Lore

Pinterest reference-image creator build

# Pinterest reference-image creator build

A way to create a new, realistic-looking AI creator who holds your product convincingly. Start from a real photo of the right type of person in the right setting, merge in the product, and use the result as the starting image for the video model.

Steps

  1. Find the base photo on Pinterest. Search for a specific person in the product's setting, for example 'UGC makeup table creator'. Keep clicking through images until you find a good example.
  2. Combine with the product in Nano Banana Pro. Drag the Pinterest photo in together with your product image and have the AI put the product somewhere in the scene.
  3. Change the face. Prompt along the lines of 'change her face so it looks like XYZ', so the result is a new creator rather than the person in the photo.
  4. Use the result as the start image in a video model: Sora 2 Pro, Seed Dance or Kling.

Why it works

The image the video model starts from already contains the product in a plausible setting, so the model doesn't have to invent how the creator holds it. This addresses the first problem in AI UGC three-level ladder (beginner, intermediate, expert), an awkward hold and a creator who doesn't match the brand.

Related

all-models-platform-sora-2-pro-creator-brief, Product-in-hand image upload, Writing the video prompt from the reference image