Kettle or air fryer product photos: cord, spout and counter scene

A kettle or air fryer photographs well on white and is dull there. Shoppers want it on a counter with a mug, a bowl of fruit, morning light. The risk is not the scene; it is the parts: a cord that vanishes, a spout that moves, a dial with the wrong number of marks.

Sume's Image API takes the studio shot as an input_references item with openai/gpt-image-2.5 and returns scenes around it.

Put the fragile parts in the prompt

Name the features that must survive: handle side, spout shape, lid, dial, power cord, base. Name the scene separately, and leave an instruction about scale: a countertop kettle is about the size of a bread loaf.

import os, requests

H = {"Authorization": f"Bearer {os.environ['SUME_API_KEY']}"}
resp = requests.post("https://api.sume.com/v1/images", headers=H, timeout=90, json={
    "model": "openai/gpt-image-2.5",
    "prompt": "Place this exact electric kettle on a kitchen counter next to a mug and a bowl of lemons, morning light from a window. Keep the handle side, the spout, the lid and the base exactly as in the reference. The power cord runs behind the kettle to a socket.",
    "input_references": [
        {"type": "image_url", "image_url": {"url": "https://example.com/kettle.jpg"}}
    ],
    "aspect_ratio": "auto",
})
resp.raise_for_status()
if resp.status_code == 202:
    raise SystemExit("job envelope: poll data.status_url")
body = resp.json()
print([d["url"] for d in body["data"]], body.get("usage"))

What goes wrong first

A note on safety imagery

Do not generate scenes that show a hot appliance in unsafe use, such as a kettle on an unlit stove or an air fryer near a curtain. Store the final file with the billed usage.cost, and keep a plain studio shot on the page for anyone who needs the real thing.

Before you run a catalog

Run one SKU end to end first. POST /v1/images returns the images directly when it finishes within 30 seconds; past that it returns a 202 envelope with status_url and result_url, which the jobs and results guide explains. Handle that branch before you loop over a catalog, and write each result's URL and usage.cost to a file keyed by SKU, so a failed run restarts where it stopped and nothing is paid for twice.