Your images. Your device.

imgy is AzenTool’s free browser-based image workspace.

Processing & privacy

Image files and results stay in this browser. Editing uses memory; continuing into another tool uses temporary IndexedDB storage (one handoff, read once, expires after one hour). Saved setting presets use localStorage and contain no images. We do not upload them or save them to an account. Closing or refreshing the page clears the active workspace. Temporary handoff data is replaced by the next handoff or deleted when opened. Browser site-data controls can clear any remaining stored data. Download anything you want to keep.

The web server receives normal page and asset requests, including IP address and browser headers. Local models and libraries are served from AzenTool. There are no analytics or ad scripts in this release. Links to other AzenTool apps open their own services and policies.

Models & accuracy

Background removal uses portrait matting or general-subject segmentation, and face detection uses UltraFace, on your device. Both can make mistakes. Always review the output, especially redactions and application photos. AI enlargement uses ESRGAN Slim locally. It predicts details and may alter faces or textures; inspect your result. Smooth resampling is also available.

Practical limits

50 images per batch; 50 MB per file; 24 MP per input or output; 80 MP total input per batch; 120 MP total output per batch. GIF creation supports 40 frames at up to 2 MP each. These limits protect browser memory; slower devices may need smaller jobs. Animated input is flattened to one frame, TIFF to its first page, and SVG to pixels. No RAW, CMYK preservation, live cloud drive imports, URL screenshots, identity recognition or public processing API.

Free without a subscription

There is no signup, added watermark or paid export tier. Computation runs on the visitor’s device, so no paid image provider is required. Hosting and bandwidth still have costs; this is not a promise of unlimited compute or support.

Open-source components

React, Vite, Lucide, JSZip, pdf-lib, gifenc, UTIF.js, ag-psd, heic-to and ONNX Runtime power this workspace. U²Net is licensed under Apache-2.0; portrait matting uses MODNet under Apache-2.0; UltraFace under MIT. See the component notices for attribution and source links.