Purpose
Wipe (wipe.photos) is a free AI object removal tool focused on erasing unwanted content from photos. It uses AI image inpainting technology, allowing users to remove people, objects, text, logos, wires, and other elements from photos through a simple brushing operation, with the AI automatically reconstructing the background behind the removed area.
Main Capabilities
- Object removal: Can remove people, objects, logos, text, signs, wires, and scattered clutter from photos.
- AI background reconstruction: After brushing over the target area, the AI fills in and reconstructs the background, making the image look natural.
- Local processing: The entire editing process is completed in the browser, and photos never leave the user's device.
- Free and unlimited: No subscription, no registration, no watermark, no usage limit.
- Fast processing: With WebGPU acceleration, editing speed can be under 1 second on computers equipped with a GPU (such as an M3 MacBook with Chrome).
- Cross-device availability: Can run in modern browsers on desktop and mobile.
How to Use
- Upload a photo.
- Use the brush to paint over the objects, people, or text you want to remove.
- AI image inpainting removes the target and reconstructs the background within seconds.
- Download the processed photo.
Use Cases
- Remove passersby or clutter that intrude into travel photos.
- Erase text, logos, or signs from images.
- Clean up distracting elements such as wires and scattered objects in the frame.
- Editing needs that are sensitive to privacy or copyright and where users do not want photos uploaded to a server.
Limitations and Notes
- The result depends on the background behind the removed object. Official information states that it works best on backgrounds such as fabric, grass, pavement, and walls.
- Supports JPG, PNG, and WebP input, and results are downloaded in PNG format.
- A high-performance computer or dedicated GPU is not required, but ideally the computer has a GPU and uses the Chrome browser; on first visit, the model is downloaded once and saved in the browser.
- JavaScript must be enabled to use it.
- The model is based on Samsung Research's LaMa (Large Mask inpainting), licensed under Apache 2.0, uses Fast Fourier Convolutions to fill large areas and generate coherent textures, and runs in the browser through onnxruntime-web; the officialprovides a self-converted big-lama ONNX version under the same license.
Pricing and Platform
The official information clearly states that the tool is free, with no subscription, no account, no login, no download, no watermark, and no usage limit. Other pricing or platform details are not yet provided.
