---
title: "Generate an AI video avatar presenter from an image"
slug: "/help/demos/edit/guides/ai-video-avatars"
topic: "demos"
last_updated: "2026-09-02T11:32:54.641Z"
---

## About This Feature

The **AI Video Avatar Presenter** turns a still image of a person (or a mascot) into a lifelike video presenter that appears alongside your demo, speaking a script you provide or one generated from your guide narration. Perfect for onboarding demos, personalized outreach, and hands-off product tours where a human recording isn't practical.

### Snapshot 📸

**Feature Status:** AI feature — consumes AI credits

**Role Access:** Admin, Editor, Presenter

**Location:** **AI Mode** in the editor → **Generate presenter**

---

## Before You Start

You need:

- **A source image** — a clear, forward-facing headshot works best. Uploading someone else's face requires their **consent** (see below).
- **A script** — either write one, paste one, or let Walnut generate it from your guide narration.
- **Available AI credits** — video generation is a paid AI feature; the exact cost is shown before you generate.

---

## Quick Start

### 1. Open the presenter panel

In the editor, open **AI Mode** and pick **Generate presenter**.

### 2. Upload the source image

Drop in a headshot or pick one from your workspace assets. Walnut warns you if the face is off-center or too small.

### 3. Give consent (if the face isn't yours)

Before Walnut can generate video from an uploaded face, you're asked to confirm you have permission from the person in the image. This is a hard gate — generation cannot proceed without it. The consent record is stored with the avatar for audit purposes.

### 4. Choose a script source

Pick one of:

- **Write it yourself** — type or paste a script.
- **From guide narration** — Walnut uses the demo's guide text as the presenter's script, one segment per step.
- **From presenter notes** — Walnut reads your existing presenter notes and rewrites them as spoken script.

### 5. Generate

Click **Generate**. Walnut renders a preview clip. If you're not happy, tweak the script or swap the image and regenerate.

### 6. Save the avatar

Save the finished avatar to your workspace. It appears in your avatar library alongside static image avatars and can be reused on other demos.

---

## Publish-Time Overrides

When you publish a demo, you can override which avatar plays for that specific published version:

- **Keep the template's avatar** — inherit whatever's on the template.
- **Swap for a different saved avatar** — pick another AI presenter or a static avatar.
- **Turn the presenter off** — publish without any avatar overlay.

This override lives on the published demo, so different customers can get different presenters from the same template.

---

## Viewer Experience

Viewers see the video presenter as an overlay on the demo. They can:

- **Toggle captions** on or off from the presenter's controls. Captions render the exact script that's being spoken.
- **Pause or mute** the presenter without stopping the demo.

---

## Tips

- **Match your brand.** For company demos, use a mascot or an authorized company spokesperson's headshot so the presenter reads as an intentional part of your brand, not an uncanny stranger.
- **Keep scripts short.** Long paragraphs feel unnatural spoken. Aim for 1–2 sentences per step.
- **Preview on mobile.** The overlay renders at a fixed corner size — check that key demo content isn't hidden behind it on small viewports.

---

## FAQs

**Do I need consent to use my own face?**
No — the consent gate only fires when the uploaded image isn't verified as belonging to the signed-in user. Consent is always required when using someone else's face.

**How much does it cost?**
Video generation consumes AI credits per second of rendered video. The exact cost is shown before you click Generate.

**Can I edit the script after generation?**
Yes — open the avatar, edit the script, and regenerate. The old clip is replaced.

**Does this replace static avatars?**
No. Static image avatars ([Create avatars](/help/demos/edit/guides/avatars)) still work exactly as before. Video presenters are an additional option in the same avatar library.