Image

Remove Background from Portrait

Free, private, and instant — your image never leaves your device.

Loading tool…

Portrait cutouts are the most common background removal task: LinkedIn headshots, team page photos, ID-style images, website profile pictures, and stickers all require the person cleanly separated from their background. The People & portraits mode uses MODNet — a portrait matting model specifically trained on human subjects. It understands skin tone gradients, handles hair strands, and uses WebGPU on supported hardware for fast processing. The result is a transparent PNG that preserves natural edges without a hard clipping path.

Tips for the cleanest cutout

Use "People & portraits" mode — MODNet is trained specifically for human matting and handles hair and skin gradients far better than the general-purpose model.
A plain, evenly lit background that contrasts clearly with the subject gives the cleanest hair and shoulder edges.
Even front lighting (or diffused natural light from a window) that eliminates harsh shadows behind the person helps the model identify the subject boundary accurately.
For group shots, individual people near frame edges may be partially cropped — single portraits always give the best results.

What to watch for

Fine or curly hair at the edges is the hardest case for any AI portrait matting — results depend heavily on background contrast. On a plain, strongly contrasting backdrop, individual strands are often preserved; on complex or similarly-coloured backgrounds they may be simplified.

Very dark hair against a dark background, or very light hair against a white background, reduces edge contrast and may cause hair to be clipped. Shooting against a contrasting colour avoids this.

How to Use

1
"People & portraits" mode is pre-selected — specifically trained for human subjects with accurate hair and skin edge handling.
2
Drop your portrait (PNG, JPEG, or WebP) onto the upload area, or click to browse.
3
On the first run, the MODNet model (~25 MB) downloads and caches permanently. It is the smaller of the two models, so it fetches quickly.
4
Check hair and shoulder edges in the "After" preview, then download the transparent PNG.

Other Background Removal Uses

Remove Background from Logo Remove Background from Product Photo Remove Background from Signature Remove White Background from Image Make Image Background Transparent

Frequently Asked Questions

Why is "People & portraits" better for portraits than "Any subject"?

"People & portraits" uses MODNet — a portrait-specific matting model trained on human images. It has a learned understanding of face and shoulder shapes, skin tone gradients, and hair boundaries. "Any subject" uses Silueta, a general salient-object detector trained on a much broader category of subjects, so it is less precise on fine portrait details like individual hair strands.

Will this work for group photos?

The model handles groups, but accuracy depends on overlap and background complexity. Tightly packed people with overlapping hair, or subjects wearing colours similar to the background, are harder to separate cleanly. Individual headshots with a plain backdrop give the sharpest results.

Can I use the cutout as a virtual background in Zoom or Teams?

Zoom and Teams perform their own background removal in real time during calls — they do not use a static transparent image for that feature. To use your cutout as a video call background, place the transparent PNG on a new scene image in Canva or a similar tool, then upload that composite image as your virtual background.

Try the full background remover

Switch modes, compare before/after, and process as many images as you like — all free, all in your browser.

Remove Background