Battalion - My First Ever 5 Minute Gen AI Short Film
LumaQueenReverse-Engineered
The Recipe
Prompt
Original prompt unavailable. Analysis provided by AI.
A cinematic military science-fiction short film set in a dystopian future where massive battalions march toward an inevitable conflict. Futuristic soldiers in worn armor move through industrial battlefields filled with smoke, debris, and distant explosions. Wide establishing shots show the scale of the battlefield and endless ranks of troops. Medium shots focus on individual soldiers preparing for combat, checking weapons, standing silently in formation. Close-ups capture tension through body language rather than dialogue. Camera movement is cinematic and grounded, slow tracking shots mixed with restrained handheld motion. Lighting is harsh and dramatic, with strong contrasts from fire, searchlights, and overcast skies. The tone is serious and somber. No hero glorification. War feels mechanical, relentless, and impersonal. Visuals are ultra-realistic with filmic color grading, subtle grain, and consistent military design across scenes. Emphasize scale, repetition, and inevitability. End on an unresolved moment that suggests the battle is only beginning.
Workflow & Recipe
The film began with a simple military sci-fi concept designed to scale visually rather than rely on complex character arcs. Key characters, uniforms, and environments were first established using text-to-image generation to lock visual identity. These images were then animated into action sequences using video models, while large-scale battle scenes were generated directly from text prompts to prioritize scope and movement. Dialogue and narration were created using AI voice tools, and music was generated separately to drive pacing and emotional weight. The final short was assembled by selecting the most coherent generations, tightening pacing, and prioritizing visual continuity where possible, accepting minor inconsistencies as part of a first long-form AI filmmaking experiment.
The Stack
More Like This
