You saw the demo: a matte-black Lamborghini that assembles itself from parts as you scroll. This is the exact pipeline that built it — no code written by hand, no site builder. Three AI tools, ~30 minutes: an AI image model for the hero shot, an AI video model for the cinematic clip, and an AI assistant (I used Claude) that writes the entire landing page as one HTML file.

What you'll build

A dark, premium landing page where scrolling drives a cinematic video of your product (scroll down = the video plays forward, scroll up = it rewinds), plus a manifesto that lights up word by word, stats that count up, a horizontal gallery, pricing and a CTA.

You need: an AI image generator, an AI video generator (Kling, Seedance, Runway — any), an AI assistant that writes code (Claude, ChatGPT), and a browser. That's it — the site is a single HTML file you open by double-clicking.

Step 1 — Create the hero image of your product

Don't jump straight to video. Generate a still image first — it locks the look of your product, the lighting and the composition. Then you animate that image.

Cinematic studio photography: [YOUR PRODUCT], [one accent color detail], low dramatic angle, vast dark concrete studio, single overhead light beam, sharp rim lighting, light haze in the air, empty dark space above the product for typography, ultra sharp, high-end commercial hero shot

The exact prompt that produced the Lamborghini demo (Nano Banana, 16:9, 2K):

Cinematic automotive studio photography: matte black Lamborghini Aventador with subtle orange accents, low three-quarter front view, parked in a vast dark concrete studio, single dramatic overhead light beam, sharp rim lighting tracing the angular bodylines, light haze in the air, empty dark space above the car for typography, ultra sharp, high-end car commercial hero shot

Keep it: dark background, one light source, empty space above the product (that's where the website headline will live).

Step 2 — Turn it into a cinematic clip (THE step most people get wrong)

Feed your image into the video generator and generate a 10-second clip.

⚠️ The #1 mistake: writing "cinematic video of my product" and hoping for magic. The AI doesn't know what should HAPPEN. You must describe the exact visual event happening to your product, like you're directing a commercial.

Bad: cinematic video of a lamborghini, epic, 4k

Good: dozens of car parts — carbon fiber body panels, angular headlights, orange brake calipers, black alloy wheels — float in from the darkness in slow motion and precisely assemble themselves piece by piece into the complete matte black Lamborghini

See the difference? The second one is a scene direction, not a vibe. Pick ONE clear transformation or camera move and spell it out:

• Assembly (what the demo uses): "the product assembles itself from its parts, pieces flying in and snapping into place, the final panel locks in and it stands complete"
• Orbit: "the camera glides in a slow constant-speed arc from the front three-quarter view to the side profile, the product stays perfectly still"