On-device batch workflow

Remove backgrounds from multiple images at once

Add a folder of images, run Remove Background once, and export the results as transparent PNGs. The operation runs on your Mac, so it does not use Gemini, Grok, or OpenAI credits and has no per-image provider fee.

Runs on your Mac Transparent PNG output No API key or model credits

· Tested with the current babbleBrush batch workflow

Remove the background from every image in a Batch Canvas, review the cutouts together, and export the completed set from one workspace.

The workspace

What is a Batch Canvas?

A Batch Canvas is a workspace for applying the same operation to a group of images. Your source files live together under Originals, and each batch edit creates a new set of results without replacing those files. You can see the status of every image, retry only the failures, and use the successful results as the starting point for another batch edit.

For background removal, add the images once, run Remove Background across the whole set, inspect the transparent cutouts together, and export them when they are ready.

Does batch background removal cost anything per image?

No. Remove Background runs locally on your Mac and does not send a request to an image-generation provider. You do not need an API key, and Gemini, Grok, or OpenAI will not charge you for the operation. Background removal is included as a babbleBrush feature; the babbleBrush app itself is a paid, one-time purchase.

Common uses

When batch background removal is useful

Product catalogues

Prepare a consistent set of transparent product cutouts for a shop, catalogue, or marketplace listing.

Mockups and compositions

Remove the original surroundings before placing several products into new scenes, layouts, or device mockups.

Marketing assets

Turn a folder of photos into reusable cutouts for social posts, advertisements, presentations, and campaign graphics.

Repeated client work

Process an entire delivery in one visible workflow instead of opening, editing, and exporting every image separately.

Step by step

How to remove every background in one batch

  1. Create a Batch Canvas

    Open the new-canvas menu, choose New Batch Canvas, and give the batch a useful name such as “Spring catalogue cutouts.”

  2. Add the original images

    Drop your files onto Originals or choose Add Images. Each source image gets its own place in the batch and remains unchanged.

  3. Choose Remove Background

    Create an edit from Originals and select Remove Background. babbleBrush shows how many images will be processed and marks the operation as on-device with no provider cost.

  4. Run the batch

    Start the operation and follow each image separately. Successful results appear as transparent PNGs, while any image that could not be processed is clearly marked as failed.

  5. Review and retry

    Inspect the cutouts at full size. If some items failed, choose Retry Failed to run only those images again without touching the successful results.

  6. Export the transparent PNGs

    Choose Export, select a folder, and save the completed set together. babbleBrush keeps recognisable source names and avoids overwriting files that already exist.

Why use a Batch Canvas?

  • Originals remain intact. Each processed file becomes a revision associated with its source image.
  • Progress is visible. Every image has its own queued, running, succeeded, failed, or cancelled state.
  • Failures are isolated. One difficult image does not invalidate the rest of the set.
  • Follow-up work can branch. A later batch edit can start from the background-free results instead of the originals.

Why there is no per-image fee

babbleBrush uses Apple's on-device Vision framework to find the foreground subject, places that subject over transparency with Core Image, and saves the result as a PNG. The image does not need to be sent to Gemini, Grok, OpenAI, or a babbleBrush server for this operation.

Apple Vision foreground-mask documentation

You can also ask an agent to run the batch

If Claude, Codex, OpenClaw, Hermes, or another MCP-compatible agent is connected to babbleBrush, it can create the Batch Canvas, add the images, start background removal, watch the progress, and retry failed items. You still see the same batch in the app and can review the results before exporting them.

Example request

“Create a Batch Canvas called Summer catalogue cutouts, add these product photos, remove every background, wait for the run, retry any failures once, and show me which images are ready to export.”

MCP is an open standard for connecting AI applications to tools and workflows. Official MCP introduction · babbleBrush agent capability reference

Where automatic removal needs review

Automatic foreground masks work best when the subject is visually distinct from its background. Closely matching colours, transparent or reflective packaging, soft shadows, fine fur or hair, holes in an object, and crowded scenes can create imperfect edges. Inspect important images at full size. If no foreground subject is detected, babbleBrush marks that item as failed so it can be reviewed or retried instead of exporting a misleading blank result.

Continue exploring

Edit images with Claude Code

Connect Claude to the same live babbleBrush canvas.

Remove image metadata

Inspect and remove EXIF and location data locally in the browser.

Get babbleBrush for Mac

One price, yours forever. A canvas for you and your agents.

Download for Mac