This workflow generates short videos directly from text using the MiniMax H3 (Hailuo-03) video model. The MinimaxHailuo03TextToVideoNode sends your prompt to the MiniMax API, which renders a 5–15 second clip with native stereo audio. The node returns the finished video payload, and SaveVideo writes it to disk as a single file—no frame assembly required. Because computation happens in the MiniMax cloud, you don’t need a local GPU to create results.

Technically, you control the generation by setting the text prompt (and optional negative prompt), duration, aspect ratio or resolution, and seed for reproducibility. Short, directive prompts with clear subjects, actions, and camera cues tend to perform best (for example: “A futuristic drone flies over neon city streets at night, smooth tracking shot, cinematic lighting”). The SaveVideo node finalizes the output path and filename, preserving the audio track produced by MiniMax H3, so you get an immediately shareable MP4 without extra post-processing.