Write the shot as one line — who, doing what, where — and Seedance 2.5, Veo 3.1 or Wan 3.0 turns it into a clip. Camera movement, light and ambience come from the model.
The more specific the description,the closer the shot lands to the one in your head.



Paste them in order and you have a short film. Each one notes why it is written that way.
An old downtown street at dusk, a few parked cars, two pedestrians walking toward camera from deep in the frame. Low angle from the road surface, tilting up slowly, warm sunset raking across, 35mm film grain.
Open low and wide so the place registers first.

A man in a dark coat walks in from frame right, slows, turns his head toward the lens. Traffic behind him goes softly out of focus, the coat hem moves in the wind. Static camera, shallow depth of field.
Give the subject one action only so the frame stays clean.

Close on his profile and eyes, rain sliding down his brow. Slow push in, his gaze coming back from somewhere far and landing toward the lens. Soft top light, warm skin.
A close up needs the light source named.

He turns into an alley, neon smeared into blotches by the rain. The camera pulls back and drifts slightly out of focus, the frame going dark.
Ending on open space beats spelling everything out.

Out of the bokeh the next scene surfaces: a window at dawn, gauze curtains lifting in the wind. A seamless hand off from the previous clip.
A light transition cuts more naturally than a hard cut.

Subject, action, setting. Skip words like “high quality” or “8K”; the model cannot read them.
Vertical for short video, horizontal for a reel. Get five seconds right, then continue.
Download the one that works, or use it as the first frame for the next shot.
“Her hair moves in the wind” is easier to execute than “a windy feeling”, because the change is visible.
Push in, orbit and crane at the same time fight each other; the usual result is a jitter in place.
“Side backlight through a window” or “neon lighting the face from below” beats “good lighting” by a wide margin.
Longer shots fail more often. Confirm the style, then continue from the last frame.
Every model is priced per second of output, and the price changes with resolution, audio and quality tier. The generator shows the exact credit cost before you submit.
It depends on the settings. Clips can run from a few seconds up to 30 seconds, and the duration selector only offers lengths the current setup supports.
If the provider rejects the job before accepting it, the credits are refunded automatically. Once the provider has accepted the job and started rendering, the run is charged.
No. The generator runs in the browser. Sign in, describe the shot and generate; there is no timeline, plugin or desktop app to install.
Yes. Adjust one variable — the prompt, duration or resolution — and regenerate. The composer keeps your previous takes in history so you can compare them side by side.
Change one thing at a time: the prompt, the duration or the resolution. Most odd results come from a prompt that describes several actions at once; split it into one clear movement and regenerate.
Yes. Keep the prompt and change one thing at a time - camera move, lighting or duration - so the takes are comparable. Save the version you like before you keep experimenting, and reuse the saved prompt when you come back to the idea later.
Not always. Many clips are usable as they come out of the generator. An editor helps when you need to trim the ends, join several shots into a sequence, or add music, captions and your own sound design.
The generator fills the gaps with its own choices, so you get a clean clip that is not the shot you imagined. Name the subject, the setting, one camera behaviour and a light direction; that is usually enough to pin the result down.
Try a generation, then decide whether to pay.