Skip to main content
This node uses MiniMax H3 Context IR to analyze your text description and any attached media, then produces a stronger, structured video prompt. The returned prompt is designed to be connected to the prompt input of a MiniMax H3 video node; if you attach media there, attach the same media in the same order, because the enhanced prompt refers to media by position.

Inputs

Parameter constraints

  • The prompt, duration, ratio, reference_images, reference_videos, and reference_audios inputs are part of the model option group.
  • first_frame and last_frame cannot be combined with any reference media.
  • reference_audios cannot be used unless at least one reference_image or reference_video is also connected.
  • When no frames and no reference media are connected, ratio cannot be set to "adaptive".
  • Reference videos must be approximately 2-15 seconds each, with a total duration of no more than 15 seconds. Their frame rate must be between 23.9 and 60.5 FPS.
  • Reference audios must be approximately 2-15 seconds each, with a total duration of no more than 15 seconds.
  • first_frame, last_frame, and every reference image must be at least 256x256 pixels and have an aspect ratio between 0.4 and 2.5.

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): 73015517f9c0f55f0aceeef935508a372e0d95668e4733d1c8100b53e4afa7e2