Published January 15, 2025 · 6 min read
How to Remove White Background from an Image
White backgrounds are the most common type in e-commerce and design. Removing them gives you transparent PNGs ready for any use — marketplaces, social media, presentations, or print on demand. Here's how to do it for free, without Photoshop.
Two ways to remove a white background
Method 1 — Color-based removal (fast, free, local)
If your image has a solid white background (not a photo of a white wall, but true white #FFFFFF), the fastest method is color-based removal. Every pixel that's within a tolerance of white gets made transparent.
Best tool: BG Remove's Transparent PNG Maker
Pros: Instant, runs in your browser, no upload, free, adjustable tolerance.
Cons: Works poorly on JPEG images with compression noise, doesn't handle shadows or anti-aliased edges perfectly.
Method 2 — AI background removal (higher quality, needs connection)
If your white background isn't pure white — maybe it's a photographed white wall, has shadows, or the subject has soft edges (hair, fur, translucent objects) — use an AI background remover instead.
Best tool: BG Remove's AI Background Remover
Pros: Handles soft edges, shadows, and imperfect backgrounds. Works on any subject.
Cons: Requires upload and processing time (5–15 seconds).
Step-by-step: color-based removal
- Open the Transparent PNG Maker.
- Upload your image.
- The tool auto-samples the top-left corner as the default background color. If that's not white, use the color picker to select white (#FFFFFF).
- Adjust the tolerance slider. Start at 30. If edges still show white, increase to 40–50. If it's eating into your subject, lower it.
- Adjust edge softness to blend the boundary. Start at 20.
- Download the transparent PNG.
Step-by-step: AI removal
- Open the AI Background Remover.
- Upload your image.
- Wait 5–15 seconds for processing.
- Download the transparent PNG.
When white background removal fails
Here's why removing a white background sometimes looks bad, and what to do about it:
White fringe around the subject
White pixels bleed into the subject during JPEG compression. The color-based tool leaves them, creating a light halo. Fix: increase tolerance, or increase edge softness. Or use AI removal which handles this better.
Subject contains white pixels
If part of your subject is white (a white shirt, a white label), the tool may accidentally make that transparent too. Fix: lower the tolerance. Use color-based removal only on images where white is exclusively the background.
Background isn't truly white
A photographed white wall has shadows, gradients, and color casts — not pure #FFFFFF. Color-based tools can't remove gradients in a single pass. Use AI removal instead.
Shadows on the background
If the subject casts a shadow on the white background, color-based removal keeps the shadow (it's grey, not white). AI can be told to remove shadows if you use a good model. Alternatively, remove shadows manually after removal.
Before and after — what clean output looks like
A good cutout:
- Has no visible white edges or halos
- Preserves fine details like hair strands and text
- Shows a checkerboard pattern in any image viewer (indicating transparency)
- Downloads as PNG (not JPG)
Common mistakes
- Downloading as JPG. JPG doesn't support transparency. The white background comes right back. Always download as PNG.
- Using too high tolerance. This removes subject pixels too. If parts of your subject disappear, lower the tolerance.
- Ignoring JPEG artifacts. JPEG compression creates noise around edges. Either start with a PNG source or use AI removal.
- Not checking at full size. Preview is rendered small. Download, open in an image viewer, and zoom to 100% to verify no fringe remains.
Alternative tools
- remove.bg — AI-based, higher quality on hard cases, paid for full resolution
- PhotoRoom — AI-based, e-commerce focused, watermarks free tier
- Photoshop — manual selection, precise but slow, requires subscription
- GIMP — manual selection, free, steep learning curve
- Photopea — manual selection, free, browser-based