A video model reads your prompt the way a camera crew reads a shot list. It does what is written and guesses the rest. These five habits leave it much less to guess, whichever model you pick in Skitify.
1. Start with the subject
Say who or what the shot is about before anything else. "A young woman in a yellow raincoat" works far better than "it is raining and somewhere there is a woman". Models give the most weight to what comes first.
2. Describe one clear action
"She opens the umbrella and looks up at the sky" works. Five things happening at once turn into a blur. One shot, one action, and the rest of the story goes into the next scene.
3. Direct the camera
Name the move and the framing: close-up, wide shot, slow push-in, handheld, orbit. Without it the model picks for you, and it usually picks a static mid shot.
4. Set the light and the look
"Golden hour", "neon at night", "soft window light", "film grain", "35mm". Two or three words of style do more than a paragraph of adjectives.
5. Keep it short and concrete
One to three sentences is plenty. Words like "beautiful" or "amazing" mean nothing to a model, so cut them and keep the ones it can actually draw.
A young woman in a yellow raincoat opens an umbrella on a neon-lit street at night, slow push-in, rain on the lens, film grain.
Stuck for an idea? Open any example on the Skitify home screen and press Repeat. Its prompt and settings go straight into your composer, ready to change.
