How to prompt HeyGen Video 1
HeyGen Video 1 writes the picture and the sound in one pass, so the prompt has to describe both. Name the subject, the action and the setting for the shot, then name the sound you want to hear with it. The clearer the sound cue, the closer the soundtrack lands to the action.
A street-food scene described for both picture and sound: the sizzle, the spatula and the market crowd come with the shot
A prompt that covers these six parts gives the model enough to work with. Most clips need only the first four.
| Part | What to write |
|---|---|
| Subject | Who or what is on screen, and what they look like in plain terms |
| Action | What happens across the clip, one clear beat is enough for 5 to 15 seconds |
| Setting | Where it is and the light, so the look and the ambience match |
| Sound | The sounds you want to hear: dialogue, ambience, effects, named plainly |
| Camera | The shot and any move, such as a slow push in or a locked wide |
| Framing | The aspect, from 21:9 widescreen to 9:16 vertical |
Write the sound into the prompt
This is the part most video prompts skip, and it is where HeyGen Video 1 earns its place. Because the audio is made with the picture, a clip is only as good as the sound you ask for. Name sounds, not feelings. For a spoken line, write the exact words in quotes, and keep them short enough to finish inside the clip.
The exact line written in the prompt, in quotes: the mouth shapes and the room tone come with it
| You want | Write |
|---|---|
| A spoken line | She looks to camera and says: "Here's how it works in three steps." |
| Room ambience | A quiet office, soft keyboard typing and a low hum under the voice |
| Effects on the action | Boots crunching on gravel, timed to each step |
| A busy scene | Market crowd murmur, a sizzling griddle, a spatula scraping metal |
| A calm scene | Light rain on the window, distant traffic, nothing else |
Settings that matter
Four choices shape the output. Draft at a low resolution, then raise it once the shot is right, since the higher tiers cost more.
| Setting | What to pick |
|---|---|
| Resolution | 480p or 768p for drafts, 1080p or 2K for the finished clip. Higher tiers are billed at a higher rate |
| Length | 5 to 15 seconds. Longer clips fit a full line or a complete beat, but give the model more to keep consistent |
| Aspect | 21:9 or 16:9 for landscape, 9:16 for vertical, plus square and portrait. Pick it for where the clip will be seen |
| Reference | Give an image in reference-to-video mode when you want the clip to follow a specific subject or look |
Source: the HeyGen Video 1 model listing and HeyGen's documentation.
Worked examples
Each clip below was made from a prompt that named both the picture and the sound. Turn on sound to hear how the audio follows the action.
Ambience from the setting: birdsong, a breeze through pines and boots on gravel, no dialogue
Effects plus a short line: tape ripping, cardboard and crinkle, with one spoken aside
Use a reference image
Text-to-video is enough for most clips. When you need the shot to follow a specific subject or a particular look, give HeyGen Video 1 an image in reference-to-video mode. It composes the scene around the reference and adapts the framing to it, so a product, a face or a style carries through instead of being described in words and reinvented each run.
Keep a scene consistent
HeyGen Video 1 holds a scene together well, but a clip asks a lot of a single run when the prompt is overloaded. A few habits keep the result steady.
| Habit | Why |
|---|---|
| One beat per clip | A single action in 5 to 15 seconds stays cleaner than three that fight for time |
| Short spoken lines | A line that overruns the clip gets cut off, so keep it to what fits the length |
| One sound to lead | Name the sound that matters most, then the background, so the mix is not muddy |
| Match aspect to the subject | A wide landscape in a vertical frame crops the sides, so pick the framing first |
| Draft small, finish large | Settle the shot at a low resolution, then raise it, instead of paying 2K to iterate |
Make a video with native audio on Morphic
HeyGen Video 1 is not on Morphic, but the same one-pass idea is. Morphic H3 writes the picture and a synced soundtrack together, and Veo 3.1 is the pick when a clip turns on a spoken line.
- Open Copilot and describe the scene, the action and the sound you want to hear.
- For a spoken line, write the exact words and ask for Veo 3.1. For ambience and effects, Morphic H3 generates the audio with the video.
- Pick the aspect and resolution for where the clip will be seen, and download the result.
FAQs
Describe the shot (subject, action, setting, camera) and then the sound you want with it (dialogue, ambience, effects). Because the audio is made with the picture, the sound cue matters as much as the visual one. Keep a spoken line short enough to finish inside the clip.
Write the exact words in quotes and say who speaks them, for example: she looks to camera and says, "Here's how it works." Keep it short, since a line that overruns a 5 to 15 second clip gets cut off.
Draft at 480p or 768p while you settle the shot and the sound, then render the final clip at 1080p or 2K. The higher tiers cost more on the services that host HeyGen Video 1, so there is no reason to pay for them while you iterate.
Instead of a text prompt alone, you give HeyGen Video 1 an image to follow. It composes the scene around that reference and adapts the framing to it, which keeps a specific subject or look consistent across the clip.
HeyGen Video 1 is not on Morphic, but the same approach works with Morphic H3, which generates native audio with the video, and Veo 3.1 for spoken dialogue and lip-sync. The prompt structure in this guide applies to both.
No. HeyGen Video 1 is not on Morphic, so the clips here were made on Morphic: the ambience and effect clips with Morphic H3, and the spoken-dialogue clip with Veo 3.1.