Scene Builder
The Scene Builder is the hands-on way to make an AI video: you build it scene by scene, mixing a talking AI avatar with your own photos and video clips, and you're in control of exactly what's said and shown at every step.
If you'd rather just describe the video you want in a sentence and let AI do the rest, see AI Auto-Generate instead. This guide covers the Scene Builder.
Opening the Scene Builder
- Go to your Media Library.
- Click + Add Media.
- Choose Generate AI Video.
- Make sure the Video Builder tab is selected at the top of the panel (it's the default).
That's it — you're ready to start building.
The basic idea
A video is just an ordered list of scenes. Each scene is one of:
- An avatar scene — your AI avatar on screen, speaking a script you write.
- An image or video scene — a photo or clip from your Media Library (or a fresh upload), which can either play silently or be narrated by a voice.
You can mix these freely and in any order — for example: avatar intro → product photo → avatar close → short video clip → avatar outro. Add as many scenes as you like.
Adding a scene
On the left-hand side you'll find + Add a Scene with four buttons:
| Button | What it does |
|---|---|
| + Avatar scene | Adds a new scene where your AI avatar speaks |
| + From Library | Picks an existing photo or video from your Media Library as a scene |
| + Upload Photo/Video | Uploads a new image or video straight from your device |
| + Create image with AI | Generates a brand-new image with AI (pick a model, describe it, and add it as a scene) — it's generated at your video's aspect ratio automatically |
Every new scene appears in the Scenes list on the right, ready to fill in.
Setting up an avatar scene
Each avatar scene has three things to set:
1. Script
Type what you want the avatar to say. Not sure what to write? Click ✨ Generate with AI — pick whether this scene is the Intro, a Feature / Benefit, the Outro, or Other, describe the topic, and it'll write just that one beat for you (not a whole video's worth of script). This needs a Brand Voice set up — see Settings & Brand Setup.
2. Avatar
Click Choose next to "Avatar" to pick who's talking. You have five options:
- Brand Face — instantly uses your zone's designated brand photo (see Settings & Brand Setup)
- From Library — pick a photo already in your Media Library
- Upload Photo — upload a new photo from your device
- My Avatars — reuse any avatar you've prepared before (managed in Settings)
- HeyGen Avatars — professionally filmed stock avatars, ready instantly with no waiting
For your own photo (Library/Upload), a clear, front-facing portrait with good lighting and a neutral expression works best — avoid sunglasses, hats covering the forehead, or extreme angles.
Once you pick an avatar it's applied to the scene straight away — no extra confirm step. (Photos from Library/Upload take a little time to prepare first — usually under a minute — and apply as soon as they're ready.) Want to fine-tune it afterwards? Click Change on the scene to reopen the picker, where you can:
- ✨ Edit with AI — describe a change (e.g. "change the background to a bright modern office", or "have them hold this product") and optionally attach reference graphics like a logo or product shot, then Generate a new version. Accept it to add it as a brand-new avatar look (your original is untouched), or Discard to keep what you had.
- Remove background — opens Adobe Express's background remover on the avatar's photo. Save the result and it replaces the avatar in your scene (the original stays in My Avatars).
- Background & motion (optional) — set a background color and motion for this scene, then click Done.
3. Voice
Click Choose next to "Voice" the same way. You'll see your Saved voices by default, or switch to All voices to browse the full catalog with language/gender filters. Tap the play icon to preview any voice before picking it.
Setting up an image or video scene
Image and video scenes are simpler:
- Narrated — tick this to have a voice read text over the scene. When it's on, a narration box and a Voice picker (same as above) appear.
- If it's not narrated, an image scene just needs a Duration in seconds (1–300), and a video scene has Volume (0–1) and a Mute checkbox instead — the clip plays for its own natural length.
Background & motion (optional, avatar scenes)
Inside an avatar scene's Avatar section, click + Background & motion (optional) to reveal:
- Replace this scene's background with a solid color — HeyGen only supports a flat color here, not an image.
- A free-text motion prompt (e.g. "gesturing naturally while speaking, nodding occasionally") and an Expressiveness level (Low / Medium / High).
- Copy settings from another scene, if you've already dialled these in elsewhere and want to reuse them.
Reordering and managing scenes
Each scene has four small controls on its right edge:
- ↑ / ↓ — move the scene up or down in the sequence
- ⧉ — duplicate the scene
- 🗑 — remove it
Video-wide settings
On the left, below "+ Add a Scene":
- Polish avatar scripts with brand voice — automatically refines every avatar script to match your Brand Voice's tone. Requires a Brand Voice — see Settings & Brand Setup.
- Use Brand Glossary for correct pronunciation — appears once you've set up a Brand Glossary, and fixes how specific words/names are pronounced.
- Add burned-in captions — subtitles baked directly into the video.
- Aspect ratio — Portrait 9:16, Landscape 16:9, or Square 1:1.
Generating your video
- Give it an optional Title.
- Click Generate Video.
You'll see a progress screen while HeyGen renders it — this can take a few minutes, and you're free to close the panel and carry on with other things. You'll get a notification the moment it's ready, and it lands automatically in your Media Library.
My Videos & saving your work
- My Videos (top of the panel) shows every video you've ever generated or attempted, with its status, script, and a Check status button for anything still processing — handy if a video finished but you never got the notification.
- Save Progress / Open Saved let you save a half-built video and come back to it later without losing your scenes, avatars, or settings.
Quick tips
- You don't need a Brand Voice or Brand Face to use the Scene Builder — HeyGen's own stock avatars and manual scripts work perfectly well on their own. Setting those up in Settings just automates more of the process for you.
- A single avatar-only scene makes a simple "talking head" video; bookending product photos or clips with an avatar intro/outro makes a proper mini product ad — both are just different ways of using the same scene list.