The prompt
Score 10 of 10, Excellent
Use it when You need a short clip showing one step of a real task, such as lifting a box safely or scanning an item at a checkout counter, for a training module, from a text-to-video model.
- {{person_description}}
- {{setting}}
- {{action}}
- {{training_topic}}
- {{key_object}}
- {{final_detail}}
- {{duration_seconds}}
- {{action_sound}}
Full prompt and why it works
Prompt
A medium shot at eye level of {{person_description}} in {{setting}}, {{action}}. The clip is for a workplace training module on {{training_topic}}, so the action is shown slowly, clearly and exactly as described, with the person's hands and {{key_object}} fully visible throughout. Camera: a static tripod shot, 35mm lens look, deep focus so the whole action is sharp, no cuts. In the final 2 seconds the camera slowly pushes in to a close-up of {{final_detail}}. Style and light: photorealistic, even, bright overhead lighting typical of {{setting}}, natural colors, realistic textures, plain work clothing with no logos or readable text. The look should be calm and documentary-like, in the style of a corporate training film rather than an advert. Timing and format: {{duration_seconds}} seconds, one continuous shot, 16:9. Sound: the quiet ambient sound of {{setting}} and {{action_sound}}. No music, no dialogue and no voice-over, because narration will be added in editing. Show only the steps described; if anything is unclear about the action, keep the movement simple and slow rather than adding steps. The frame contains only the one person and the objects the task needs, against a tidy, uncluttered background.
Variables
- {{person_description}}
- A specific but fictional person, for example "a warehouse worker in a high-visibility vest and work gloves". Do not describe a real, recognizable person.
- {{setting}}
- Where it happens, for example "a warehouse aisle with metal shelving".
- {{action}}
- The step to show, precisely, for example "bending at the knees to lift a cardboard box, keeping it close to the body".
- {{training_topic}}
- What the module teaches, for example "manual handling".
- {{key_object}}
- The object being handled, for example "the box".
- {{final_detail}}
- Where the camera ends, for example "the gloved hands gripping the box".
- {{duration_seconds}}
- A clip length your video model supports; check its settings.
- {{action_sound}}
- The sound of the action, for example "the scrape of the box sliding off the shelf".
Why it works
- Format. A static tripod shot with deep focus and one slow push-in keeps the whole action visible, which matters more in training than a cinematic move.
- Background facts. Sound gets its own sentences after the visuals, as Google's Veo guidance recommends, and leaves out music and voice-over so your narration can go on top.
- What good looks like. Generated footage can get the details of a procedure wrong, so the prompt asks for the action exactly as described; check each clip against your written procedure before using it.