Methodology
How Image Wizard processes your images
The short version: your browser reads the selected file, performs the requested operation locally, and creates a new file for you to download. Image files are not sent to an Image Wizard server.
Conversion, crop, and resize
Standard browser image decoding and Canvas APIs handle common formats and pixel operations. Additional WebAssembly codecs are used where browser-native encoding is not sufficient. Output quality, transparency support, and metadata preservation depend on the selected format.
On-device AI
Background removal and restoration use model files downloaded to your browser. Inference then runs locally through browser-compatible machine-learning runtimes. The first run can take longer because the model must be downloaded and initialized.
What “private” means here
Image pixels and image files stay in your browser during processing. The site does use analytics, so ordinary website information—such as page views, browser type, and device information—may be sent to the analytics provider. Analytics events are not designed to contain your image files.
Limitations and verification
Results vary with the source image, browser support, available memory, and selected settings. AI restoration estimates missing detail; it cannot recover facts that were never captured. For sensitive work, inspect the downloaded output before relying on it and remove metadata separately if that matters to your use case.