Skip to main content
HeyGen Avatar 6 turns a photo, a short character video, or a HeyGen avatar look into a talking clip, lip-synced to speech you supply or write. In ComfyUI it runs through two partner nodes. HeyGen Avatar 6 Reference to Video animates a photo or a stylized character with your own audio track. HeyGen Avatar 6 Identity to Video renders a stock or custom avatar from a written script, in a voice you pick. Both are cloud API nodes: they need a Comfy account with credits and download no local model. Rates are on the partner node pricing page. Every run renders at 1080p or 2K, in 16:9, 9:16, 1:1, 4:5, or 5:4, and the clip length follows the speech. heygen-avatar-6 is the default engine; heygen-avatar-6-turbo is the faster, cheaper option.
To use the Partner Nodes, you need to ensure that you are logged in properly and using a permitted network environment. Please refer to the Partner Nodes Overview section of the documentation to understand the specific requirements for using the Partner Nodes.
Make sure your ComfyUI is updated.Workflows in this guide can be found in the Workflow Templates. If you can’t find them in the template, your ComfyUI may be outdated.If nodes are missing when loading a workflow, possible reasons:
  1. You are not using the latest ComfyUI version (Nightly version)
  2. Some nodes failed to import at startup

Available workflows

Photo or video to video

Turns one photo, one short video of a stylized character, or both, into a talking clip in your own audio. The audio sets the length of the render.
  • The photo sets the identity and the scene. It is not used as the first frame: the model re-stages the pose, so a photo with arms crossed comes back re-posed.
  • The reference video shows how the character talks and moves, at least 2 seconds long. Longer clips are trimmed to the first 30 seconds. Videos of realistic people are refused; connect a photo of the person instead.
  • Audio runs from 1.5 seconds to 10 minutes.

Run on Comfy Cloud

Open in Comfy Cloud

Download Workflow

Download JSON or search “HeyGen Avatar 6: Photo to Video” in Template Library
Input materials

pink_hair_portrait.png

Load into the image input

avatar6_narration.mp3

Load into the audio input

Identity to video

Renders a presenter without a shoot: pick a stock look or an avatar you created, write the line, and choose a voice. With speech set to script the node needs nothing connected; with audio, connect a Load Audio node to the audio input. speech has two modes. script writes the line into text (up to 5000 characters) and speaks it in the selected voice, with an optional voice_speed multiplier. audio takes your own clip of 0.5 seconds to 10 minutes, and the clip length becomes the render length.

Run on Comfy Cloud

Open in Comfy Cloud

Download Workflow

Download JSON or search “HeyGen Avatar 6: Identity to Video” in Template Library
The template also ships two bypassed groups that create your own avatar and voice. Select a group’s nodes, press Ctrl+B to clear the bypass, then run the graph with Run or Ctrl+Enter and reuse the saved ID. Clear the string_a prefix in Concatenate Text first, or copy the raw ID from the creation node’s output, because the saved text is the prefix followed by the ID:
  • Avatar: HeyGen Create Avatar with source set to photo returns a look ID. Set avatar to custom avatar on the main node and paste the ID into custom_avatar_id. pink_hair_portrait.png ships with the template.
  • Voice: HeyGen Voice 1.0 Instant Clone returns a voice ID from at least 10 seconds of clean single-speaker audio. Paste it into custom_voice_id; it overrides the voice dropdown whenever it is not empty. avatar6_voice_sample.mp3 ships with the template.

Node controls

Reference to Video needs an image, a video, or both connected. Identity to Video needs no inputs in script mode; audio mode needs an audio input.

Prompting tips

  • Direct the performance in motion_prompt. Say how the avatar speaks and moves: gestures on the key words, relaxed shoulders, a slight head tilt, and a slow push-in on camera. Leave it empty for a neutral delivery.
  • Keep gestures small when the frame is tight. Set motion_expressiveness to less for head-and-shoulders crops, where wide gestures leave the frame.
  • Use clean single-speaker audio. Background music or a second voice in the track shows up in the lip sync. For a voice clone, 10 seconds of continuous speech is enough.
  • Frame the photo before connecting it. auto keeps the photo’s shape, so crop to the aspect ratio you want instead of re-framing afterwards.

Get started

  1. Update ComfyUI to the latest version
  2. Go to Template Library, search for HeyGen Avatar 6