Prompts are the difference between a video that ships and a video that goes in the trash. The good news is that Videna rewards short specific prompts more than long flowery ones. Here is the structure that works.
The three sentence formula
Sentence one names the setting. Sentence two names the action with the product. Sentence three names what the camera does. Three sentences is enough. Five is fine. Ten is too many because the model starts ignoring details when the prompt gets long.
What to be specific about
Be specific about the time of day, because lighting changes everything. Morning kitchen, late afternoon park, evening living room. Be specific about the person, if there is one. Age range, mood, what they are wearing. Be specific about how the product is used. Twists, pours, slides, places. Be specific about the camera. Steady, push in, pull out, handheld follow.
What to leave out
Leave out adjectives that describe quality, like beautiful, stunning, perfect. The model already tries to make things beautiful, and adding the words mostly produces saturated grading you do not want. Leave out brand names and celebrity names. Leave out anything you cannot point at on screen.
A working example
Bad prompt: A beautiful woman uses our amazing skincare product in a luxurious bathroom. Good prompt: A woman in her late twenties stands at a marble bathroom counter in soft morning light. She unscrews the dropper, places three drops on her fingertip, and presses them gently into her cheek. The camera stays at face height, then pushes in slowly on her hand.
The good prompt does not promise quality. It describes a moment. The model handles the quality.
When to iterate
If the first clip is close but the camera is wrong, only change the camera sentence. If the person looks wrong, only change the person sentence. Changing one variable at a time turns each generation into a real test. Changing three variables at once turns generations into a roulette wheel that burns credits without teaching you anything.
Save your best prompts as templates. Most merchants end up with two or three prompt formulas that they reuse for every product, swapping out the setting and the action while keeping the camera language identical.