening frame. That decision determines which input is worth preparing.

What should I check before downloading or editing?

Watch the clip once at its intended display size and again with your attention on the cut points. The first frame should establish the subject quickly enough for the surrounding edit. The last should leave you somewhere useful to cut. Check hands, object contact and changes in direction at full speed, then pause on anything that looks wrong. Watch without sound to see whether a viewer can still follow the action. If the delivery needs a readable headline, add the final typography in your editor so you can control spelling, timing and placement. Keep the prompt and settings alongside your selected take for later revisions.

On this page

Start with the shot, not a bag of adjectives

Write who or what is visible, what changes during the clip, where the action occurs and how the camera observes it. Put must-have details before atmosphere and finish with only the visual constraints that matter.

  • Subject: the visible person, object or environment.
  • Action: one clear change that can happen within the chosen duration.
  • Camera: framing, movement or viewpoint.
  • Light and texture: details that support the story instead of replacing it.
Match the model to timing and framing

Wan 2.5 offers 5 or 10 seconds and includes square framing. Wan 2.6 adds a 15-second option for text work. Wan 2.7 accepts 2–15 seconds and a broader aspect-ratio set. WAN3 extends text-led output to 2–30 seconds when the scene needs a longer opening, action and ending.

Use a prompt blueprint you can revise

Write one sentence for the subject and action, one for the environment and camera, and one for the continuity constraints. For example: a ceramic robot crosses a rain-soaked market; the camera tracks at waist height; keep the blue glaze, warm shop light and measured walking pace consistent. This structure makes the next revision specific instead of random.

Review the final quote in Workspace

Carry the prompt into the studio, choose the model, duration, resolution and aspect ratio, then review the credit cost before generating. For the first pass, use only enough duration and quality to test the shot; increase either after the motion and framing work.

Plan a focused first test

Build the shot from subject, action, setting and camera movement. Begin with one action in one location; a longer prompt does not replace a clear sequence. If exact product appearance or composition matters more than inventing a scene, prepare an image first and choose the image-to-video guide instead.

  • Illustrative prompt: A red kite lifts above a grassy hill; the camera tilts upward as it catches the wind under an overcast sky.
  • Review the result: Watch the action without reading the prompt: the subject, motion and camera direction should be easy to identify.

Continue exploring