Car background removal AI: one backdrop for every vehicle

Car background removal AI takes each vehicle photo shot on the lot or the street and cuts the car out, so every car in your inventory can sit on the same showroom or plain backdrop. There are two routes: a background remover returns the car as a transparent PNG at one price per image and you place it on your backdrop, or an image-edit model redraws the photo with a new setting while your prompt says the car must not change.

Below is how each route works with Sume, from the RMBG 1.0 schema in the Sume API reference (served by the API reference docs), the Image API, and Generation admission, all read on 2026-09-29. The general how-to for any subject is how to change the background of a photo with AI; this post is the dealer-inventory version.

How do I remove the background from car photos with AI?

Send one request per photo to Sume RMBG 1.0. The only required field is image_url, a public HTTPS image URL; there is no model to pick. With mode: async the response returns at once with a status URL to poll, and the finished result is a PNG with alpha.

curl -X POST "https://api.sume.com/v1/rmbg-1.0/remove" \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "image_url": "https://example.com/inventory/stock-1042-front-34.jpg",
    "mode": "async"
  }'

How do I put every car on the same backdrop?

Pick one backdrop file for the whole lot, then choose how the car gets onto it:

• Composite the cutout yourself: place each transparent PNG on your backdrop in an image editor or a script, at the same position and scale for each angle. The car comes from your photo; only its surroundings are removed.
• Or edit with references: send the car photo to POST /v1/images in input_references on a model that takes references, such as ChatGPT Image 2.5 (up to 16), with aspect_ratio: "auto" and a prompt like “place this car in a bright white showroom; keep the car, its color, wheels, badges, and plates exactly the same”. The result is a redrawn image of the whole scene, so compare it with the original.
• Use the same backdrop and the same prompt wording for every angle of every car, so the listing grid stays consistent.
• A car pasted onto a floor with no shadow under its tires can look like it floats. Add a shadow to a product photo covers adding one back.

How do I process a whole inventory at once?

Each photo is its own paid job, so a car shot from 20 angles is 20 requests, and 50 such cars are 1,000. Jobs beyond your workspace's concurrency limit are accepted as queued rather than refused, until the queue itself is full and a new submit fails with 429 queue_full. Remove background from images in bulk covers the per-plan limits, pacing, and polling.

How much does car background removal cost?

A cutout is listed at $0.0225 per image, plus a 5.5% agent fee by default, and the price does not vary by image size. A reference edit is billed per completed image at the model's endpoint price instead.

What should I check on each car photo?

The docs make no promises about fine detail, so look at each file before it goes live:

• The edge of the car: mirrors, antennas, spoilers, and wheel spokes.
• Windows: whether the old background still shows through the glass.
• On edited images: badges, plates, trim, and paint color against the real car, since the model redraws the scene.
• Truthfulness: Sume's terms bar using the service to mislead people about the authenticity of generated media, so the listing should show the car as it is.