Make a portrait cutout
Choose a JPG, PNG or WebP photo under 8 MB. The browser loads a portrait segmentation model, estimates which pixels belong to the person, and creates a transparent PNG you can preview before saving. The first run can take longer while the model downloads.
The model is intended for people. It can miss fine hair, transparent accessories, objects held in front of the body or people in a crowded scene. It is not a general product-photo background remover.
Your photo stays in the browser
The photo is processed locally and is not uploaded to SaveFromNet. The browser downloads model files from Google and the MediaPipe runtime from jsDelivr. Close the page to discard the working image and preview.
Save the resulting PNG if you need transparent edges. JPG cannot store transparency; use the PNG to JPG converter later only if you want a solid background.
Frequently asked questions
Which photos work best?
Photos with a clear person in the foreground work best. The model is trained to separate people and may not isolate products, pets or complex objects.
Is my photo uploaded?
No. The image is processed in your browser. The segmentation model and runtime download from Google's model host and jsDelivr on first use.
Which file can I save?
The result is a transparent PNG. Check the preview because fine hair, glass and crowded scenes can produce imperfect edges.