# Video Generation

## **How to start**

Start with a description, or turn an existing image into a video.

**Open the generator.** Tap **Create** in the bottom bar (nectar.ai/create), then set **Type** to **Video**. You can also use the chat routes described in [Image Generation](/media-generation/image-generation).

![Camera menu on the Video tab with Send me a video, Send me a video of… and Generate a video, camera icon outlined in coral red](https://cdn.sanity.io/images/bf00glst/production/03c559509ab7e4935eab136624c3e0f4e92cae09-728x806.png)

**Or turn an image into a video.** Open an image your companion sent in chat. Tap the video button in the bottom-right corner.

This route offers **High** and **Standard** options. These are separate from the **Model Version** options in the main generator.

## **Using the generator**

Set **Type** to **Video**, then pick a model. Your choice changes the rest of the form.

| **Field** | **What to choose** |
| --- | --- |
| **Model** | Choose whether to use companions or a prompt alone. See [below](/media-generation/video-generation#v3-model-for-video-generation). |
| **Companions** | With **Nectar Companion Model**, select up to **10 companions** for the video. |
| **Model Version** | After selecting a companion, choose an available **Model Version**. Use **V3** for the first-video steps on this page. Video controls differ from the image options in the [model breakdown](/media-generation/image-generation). |

![Generator with the Video type selected and the Model dropdown open on Nectar Companion Model and Nectar Video Model](https://cdn.sanity.io/images/bf00glst/production/5508b0f76c21ff8ac423fa840c9ccae24b3812ea-1378x676.png)

| **Model** | **What you get** |
| --- | --- |
| **Nectar Companion Model** | A video featuring the companions you select. |
| **Nectar Video Model** | A video from your prompt alone. |

For your first companion video:

1. Choose **Nectar Companion Model**.

1. Select one companion.

1. Choose **V3**.

1. Set the length to **5 seconds** and quality to **720p**.

1. Describe one simple action.

1. Tap **Create**.

For example:

```
Mika sits at a café table in a blue sweater. She lifts her coffee mug, looks toward the viewer, and smiles. Morning window light. Keep the viewpoint still.
```

For help describing clothes, settings, and lighting, see [Image Generation](/media-generation/image-generation).

### **V3 model for Video Generation**

Use **V3**. It adds sound and lets you choose the length and quality. Other versions are silent.

- **Length:** 5, 10, 15, 20, 25, or 30 seconds.

- **Quality:** 480p, 720p, or 1080p. These numbers describe the video’s resolution.

- **Audio:** Included automatically. Describe speech or background sounds in your prompt.

Start short, then go longer.

#### Adding sound

Write what you want to hear in the same prompt box as the scene.

For example:

```
Rain taps against the café window. Mika says softly, "Stay for one more coffee."
```

Keep spoken lines short enough to fit the clip. Describe the speaker’s tone or accent when it matters.

#### Make it longer

Tap **Extend Video** on a finished clip to add **5s**, **10s**, or **15s**.

Follow the steps under **Extending a video** below.

> 💡

> For several companions, use separate descriptions for each person. See [Multiple companions](/media-generation/writing-image-prompts#multiple-companions) on Writing Image Prompts.

## **Writing better prompts**

These description tips from [Image Generation](/media-generation/image-generation) work for video too:

- **Put your companion somewhere specific:** name the place and light source.

- **Name the style:** live-action, anime, or oil paint.

- **Keep the scene focused:** one viewpoint, one time of day, and one location.

- **Describe clothes clearly:** put the outfit before less important body details.

- **Change one thing at a time:** compare results before making another adjustment.

Video adds **movement**. Start with one clear action per clip, such as "she turns and smiles."

### **Build the clip in beats**

**Power tip**

**Before you try this:** Make a short V3 video first. Then choose a longer duration when you want a sequence of actions.

A **beat** is one part of the scene. Timed blocks describe what you want to happen in each part. This technique is supported on **V3**.

Match the blocks to your selected length. For a ten-second clip, try two five-second blocks.

