Guide 01 · 9 min read
How to Write AI Video Prompts That Control the Shot
A practical framework for subject, action, camera, light, timing, audio, and model-specific constraints.
- Prompt anatomy
- Camera language
- Revision checklist
Field notes for AI video makers
Video Prompt Guides
Learn how to turn an original idea or a real reference clip into production direction an AI video model can follow. These guides focus on observable decisions: subject movement, shot order, camera, composition, light, timing, transitions, dialogue, and sound.
Guide 02 · 8 min read
Video to Prompt: A Shot-by-Shot Reference Workflow
Learn how to analyze a reference clip without confusing observation, inference, and creative rewriting.
- Evidence capture
- Timeline breakdown
- Model adaptation
Start with the decision you need to control
Prompt writing becomes easier when you stop asking for a general “cinematic” result and identify the decision that matters most. Is the shot about a person’s motion, a product reveal, a sound cue, a transition, or the continuity between several compositions? Put that decision at the center and let every other instruction support it.
When you have an existing clip, observation comes before writing. Record what is visible and audible without inventing equipment or production history outside the frame. Then translate the evidence into the syntax your model rewards. Kling needs legible physical motion; Veo benefits from sourced and timed audio; Seedance needs continuity; Sora benefits from world and composition; Runway often rewards a concise single shot.
The two guides above cover both starting points. Use the prompt anatomy guide when you are creating from zero. Use the reference workflow when you need to understand why a real edit works before designing a new version.