Build a directed text-to-video prompt.

Prompting can be difficult because there are so many variables: subject, motion, camera language, lighting, color, texture, and mood. Sometimes the fastest way to communicate the look you want is to start with a reference image.

Steps

1. Write a plain-language prompt describing the shot, action, and camera movement. Start simple, as if you were explaining the shot to another person.

2. Choose a reference image that matches the desired look. It can come from almost anywhere: a Google image search, a movie still, a screenshot, a mood board, your own photography, or any visual that captures the style you want. The reference can help define lighting, color, composition, texture, lens feel, and overall mood.

3. Use an LLM like ChatGPT or Gemini to turn your rough idea into a stronger Veo 3.1 prompt. Ask it to adapt your prompt to the style of the reference image and format it in a way Veo 3.1 is likely to understand well.

4. Keep the final prompt focused. Avoid asking for too many changes at once. Generate, choose lossless Veo export only where the selected model supports it, and review the result in your post-production workflow.