Frequently Asked Questions

Open the editor at /upload/, drag a photo into the drop zone (or tap Upload Image on mobile), and the cutout finishes in two to four seconds for a typical 12-megapixel JPEG. There is nothing to install, no plugin to enable, and no account to create. The result appears side-by-side with the original so you can flip between them, then download as a transparent PNG, recolor the background, or send the image straight into the bulk processor at /bulk-background-remover/ if you have more files queued up.

JPEG, PNG, WebP, GIF (first frame), BMP, and HEIC/HEIF photos exported from iPhone all work. There is no client-side dimension cap — phone photos at 48MP, scanned posters, and oversized product shots upload as-is. The export defaults to PNG with a transparent background; switch to JPEG with a solid colour, or WebP for a smaller file, in the Download dialogue. Animated GIFs and multi-page TIFFs are processed as a single still — for video frame extraction, use a desktop tool first and upload the frame you want.

Yes — the editor is mobile-first. iOS Safari, Chrome on Android, Samsung Internet, and Firefox Mobile are all supported on the latest two major versions of each. The interface adapts to a tab bar at the bottom, supports pinch-to-zoom, long-press for the context menu, and lets you upload directly from your camera roll or take a fresh photo. iPhone HEIC photos are decoded in-browser with no manual conversion. Add /upload/ to your home screen for a near-app launch experience that keeps your last image cached.

Yes — the bulk processor at /bulk-background-remover/ accepts up to 100 images per session and queues them through the same engine the single-image editor uses. Up to three images run in parallel and the rest wait in line so you don't overwhelm the connection. Each result is downloadable individually or as a single ZIP when the batch finishes. The queue survives page reloads via local storage, so a dropped Wi-Fi signal won't lose your work — refresh and the unfinished items pick up where they stopped.

No. There is no account, no email gate, no free-trial countdown, and no credit card on file. Land on the page, drop an image, get a transparent cutout, and leave. Settings such as your last-used background colour and the panel layout are remembered in your browser via local storage so a return visit feels familiar, but nothing about you is stored on a server. If you clear site data in your browser, the next visit starts fresh — there is no profile to restore because none exists.

The editor on /upload/ keeps the cutout on a transparent layer and exposes the full layer panel: pick a solid colour, gradient, or stock background; drop in your own photo as a backdrop; add text overlays with custom fonts; apply drop shadows, reflections, outlines, or blur; and resize the canvas to platform-specific presets such as 1:1 for Instagram or 1600×2560 for Amazon. The Magic Brush handles the last few pixels of stray hair or fur with a paint-style erase and restore tool that respects existing edges.

The PNG you download matches your upload pixel-for-pixel — there is no downscale, no quality slider locked behind a paywall, and no "HD upgrade" prompt. A 6000×4000 product photo comes back as a 6000×4000 transparent PNG. The only exception is the animated GIF preview thumbnail rendered above the canvas, which is a small derivative used for in-browser display. JPEG and WebP export options also preserve full resolution; choosing one of those just changes the container, not the pixels.

The default segmentation model is tuned for portrait hair, animal fur, plant foliage, and soft fabric edges that traditional thresholding misses. Wispy strands and motion-blurred edges retain their alpha gradient instead of clipping to a hard line. When the auto pass leaves a few pixels behind — common with white-on-white setups or chrome bumpers — open the Erase / Restore tab and use the Magic Brush to add or subtract from the mask. Brush size, hardness, and pressure all respond to a stylus on iPad and Surface.

Mostly yes for low light, partly for motion blur, and yes for noise within reason. The model is robust to ISO grain up to roughly 6400 on a phone sensor and to the kind of indoor lighting where a subject is darker than the background. Severe motion blur — a moving subject where the edge is several pixels wide — is the failure case; the cutout will be soft along the blurred axis. For best results on tricky shots, expose for the subject rather than the background and avoid HDR composites that fight the alpha edge.

Shoot on a single high-contrast background — pure white sweep, matte black, or chroma green — with even diffused lighting and no harsh reflections from the product itself. Avoid jewellery on a glossy reflective surface; the model can mistake the reflection for a second object. After the cutout, open the BG Removal accordion and nudge Edge Refinement → Edge Shift one or two pixels inward to remove residual halo, then add Feathering of one pixel to soften the transition. Save the settings as a preset so future products in the same series export identically.

Yes — save the current background, effects stack, and canvas size as a Saved Template in the Templates panel. The bulk processor on /bulk-background-remover/ then applies that template to every image in the queue, so a 50-shot product run exports with the same shadow, the same canvas crop, and the same JPEG quality without manual re-entry. Brand Kits hold your colour palette so the same brand colours appear across every project. Recipes (in beta) extend this to a full one-click state bundle.

PNG is lossless and stores a per-pixel alpha channel; JPEG is lossy and has no alpha at all. A 2 MB JPEG product shot routinely exports as a 6–10 MB PNG because the transparency information has to be stored exactly. To shrink the file: switch the export format to WebP (often half the size at visually identical quality) or flatten onto a solid background and re-export as JPEG. For e-commerce platforms that require small file sizes, the Resize panel lets you crop and downscale before export rather than after.

The default path runs on a server at api.remove-bg.io because the production segmentation model is too large to ship to every browser. The image is uploaded over HTTPS, processed, returned as a transparent PNG, and immediately deleted — there is no persistent storage of the original or the result on the server. If your network is offline, the editor automatically falls back to a smaller U2-Net model (about 175 MB) that runs entirely in your browser via WebAssembly, so the photo never leaves the device. The fallback is also used when our server is rate-limited.

