Photo to Video

Turn a single headshot into a talking avatar video — upload, create, voice, render. All from the CLI.

Prerequisites

• HeyGen CLI installed and authenticated (heygen auth login)
• A headshot photo (PNG or JPEG, max 32 MB). Front-facing, good lighting, and a neutral expression works best.

Steps

1. Upload the photo

   Upload your image to get an asset_id:

   heygen asset create --file ./headshot.jpg
   

   {
     "data": {
       "asset_id": "ast_abc123"
     }
   }
   
2. Create a Photo Avatar

   Pass the asset to avatar create. This trains a Photo Avatar from your image:

   heygen avatar create -d '{
     "files": [{"type": "asset_id", "asset_id": "ast_abc123"}]
   }'
   

   {
     "data": {
       "avatar_id": "avt_xyz789",
       "avatar_group_id": "grp_def456",
       "status": "processing"
     }
   }
   

   Avatar training takes a few minutes. Poll the status:

   heygen avatar looks get avt_xyz789
   

   Wait until status is completed before proceeding.

   Use --request-schema on any command to discover all available fields: heygen avatar create --request-schema

3. Pick a voice

   Browse available voices:

   heygen voice list --language English --gender female --limit 5
   

   {
     "data": [
       {
         "voice_id": "1bd001e7e50f421d891986aad5e3e5d2",
         "name": "Sara",
         "gender": "female",
         "language": "English"
       }
     ]
   }
   

   Copy the voice_id you want. If none of the stock voices fit, see the Design a Custom Voice recipe.

4. Generate the video

   heygen video create -d '{
     "type": "avatar",
     "avatar_id": "avt_xyz789",
     "script": "Hi there! I was created from a single photo using the HeyGen CLI.",
     "voice_id": "1bd001e7e50f421d891986aad5e3e5d2",
     "aspect_ratio": "16:9"
   }'
   

   Add --wait to block until the video is ready, or poll manually with heygen video get <video_id>.

5. Download

   heygen video download vid_qrs321 --output-path ./my-avatar-video.mp4
   

Full shell script

Chain everything together in one script:

#!/bin/bash
set -e

# 1. Upload
ASSET_ID=$(heygen asset create --file ./headshot.jpg | jq -r '.data.asset_id')
echo "Uploaded: $ASSET_ID"

# 2. Create avatar
AVATAR_ID=$(heygen avatar create -d "{\"files\": [{\"type\": \"asset_id\", \"asset_id\": \"$ASSET_ID\"}]}" \
  | jq -r '.data.avatar_id')
echo "Avatar: $AVATAR_ID (training...)"

# 3. Wait for avatar training
while true; do
  STATUS=$(heygen avatar looks get "$AVATAR_ID" | jq -r '.data.status')
  [ "$STATUS" = "completed" ] && break
  [ "$STATUS" = "failed" ] && echo "Avatar training failed" && exit 1
  sleep 10
done
echo "Avatar ready"

# 4. Create video and wait
VIDEO=$(heygen video create --wait -d "{
  \"type\": \"avatar\",
  \"avatar_id\": \"$AVATAR_ID\",
  \"script\": \"Hello! This video was generated from a single photo.\",
  \"voice_id\": \"1bd001e7e50f421d891986aad5e3e5d2\"
}")
VIDEO_ID=$(echo "$VIDEO" | jq -r '.data.id')
echo "Video ready: $VIDEO_ID"

# 5. Download
heygen video download "$VIDEO_ID" --output-path ./result.mp4
echo "Done: ./result.mp4"

Optional parameters

Parameter	Description
aspect_ratio	16:9 (landscape) or 9:16 (portrait)
background	Set a solid color or image background
callback_url	Webhook URL — skip polling and get notified when the video is ready

Photo Avatars use the avatar_iv engine. The avatar_id you pass to video create is the look ID from avatar looks list, not the group ID.