This ComfyUI workflow turns two still images—a defined first frame and last frame—into a single, smoothly animated video using the MiniMax H3 (Hailuo-03) first–last frame API. Two LoadImage nodes feed your start and end images into the MinimaxHailuo03FirstLastFrameNode, which sends them to the MiniMax service to predict the in-between frames. The node returns the generated sequence back to ComfyUI, and SaveVideo encodes it as a video file at your chosen frame rate.

Technically, the MinimaxHailuo03FirstLastFrameNode performs an API call to MiniMax H3’s First–Last Frame to Video endpoint, conditioning motion on the content of both input images to synthesize temporally consistent intermediate frames. Because the motion is guided by your exact start and end frames, you get a coherent transition without training or manual keyframing. This makes the setup ideal for before/after reveals, scene transitions, and quick animatics where you want subject consistency and controlled start/end states.