Use this format for each block. Replace the bracketed text with your details:

```
[Start time] - [End time]: [place, lighting, viewpoint].

[Companion name]
[Clothes, position, and action.]

[Speaker name, tone]: "Short line."
```

Here is a complete ten-second example:

```
0s - 5s: Café table in morning window light. Still viewpoint.

Mika
Wearing a blue sweater. Seated opposite the viewer, with both hands around a white coffee mug. She looks up and smiles.

[Mika, warm tone]: "You made it."

5s - 10s: Same café table and morning window light. Still viewpoint.

Mika
Still seated in her blue sweater. She sets the white mug on the table, then nudges the menu toward the viewer.

[Mika, playful tone]: "Breakfast first?"
```

**Re-anchor each companion at every beat.** This means restating their clothes, position, and current action so the next part has clear direction.

**Keep lines short.** Leave enough time for both speech and movement.

Use timestamps to organize the sequence. Keep the actions simple enough to fit each block.

### **Directing dialogue**

**Power tip**

**Before you try this:** Use **V3**, which includes audio. Name each speaker and allow enough time for their line.

Put the speaker’s name and delivery notes before the dialogue:

```
[Speaker, accent, tone, gaze]: "Dialogue."
```

For example:

```
[Alice, slight Swiss accent, confident tone, looking toward the viewer]: "Okay, let's go! Zero deaths!"
```

This separates who speaks from how they speak. Use the delivery notes to guide the accent, tone, and gaze.

For several companions, put each speaker’s line on its own line.

---

## **Extending a video**

Continue a finished clip from where it ends:

![Extend Video dialog (Extend by +5s) with a grid of eight preset scene thumbnails, an optional prompt box and Extend Video button](https://cdn.sanity.io/images/bf00glst/production/f91c97ee9336c9a804fcc27194810a08d9d84928-1236x1460.png)

1. Open the video and tap **Extend Video**.

1. Pick one **Customization**.

1. Choose **5s**, **10s**, or **15s**.

1. Describe what happens next. Tap **Enhance** to polish it.

For example:

```
Mika keeps her blue sweater on. She sets the coffee mug down and turns toward the window. Keep the same café lighting and viewpoint.
```

**Watch the clip before extending it.** Check that:

- The intended companions are present.

- Their appearance and clothes match the scene.

- The movement ends where you want the next action to begin.

- Any speech sounds the way you want.

Choose a clip you want to continue, then describe the next action.

## What to do with a finished video

### **Remix**

To remix a video:

1. Open it from the **Create** page.

1. Tap **Remix - Try It Yourself!** (lower right).

1. The form fills in. Change one thing and create.

![Generated video view starring a companion, with like, favorite and bookmark buttons above Extend Video and Remix - Try It Yourself!](https://cdn.sanity.io/images/bf00glst/production/a57284679df60db7a88443c796c63353f465f16f-998x1488.png)

For example, keep the café scene and change Mika’s action from lifting her mug to waving.

### Copy the prompt and other actions

Open a video to see its available actions.

| **Action** | **What it does** |
| --- | --- |
| **Copy Prompt** | Copies the text used to describe the video. |
| **Copy Icks** | Copies the details to leave out, also called the negative prompt. |
| **Submit to be featured** | Submits the video for public featuring. Use this when you want to share it publicly. |
| **Submit as profile picture** | Starts the process of using it as the companion’s profile picture. |
| **Bookmark** | Saves it for later. |
| **Thumbs up / down** | Rates the result. |

---

## **Generating in the background**

Videos generate in the background while you keep chatting or browsing. **Ultimate subscribers can run up to four video generations at once.**

The placeholder reads "Generating Image" while your video processes.

Find your videos under **My Creations** in the gallery. **My Likes** and **Suggested** are separate browsing views.

## **Video messages**

Ultimate subscribers can request videos in chat. Tap the **camera icon** in the message bar, then choose the **Video** tab.

You can also attach a video to a companion’s greeting through [Greeting Media](/companions/creating-a-companion).

Next: [Voice Chat and Voice Messages](/media-generation/voice-chat-and-voice-messages) · [Image Generation](/media-generation/image-generation) · Stuck? [Troubleshooting and Getting Help](/help-center/troubleshooting-and-getting-help)