> ## Documentation Index
> Fetch the complete documentation index at: https://docs.tryvinci.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Make four product stills from one prompt

> Write one prompt, make four takes at once, and keep the best one. Each take has its own seed, and the price shows before you generate.

The first take of an AI still is rarely the best take. The light, the steam and the angle change from one generation to the next. This guide makes four takes from one prompt on the **Image** page of the studio, so that you choose the best take, not the first one.

> [card] What you get
>
> Four 2560×1440 stills of one product scene, from one prompt, for \$0.16.

> [card] Time to read
>
> About 4 minutes. The steps take about 5 minutes.

## The example

One prompt and four takes on Seedream 4.5. Each take has its own seed, so the kettle, the cup and the steam are a little different in each picture.

![Four stills of a black pour-over kettle and a white cup on a walnut counter in morning light. Each take is a little different.](https://mintcdn.com/vinci/LqVJF0s-dAQonWva/images/learn/four-stills-takes.jpg?fit=max\&auto=format\&n=LqVJF0s-dAQonWva\&q=85\&s=65ebf9275bb7299a656811a4e63695e1)

Four takes from one prompt on Seedream 4.5, 2560×1440 each. Model spend: $0.16 ($0.04 for each still) at the prices of November 2026.

And this is the process in the studio, with Seedream 4.5. This run made four new takes from the same prompt.

<video controls playsInline src="https://mintcdn.com/vinci/LqVJF0s-dAQonWva/videos/four-stills-walkthrough.mp4?fit=max&auto=format&n=LqVJF0s-dAQonWva&q=85&s=345cd156032e9f2c2723cd6f3758933a" poster="/images/learn/four-stills-walkthrough-poster.jpg" />

The steps on the Image page with Seedream 4.5. Model spend: \$0.16. The wait for the four takes (about 50 seconds) is cut. 36 seconds, no sound.

## What you need

* The free and open source Vinci studio (the install is in step 1)
* A model provider key (this example uses Wavespeed)
* One sentence about the product, the place and the light

## Steps

**1. Start the studio with live spend on**

```bash theme={"system"}
git clone https://github.com/tryvinci/vinci && cd vinci
python3 -m venv .venv && . .venv/bin/activate && make install
./scripts/vinci init
VINCI_ALLOW_LIVE_SPEND=1 ./scripts/vinci studio
```

The studio opens at `http://127.0.0.1:4311`. Open **Connect** on the left rail and add your provider key. Then pick **Image** on the left rail. Live spend stays off until you start the studio with `VINCI_ALLOW_LIVE_SPEND=1`. If the studio already runs, stop it first with `./scripts/vinci stop`.

**2. Write what you want to see**

Write one prompt: the product, the place, the light and the lens. Use the same words for the product in every prompt. The example prompt:

```text theme={"system"}
A matte black pour-over coffee kettle on a walnut kitchen counter, a white ceramic cup beside it with a thin curl of steam, soft morning light from a window on the left, shallow depth of field, product photography, 50mm lens.
```

**3. Pick a model**

Click the model chip next to the prompt. The list shows the models of the providers that you connected: the featured models at the top, and all models below them. The example uses **Seedream 4.5**.

**4. Pick the shape**

Click the aspect chip. Pick **16:9** for a web page or a video, **9:16** for a phone screen, or **1:1** for a feed.

**5. Make four takes at once**

Click the batch chip and pick **4**. Each take gets its own seed, so you get four different pictures, not one picture four times.

**6. Check the price, then generate**

The price is on the **Generate** button. Four takes on Seedream 4.5 show \$0.160. Click **Generate** one time. The takes show at the top of the page.

**7. Keep the best take**

Hover a take to see its actions. You can mark it as a favorite, download it, or click **Animate** to make a clip from it. The next guide starts here: [Turn a still into a 6-second clip](https://docs.tryvinci.com/learn/animate-a-still.md).

> [!TIP]
>
> **The same prompt again costs \$0.** With a batch of 1, the same model, prompt and settings give the saved picture again, and nothing is billed. To get new takes, make a batch of 4, or change the prompt.

> [!WARNING]
>
> **A model does not obey every word.** In the example, the second take (top right) put the kettle on an electric base, and the prompt did not ask for one. Look at every take before you use it.

## Before you use a still

* [ ] The product has the same shape, color and label as the real product
* [ ] The shape fits where the still goes: 16:9, 9:16 or 1:1
* [ ] There is nothing in the picture that the prompt did not ask for

## Two ways to go further

> [card] [Let Vinci make it](https://tryvinci.com/meet?utm_source=guide\&utm_medium=docs\&utm_campaign=four-stills)
>
> Send us your product photos, and we return a set of stills in your brand style. Book a demo to get started.

> [card] [Free and open source Vinci](https://github.com/tryvinci/vinci)
>
> Run the Vinci studio on your own machine with your own model keys. It shows the price before every run.

## More guides

> [card] [Turn a still into a 6-second clip](https://docs.tryvinci.com/learn/animate-a-still.md)

> [card] [Chain a still and a clip on the canvas](https://docs.tryvinci.com/learn/still-to-clip-on-the-canvas.md)


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.