Background Remover
Remove image backgrounds with HD quality in your browser. Preserves hair, soft edges, and fine detail. Full resolution, transparent PNG, no signup, no watermark.
Your photo stays inside this browser tab, never reaches a server, and nothing is saved to your computer. The first visit prepares the AI model in your browser cache (about 115 MB, one-time). After that, every photo processes instantly.
Drop your image here, paste, or click to browse
JPG, PNG, WebP, AVIF, GIF, BMP. Larger photos are auto-resized to 2048 px on the longest side for speed (mask quality is unaffected).
Lossless, transparent background supported.
lightbulbTip
Drop your image, then play with the background presets to see the cutout against transparent, a brand color, a gradient, or another photo. Slide the divider on the preview to compare against the original. PNG keeps transparency; pick JPG for the smallest file with the chosen background baked in.
What is the Background Remover?
A background remover detects the main subject of a photo (person, product, animal, object) and erases everything else, leaving a transparent or replaceable background. This tool uses BiRefNet, an open-source segmentation model that runs entirely in your browser via WebGPU. Hair, fur, soft edges, and translucent areas are preserved using foreground color decontamination, the same technique cloud services charge for behind paywalls.
How to use the Background Remover
- 1
Upload your image
Drop a JPG, PNG, or WebP file onto the upload zone, or click to browse. Up to 4096 pixels on the longest side. Photos never leave your browser.
- 2
Wait for the AI to process
On first use, the BiRefNet model downloads from Hugging Face (~115 MB, cached after that). Removing the background takes 1-3 seconds per image once the model is ready.
- 3
Choose a background
Pick transparent (default, downloads as PNG), solid color, two-color gradient, or upload a custom background image to composite the subject onto.
- 4
Download the result
Save as PNG (transparent), WebP (smaller file), or JPG (with chosen background color). Full resolution, no watermark.
Frequently Asked Questions
Is my photo uploaded to a server?
No. The AI model runs inside your browser using WebGPU or WebAssembly. Your photo stays on your device the entire time. You can verify by opening DevTools and watching the Network tab during processing.
How is this different from remove.bg or Photoroom?
remove.bg and Photoroom both cap their free tiers (remove.bg limits free downloads to 0.25 MP, Photoroom adds watermarks above SD). This tool is fully free at any resolution up to 4K, no watermark, no signup. The tradeoff is the first-use download (~115 MB) and that the inference is local rather than cloud-batched.
How good is the hair and edge quality?
BiRefNet plus a foreground decontamination pass keeps fine hair, fur, and soft edges intact. Most free open-source background removers skip the decontamination step, which is why their outputs show green or blue color bleed around hair. We do that step locally, the same way the paid services do server-side.
What input formats are supported?
Anything your browser can decode as an image: JPG, PNG, WebP, GIF (first frame), BMP, AVIF on supporting browsers. The largest dimension is capped at 4096 pixels; larger images are downscaled before processing to avoid running out of memory.
What can I export as?
PNG with transparent background (default), WebP (smaller file size at the same quality), or JPG with the chosen background color or image baked in. All exports are full resolution.
Why does the first run take a while?
The browser downloads the BiRefNet model (~115 MB) from Hugging Face on first use. After that the model is cached in your browser's IndexedDB and loads in seconds on return visits. Browsers without WebGPU fall back to a much smaller portrait-focused model automatically.
Does it work on mobile?
Yes, but a smaller model is selected on devices with less than 4 GB RAM to avoid memory pressure. Quality is still good for most photos. For best results, use a desktop or laptop with WebGPU enabled (modern Chrome, Edge, or Safari).