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

# Step 5: Create the gallery

> Create scene prompts that show the character's range and build a strong four-item gallery.

Once the character's permanent appearance is stable, use the Generate tool for a small “photo shoot.” Each image should reveal something about the character or story while preserving the same visual identity.

## Build one scene at a time

Open [Generate](https://ourdream.ai/generate), choose the character, and use this order:

```text theme={null}
outfit + pose or action + setting + lighting or mood
```

For example:

```text theme={null}
(black button-up shirt), sleeves rolled up, standing behind a
polished mahogany bar, glowing bottles in background,
dim warm lighting, knowing smile
```

The character's permanent appearance prompt is already part of the generation. The scene prompt should describe only what changes for this image.

| Put in the character           | Put in the scene                         |
| ------------------------------ | ---------------------------------------- |
| Stable face and body traits    | Outfit                                   |
| Core hair color and style      | Pose or action                           |
| Defining markings              | Location and props                       |
| Features needed in every image | Lighting, mood, and temporary expression |

## Plan the gallery before generating

Character Studio supports up to four Display Media items. A useful set is:

1. **Cover:** the clearest summary of the character and premise.
2. **World:** a wider scene that establishes their environment.
3. **Personality:** an action or expression that reveals who they are.
4. **Range:** a different outfit, mood, or story beat that still feels like the same person.

The first item is the cover, so choose the image that reads well at a small size and accurately represents the experience.

<Tip>
  Four similar close-ups add less value than a deliberate sequence. Let each
  slot answer a different question about the character.
</Tip>

## Keep prompts visual and focused

Describe only details a viewer could see. Remove biography, sound, smell, and invisible emotion unless you translate them into expression, posture, lighting, or environment.

For a bartender:

```text theme={null}
simple black button-up shirt, dark gray waist apron,
polishing a cocktail glass, behind a mahogany bar,
warm practical lighting, soft background blur
```

For a florist:

```text theme={null}
dark green linen shirt, worn brown canvas apron,
arranging delphiniums at a market stall,
buckets of flowers, bright morning light
```

See [Image prompting basics](/characters/image-prompting) for weights and model-specific syntax.

## Generate for consistency, then for variety

Start with a conservative scene close to the base image. Once the face and body remain stable:

1. Change the outfit.
2. Change the pose or activity.
3. Change the environment.
4. Try a different lighting setup.

When a generation drifts, compare it with the last successful result. Simplify the newest prompt or use image editing from a clean base.

## Add the final media

In **Character Studio → Profile → Display Media**:

1. Add or replace generated images or videos.
2. Arrange the strongest cover first.
3. Review the four items as a set.
4. Save the character changes.

Before publishing, check that:

* The same character is recognizable in every item.
* The cover matches the Public Description and Greeting.
* No temporary visual accidentally became part of the permanent appearance.
* Backgrounds and poses look deliberate rather than generic.
* Every image follows the current [content policy](https://safety.ourdream.ai).

<Card title="Next: Test the character" icon="arrow-right" href="/characters/test-and-release">
  Check the opening, rules, progression, and visuals under different play
  styles.
</Card>