Autonomous Detective Storytelling Engine Submission for the #Melius Challenge. The idea was to build an automated storytelling engine natively inside Melius Canvas that orchestrates narrative generation and multi-model media pipelines in one connected system. How it works: Narrative Orchestration: Gemini dynamically structures a 3-scene mystery — defining the suspects, plot twists, and clues. Different story every run. Multi-Model Pipeline: The master script automatically feeds data downstream into specialized image (GPT) and video (LTX Video) nodes. Synced Narration & Visuals: ElevenLabs generates the voiceover first, which then becomes the timing reference for the visual pipeline. Fully modular — swap the LLM or video model anytime, and the pipeline adapts automatically. Overall the first experience using Melius was fairly smooth, especially for building multi-node pipelines like this. The canvas is intuitive and the node connections are easy to understand. https://x.com/decodexai/status/2056985544048148842?s=20