# Studio

> Make one video clip from a sentence: the prompt, model, length, shape and resolution, start and end frames, reference pictures, sound, Taste and the price.

## What you will make

A clip of 4 to 30 seconds, depending on the model: a shot for a story, an ad, a reel, a moving picture of something you made in [Images](https://studio.oneinfer.ai/docs/images), or a test of how a new model sees your idea. Studio is the place for one shot. When you want a story made of many shots with the same cast, use [Movie and Film](https://studio.oneinfer.ai/docs/movie-and-film).

Here are clips made this way, with the words that made them.

<div class="samples">

![A woman in a trench coat walks toward the camera down a long arched corridor](https://oneinferdata.nbg1.your-objectstorage.com/media/samples/ast_4a45810850.mp4 "Arched corridor")

![A hummingbird hovers at a red hibiscus flower in a rainforest](https://oneinferdata.nbg1.your-objectstorage.com/media/samples/ast_d016414372.mp4 "Hummingbird")

![A blue morpho butterfly rests on orange orchids in a rainforest](https://oneinferdata.nbg1.your-objectstorage.com/media/samples/ast_99dcbbbc35.mp4 "Morpho butterfly, made with Wan 3.0")

</div>

Prompt for the corridor: "A woman in a camel trench coat walks toward the camera down a long arched corridor. Late sun through tall windows, slow dolly back, soft haze, 35mm."

Prompt for the hummingbird: "Macro shot of a hummingbird hovering at a red hibiscus in a rainforest. Wings blurred, shallow depth of field, morning light."

Prompt for the butterfly, made with **Wan 3.0**: "Cinematic nature commercial, 16:9, 24fps, 85mm prime lens, shallow depth of field. A brilliant metallic blue morpho butterfly gently rests on a bright orange tropical orchid dripping with morning rain. Sunlight breaks through a dense rainforest canopy, casting dappled, warm volumetric sunbeams across the scene. The camera executes a slow, buttery smooth 180-degree orbit around the flower. Dewdrops glisten on the vivid orange petals, and the butterfly's wings open and close slowly with micro-flutters, revealing shimmering structural color. Pristine natural realism, vivid color contrast, zero frame jitter."

The first two are a sentence or two: who or what, where, the light, and what the camera does. The third shows how much detail a model will take when you give it.

## A clip from start to finish

Say you want a short establishing shot for a film about the monsoon.

1. Open **Studio** in the menu and check the project in the left menu.
2. Click the prompt box and write: "A tea stall at dusk in heavy monsoon rain, steam rising from the kettle, slow push-in, warm lamp light."
3. Look under the box. It reads "5s at 720p · about 116 credits", and the same number is on the **Generate** button. Those are the defaults: **Seedance 2.5**, 5 seconds, 720p, 16:9, with sound.
4. Press **Generate**. The player shows progress while the model works. When it is done the clip plays, and it is saved to your Vault and listed in **Recent renders**.

![Studio with a prompt written, the Seedance 2.5 model and the price on the Generate button](https://studio.oneinfer.ai/docs/images/studio-start.webp "Studio, demo data")

If you are not signed in, **Generate** asks you to sign up first. If you do not have enough credits, the line under the box turns amber and says how many you need.

## What you will see

The player ("Studio Monitor") is on top and says **Your clip plays here** until there is a clip. The prompt box is under it, and **Recent renders** is on the right (on a narrow screen it is a strip above the box).

Above the box, three pills choose the kind of shot:

| Pill | What it does |
| --- | --- |
| **Text** | A clip from words alone. The default. |
| **Image** | A clip that starts from a picture you add. |
| **Restyle** | Changes the look of a video you already have. It needs a source video and, if you like, a style picture. |

Under the box, left to right: **Add to shot**, **Taste**, **Shot**, **Takes**, the model, and **Generate**.

## The prompt

Describe one continuous scene: the subject, the place, the light and the camera. A few rules that work for every model:

- Keep it to one moment. Studio makes one shot, not a cut between places.
- Name the camera: "slow push-in", "handheld", "tracking shot from behind".
- Type **@** and a name to put one of your cast in the shot. See [Cast and pictures](#cast-and-reference-pictures).
- **Polish prompt** rewrites your words into a fuller prompt for the model, shape and Taste you picked. It shows "Polishing…" while it works.
- **Clear** empties the box. The box takes up to 12,000 characters.
- You can paste a picture into the box (Ctrl+V) to add it to the shot.

The grey words that type themselves in the empty box are only examples. **Generate** stays off until you write something yourself.

## Pick the model

Press the model button (it shows **Seedance 2.5** at first). The list shows each model's name, a short note, what it can do and its price for a clip at its default settings. At the time of writing, prices in the list ranged from about 25 credits for **MiniMax H3** and 50 for **Wan 3.0**, to about 116 for **Seedance 2.5**.

![The model list with a price on each row](https://studio.oneinfer.ai/docs/images/studio-model.webp "Choosing a model, demo data")

- The list depends on the pill: models that cannot start from a picture do not show under **Image**.
- A model with no published price is not offered, so every row has a price.
- **Open full video catalog** opens **Models**, with what each model is for and what it can't do.

Different models want different amounts of detail. Seedance 2.5 takes up to 30 reference pictures and several shapes; others take frames only. The sections below say what each control does and when it is missing.

