Pixelate Image

Pixelate or blur faces, text and secrets in images and screenshots. Select areas or the whole image, black out passwords, strip metadata. No upload.

  • Runs in your browser
  • Tool input is not uploaded to ZeroTool servers
  • Free · No Sign-Up
Select an image file Choose, drop, or paste a local image. GIF input uses its first frame.

Drop an image here, or click to select

You can also paste a screenshot with Ctrl/Cmd+V

PNG / JPEG / WebP / GIF / BMP / AVIF · first GIF frame only

Image preview Drag rectangles over the preview. Coordinates map to the original image; areas smaller than two pixels in either dimension are ignored.

Load a local image to preview the effect and select areas.

Examples, details and FAQ Worked examples, how it compares with other tools, and answers to common questions.

Which Effect Should You Use?

  • Pixelate — averages each block of pixels into a flat color. Good for a deliberate mosaic look or casual obscuring. Not safe for text, and a large block size does not guarantee a face cannot be recognized.
  • Blur — scales the area down and back up, producing a soft, smeared look. Reads as less "censored" than pixelation but is just as reversible with strong enough deblurring, and equally unsafe for text.
  • Solid fill — replaces the area with one flat color. The only option here that destroys the underlying pixels completely. Always use this for passwords, tokens, API keys, and anything you paste into an AI chat.

What Each Effect Does to the Pixels

Each time you change a setting, the tool redraws the original image and applies the effect again, so effects never stack. The numbers below come from the tool's own pixelate function.

  • Pixelate (block size 4–64 px, default 16) splits each area into square cells, starting at the area's top-left corner, and paints each cell with the average color of its own pixels. A 2 × 2 cell with two black and two white pixels becomes rgb(128, 128, 128). Color is weighted by alpha: a cell with two opaque red pixels and two fully transparent pixels becomes red at half opacity (255, 0, 0, alpha 128), not dark red.
  • Edge cells shrink to fit. An area 100 px wide at block size 16 gives six 16 px columns and a last column 4 px wide (pixels 96–99). Edge cells average only their own pixels and never read pixels outside the area.
  • Blur (radius 2–40 px, default 12) draws the area onto a small canvas of ceil(width / radius) × ceil(height / radius) pixels with smoothing on, then scales it back up. A 300 × 40 area at radius 12 keeps 25 × 4 = 100 samples. The same area pixelated at block 16 keeps 19 × 3 = 57 cells. Blur keeps about as much information as pixelation; the smoothing only hides the cell edges.
  • Solid fill paints the area with one opaque color (default black). Nothing of the original pixels remains.
  • The dashed outlines are drawn only in the preview. Download and Copy image render the image again without them.

Why Pixelated Text Can Be Recovered

Pixelation and blur are lossy, but an attacker does not need to reverse the math. They render guesses in the same font, apply the same filter, and compare the blocks. Text in a known font is a small set of guesses, so this works well:

  • Depix (2020) recovers text from screenshots pixelated with a linear box filter, the same kind of filter this tool's Pixelate uses, by matching blocks against a rendered character sequence.
  • Bishop Fox's Unredacter (2022) guesses one character at a time and also tries every grid offset, so a grid that does not line up with the text does not help.
  • Hill, Zhou, Saul and Shacham (PoPETs 2016) recovered readable text under a mosaic coarser than the font, after heavy JPEG compression, and from blurred text. Their conclusion: mosaicing and blurring "are not viable approaches for text redaction."

For faces, McPherson, Shokri and Shmatikov (2016) trained models that identify people in mosaiced and blurred images. Use Solid fill when identity matters.

