To ensure your Digital Twin ( Video Avatar) looks and performs its best, follow these friendly and simple guidelines when shooting your video footage. Our AI needs specific conditions to work its magic and generate an accurate avatar.
🛎️ Don't feel like reading? Here's a quick tips video on how to create your first Digital Twin avatar! Click Here
Your Video Does Three Jobs at Once
Every avatar is made of three separable ingredients:
Motion comes from your video — how you move, gesture, and blink.
Look comes from a photo — your outfit, setting, and framing.
Voice comes from your audio — your tone and accent.
A single video delivers all three at once — it's audio plus thousands of photo frames. This is why you should always record a strong video reference first, even when you plan to use a photo or AI-generated Look afterward. The video teaches the AI how you move and express yourself across a range of emotions. Once it knows that, you can swap the Look or Voice separately — but the motion stays anchored to you.
💡 Think of it like training a really good impersonator: the video is you showing them how you move, sound, and look. After that, they can perform any new script as you — they're not replaying your clip, they're acting it out fresh every time.
Uploading Guidelines
When creating your Still Avatar, we offer two options to record your footage:
Upload a Google Drive or local video file
Record with your computer's webcam
Please keep in mind that uploading your footage from a higher quality camera is the better option to create a higher quality avatar.
Filming Tips
Camera Quality: Use professional cameras for the best results. Mobile phones or webcams can also work.
Environment: Choose a quiet space with good lighting.
Video Length: Record at least one 2-minute long uninterrupted speech.
Resolution: We recommend filming in 1080p for higher avatar quality. 4K at 60FPS is even better.
Continuous Shot: Do not edit or cut the footage in the middle. It should be a continuous shot.
Stability: Stand still and avoid sudden head motions or unnatural body movements. Be natural and calm. Your avatar will perform exactly as you do.
Eye Contact: Maintain direct eye contact with the camera at all times. Avoid looking around or up and down during the recording.
Hand Movements: Keep hand movements below chest level and avoid raising them above your chest.
Pauses: Take pauses with your lips closed between sentences while maintaining eye contact with the camera. Start the video with a 2-3 second pause. Insert a 1-2 second pause between longer sentences.
Smartphone users: Before you start recording, tap and hold your face on the screen until a lock icon appears — this locks both focus and exposure, preventing the camera from drifting mid-clip. Do a quick 5-second test recording with some movement to confirm focus and brightness stay consistent before your full take.
When moving your head, limit turns to no more than 30 degrees in either direction. Quick or wide head turns reduce motion quality in your avatar's output.
Avoid clothing with large logos, text, or busy patterns. These can cause visual artifacts or distortions when HeyGen generates Looks from your avatar.
When you pause between sentences, keep your lips closed at natural rest. This gives the avatar a clear break point and significantly improves lip-sync accuracy.