AI Ad Creative — turn a product photo into a finished ad with one API call

Every store has the same problem. You have a clean product photo. What you actually need to run is an ad — that product sitting in a scene that feels premium, with a headline, a call-to-action button, a brand name and maybe a "20% OFF" badge, sized correctly for Instagram, a Google banner or a Flipkart listing. Getting from the first thing to the second normally means a designer, a Canva afternoon, or a $1–4-per-creative SaaS subscription. We built ad-creative so it's a single HTTP call that costs one cent.

What it does

POST /v1/image/ad-creative takes your product image plus a few fields of ad copy and returns a finished, ready-to-run ad. Under the hood it runs a real compositing pipeline, not a one-shot text-to-image guess:

- Cut-out. We matte your product off its background cleanly. - Scene. An AI generates a premium environment around your product — grounded, lit and scaled to match, with a defined contact shadow so the product sits in the scene instead of floating on top of it. - Your exact product, pasted back. The pixels of your product are preserved. We never re-imagine your item, so the watch face, the logo and the stitching are exactly what you uploaded. - Real typography. Headline, sub-headline, CTA button and badge are rendered as real HTML/CSS text — crisp, legible and on-brand. No AI-garbled lettering.

Why we built it

The two existing options are both bad in different ways. Pure text-to-image ad generators ("describe your ad") are fast but they re-draw your product into something that is like your product but not actually it — wrong logo, wrong shape, and ad text that comes out as melted nonsense. Traditional ad-creative SaaS gets the text right but charges $1–4 per creative and still makes you art-direct every variation by hand.

Our angle is to treat it as a compositing problem with a generative scene in the middle, so two things are true at once: the product is pixel-exact, and the text is real and editable. That combination is what makes the output something you can actually run, not just a pretty mock-up.

Quickstart

Grab an API key from the dashboard, then post your product image and copy:

curl -X POST https://api.pixelapi.dev/v1/image/ad-creative \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -F "[email protected]" \
  -F "headline=CARRY MORE, EASIER" \
  -F "subheadline=Everyday backpack, built to last" \
  -F "cta=Get yours" -F "brand=NOMAD" -F "badge=20% OFF" \
  -F "ad_format=meta_1080x1080" -F "style=luxury"

You get back a generation_id; poll GET /v1/image/{id} and the output_url appears when it's done, about twenty seconds later. Formats: meta_1080x1080 (square), meta_1080x1920 (story), google_1200x628, flipkart_1080x1080, meesho_1080x1080. Styles: luxury, minimal, vibrant, dark_luxury, festive and more. No product image? Omit it and you get a typographic sale/announcement ad on a generated background.

Prefer not to touch code? The same tool is in the web app and in the Lensora Android app — upload, type your headline, download.

Pricing

At 10 credits a call, generating a hundred ad variations for an A/B test costs a dollar. That's the only price at which this kind of thing actually gets used in production instead of reserved for one hero asset. Full reference on the AI Ad Creative page.