Skip to main content
Generate a video with synchronized dialogue and sound from a text prompt using HeyGen Video 1.0, optionally guided by connected reference material. Images of people, products or places, videos to reuse and audio clips that provide a voice can all be used as references. Mention each reference in the prompt as @Image1, @Video1 or @Audio1, numbered per type in the order the inputs are connected: these tags are rewritten into the labels HeyGen expects, and a tag that points to a reference which is not connected raises an error. Without any reference image or video the node runs as text-to-video generation.

Inputs

Parameter Constraints

  • Reference limit: at most 15 references in total: up to 9 images, 3 videos and 3 audio clips. Connecting more raises an error.
  • Audio needs image or video: reference audio is rejected when no reference image and no reference video is connected.
  • Reference image rules: each reference_images input must hold exactly one image (a batch is rejected), and each image must be between 1:4 and 4:1 in aspect ratio.
  • Prompt tags: @ImageN, @VideoN and @AudioN are matched case-insensitively. The number must not exceed the count of connected references of that type, and the prompt must be non-empty after trimming whitespace.
  • Mode: with at least one reference image or video the request is a reference-to-video run; with none it is a plain text-to-video run.
  • 1080p and 2k resolution: resolution "1080p" or "2k" requires aspect_ratio "16:9" or "9:16"; "auto" is accepted at these resolutions only when no reference image and no reference video are connected.
  • Duration: with reference images or videos the generated video stays at 15 seconds or less. Longer videos must have no reference images or videos; auto_duration only makes duration ignored (8 to 30 seconds from the dialogue, 8 to 15 with references), it does not raise that cap.

Outputs

This documentation was AI-generated. If you find any errors or have suggestions for improvement, please feel free to contribute! Edit on GitHub

Source fingerprint (SHA-256): 415ffca1c385936e35bdb0451a23f03756e936b3b871d8d6a958da1ef8c56c36