Useful photo tools should solve the job before asking for your data.

BackgroundBlur is a browser-first image editing project built around a simple principle: separate the foreground accurately, make background changes easy to control, and keep the normal source-image workflow on the user's device.

Three quality modes instead of one fake “best” model

Universal AI is the balanced default for most subjects. Portrait AI prioritizes fast human matting. High Quality AI uses a heavier WebGPU model when difficult edges justify the cost. If a requested engine cannot run reliably, the editor falls back instead of trapping the user behind an error.

A real editor behind every tool page

Blur, removal, white or black backgrounds, product cleanup, profile-picture depth, and custom image replacement all use executable presets of the same editor. We add landing pages when the corresponding capability exists, not merely because a keyword has search volume.

Built to improve

The inference layer is replaceable, the compositor is independent of the AI model, and the mask can be repaired manually. That lets BackgroundBlur improve model quality without rebuilding the entire product.