How to remove a photo background (and when you can’t)

Automatic cutout uses a segmentation model trained to split “main subject” from “everything else”. It does not understand what the object is — it recognises shape, outline and contrast patterns, which is why it struggles with same-colour backgrounds, hard shadows, glass and reflections.

• Contrast between subject and background.
• Diffuse light — hard shadows create false edges.
• Focus on the subject, with a slightly blurred background.
• Enough resolution, but not excessive.

Loose hair and fur remain the hardest test: good at a glance, imperfect at 100% zoom.

Remove the background, place a solid white one, then crop to the required ratio. Shoot with frontal light and no shadow on the wall behind — that shadow is the number one reason ID photos get rejected.

The Localside background remover downloads the AI model into the browser and runs the cutout on your device — no upload, no watermark, no usage limit. Follow up with the image compressor if the file needs to be lighter.