Bulk remove backgrounds for Etsy and e-commerce listings

The fastest free workflow for a product-photo queue

Start with folders, not an upload button. Put the original photographs in a read-only source folder and make a working folder for each SKU, colour, or collection. Consistent filenames such as linen-pouch-blue-front.jpg and linen-pouch-blue-detail.jpg survive ZIP downloads and make mistakes easier to catch than camera names such as IMG_4821.jpg.

Open FileGizmo’s background remover, choose up to 20 JPG, PNG, or WebP photographs, and select the edge model. Best edges uses the locally packaged BiRefNet Lite model and is the stronger default for products, handles, hair, irregular outlines, and difficult gaps. Fast portraits is a smaller model intended mainly for people and can help on a lower-powered device.

The queue runs sequentially. Each row moves from Waiting to Processing and then Done or Failed. That is deliberate: running several local AI inferences simultaneously could exhaust browser memory. One failed photograph does not erase completed work. When the queue finishes, download one ZIP containing every successful transparent PNG.

For a catalogue larger than 20 images, repeat the process. A 120-photo set is six batches of 20. Process, download, inspect, and archive each group before starting the next. The limit keeps local memory predictable; it is not a daily allowance, account quota, or credit counter.

What 120 background removals cost

Competitor prices, currencies, taxes, packages, and commercial-use terms change. The table below is an illustrative comparison checked on 19 July 2026, using the benchmark rate of approximately €0.36 per paid full-resolution image. Confirm the live checkout and licence before purchasing or publishing.

Workflow for 120 photos	Tool fee	Batch shape	Important condition
remove.bg at ≈ €0.36/image	≈ €43.20	Depends on the current paid plan or credit package	Official help distinguishes low-resolution free previews from paid full-resolution output; commercial use requires an eligible paid plan.
FileGizmo	€0	6 free batches × 20 images	Your device performs the work; processing time, memory, and battery are the practical costs.

The arithmetic is 120 × €0.36 = €43.20. It is not a promise that remove.bg will charge every buyer exactly that amount. Subscription tiers can lower per-image cost, pay-as-you-go packages can differ, and regional pricing can change. Follow the official pricing page for a purchase decision. FileGizmo’s current batch has no payment step, but it also has no hosted API, service-level agreement, or cloud automation.

Use transparent cutouts without making the listing misleading

Background removal is presentation work, not permission to invent the product. Etsy’s image rules emphasize original photographs that accurately represent the physical item. Keep the colour, texture, dimensions, inclusions, wear, and quantity truthful. Do not erase damage, alter a gemstone, hide a seam, or make a set look larger than what the buyer receives.

Etsy’s Seller Handbook describes a studio shot as a product on a plain background with plenty of light, but it also recommends a variety of photo types. A clean hero image can make a search thumbnail easier to read. Contextual views show scale and use. Close-ups reveal materials and workmanship. Variation photos prevent a buyer from confusing colours or sizes.

A transparent PNG is therefore an asset, not necessarily the final listing image. Place the cutout on a consistent brand background, export the final storefront format, and keep other photos that show the real setting and scale. Test the crop at thumbnail size because marketplace layouts may display several aspect ratios.

Inspect every edge before publishing

Automatic segmentation estimates which pixels belong to the foreground. Product photography contains difficult cases: translucent glass, reflective metal, fine chains, lace, fur, steam, shadows, pale objects on white, and holes inside handles. A batch tool can save repetitive waiting, but it cannot approve 120 images responsibly.

Review each PNG over both a light and dark background. Look for pale halos, clipped corners, missing straps, filled-in holes, and retained pieces of the sweep or tabletop. If one important image needs correction, process that photograph alone and open Refine edges. Restore paints original pixels back into the mask; Erase removes unwanted remnants. Zoom, pan, compare, undo, and redo are available for a single result.

Keep the originals. A later marketplace crop, seasonal redesign, or improved segmentation model may justify a fresh export. Editing an already cut-out PNG cannot reconstruct pixels that were not preserved.

Prepare the source photos for a cleaner batch

A model performs better when photography makes the boundary visible. Use even, diffuse light. Leave space around the complete product. Avoid motion blur and compression artifacts. Choose a temporary backdrop that contrasts with the product without casting a strong colour reflection onto it.

Photograph variations from a consistent camera position and distance. This makes a catalogue grid look intentional and helps you spot a result that is scaled or cropped differently. Do not rely on the cutout to repair bad focus, inaccurate colour, or an angle that hides important construction.

Large images increase local inference time. Close unrelated tabs and keep the FileGizmo tab active on a memory-constrained device. If quality mode is too slow, try a smaller batch before switching models. Resizing the source first can improve speed, but retain enough resolution for the marketplace’s current recommendation and your own zoom/detail needs.

Understand the privacy and performance trade

FileGizmo loads its model assets from the FileGizmo origin and then processes selected photographs inside a browser worker. The image queue is not posted to a background-removal API. That architecture helps with unreleased products, personalised items, customer photographs, and catalogue plans that a seller may not want stored by another processor.

Local does not mean costless. The first model load is large, inference consumes CPU or GPU resources, and a 20-image queue can take time. A hosted specialist may be the better choice for an automated inventory pipeline, an API integration, formal support, or a weak device. Choose based on the full workflow rather than a single price headline.

For a seller who needs a manually reviewed batch and can keep a tab open, the free path is straightforward: organize the source set, process up to 20 at a time, download the ZIP, inspect every cutout, and publish only images that represent the product accurately.