Use case · Marketing

Turn a campaign brief and a single product photo into a cinematic 16:9 product video. Your inputs feed a prompt, an agent directs the shot, and a text-to-video model renders a ready-to-post clip.

How it works

Four inputs and four nodes: a brief, an audience, a feature list, and the hero photo go in; a finished clip comes out.

graph LR brief["Brief (String)"] audience["Audience (String)"] features["Features (String)"] photo["Product photo (Image)"] prompt["Prompt"] agent["Director (Agent)"] i2v["Render (ImageToVideo)"] brief --> prompt audience --> prompt features --> prompt prompt --> agent photo --> agent agent --> i2v photo --> i2v
  1. Feed in the campaign. Three text inputs hold the brief, the target audience, and the key features. A product photo goes in alongside them as the hero shot. (e.g. “Aurora Trail smart fitness watch · active millennials who hike · GPS, heart-rate, adaptive coaching, water resistance”)
  2. Assemble the prompt. A Prompt node templates those inputs into a precise, rule-based request: one camera move, one subject action, lighting, color anchors, and hard constraints (no on-screen text, logos, or watermarks).
  3. Direct the shot. An agent turns the request into a concrete, cinematic prompt with framing, lens, and motion cues — the part most people get wrong by hand.
  4. Render the video. A text-to-video model animates the product photo from the agent’s prompt into a finished, ready-to-post clip.

One photo in, a clip out

The whole pipeline runs from a single product render. Same graph, any product.

Make it yours

  • Swap the video model. Veo, Seedance, Kling, Runway. Change one node, the rest of the graph stays exactly the same.
  • Retune the agent. Rewrite the system prompt for a luxury, playful, or hyper-technical tone without touching the pipeline.
  • Change the format. Aspect ratio, resolution, and duration all live on the video node. Go vertical for social in one click.
  • Reuse for any product. Drop in a new photo and brief, run it again. The workflow is the reusable part, not the output.

Models in this workflow

Called with your own keys. The bill comes from the provider, and you can switch any of them for a better model the day it ships.

Model Role Provider
Gemini 3.1 Pro Preview Writes the video prompt Gemini
Veo 3.1 Preview Renders the clip from the product photo Gemini

See Models & Providers to set up keys.

Next steps

</content> </invoke>