Say the shot the way you see it.

Sometimes the shot you want is easier to say than to pick. Write it as a sentence, the way you see it, and the studio uses your words as written. Nothing is rewritten behind your back, so the detail you cared enough to type is the detail the render gets.

Words and cards work together. Whatever your sentence leaves out, the studio fills by choosing the matching cards around it: the outfit, the pose or the place. You get the freedom of describing a scene in your own words and the speed of picks for everything you did not mention.

Name what matters and leave the rest. "She laughs and looks over her shoulder on a beach at golden hour" gives the studio an action, a place and a light, and lets it pick a fitting outfit and pose. Keep the sentence about one moment, since a take runs up to ten seconds. If she is one of your characters, she keeps her accepted face.

Then choose the length and shape as you would with cards, see the price in credits and render. If the result is close but not right, change a few words or swap one of the cards the studio chose, and shoot again. Short, concrete sentences tend to land better than long lists of adjectives.

A take you write is kept the same way as a take you picked. Your gallery saves the clip with the sentence and the cards around it, so you can open it later and shoot the same scene again, somewhere new or in a different outfit. The words you wrote are part of the recipe, not lost after the render.

This is what makes a sentence useful for fan requests. Paste a request as written, keep her face, and render it. If a request sells, it is already a recipe you can run again. Developers get the same behavior through the API, by sending a sentence in the describe field of POST /v1/videos.