Video Prompts

Text-to-video went from party trick to production tool in about eighteen months. The models have names you'll need to keep up with — Veo, Kling, Pika, Runway — but the skill that matters hasn't changed: describing a shot so precisely that a model can film it. That's what this collection is: shot lists, camera language, and pacing notes that turn "a cool video" into an actual video.

Tool prompts vs. use-case prompts

Here's the trap with video prompts: models churn. Sora dominated the conversation, then shut down in March 2026 and orphaned a mountain of "Sora prompts" overnight. The prompts that survive model churn are use-case prompts — "UGC-style product testimonial," "ASMR process video," "cinematic product ad." Those describe durable formats, and they port from Veo to Kling to whatever launches next year with minor tweaks. This collection pairs tool-specific prompts (for the models that matter right now) with use-case twins built to outlive them.

How to write a video prompt that actually renders

• Lead with the shot, not the story. "Slow dolly-in on a perfume bottle" beats "a luxurious feeling." Models render cameras, not adjectives.
• One shot per prompt. Video models handle 5–10 second single shots far better than multi-scene epics. Chain shots together in editing, not in the prompt.
• Specify camera language: shot size (close-up, wide), movement (dolly, pan, orbit, static), lens feel (shallow depth of field), and frame rate feel (slow motion vs real-time).
• Lock the look: lighting, color palette, film grain, era. "Golden-hour, 35mm film grain, warm palette" is a complete look in nine words.
• Describe motion explicitly. What's moving, how fast, in what direction. Still-image habits ("beautiful landscape") produce beautiful slideshows.
• For image-to-video: describe the motion you want from the still — "slow push-in as steam rises, camera drifts left." The model needs a direction, not just a scene.

The formats people actually want

The Veo cinematic product ad is the commercial workhorse — a 6-second spot structure any brand can use. Kling talking-head avatars cover the explainer/UGC lane where natural lip-sync matters. The UGC testimonial prompt is the TikTok-native format: shaky, authentic, high-converting. Image-to-video zoom turns any still into a cinematic opener. ASMR process videos are the satisfying-loop genre that prints watch time. And the parody infomercial is pure shareability — the format behind half the viral AI ads you've seen this year.

A note on faces and consistency

Video models still struggle with the same face across multiple shots and with readable text in-frame. Plan around it: keep faces small or in profile for multi-shot sequences, and never rely on the model to render your brand name or a phone number — composite text in editing.