## Length, shape, resolution and sound

Press **Shot**. Its label summarises the choice, for example "720p · 5s · 16:9 · sound".

![The Shot menu with Shape, Length, Resolution, and Sound and camera](https://studio.oneinfer.ai/docs/images/studio-shot.webp "The Shot menu, demo data")

- **Shape**: the shapes the model can make, such as **Widescreen 16:9**, **Story 9:16** or **Square 1:1**. The default is 16:9.
- **Length**: a slider and a box in seconds. The range is the model's own (Seedance 2.5 runs from 4 to 30 seconds). The default is 5.
- **Resolution**: for example **480p**, **720p** and **1080p**. The default is 720p. A size above your plan's is shown with a lock; pressing it points you to the plan.
- **Sound and camera**: **With sound** (on by default) or **Silent**, and **Camera moving** or **Camera still**. A model with no choice says so, for example that it always makes clips with sound or without.

Price follows all of these. A longer clip, a bigger size and sound all cost more, and the number on **Generate** changes as you move them.

### Takes

**Takes** makes the same prompt more than once (1 to 4; the default is 1). Each take is charged, and they run one after another, so you can pick the best. With more than one take the button reads, for example, **Generate 3 takes · N credits**.

## Cast and reference pictures

Press **Add to shot** (it shows how many pictures it holds, such as 0/30). A panel opens with **Cast**, **Places** and **Props** from **Cast & World**, and the ways to add a picture: **From device**, **Paste**, **Camera**, **From Vault** and **Make a still**.

![The Add to shot panel with the cast of the project](https://studio.oneinfer.ai/docs/images/studio-frames.webp "Add to shot, demo data")

Two modes, shown as **Frame to frame** and **Mix references**:

- **Frame to frame**: the first picture is what the shot **Opens on** (the start frame). Add a second for what it **Ends on** (the end frame). Not every model takes an end frame; where one does not, the button says **No end frame**.
- **Mix references**: pictures guide the shot without being its first frame. Each picture can be **Opens on**, **Look** (guides style, places and props) or **Ends on**.

How many pictures a shot can take depends on the model: **Seedance 2.5** up to 30, **Seedance 2.0** and its Fast and Mini versions 9, **MiniMax H3** 9, **Wan 3.0** 5, and models that take only frames, one or two. The counter reads "N of M pictures", and if you add too many it tells you to remove some. If the model takes only a first frame and you tagged an actor, Studio first draws a still of the actor and then animates it. That adds the cost of the still, and it is included in the price.

Studio has controls for reference pictures only; it has no place to add a reference video or audio. An actor's own voice is sent along by itself when they speak.

In **Restyle**, the panel shows **Source video** (**Use the video in the player**, **From device** or from the Vault) and an optional **Style picture**.

## Taste

**Taste** is a ready-made look: **Film noir**, **1980s VHS / retro TV**, **Liquid splash commercial**, **Glossy product studio**, **Creator on a phone**, **Golden-hour documentary**, **Neon night**, **Pastel symmetry**, **Claymation**, **Anime cel**, **Storybook watercolour**, **1980s cinema** and **1982 disco**. The default is **No taste**.

![The Taste list, with the Make an ad and Make a still links at the bottom](https://studio.oneinfer.ai/docs/images/studio-taste.webp "Taste, demo data")

A Taste adds its words to your prompt only when you press **Generate**. Your own text stays as you wrote it. A few tastes say **suits 9:16**. The links at the bottom of the list take your words to **Business** (**Make an ad**) or [Images](https://studio.oneinfer.ai/docs/images) (**Make a still**).

## The price before you spend

Studio shows the price three times before you press **Generate**: on the model's row in the list, on the **Generate** button, and in the line under the box ("5s at 720p · about 116 credits"). With **Takes** above 1 it multiplies, and if Studio has to draw a first frame first, the line says the number includes it. The wallet line on the right shows your **plan** and **balance** credits.

The price is the platform's published rate for that model at that length and size, so it matches what is taken.

## Recent renders

**Recent renders** lists this project's latest clips (up to 30), newest first. Press one to play it. Each card has **Share** and **Download**. Only videos from this project show here, not stills or audio.

After a clip is made, the player has: **MP4** (download), **Share** (to another app), **Capture Frame** (grab a still to use as a start frame), and **Add to Director** (put the clip on the timeline). A clip is saved to your Vault automatically. If saving fails, Studio says so and tells you to download it right away because the link expires within a day.

> [!NOTE]
> A model's safety checks can refuse a prompt. The message says "Responsible AI Policy" and gives the reason; change the words and try again.

## Questions

### Why is Generate greyed out?

You have not written a prompt, or the model needs a start frame (**Image** mode), or you do not have enough credits. The reason is written in amber under the box.

### Can I make the clip longer than the model allows?

No. The **Length** range is the model's own. For a longer piece, make several clips and join them in **Director**, or write a story in [Movie and Film](https://studio.oneinfer.ai/docs/movie-and-film).

### Which model should I use?

Start with the default, **Seedance 2.5**. If you need a lower price, try **Wan 3.0** or **MiniMax H3**. **Models** says what each is for.

### Does my Taste change my prompt?

No. It is added on the way to the model, and your words stay as you wrote them.

### What happens to the clip if I leave the page?

It keeps making, and when it is done it is saved to your Vault with the project you were in.