Uploads are processed in memory and deleted as soon as the response is sent — typically within a few seconds. There is no archive, no analytics dataset built from your photos, and no human review queue. Logs retain only the request size, latency, and HTTP status code for capacity planning; the image bytes themselves are never written to disk. Browser-side, the original and the cutout are kept in IndexedDB so you can switch between recent images in the editor, but that storage is local to your device and clearable from your browser's site-data settings.

No. Your uploads are not retained, not labelled, and not added to any training corpus. The segmentation model used for production runs is trained on licensed and publicly available datasets, not on user submissions. There is no opt-in or opt-out toggle for training because there is no training pipeline that touches your data. If a future model improvement requires user-contributed examples, that program would be opt-in, separately consented, and disclosed in the Privacy Policy at /privacy-policy/ — never silently enabled.

Yes. The service collects no account information, no email address, and no payment details, which removes most categories of personal data from scope by design. Cookie-based analytics are limited to anonymized page-view counts with IP truncation and respect Do Not Track. Per Article 17 of the GDPR you have a right to erasure, but because nothing about you is stored server-side there is nothing to erase — clearing your browser cache removes the only local copy. Read the full breakdown at /privacy-policy/ and contact privacy@remove-bg.io with formal requests.

You do. The cutout is a derivative of the image you uploaded, and copyright in the result follows copyright in the source — if you owned the original, you own the result, and you can use it commercially without paying us a royalty. The service does not assert any rights over your output, attach any watermark, embed any tracking metadata, or require attribution. The only restriction we ask you to respect is that you upload images you have the right to edit; uploading copyrighted photos you do not own is a misuse of the service, not an issue with the cutout itself.

Genuinely free for personal and commercial use, with no credit card, no trial countdown, and no daily quota for typical browser use. The free tier covers full-resolution downloads, the bulk processor up to 100 images per session, every editor feature, and every output format. There is no "premium" plan being held back; the entire product is the free product. The infrastructure is funded by anonymous, contextual display advertising on the marketing pages — never on the editor itself, never on the download — and by the API tier for businesses that need volume.

No. The PNG, JPEG, and WebP exports are clean — no watermark, no corner logo, no embedded credit, no metadata stamp pointing back to remove-bg.io. The only thing in the file is the image data and a standard colour profile. This is a deliberate product commitment: a free background remover that adds a visible mark would be useless for the e-commerce, marketing, and personal-content workflows the tool exists to serve. If you want to add your own logo to the result, the editor supports text and image overlay layers in the Layers panel.

No. There is no account gate on full-resolution downloads, no email-for-HD prompt, and no "upgrade for the original size" upsell. The image you upload comes back at the same pixel dimensions, in your chosen format, with the Download button reachable in one click from the editor. Many free background removers downgrade the resolution unless you create an account or pay; this one does not. The non-negotiable promise on the homepage — "no sign-up, no watermarks, HD downloads free" — is what the product actually delivers.

Yes. The terms allow commercial use of cutouts you generate from images you have the right to edit — product photography for your store, marketing assets for paid campaigns, design work for paying clients, social media for a monetized account, prints sold at a market. There is no separate commercial license to purchase. The only category that needs explicit permission is reselling the service itself (for example, wrapping the API and charging end users) — for that, contact api@remove-bg.io. See the full Terms of Service linked from the footer for the exact wording.

Yes — the API runs at api.remove-bg.io and accepts a base64-encoded image with HMAC-signed headers and a small proof-of-work token to deter automated abuse. Response times average two to four seconds for a typical 12 MP image and the endpoint returns a transparent PNG plus a JSON metadata payload. There is a free tier for development and several paid tiers for production volume. Contact api@remove-bg.io for credentials and full reference documentation; example integrations in Node, Python, and PHP are linked from /developers/.

The latest two major versions of Chrome, Edge, Safari, Firefox, and Opera on desktop, and the equivalent mobile builds on iOS 16+ and Android 10+. The editor needs WebAssembly, IndexedDB, and Canvas 2D — all standard since 2020 — and uses WebGL for accelerated filters when available. Internet Explorer is not supported and never will be. Older Chromebooks and entry-level phones that struggle with the in-browser fallback model still get the cloud path, which carries almost all of the work, so the experience degrades gracefully rather than failing outright.

Yes — the editor targets WCAG 2.1 AA. Every interactive control has a visible focus ring, a descriptive label, and a keyboard equivalent. Sliders accept arrow-key fine adjustment, the layer panel exposes a roving tab index, the canvas has an alt-text region that announces the current selection, and modal dialogues trap focus correctly with Escape returning to the trigger. Reduced-motion preferences are honoured for slide-in animations. If a specific assistive-technology pairing breaks, file a report at /help/ and tag it "accessibility" — these issues are prioritized.

Open the Resize panel in the editor and pick from 26 platform presets (Instagram square, Amazon main image, Etsy listing, LinkedIn cover, YouTube thumbnail, Shopify, eBay, plus print sizes) or type custom width and height between 50 and 8000 pixels. The cutout reflows to the new canvas, and the export keeps the alpha channel as a transparent PNG so you can drop the result onto any backdrop later. For platform-specific guides, see /tools/ for the full per-platform tool list and /blog/ for walk-throughs of e-commerce listing requirements.

Yes — the transparent PNG output drops into any image editor that understands alpha channels, which is every modern one. In Photoshop, drag the PNG onto the document and it imports as a layer with transparency intact. In Figma, paste it directly into a frame. In Canva, upload it through the Uploads panel. For automated workflows, the API can be called from a Photoshop script via Adobe ExtendScript or from a Figma plugin via the standard fetch API. There is no proprietary plugin to install and no file format conversion required.