Does this tool upload my photos to a server?
No — never. All processing happens entirely in your browser using WebGPU and ONNX Runtime Web. Your photos are read from your device, processed on your local GPU, and the result is saved back to your device. Nothing is transmitted over the network. You can verify this yourself: open your browser's DevTools Network tab and watch — no image upload request is ever made.
Does it work without an internet connection?
Yes, after the initial model download. The first time you select a model, it downloads once and is stored permanently in your browser's IndexedDB cache. From that point on, you can disconnect your Wi-Fi, process images offline, and everything works exactly the same. The app also works as a PWA — install it to your home screen for a fully offline experience.
Is it really free? Are there hidden costs?
Completely free. No subscription, no watermark, no account required, no export limits. Because the AI runs on your hardware rather than our servers, there are no cloud GPU costs to pass on to you. Full-resolution downloads are always free.
Which AI model should I choose?
Fast Mode (4 MB) is best for quick edits on older devices or simple subjects. Balanced Mode (43–88 MB) handles most portraits, products, and animals well. Maximum Quality (176 MB) is for professional use — fine hair, fur, translucent materials, and e-commerce where every pixel matters. Start with Balanced; upgrade to Max if you need finer edges.
How does browser-based background removal work?
WebGPU gives your browser direct access to your graphics card. ONNX Runtime Web runs a segmentation neural network (the same architectures used by cloud services) on that GPU locally. The model identifies which pixels are foreground and which are background, produces an alpha mask, and composites it with your original image to create a transparent PNG — all without leaving your browser tab.
Can I use it for passport photos?
Yes. After removing the background, click the 'Passport Photo' button in the toolbar to open the passport photo editor. It supports US (2×2 in), UK (35×45 mm), EU/Schengen (35×45 mm), Australian (35×45 mm), Indian (51×51 mm), and China Visa (33×48 mm) templates. You can set the correct background color per country, position and resize your face, and download either a single photo or a print-ready collage sheet for a 4×6 or A4 print.
What happens to my photos after I close the browser tab?
They are gone. Processing happens in temporary browser memory (RAM and GPU VRAM). When you close the tab, that memory is released. The only data that persists in your browser is the AI model weights in IndexedDB — not your images. No copy of your photo is retained anywhere, on your device or otherwise, after the session ends.
Does it work on mobile devices?
Yes. WebGPU is supported on Android Chrome and iOS Safari 18+. On phones with a dedicated GPU or NPU (Apple A-series, Qualcomm Snapdragon 8-series), processing is fast. We recommend the Lite or Balanced model on mobile — the 176 MB maximum quality model is better suited to desktop hardware.
Can I process multiple images at once?
Yes — use the Image Compressor at /compress for batch resizing and compression of up to 10 images. For background removal, the model loads once and processes each image in sequence. Batch background removal processes images one after another without re-downloading the model between runs.
What image formats are supported?
You can upload JPEG, PNG, and WebP images. Output is always a transparent PNG (the only format that supports an alpha channel for transparency). The image compressor additionally supports JPEG and WebP output.
Can I use the results commercially?
Yes. There are no restrictions on how you use the output images. The tool is free for personal and commercial use — product photos, marketing materials, client work, e-commerce listings, anything. No attribution required.
Why does the first run take longer than subsequent ones?
The first time you use a model, the browser downloads the ONNX model file (4–176 MB depending on the model you choose) and stores it in IndexedDB. This download is a one-time cost. On all subsequent runs — even across sessions and even offline — the model loads from the local cache in seconds.
Which browsers are supported?
Chrome 113+ and Edge 113+ have full WebGPU support and are recommended. Safari 18+ on macOS and iOS also supports WebGPU. Firefox supports WebGPU behind a flag. If WebGPU is unavailable, the tool falls back to WebAssembly (WASM) execution on your CPU — slower, but still fully local with no upload.
How is this different from remove.bg or Canva?
The fundamental difference is where the AI runs. remove.bg, Canva, and Adobe Express all send your image to their servers for processing — and charge for high-resolution exports. BG Remove Free runs the AI on your device, so your photos never leave your machine, full-resolution exports are always free, and it works offline. The underlying model architectures are comparable; the privacy and cost model are completely different.