I’m building an Android game whose main hook is an AI-generated, video-driven narrative that changes on the fly. The gameplay element I have in mind is exactly what I wrote in the brief: we will create AI-powered video, letting the system assemble and stream short cinematic sequences that adapt to every choice the player makes. All action takes place from a First-person viewpoint, so the code must keep camera movement, UI overlays, and interaction prompts compatible with that perspective. The setting is intentionally marked “public variable” because I want the world-building to stay flexible: the AI should be able to pull in fantasy, sci-fi, historical, or any blend we later decide, without needing a full rebuild. Think of it as a sandbox framework that can swap art packs, dialogue datasets, and rule presets through configuration files. Here’s what I need delivered: • A working Android build (APK + project files) created in Unity or Unreal with C# / Blueprint scripts that call the chosen ML model or API for on-device or cloud-side video generation and branching logic. • Core systems for dialogue, decision tracking, and scene assembly, tested in a simple proof-of-concept level showing at least three meaningful branches. • Clear documentation explaining how to plug in new theme assets and how to train or fine-tune the AI module for additional storylines. • Source code, scene files, and instructions packaged so I can open, edit, and export future builds myself. The scope is limited to the gameplay framework; I’m not asking for full asset libraries yet, just placeholders that prove the pipeline works. Once the foundation is stable, we’ll discuss expanding content and polishing visuals. If you already have experience mixing Unity/Unreal with GPT-style text generators, Stable Diffusion–style video, or similar ML models, you’ll be able to hit the ground running. Let me know which tools you prefer and any prior work that shows dynamic content generation on mobile.