A practical pre-generation checklist
Before opening an AI video generator, write a shot brief. A clear brief makes it easier to compare outputs without changing several variables at once.
1. Define one visible action
Choose one subject, one setting and one observable movement. Example: a ceramic mug on a wooden desk beside a window; steam rises gently; the camera stays still. Avoid combining a spinning product, a moving camera and a changing background in your first test.
2. Separate visuals from sound
Write the visual prompt first. Keep dialogue, music and sound-effect notes separate unless the selected model explicitly supports them. A quiet desk scene might use room tone and a short ceramic tap. Not every model generates synchronized audio.
3. Build a continuity sheet
Shot A establishes the mug and window. Shot B moves closer to show steam above the rim. Preserve the mug color, handle direction, lighting and desk surface in both prompts. Use reference images only when you have permission.
4. Compare one variable
Save each prompt and its settings. Compare a static-camera version with one slow push-in while keeping the subject and lighting unchanged. Inspect object shape, movement and continuity before selecting an output.
5. Review before sharing
Check hands, text, logos, reflections and sudden changes frame by frame. Leave space for captions. Confirm rights to images, music and likenesses. Label synthetic footage where appropriate and never present it as evidence of an actual event.
Reusable brief
Subject: [one subject]. Setting: [location and light]. Action: [one action]. Camera: [static or one movement]. Continuity: [details to preserve]. Sound: [separate notes]. Delivery: [aspect ratio and duration].
About this guide
Created by the team behind Flow AI Video. This independent planning guide does not promise that every model supports every feature. Visit the official website with the flow ai video button above. We are not affiliated with Google.