Add an image
Choose or drop a browser-supported image. The upload area is replaced by the editing workspace.
Remove a photo background with browser-based AI, keep a transparent cutout or add a colour or replacement image, then download PNG or JPG.
Choose a JPG, PNG, WebP, GIF, BMP, AVIF or HEIC image supported by your browser.
Preparing the image…
Continue editing the downloaded image with another free CorrectionCopy tool.
Create a transparent cutout or replacement background in three steps.
Choose or drop a browser-supported image. The upload area is replaced by the editing workspace.
Use Portrait mode for people or Object mode for common objects, then select transparency, colour or a photo background.
Adjust edge smoothness, inspect the preview and download the validated result in your preferred format.
Use one private browser workflow for profile photos, products, thumbnails and design assets.
A browser-optimized detail-aware matting model identifies the foreground, while portrait and object tuning refine the final edge treatment.
Keep the removed area transparent for logos, e-commerce images, thumbnails and presentation graphics.
Place the cutout on a solid colour or upload a replacement image that fills the background area.
Feather the mask boundary to soften rough edges around hair, clothing and curved products.
Inspect portrait, landscape, square and panoramic images without changing the final output dimensions.
Your photo is decoded, segmented and composed locally. CorrectionCopy does not receive the image file.
This image background remover creates a foreground cutout from portraits, product photos and common objects. It can produce a transparent PNG, replace the original background with a solid colour, or place the subject over another photo. JPG export is useful when transparency is unnecessary, while PNG preserves transparent pixels.
Portrait mode preserves softer hair and clothing transitions, while Object mode creates firmer product and general-object contours. The high-quality matting model works best when the main subject is clearly separated from the background. Fine hair, transparent materials, motion blur, shadows and highly detailed backgrounds can require more edge smoothing or a different source photo.
The uploaded photo is processed in your browser and transferred only between the page and a local Web Worker. The first use requires internet access to load the pinned browser runtime and high-quality matting model. Those code and model files may be cached by the browser; the image itself is not sent to those hosts.
Continue with active image, media and document tools selected for this workflow.
Answers about transparency, privacy, formats, quality and common processing problems.
Add an image, wait for the foreground mask, choose Transparent, Colour or Photo, adjust edge smoothness and download PNG or JPG.
Yes. Download PNG when you need transparent pixels. JPG cannot store transparency, so the tool places a white background behind a transparent result.
Yes. Select Photo under New background, choose an image and the tool scales it to cover the output before placing the cutout above it.
Use Portrait / People for selfies, headshots and full-body photos. The general object mode is intended for products and common non-person subjects.
Segmentation is an estimate. Fine hair, transparent objects, motion blur and low contrast can confuse the model. Try a clearer source photo and adjust Edge smoothness.
No. The image is processed locally in your browser and a local worker. The runtime and model files load from external hosts, but your image pixels are not uploaded to CorrectionCopy or those model hosts.
The first run must download and initialize the browser AI runtime and high-quality matting model. Later runs in the same session are usually faster because the worker reuses them.
The page accepts common browser-supported images including JPG, PNG, WebP, GIF, BMP, AVIF and HEIC when the browser can decode them. Output is PNG or JPG.
Check the internet connection, disable a network rule that blocks the pinned CDN or Google model host, and try again. Your selected image remains available after a recoverable error.
Yes on modern mobile browsers with JavaScript, WebAssembly, Web Workers and sufficient memory. Large photos may process more slowly on lower-memory devices.