Skip to content
Scalpels home

Scalpels is officially launching on October 31, 2026. Special early adopter pricing is available now.

Matte

Any stack MIT licensed

Background removal on hardware you already pay for. Post an image, poll the job or request a signed callback to a registered destination, and fetch a transparent PNG from a Laravel Cloud app you own, on your bucket and your queue, with no per-image invoice. A classical engine and an ML model both ship in the binary.

Deploy Matte for free Or run it yourself, free Read the Docs
Replaces
Per-image background removal APIs
Runs on
Your Laravel Cloud account
Needs
Compute · Database · Queue · Object storage
Matte, installed and cutting out Placeholder — the launch recording replaces this

What it does

Three endpoints, callable from anything that speaks HTTP

  • Async by default

    Submit an image and get a job id; poll its status, then stream the finished PNG.

  • Sync for small images

    One flag runs the conversion in the request and returns the PNG inline.

  • Two engines, one binary

    An ONNX model by default, or classical GrabCut with no model, with fast, balanced and quality presets and edge tuning.

  • Signed completion callbacks

    Choose a registered callback destination and the worker posts the result with a bound HMAC signature.

  • Deterministic outputs

    Results are keyed by a hash of the input and its options, so a retried job overwrites rather than duplicates.

  • Per-app credentials

    Hashed bearer credentials per client, with plaintext revealed only when each credential is issued.

  • Laravel client included

    SDK

    Install the client and a facade gives you remove and removeSync, a polling job, and a completion event. Webhook receipt is opt-in and verifies the signature in constant time.

  • Runs on a plain Cloud worker

    The statically linked binary, the ONNX runtime and the default model are baked into the build artifact, so every instance carries them with no container or sidecar. A doctor command runs a real conversion to prove the runtime.

Matte vs. the incumbents

We rebuilt one row of their feature matrix

remove.bg, Photoroom and Cloudinary are good products, and Matte is not trying to replace them. It does one slice of what they do — plain background removal, on infrastructure you own — and deliberately does not do the rest. The question is not whose model is better. It is what you are actually paying for.

Matte feature comparison
Feature Matte Self-hosted, MIT remove.bg Hosted Photoroom Hosted Cloudinary Hosted, with CDN
Plain background removal over HTTP the row we rebuilt Does this well:Yours, unmetered Does this well:Included Does this well:Included Does this well:Included
Async jobs with signed webhooks the row we rebuilt Does this well:Poll or webhook Not built:Sync only Partly, or with a caveat:Async, no webhooks Does this well:Notification URL
Edge quality on hair and clutter we did not build this Partly, or with a caveat:Open model or GrabCut Does this well:Tuned model Does this well:Tuned model Does this well:Fine edges, hair
Output formats and sizing we did not build this Not built:PNG with alpha only Does this well:PNG, JPG, WebP, ZIP Does this well:Formats, resize, pad Does this well:Any format
Editing and generation we did not build this Not built:Removal only Partly, or with a caveat:Brush editor Does this well:Full editing API Does this well:Generative fill
Batch endpoint and CDN delivery we did not build this Not built:One image, you serve it Partly, or with a caveat:Desktop bulk, no CDN Partly, or with a caveat:No batch, no CDN Does this well:Bulk and global CDN
Where your images live Does this well:Your bucket Works, but on their terms:Theirs, briefly Works, but on their terms:Theirs, may train Works, but on their terms:Their storage
Cost of the next thousand images Does this well:Queue time Works, but on their terms:Per image Works, but on their terms:Per image Works, but on their terms:Usage plan
Rate limits Does this well:Only your queue depth Works, but on their terms:Per minute, by size Works, but on their terms:Not published Works, but on their terms:By plan
Who is on call for the tool itself Partly, or with a caveat:You are Does this well:They are Does this well:They are Does this well:They are
  • Does this well
  • Partly, or with a caveat
  • Not built
  • Works, but on their terms

The honest test

Open your current bill and ask what you actually touched last month. If these are all true, you are paying per image for a suite to use one row of it.

  • Your images are mostly single subjects on distinct backgrounds, and good beats best-in-class on hair.
  • You only need a transparent PNG back; you resize, convert and serve it yourself.
  • You already run a Laravel Cloud app with a queue and a bucket, or are happy to.

Then you lose very little by moving. Plain removal runs on infrastructure you already pay for, however many images you send it.

Keep paying them if

These are the parts we chose not to build, and we are not going to pretend otherwise.

  • You need best-in-class edges on hair, fur and cluttered backgrounds, every time.
  • You want shadows, AI backgrounds, relighting or any editing beyond removal.
  • You need a CDN, multiple output formats or a batch endpoint out of the box.
  • You need a contract, an SLA and someone to phone.

If the model quality or the editing suite is what you are paying for, Matte will not replace it. It replaces the invoice for plain removal.

Be honest about it

You don't need us to run Matte

The repository is public and MIT licensed. These are the steps, and you are welcome to spend the afternoon doing them by hand.

  1. 1

    Fork the Matte repository into your own GitHub organisation.

  2. 2

    Deploy the fork to Laravel Cloud with its own compute, Postgres, queue and object storage, with the binary provisioned at build time.

  3. 3

    Issue a token, install the Laravel client or post straight to the API from anything that speaks HTTP.

Or let Scalpels do it, free. It forks the app, provisions exactly what it needs on your Laravel Cloud account, and deploys it. The app and the infrastructure are yours either way — the only question is who spends the afternoon.

Deploy Matte for free

Want some help?

After you buy, we will pay for an hour with one of these experts to help you get everything up and running.

  • Artisan Build

    USA

    We run Scalpels, as well as a suite of other tools for developers and development teams.

    Agency Ask for help
    Ask Artisan Build for help

Matte pairs with the rest of the suite

Matte, Hone and Sink share one token registry, one install command and one environment per client, so a team already running either has the auth and provisioning story done.

Hone Sink Crate