Redact Image

Redact screenshots and images locally with solid blackout, pixelation or blur. Export flattened PNG or JPEG and inspect the actual result.

Processed on your device

Input
Still PNG, JPG/JPEG or WebP
Output
Flattened PNG or JPEG · generated bytes reopened for supported checks
Limits
20 MiB per file · 12 megapixels · 6,000 pixels per edge
Access
Free

Drop image here

PNG, JPEG or WebP still images · 20 MiB · 12 megapixels · maximum 6000px edge

No file selected

Limits

PNG, JPEG or WebP still images · 20 MiB · 12 megapixels · maximum 6000px edge

You can also paste an image from the clipboard. Source files stay in memory on this device.

How to use Redact Image

  1. Choose, paste or drop a still image.
  2. Draw solid regions over sensitive content; move or resize them and review the plan.
  3. Export a flattened copy, then inspect it before sharing.

Supported files and limits

Still PNG, JPG/JPEG or WebP. 20 MiB per file · 12 megapixels · 6,000 pixels per edge.

These are configured admission ceilings, not measured maximum workloads on every device. Applicable file-size, pixel, edge and memory budgets apply together; not every maximum can be combined.

Output and privacy considerations

Solid regions are baked into the exported pixels. Review every sensitive area; unmarked content remains visible. Blur is not used as the redaction guarantee. Use PNG for exact opaque solid-pixel checks; JPEG is lossy and flattens transparency onto white.

Selected creative files are processed locally. The original is not overwritten; review the generated copy before sharing it. Website code and optional processing assets are still requested from the site.

Remove the visible information you have chosen to hide

Redact Image is a focused manual workflow for covering sensitive regions in a screenshot or photograph. Choose a local still PNG, JPEG or WebP, mark the areas and export a flattened PNG or JPEG. The original is not overwritten. It is useful when you already know that an address, account number, badge or other visible detail should not appear in the copy you share.

The important word is chosen. The tool cannot decide every privacy risk for you, and a solid box over one item does not sanitize the rest of the picture. Review the whole frame first. Look for repeated values, reflections, small corner details and background clues before marking the obvious central item.

Prefer solid redaction for sensitive text

Solid blackout replaces the marked region with opaque pixels. That is a more appropriate choice for genuinely sensitive information than a cosmetic blur. Blur and pixelation can conceal detail visually, but they may leave enough structure or context for someone to recognize what was hidden. Do not equate an unreadable thumbnail with a reliably removed identifier.

A screenshot of a payment page illustrates the difference. Cover the complete number, adjoining characters and any repeated appearance elsewhere. A partial rectangle can leave a final digit, a label or a hint that materially changes what is disclosed. Use padding rather than drawing exactly along the visible outline of the letters.

If a region should remain understandable in a redacted document, consider whether a short explanation outside the image would help the recipient. Do not invent content under the box or imply that the hidden information has been independently verified. The exported copy simply contains your selected visible redactions.

Place rectangles carefully

Use the available selection and numeric editing controls to define regions. Check the boundaries at a useful scale. Anti-aliased text can extend slightly beyond the dark core of a letter, and a shadow or highlight can reveal part of the shape. Thin margins are especially risky when the image will be enlarged later.

Consider the destination crop. A recipient may display the image against another background, resize it or zoom into a region. The cover should be part of the saved pixels, not an overlay that only exists in your editing view. A flattened export is intended to make the marked result independent of the editing controls.

For a photograph, a rectangle over a face may not remove identifying context such as a name badge, distinctive clothing or a nearby sign. For a screenshot, the same username can appear in a title bar and a sidebar. Treat the scene as a collection of clues rather than a single field to hide.

Manual review and assisted checks complement each other

This page is useful when you want direct control over known regions. Image Privacy Check adds optional supported metadata, text, face and QR suggestions. Those suggestions can help you find additional items, but they are not exhaustive and are not a prerequisite for making a manual redaction here.

Handwriting, tiny text, uncommon typography and reflected content can evade automated checks. A detector that returns no finding has not proved that a region is harmless. Conversely, a detected name or face may be information you have a legitimate reason to preserve. You remain responsible for the final choices in context.

When the source is a PDF, do not assume that placing a rectangle over an exported image changes the original PDF's text or objects. Use Redact PDF for that separate document workflow. Image redaction protects the resulting image copy, not every earlier file from which the pixels were obtained.

Verify the format you intend to share

The output is flattened and reopened for supported checks. PNG can support exact opaque-pixel verification within marked regions. JPEG encoding is lossy and may alter pixels near boundaries, so its output needs the corresponding checks and a careful visual review. A smaller JPEG is not automatically the appropriate choice for a particularly sensitive text redaction.

Re-encoding does not copy source identifying metadata, but the new file can still contain the ordinary structural information its format needs. Source ICC profiles are not preserved by the simple image workflow. Keep an original master privately when metadata, color handling or evidentiary provenance is important.

Check the reported verification scope. Incomplete or unavailable checks are limits, not evidence that a file passed them. Even a verified cover tells you nothing about unmarked information elsewhere in the image. The privacy task ends with reviewing the saved output as a whole.

Save a copy you can identify unambiguously

Download the result, reopen it and inspect every redacted region. Look at the complete frame again, then verify the actual format, dimensions and orientation. Make sure the file you attach or publish is the exported copy, not the original still selected in another window. Similar filenames can cause a more serious mistake than a slightly awkward rectangle.

A download request only establishes that the browser started the saving path. It does not prove completion on disk or delivery to another person. Keep your source and result distinguishable while reviewing, and retain the original according to your own legitimate record-keeping needs rather than deleting it to create a false sense of safety.

Respect the processing boundary

The configured input limit is 20 MiB, 12 megapixels and a maximum 6,000-pixel edge for supported still PNG, JPEG and WebP. Browser resources can impose additional limits. The tool does not expand its scope because a file has a familiar extension, nor does it promise universal removal of all possible identifiers.

Creative bytes stay in the browser workflow, while application code and optional resources still load from the site. Local processing does not remove the risks of shared devices, extensions or exposed downloads. Use the redaction versus blur guide when choosing the method, and use Image Inspector for a separate supported metadata review of the actual saved copy.

Continue with these tools

Open a separate workflow. Files are not transferred automatically; save your current work before leaving.

Format context: PNG