Back to blog
Transparency

How to Remove Black from an Image and Make It Transparent

Make black backgrounds and black pixels transparent: select the color, control removal with tolerance and connected mode, then export a clean transparent PNG.

removecolor.online makes black transparent right in your browser: click the color, choose connected or global mode to control how far the removal spreads, use tolerance for uneven black backdrops, and download a transparent PNG. The one-line rule: global mode when the subject has no black, connected mode for everything else. Below are the full steps, plus what to do when the subject itself contains dark detail.

How to remove black from an image

removecolor.online is free, needs no account, processes everything locally so the image never leaves your machine, and downloads a PNG at the original resolution. The five steps below work for product-photo backdrops, screenshot borders, logos scanned on black, and dark studio backgrounds.

Step 1: Open the tool and load the image

Drag your image into removecolor.online. PNG, JPG, and WebP files up to 20 MB all work. A JPEG with a baked-in black background is fine too — the tool reads the pixels and writes a fresh PNG with a real alpha channel.

Step 2: Click the black area

Click any pixel of the black background. That exact color becomes the removal target and the preview updates instantly. On photos, pick a spot in the middle of the background rather than against the subject's edge — compression noise and lighting shifts concentrate near edges.

Step 3: Choose connected or global mode

Connected area removes only the black region contiguous with your click; all matching colors scans the whole image and removes every matching pixel within tolerance. The core rule: global mode when the subject has no black, connected mode for everything else. A logo or icon on a uniform black field clears in one global click; if a photo's subject might contain dark tones, start with connected mode — it cannot reach black areas inside the subject that don't touch the backdrop.

Step 4: Adjust tolerance until the checkerboard is clean

Tolerance controls how far a pixel may drift from your sampled black and still be removed. The tool starts at a default of 20, which suits most images — treat it as a starting point, not a setting to defend. A photographed black backdrop is never perfectly black: lighting gradients and compression push nearby pixels off target. If patches remain, raise tolerance gradually; once the background reads fully transparent, step back one notch — every extra point starts removing pixels you may want to keep.

Step 5: Smooth the edges and download the PNG

If the outline looks harsh, a small bump of edge smoothing makes the transparency transition more natural. Mind what it does: edge smoothing adjusts the alpha transition, not the colors. If the edge pixels themselves have absorbed color from the old black background, smoothing alone may not fully remove a dark fringe — that calls for a dedicated edge color cleanup. When the preview is clean, download. The output is always PNG, which preserves the alpha channel you just built. Don't re-save it as JPEG; that's exactly how the black comes back.

Connected or global mode

Green screens became the film-industry standard partly because subjects rarely contain green. Black is the opposite — it's everywhere inside subjects: black hair, black clothing, black text, pupils, shadows, dark materials. That's the whole logic behind the rule "global when the subject has no black, connected otherwise."

Connected mode has one limitation worth knowing: it only removes the region connected to your click. Photo backgrounds usually wrap around the whole subject, so one click is enough. But if the subject touches the image border, or a limb crosses the frame, the black background may be split into separate regions — click each one. Connected mode clears them one at a time and never reaches the black inside your subject.

When the subject itself is dark

Dark hair or clothing against black. Connected mode with low tolerance removes the backdrop without hollowing out the subject. Edges where hair blends into the background are genuinely ambiguous pixels — edge smoothing softens the transition, but some softness is unavoidable: where two near-black tones meet, there is little contrast for any tool to separate them, and JPEG compression leaves noisy grey pixels around such edges. Working at low tolerance and cleaning up leftovers with extra clicks preserves more real detail than one aggressive pass.

Black text or thin black lines on black. When text and background share a color and touch, click-based tools can't separate them reliably. If the logo exists as a vector file (SVG, AI, PDF), re-export from the original with a transparent background — that's the clean fix. The browser route works when black text sits clearly on a white or colored plate that itself sits on black: remove the outer black, keep the plate and the text on it.

Dark-mode screenshots. The dark panels are the design, not a background — removing them destroys the screenshot. The actual goal is usually trimming the outer matte or letterbox bars: connected mode plus one click on the outer frame does exactly that, leaving the UI intact.

Night photography. The dark sky is background; the dark road is not. Click the sky region in connected mode, and everything disconnected from it — buildings, roads, lit windows — stays untouched.

And if the subject truly cannot be separated from black — a black cat on black velvet — color removal is the wrong tool: any keying method either leaves black behind or eats the cat. That's a background-replacement problem.

Why transparent images sometimes look black

An image with transparency stores four numbers per pixel: red, green, blue, and alpha. RGB describes the color; alpha describes opacity. Even a fully transparent pixel can still carry color data in its RGB channels — some processing pipelines set those values to black, others keep the original colors. Software that handles alpha correctly never shows them, but once transparency is dropped or mishandled, the hidden colors can surface.

Three situations cover most cases:

  1. The image was saved as a JPEG somewhere along the way. JPEG has no alpha channel. Exporting a transparent image to JPEG forces the editor to fill transparent areas with a solid color — black or white is what comes out. The transparency is destroyed at the moment of export.
  2. The viewer or platform ignores alpha. Some dark-mode apps, older document editors, and embedded viewers composite images onto a black surface. A healthy transparent PNG then looks like a black rectangle; the file itself is fine.
  3. Dark edges around the subject. Semi-transparent boundary pixels can carry dark color values, and on a light background they show up as a thin dark fringe — the case Step 5 flagged for edge color cleanup.

Saved a transparent PNG and it still looks black?

  1. Check where you're viewing it. Transparent pixels show whatever sits behind the image. A dark-theme editor's canvas may be near-black, and some apps draw the checkerboard in dark grey. Place the image over a white layer or open it in a browser tab: if the black rectangle turns into a checkerboard, the file is fine.
  2. Check the file extension. A .jpg or .jpeg file contains no transparency, no matter what it looked like before export. Re-download as PNG.
  3. Check what happened after the download. Some chat apps and social platforms re-encode images to JPEG on upload, flattening transparency into a solid background — white on some platforms, black on others. The original file on disk is still transparent; verify it in a browser before blaming the export.
  4. Dark fringe only, after compositing. The boundary pixels are semi-transparent with color values inherited from the old black background. Transparency tweaks alone rarely clear this fully: edge smoothing softens the transition, a real fix needs edge color cleanup — and until that exists in your workflow, compositing onto a mid-tone background hides a slight fringe.
  5. The preview looked clean but the download shows the background. The on-screen preview is scaled down for performance while the download is processed at full resolution — the two can disagree at tiny scales. Open the downloaded PNG in a browser tab: a checkerboard there means the export is fine and the preview was simply too small to judge.

Related articles