- How does this differ from cloud background removers?
- This tool downloads its model and processes the selected image in your browser. Cloud services may offer different models, edge quality, batch features, and pricing. Compare current provider terms directly before choosing a workflow, and review fine hair or transparent edges in every result.
- What output resolution will I get?
- The tool exports a transparent PNG at its working resolution. Browser memory, device performance, and model preprocessing can affect very large images, so verify the downloaded dimensions before production use.
- How does it work?
- The BEN2 model (about 208 MB) runs through Transformers.js using WebGPU when available and WebAssembly as a fallback. It detects the foreground subject and produces a transparent PNG. Fine hair, glass, smoke, and complex edges may still need manual correction.
- Is my image uploaded anywhere?
- The selected image is processed in the current browser session and is not sent to an EaziApps image-processing server. Your browser still downloads the AI model files needed to run the tool.
- Why is the first run slow?
- The first time you use the tool, the AI model is downloaded to your browser and may be cached in IndexedDB. Later runs can reuse that cache when browser storage is available. Download and inference time depend on your connection, device, browser, and image size.
- What images work best?
- Photos with a clear subject and a contrasting background work best. Portraits, products, animals, food, isolated objects. Tricky cases: very fine hair, transparent objects (glass, mesh), extreme low-contrast backgrounds, abstract art.
- Does it work on mobile?
- It can work in modern mobile browsers, but the model download and memory use are substantial. Results depend on browser WebGPU or WebAssembly support, available storage, and device memory.
- What is the model license?
- The tool uses the MIT-licensed BEN2 model through the Apache-2.0-licensed Transformers.js library. Review the linked project licenses if you plan to redistribute model files or modified software.