This ComfyUI workflow turns a single reference image into a cinematic, multi-shot video using the LTX-2.5 fast distilled model. It links three nodes end-to-end: LoadImage brings your source still into the graph, LtxApi25ImageToVideo sends that image to the LTX-2.5 API for motion synthesis and shot generation, and SaveVideo writes the returned clip to disk. Because LTX-2.5 supports native multishot generation and automatic duration prediction, the result feels like a short edited sequence rather than a single continuous pan or zoom.

Technically, the LtxApi25ImageToVideo node conditions all shots on the same input frame, which helps maintain character, lighting, and style consistency across cuts. The node handles request/response with the remote service, queues generation, and outputs a ready-to-save video. SaveVideo then encodes the result at your chosen filename and frame rate. This makes the workflow ideal for rapid storyboarding, social posts, and pre-visualization: drop in one clean reference image, let the API infer pacing and camera moves, and export a cohesive multi-shot clip in minutes.