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

# Avatar videos

> Create a talking presenter from a script with a pre-made HeyGen avatar or your own portrait photo, and choose voice, framing, background and captions

Ask GoFrame Editor Agent for a presenter and it produces a talking-head clip from a script: a pre-made avatar from a library, or a portrait photo of you animated to speak. Use it for product demos, tutorials, announcements, FAQ videos and multi-language versions of the same message.

## Before you start

* **A script.** 50 to 300 words works best. Split longer scripts into several clips.
* **A plan that includes the model.** **HeyGen Avatar 4 Digital Twin**, the default, carries the **pro** badge and needs a paid plan.
* **Credits.** Avatar video is billed per second of finished video. The agent shows the cost and asks for approval before running, unless you've set avatar-video to auto-approve in **Settings → Agent → Permissions**.
* **Time.** Generation takes a few minutes; the job card shows progress.
* **A portrait (Photo Avatar only).** A clear, well-lit face, frontal or at a slight angle. Attach it to the chat or reference it in the Media pool with `@`.

## Ask the agent

```text theme={null}
/avatar-video presenter reads our product intro script
```

```text theme={null}
Create a 20-second avatar video: business woman, sitting, front-facing, calm professional voice, dark navy background, 1080p. Script: "..."
```

```text theme={null}
Use @headshot.jpg as the presenter and have it read this welcome message, expressive style, captions on.
```

```text theme={null}
Make the same clip in Spanish with the same avatar and voice.
```

```text theme={null}
Turn this script into a three-scene presenter video, one clip per section, same avatar throughout.
```

Pick the model with the ✦ button in the composer or `/model`, in the avatar section of the picker. If you attach a photo, the agent uses Photo Avatar. If Photo Avatar is selected and you haven't given a photo, it asks for one.

## Options

| Setting           | Choices                                                                                                      |
| ----------------- | ------------------------------------------------------------------------------------------------------------ |
| **Model**         | **HeyGen Avatar 4 Digital Twin** (pre-made avatars, default) or **HeyGen Photo Avatar 4** (your portrait)    |
| **Avatar**        | Describe it (business, casual; sitting or standing) or name one; the agent suggests a match                  |
| **Framing**       | Normal (more body and room to gesture), close-up, or circle                                                  |
| **Voice**         | 70+ voices across 100+ languages, so write the script in the target language and pick the voice by character |
| **Speed**         | Slower or faster delivery                                                                                    |
| **Talking style** | Stable (minimal movement, serious content) or expressive (animated gestures, casual content)                 |
| **Resolution**    | From 360p up to 1080p                                                                                        |
| **Background**    | A solid color, an image, or a video from your project                                                        |
| **Captions**      | On or off                                                                                                    |

Ask the agent to "suggest three avatars for a healthcare explainer" if you'd like to choose from the library.

## Where the result lands and what to do next

* The clip is saved to the **Media pool** with the **Created by AI agent** badge and placed at the end of the timeline on a video track, or where you asked. An **Avatar video** job card shows progress; **Regenerate** re-runs it unchanged.
* Add your own captions: **Transcribe** the clip in the Media pool and apply a preset in the [Captions](/editor/captions) panel. This gives you word-level styling that the built-in captions option doesn't.
* Layer it: put a screen recording or B-roll on the track above and shrink the presenter into a corner with the [Inspector](/editor/inspector), or crop it into a circle with a mask.
* Multi-scene videos: the `wf-avatar-presenter` workflow plans sections, generates one clip per section with a locked avatar, voice, style, background and resolution, and assembles them in order.

## Tips

* Write for speech: a hook in the first sentence, two or three points, one call to action. Contractions and short sentences sound natural; sentences over 30 words flatten the delivery.
* Commas and dashes are where the avatar gestures. Put them where you want a beat.
* Keep one avatar, one voice and one background across a project. Any change reads as a different presenter.
* Draft at 720p, generate the final at 1080p.
* If you only need the voice, use [Voiceover](/editor/voiceover). If you need a scene rather than a presenter, use [Generate video](/editor/generate-video).

## Related

* [Voiceover and text-to-speech](/editor/voiceover)
* [Captions](/editor/captions)
* [Skills and workflows](/editor/skills)
* [AI models](/editor/ai-models)