No footage. No camera. No start frame.

Text to Video

Write the shot the way you would brief a camera operator, and four engines render it — 5 to 30 seconds, up to 4K, with sound.

Prompt only4 text-capable engines5–30 secondsNative audioSix aspect ratios

Three prompts, three clips:

Paper lanterns

Crystal city

Aurora lake

Write a promptNo experience needed · Cancel anytime

A 5-second clip starts at 5 credits and the cost of your settings is shown on the Generate button before you commit. Plans renew monthly until cancelled and can be cancelled online at any time — see pricing details.

A prompt, and what it produced

Each caption below is typed exactly as it was written into the studio, and the clip that follows is what came back. Nothing here was filmed.

AI generated

a crystal city at dusk, light trails weaving between translucent towers

Writing is the whole interface

Text to video is the purest form of the tool: the sentence is the only input, so everything on screen is a consequence of how you wrote it. That sounds limiting until you have tried the alternative. Sourcing a stock clip means accepting somebody else’s framing; filming means a location, a crew and a day. A prompt means the shot you described, in the ratio you need, in about the time it takes to read this paragraph.

It is also the mode that rewards precision most brutally. With no reference image to anchor to, every decision you leave out is one the model makes for you — the lens, the time of day, whether the camera is moving at all. The fix is not longer prompts. It is prompts that say what matters and stay quiet about what does not.

The anatomy of a prompt that works

Five slots, in roughly this order, cover almost every shot worth generating.

1. Subject, concretely

Not “a car” but “a black sedan, rain-beaded paintwork”. Materials and surfaces do more work than adjectives of quality — glass, brushed steel, wet asphalt, worn linen all give the model something to render.

2. What the subject does

One action per clip. “Turns into the corner” is a shot. “Drives through the city, parks and the driver gets out” is three, and asking for all of them in five seconds produces a muddle of all three at once.

3. Where the camera is, and what it does

This is the single highest-value line in the prompt and the one most often missing. Low tracking shot, slow push in, locked-off wide, handheld follow, crane down, orbit. Seedance 2.5 responds to this vocabulary more reliably than the rest of the line-up.

4. The light

Where it comes from and what kind it is: hard rim light from behind, overcast diffusion, a single practical in frame, dusk with headlights flaring. Light is what makes a generated clip read as photographed rather than illustrated.

5. The mood, in one or two words

Tense, serene, playful, clinical. Put it last — leading with mood tends to give you an atmosphere and no shot.

Put togetherLow tracking shot alongside a black sedan on wet tarmac at dusk, headlights flaring, shallow depth of field, tense.

Which engines take text alone

Four of the six engines in the studio generate from a prompt with nothing attached. The two WAN engines are built around an input frame or reference media, so they sit on the image to video side instead.

Engine	Length	Resolution	Sound	Use it for
Seedance 2.5	5 / 10 / 15 / 30s	480p – 4K	Native audio	The finished take
Seedance 2.5 Turbo	5 / 10 / 15 / 30s	720p / 1080p	Native audio	Fast iteration
Seedance 2.0	5 / 10 / 15 / 30s	480p – 4K	Native audio	Big-screen output
Seedance 2.0 Mini	5 / 10 / 15 / 30s	480p – 4K	Native audio	Cheap first drafts

Measured generation times and cost per second for each engine are on AI video models compared.

The practical pattern: draft on Mini or Turbo at 480p or 720p, where a run is cheap enough to make five of, and re-run only the prompt that won on Seedance 2.5 at the resolution you actually need. Most of the money people waste on generative video is spent rendering early drafts at final quality.

From one clip to a finished piece

Generate a few

Same prompt, cheap engine, low resolution. You are choosing a direction, not a final frame.

Re-run the winner

Same words, flagship engine, the resolution and aspect ratio the piece actually ships in.

Extend and finish

Continue the clip rather than regenerating it, upscale if it is soft, and score it with a generated track.

A sequence is built the way a cut is: shot by shot. Generate each one separately, keep the wording of the camera and light lines consistent between them so the look carries, and join them in your editor. When a shot needs to continue rather than cut, Extend is cheaper and more consistent than regenerating at a longer duration. A clip that is right but soft goes through the video upscaler, and a finished cut can be scored with the AI music generator.

When to stop writing and start with a picture

Text is the wrong entry point for some jobs, and knowing which saves a lot of credits. If the subject is a specific thing that already exists — your product, your packaging, a location you photographed — no prompt will describe it as accurately as the photograph does. Generate or upload the frame, then animate it: that is the image to video route, and it is how most product and brand work is done here.

The same applies to consistency. If a subject has to appear in five clips and look like itself in all five, anchor it — make the frame once in the AI image generator, save it as a character or location, and pull it into each prompt with an @ mention.

AI generated

Formats, limits and labelling

Six aspect ratios — 16:9, 9:16, 4:3, 3:4, 1:1 and 21:9 — are set before generation rather than cropped afterwards, so a vertical piece is composed vertically rather than trimmed from a widescreen frame. Durations run 5, 10, 15 or 30 seconds, and resolution from 480p to 4K depending on engine.

Every clip generated here carries a signed C2PA provenance mark identifying it as AI-generated, in line with Article 50 of the EU AI Act; the mark travels with the file and anyone can read it back on the verification page. Prompts are screened before generation, and the Content & Safety Policy sets out what may not be created here. Plans renew monthly until cancelled and can be cancelled online at any time; the full terms are on pricing details.

Frequently asked questions

How does text to video work?

You write a description of a shot and a generative model renders the frames for it. Nothing is retrieved from a footage library and nothing is edited — the video is synthesised from the description, which is why the wording of the prompt is the main thing you control. On WowMade you write the prompt, pick an engine, a duration, an aspect ratio and a resolution, and the clip renders in a queue on the page.

Which WowMade engines generate video from text alone?

Four: Seedance 2.5, Seedance 2.5 Turbo, Seedance 2.0 and Seedance 2.0 Mini. The two WAN engines — WAN 3.0 Prime and WAN 2.7 — need a start image or reference media to animate from, so they belong to the image-to-video route instead.

How long should a text-to-video prompt be?

Long enough to make the decisions you care about, and no longer. The field accepts up to 5,000 characters, but two well-chosen sentences that name the subject, the camera move, the light and the mood beat a paragraph of adjectives. If the shot needs more than one action, generate it as two clips and join them rather than asking one prompt for both.

Can a text-to-video clip have sound?

Yes. Seedance 2.5 and Seedance 2.0 generate native audio alongside the picture, and you can turn it off if you would rather score the clip yourself. For a composed soundtrack, generate a track with the AI music generator and lay it over the finished cut.

Why does the same prompt give different results each time?

These models sample rather than look up, so two runs of one prompt are two interpretations, not a repeat. That is useful: generate several and pick, rather than expecting the first to be final. When you need consistency instead of variety, give the model a fixed starting point — a start frame, or a saved character dropped into the prompt with an @ mention.

How much does a text-to-video generation cost?

It depends on the engine, the resolution and the duration, and the exact figure for your settings is printed on the Generate button before you spend anything — a 5-second clip starts at 5 credits. Drafting on Seedance 2.0 Mini or 2.5 Turbo at a low resolution and re-running only the winner at full quality is the cheapest way to work.

Can I keep going past the maximum length?

Yes. Extend continues a clip you have already generated instead of regenerating it from the start, so the look carries over and you pay for the new seconds rather than the whole thing again. Every engine in the line-up has a matching Extend variant.

Where to go next

One credit balance across video, image, music and the editing tools.

Describe the shot. See it in seconds.

Write a prompt