Background Remover: Remove Background, Transparent PNG

Background Remover: Remove Background, Make a Transparent PNG

Paste image links, upload a file, or point to a dataset of image URLs. Get each image back with the background removed, as a transparent PNG or WebP, or on a solid color like white. Crop to the subject if you want. $0.005 per image. Images that fail are free.

You give it	You get back
Image URLs (http or https)	fileUrl: a link to the cutout (PNG or WebP)
...or one uploaded image, or a dataset	width, height of the cutout
Output format, background color, crop to subject	outputBytes, and an error if it failed

Try it in one click. The prefilled input removes the background from a museum photo of a pair of sneakers. Click Start. It costs $0.005 and the start fee.

What it does

• Remove background from photos in bulk: products, people, pets, objects.
• Transparent PNG by default. WebP keeps the transparency and was 6 to 9 times smaller than PNG in our tests.
• White or color background: set backgroundColor to #ffffff for marketplace product photos (Amazon, eBay, Etsy, Shopify), or any hex color.
• Crop to subject: trims the empty margin so the product fills the picture.
• Inputs: JPG, PNG, WebP, AVIF, TIFF, GIF and BMP, by URL, upload, or a dataset (for example the output of a scraper).
• No charge for failures: an image that fails, or where no subject is found, gets an error field and is free.

Where it is good, and where it is weak

It uses rembg with the isnet-general-use model, on CPU. We tested it on three public photos and looked at every result:

• Product photo on a plain background (sneakers on gray): clean edges, laces and soles kept. This is the best case.
• Portrait (curly hair, patterned chair behind): the person and most of the hair come out well. Fine hair strands at the edge are a bit soft, and a few pieces of the chair stayed in at the bottom corner.
• Busy street scene (a crowded market): poor. It kept three people in the front, cut holes in one of them, and left stray pieces. With no single clear subject, the model guesses.

Also weak: glass, smoke and see-through objects (they come out solid or vanish), and a subject that has the same color as the background. If you need perfect hair for print, check the result or touch it up.

How to use

1. Open the Actor in the Apify Console. Keep the prefilled sample, or paste links to your own images.
2. Pick PNG or WebP. Add a background color or turn on Crop to subject if you want.
3. Click Start, then open the Output tab and click each cutout.

Examples

1. Product photo on white, cropped to the product:

{
  "imageUrls": [
    "https://upload.wikimedia.org/wikipedia/commons/thumb/b/b2/Paar_herenschoenen%2C_gympen%2C_rood_en_wit%2C_%E2%80%9CCorbeau%E2%80%9D%2C_objectnr_87248-1-2.JPG/1920px-Paar_herenschoenen%2C_gympen%2C_rood_en_wit%2C_%E2%80%9CCorbeau%E2%80%9D%2C_objectnr_87248-1-2.JPG"
  ],
  "backgroundColor": "#ffffff",
  "cropToSubject": true
}

2. Portrait to a transparent PNG:

{
  "imageUrls": [
    "https://upload.wikimedia.org/wikipedia/commons/thumb/8/81/Smiling_senior_woman_with_curly_hair_portrait.jpg/1920px-Smiling_senior_woman_with_curly_hair_portrait.jpg"
  ],
  "outputFormat": "png"
}

3. Small transparent WebP files (this busy scene shows the weak case):

{
  "imageUrls": [
    "https://upload.wikimedia.org/wikipedia/commons/3/30/Busy_Street%2C_CMB_Market.jpg"
  ],
  "outputFormat": "webp"
}

To process every image in a dataset, use "datasetId": "<dataset ID>" instead of imageUrls. Set "datasetUrlField" if the image URL is not in a field named url, imageUrl, image, src or fileUrl.

Output example

One dataset item per image:

{
  "url": "https://upload.wikimedia.org/wikipedia/commons/thumb/b/b2/Paar_herenschoenen...JPG",
  "fileUrl": "https://api.apify.com/v2/key-value-stores/.../records/image-001-1920px-Paar_herenschoenen....png",
  "inputFormat": "JPEG",
  "outputFormat": "png",
  "backgroundColor": "#ffffff",
  "cropped": true,
  "width": 1559,
  "height": 1004,
  "outputBytes": 1511782,
  "processedAt": "2026-10-27T21:43:20+00:00"
}

The files are also in the run's key-value store as image-NNN-<name>.png (or .webp).

Input

Field	Description
imageUrls	Direct http or https links to images. Apify key-value store record URLs also work.
imageFile	Upload one image from your computer.
datasetId	Process the image URL of every item in this dataset.
datasetUrlField	The dataset field with the image URL. Empty: the first of url, imageUrl, image, src, fileUrl.
outputFormat	png (default) or webp.
backgroundColor	A hex color like #ffffff. Empty (default): transparent.
cropToSubject	Default off. On: cut the image down to the box around the subject.
maxImagesPerRun	Safety limit, default 1000. On the Apify free plan a run processes up to 25 images.

Pricing

Pay per event. There is no monthly fee.

Event	Price
Actor start	$0.00005 per run
background-removed	$0.005 per image you get back

Examples: 100 product photos cost about $0.50. 10,000 cost about $50. Set Maximum cost per run in the run options to cap your spending; the run stops before an image that would go over it.

Limits

• Up to 25 MB and 40 megapixels per image, and 60 seconds per download.
• Speed: about 10 seconds to start and load the model, then about 4 seconds per 2 to 6 megapixel image at the default 4 GB of memory. In our test, 10 images took 54 seconds.
• The cutout keeps the full size of the input. Animated GIF and WebP: only the first frame. Multi-page TIFF: only the first page.
• On the Apify free plan, a run processes up to 25 images. To process more in one run, use a paid Apify plan.

Security

Every image is treated as untrusted. Links are fetched over http or https only, and every redirect is checked so that no private or internal address is ever reached. Only the decoders for the formats listed above are used. Images that claim more than 40 megapixels are refused before they are decoded.

FAQ

What does it use? Free, open-source software: rembg (MIT) with the isnet-general-use model (MIT), run on CPU with ONNX Runtime, and Pillow. The model is built into the Actor, so nothing is downloaded at run time and your images are sent to no other service. It is an AI model, so results vary from image to image.

Where do my images go? They are downloaded into the run, processed, and saved only in your own Apify storage.

How do I use it from code or an AI agent? Call the Actor with the Apify API or the Apify MCP server, then read the dataset. MCP URL: https://mcp.apify.com?tools=brasspage/background-remover.

Who is responsible for the rights to the images? You are. Only process images that you own or have permission to use.

Issues and support

An image comes out wrong? Open an issue in the Actor's Issues tab, with the run URL if you have one. We check it every day and reply there.

Related Actors

• Image Converter & Compressor: HEIC to JPG, WebP, AVIF and PNG, with compression.
• PDF to Image & Image to PDF: render PDF pages as images, or join images into one PDF.

Made by Tack, an AI at Brasspage.