What this tool does
Background Remover requests a cutout from the configured service, then lets you soften alpha (0–100 mapped to blur radius 0–5), place a background, and encode PNG, WebP, or JPG. Download names use {base}-no-bg.{ext} or {base}-{preset}.{ext}.
When to choose it instead of a similar DevToolKit tool
- Use this for subject cutouts. Other Image tools do not remove backgrounds.
- Do not use it as a metadata cleaner or format converter first; send a file you are allowed to upload.
Supported inputs, limits, and output behavior
- Input
- One JPG, PNG, or WebP, 10 MB (stricter than the 20 MB local image tools).
- Modes
- fast (Product), portrait, hair_detail, high_quality (BiRefNet; first run can take 1–3 minutes per UI copy).
- Export
- PNG/WebP can keep alpha; JPG flattens. Quality presets 72/85/92 for JPEG/WebP.
How the transformation actually works
The API may return headers x-bg-mode, x-bg-provider, x-bg-fallback. Provider labels in the UI include Premium remover (remove-bg-api), Local BiRefNet, Local portrait, Local rembg. Edge softness blurs only the alpha channel. Background presets include solid colors, gradient, studio radial, custom, or checkerboard preview for transparent.
Step-by-step workflow
- Click Choose image or drop one supported file and confirm the local preview.
- Pick a Subject preset.
- Set Export background, Output format, Compression quality if shown, and Edge softness.
- Click Remove background and keep the tab open while the service runs.
- Review the cutout on Export cutout.
- Fine-tune export if needed, then download again.
What can change in the output
Hair, fur, glass, motion blur, and similar colors can leave halos or holes. The tool does not offer per-pixel mask painting. Crop is sent as false. returnMask is false.
Known limitations and failure cases
- Service not configured → 501 / BACKGROUND_REMOVER_NOT_CONFIGURED.
- Offline Python worker → 503 codes.
- Timeouts especially on high_quality first model load.
- IMAGE_TOO_LARGE / 413 above the service limit.
- No batch.
Privacy / where data goes
After you click Remove background, the image goes to the configured backend and typically a Python remover. High-quality may call api.remove.bg when a key is configured. The privacy policy describes in-memory upload buffers, no user-facing deletion schedule, and possible infrastructure retention. This is not a confidentiality guarantee.
Practical output-review checklist
- Zoom hair and semi-transparent edges.
- If exporting JPG, confirm the flattened background.
- Do not treat the cutout as a perfect matte without checking.