What Still Leaks After a Correct Redaction

  • Length. A black bar that fits the text exactly shows how many characters were there; with a monospace font, the exact count. Draw the bar wider than short secrets such as PINs.
  • The file name. The download keeps your base name and adds -pixelated, so passport-jane-doe.jpg becomes passport-jane-doe-pixelated.png. Pasted images are saved as pasted-image-pixelated. Rename the file if the name identifies someone.
  • The rest of the screenshot. Tab titles, the address bar, notification toasts, autocomplete lists and the same key further down a log often show the secret again.
  • The original file. Redaction makes a new file; the original stays on disk and in any photo backup. The export is built from canvas pixels only, so it has none of the original's EXIF or GPS data and no leftover bytes from the original file, unlike the 2023 aCropalypse bugs in Google Pixel Markup (CVE-2023-21036) and the Windows Snipping Tool (CVE-2023-28303).

Common Use Cases

  • Sanitizing screenshots before pasting into an AI chat, a support ticket, or a public forum — black out API keys, tokens, account IDs, email addresses, and internal URLs with Solid fill.
  • Blurring faces or license plates before publishing a photo. Use Solid fill when identity must stay hidden — recognition models can still match pixelated or blurred faces.
  • Retro pixel-art effects — check "Apply to entire image" with a large Pixelate block size for a deliberate low-res look.

Privacy: Nothing Is Uploaded

Decoding, effect rendering, and export all run through createImageBitmap and the Canvas 2D API directly in your browser. There is no server endpoint involved — try it in airplane mode to confirm. This tool's persistence policy is disabled: it never writes your image to local storage, and this page does not load analytics or ad scripts.

Limits

  • No automatic face or text detection — you draw the areas yourself. Detecting faces or text reliably needs a machine-learning model, which is a better job for an AI image tool than a lightweight browser utility.
  • Areas cannot be moved, resized, or given independent effects after being drawn — undo and redraw instead.
  • One image at a time; there is no batch mode for multiple files.
  • Animated GIFs export as a single static frame, by design. To get every frame as a separate image, use the GIF Splitter.
  • One effect applies to every area at once. To pixelate faces and black out a password in the same image, download the first result, load it again and apply the second effect.
  • Areas smaller than 2 px in either direction are discarded, so a mis-click does not add an empty area. While Apply to entire image is on, drawing is disabled.
  • JPEG has no transparency: transparent areas are flattened onto white. If the browser cannot encode WebP (Safari, per MDN's compatibility data), the WebP choice is saved as a PNG file. Copy image always puts a PNG on the clipboard.
  • Browsers limit canvas size. If an export fails, the tool shows "Could not export — try a smaller image".
  • HEIC photos open only in Safari 17 and later (Can I use). In other browsers, convert them to JPEG first.
  • PDF redaction is out of scope — use a local PDF editor's redact or black-out feature instead.

Related Tools

  • Secret Redactor — mask API keys and tokens in text and logs the same way, then restore them later.
  • EXIF Metadata Viewer — inspect or strip EXIF/GPS data without altering the image itself.
  • Image Compressor — shrink a screenshot's file size after redacting it.

FAQ

Is pixelation safe for hiding passwords or API keys?

No. Small-block pixelation and blur are reversible — the public tool Depix has demonstrated recovering readable text from pixelated screenshots by matching the averaged blocks against known font patterns. For passwords, API keys, tokens, or any text secret, use Solid fill instead. Pixelate and blur suit a mosaic look or casual obscuring. When a face, license plate or any text must stay unrecoverable, use Solid fill.

Are my images uploaded anywhere?

No. Decoding, the effects and the export run with the Canvas 2D API in this tab. The tool does not save the image or the areas you cover, and stored values for this page are deleted each time it loads. The page loads neither Google Analytics nor AdSense. Download writes a new file, and Copy image puts the result on your system clipboard.

Does it remove EXIF and GPS metadata?

Yes. The exported file is a fresh re-encode produced by the Canvas 2D API, so none of the original file's EXIF, GPS, or camera metadata carries over — regardless of which area you pixelate.

Which image formats are supported?

Any format your browser's createImageBitmap can decode — typically PNG, JPEG, WebP, GIF, BMP, and AVIF. HEIC only decodes in Safari. Animated GIFs are read as their first frame only and exported as a static image, by design. To get every frame, use the GIF Splitter tool.