How to Remove a Background from a Photo (and Get Clean Edges)
A guide from the NextG Design team
Automated background removal has gotten very good, but it's not magic — it's a model guessing, pixel by pixel, where your subject ends and the background begins. Most of the time that guess is close enough. The difference between a cutout that looks professional and one that looks obviously cut out usually comes down to a handful of predictable trouble spots, and knowing how to fix them.
Start with a photo that gives the model a fair chance
Segmentation models do best when there's real contrast between your subject and the background — different colors, different brightness, a bit of separation in focus. A subject wearing a color close to the background, or standing right up against a cluttered, similarly-lit backdrop, is the single biggest cause of a rough cutout. If you have any control over the shot, a plain or contrasting background behind the subject will save you touch-up work later.
Know where AI cutouts usually struggle
- Fine hair and fur. Individual strands are thinner than the model's effective resolution, so edges around hair are the most common place for slight haloing or over-smoothing.
- Semi-transparent or reflective material. Glass, sheer fabric, and smoke don't have a single clean edge, so the model has to pick one — it won't always pick the one you'd want.
- Low contrast boundaries. A dark jacket against a dark background, or pale skin against a light wall, gives the model the least information to work with.
- Small gaps and holes. The space between someone's arm and their body, or inside the handle of a mug, can get filled in or left transparent inconsistently.
Use the touch-up brushes instead of starting over
Rather than accepting an imperfect result or trying a dozen different source photos, the Background Remover's Erase and Restore brushes let you correct the specific areas the model got wrong, directly on top of its first pass:
- Erase removes any background pixels that leaked into your subject — useful for cleaning up stray bits near hair or busy edges.
- Restore brings back subject pixels the model mistakenly cut away — useful for thin details like glasses arms, wisps of hair, or a finger that got clipped.
- Zoom in before brushing. Edge mistakes are usually only a few pixels wide, and it's much easier to be precise at higher zoom than to guess at full size.
- Work in short strokes near the boundary rather than one long sweep — it's easier to stop exactly where you meant to.
Check your export, not just the preview
Before downloading, use the tool's Review & Export step to see your result against both light and dark backgrounds. A cutout that looks fine on a white preview can reveal a faint halo once it's placed on a dark or colored background — and vice versa. If you notice that, a quick Erase pass around the affected edge usually clears it up.