Skip to main content

Background Remover Guide

Remove Background from Image — Format-Aware Cutouts

Camera JPEGs, web PNGs, and Apple HEIC land in the same local pipeline. Odd formats rasterize first, then you export alpha.

Supports JPG, PNG, WebP, HEIC, GIF, BMP, TIFF, AVIF, and SVG. Apple HEIC photos decode on-device before cutout — no convert-first step required.

Loading background remover…

Prefer the main tool? Open the free background remover to remove background in one click.

Written by Safaat Rahman Siham, Founder & Developer at Convertify4u. Contact us.

Most “unsupported format” errors are a decoder problem, not a mask problem. iPhones still write HEIC. Designers still email TIFF. Browsers still prefer JPEG, PNG, and WebP for Canvas. This page exists so you know what happens before the model runs: HEIC becomes a raster in WASM, SVG is drawn to pixels (strokes can fatten), animated GIF keeps a frame — usually the first — then isolation proceeds. Keep a copy of the original container. A 16-bit TIFF may posterize when it becomes 8-bit PNG. An SVG logo is better traced or exported from the vector app; rasterizing tiny art then isolating it will look soft. JPEG ringing around text will be treated as “edge,” which is why screenshots of UI often need the PNG spoke instead. After a clean mask, you still choose an output. Alpha wants PNG. Marketplaces that reject transparency want a white composite, then JPEG. Bytes after that are Image Compressor. This spoke does not change those rules; it only gets more containers into the pipeline without a separate desktop converter.

What actually decodes, and what you should convert first

Primary stills: JPEG/JPG from cameras and exports, PNG with or without alpha already, WebP from modern sites, HEIC/HEIF from iPhone and iPad. Drop HEIC here — decode is local; you do not need HEIC to PNG first unless you want a converted archive. Secondary: GIF (one frame), BMP, TIFF (expect 8-bit), AVIF where the browser can decode it, SVG (rasterized). If AVIF fails, convert to PNG elsewhere, then retry. SVG text and hairline strokes will not stay infinitely sharp after rasterize. Color: sRGB is assumed; Display P3 iPhone shots may shift slightly. EXIF orientation is respected on load so a sideways Camera photo should preview upright. Object isolation is the same model as the hub — this page is the formats briefing.

How it works

  1. 1

    Match the container to a decoder

    Camera JPEG and HEIC can go straight in. If the picker shrugs at TIFF or AVIF, export PNG from Preview or Photos, then drop that raster.

  2. 2

    Isolate after decode

    The mask never sees the original HEIC bitstream — it sees pixels. Soft HEIC grain can look like hair; preview at 100% before you trust the silhouette.

  3. 3

    Pick PNG versus a flatten

    Keep PNG when you will composite. Flatten with PNG to JPG only if a form forbids alpha. Slim afterward if the theme is heavy.

Use cases

iPhone Camera roll

HEIC from Photos without a round-trip through a Mac Preview export, unless you want a JPEG archive anyway.

Web assets already in WebP

Heroes saved from a CMS as WebP can be isolated without converting to JPEG first (JPEG would crush the alpha you are about to create).

Legacy scans

BMP and TIFF from old scanners decode, then you get a modern PNG. Expect dust in the mask — clone it in an editor.

Vector marks that arrived as SVG

Rasterize at a large size first if you can. Isolating a 64px SVG export will never look like a 2× logo.

Problems we solve

Tools that refuse HEIC

Many cutout sites want JPEG only. Decode here locally instead of emailing the still to a convert bot.

Assuming SVG stays vector

Once drawn to Canvas, paths are pixels. Edit the SVG in Figma if you still need infinite scale.

Private camera rolls on unknown hosts

Family HEIC should not hit a GPU queue. This spoke keeps decode and mask in the tab.

Trust, speed & privacy

No convert-first tax for HEIC

iPhone stills decode here. Optional HEIC to PNG remains if you want a second file type.

Same silhouette model

Format does not buy a different neural net — it only changes the decode step.

Local rasters

Decoded frames are not uploaded. Close the tab to drop them.

Frequently asked questions