Prompt for AI video generation

Act as a director of photography who's an expert in AI video generation. Write prompts to create a clip of [length, for example 8 seconds] in [tool, for example Sora, Veo, Runway or Kling]. Context: - Scene: [what needs to happen] - Use: [for example an ad, a website background or a social video] - Style: [for example realistic cinematic, 3D animation or an 80s look] - Format: [aspect ratio, for example 9:16 or 16:9] Task: 1. Write a main prompt that describes, in this order: subject (with concrete features and wardrobe) and their action, setting, shot type and lens, camera movement with direction and speed (for example a slow dolly-in), lighting, color palette and pacing. 2. If the tool generates audio, describe the sound: ambience, effects and, if there's dialogue, short lines in quotes saying who speaks. 3. Write 2 variants: one with a different camera movement and one with a different mood. 4. If the idea needs more than one shot, split it into a list of short shots, each with its own prompt, repeating the same description of characters and location to keep continuity. Rules: - One main action per shot: models struggle with overloaded scenes. - Avoid on-screen text (add it later in editing) and complex physical movements: they often come out wrong. - If a character or product must stay consistent, suggest starting from a reference image. - Don't include real people, trademarks or the styles of living artists.