Screenshot Redaction

Screenshot redaction is the deliberate removal of sensitive visual information from an image before the image is shared. Typical targets include names, email addresses, phone numbers, account IDs, API keys, customer records, internal URLs, addresses, notifications, and other details that were not meant for the recipient.

A useful distinction is that redaction is about removing the selected information from the released image, not merely making it harder to read. This tool uses fully opaque black replacement pixels and creates a new PNG from the decoded image.

The tool is intentionally manual. It does not guess which parts of a screenshot are private, because only the person reviewing the image knows which details are sensitive in the particular context.

Choose one PNG, JPEG, or WebP screenshot and let the browser decode it locally.

Draw a rectangle around every sensitive region, or enter exact source-image coordinates when a repeatable or keyboard-driven selection is useful. Use 100% or 200% zoom when inspecting small text.

Add source-pixel edge padding when a tight selection could leave anti-aliased characters, shadows, borders, or nearby pixels visible.

Prepare the encoded release preview and inspect the actual PNG at source size. Do not rely only on the editor view: the release image is the file that will be shared.

A solid opaque box and a blur are not the same privacy operation. Blur and pixelation transform the original pixels while retaining some visual information; a solid replacement overwrites the selected pixels with one opaque colour.

For short, predictable, or high-risk values such as account numbers, passwords, access tokens, and API keys, a solid opaque replacement is the more conservative choice than relying on a visual obscuring effect.

This tool deliberately provides opaque black replacement rather than blur or pixelation. It also re-encodes the result as a new PNG so the editor's selection guides are not part of the released image.

A black shape placed above text in a layered document is not necessarily redaction. If the underlying text or image remains in the document, someone with access to the editable file may be able to move, hide, or delete the covering object.

A screenshot redaction should therefore be judged by the exported representation, not just by how the editor looks while you are working. This tool decodes the source image, paints committed regions directly onto a canvas, and encodes a fresh PNG from that canvas.

The resulting PNG contains opaque black pixels at the selected source-image coordinates. That does not, however, erase the original screenshot or any other copy that already exists elsewhere.

Start with direct identifiers such as names, email addresses, phone numbers, usernames, account numbers, customer IDs, and faces.

Then inspect technical and contextual details: API keys, tokens, internal URLs, IP addresses, hostnames, ticket numbers, filenames, browser tabs, notification previews, Wi-Fi names, maps, addresses, timestamps, and device information.

Context can identify a person or organisation even when no single field looks sensitive. A project name, conversation, date, unique incident, or combination of harmless-looking details may reveal more than the obvious text.

The most common redaction mistake is stopping after covering the field that first attracted attention. Screenshots capture the surrounding interface too: sidebars, tabs, notification banners, avatars, thumbnails, reflections, filenames, status bars, and background windows can all carry identifying information.

After creating the release PNG, inspect every edge and corner at source size. If the screenshot is going into a public post, support ticket, bug report, documentation page, classroom, or team channel, review it from the perspective of someone who has no reason to know the original context.

A screenshot displayed at a different size from its source dimensions can map pointer positions to fractional source pixels. A selection that looks perfectly tight on screen can therefore miss a narrow edge when converted back to image coordinates.

This tool can add padding in source-image pixels around each committed region. The left and top boundaries are conservatively floored, while the right and bottom boundaries are ceiled, and the result is clipped to the image.

Padding is a review aid, not a substitute for reviewing the final image. Cover too little and sensitive pixels may remain visible; cover too much and useful surrounding information may be unnecessarily removed.

The editor includes borders, numbered region labels, and a temporary translucent rectangle while you work. Those are editing guides, not part of the release file.

The release preview is generated separately from the encoded PNG Blob. That gives you an opportunity to inspect the actual output that will be downloaded rather than assuming that the editing canvas represents the final file.

Before sharing, check the covered edges, the corners of the image, the output dimensions, and any sensitive context that was outside the rectangles you originally selected.

The screenshot is decoded, edited, coverage is calculated, the PNG is encoded, and the release preview is created in the browser. The tool does not upload the image or add it to the page URL.

Local processing reduces the need to transmit the screenshot to a redaction server, which is particularly useful when the image contains customer information, internal systems, financial details, credentials, or private conversations.

Local processing is not the same as secure deletion. The original screenshot can still exist in your downloads, cloud storage, backups, messaging history, clipboard history, or other copies outside this editor.

A redacted PNG is a new file. Creating it does not modify or sanitize the original screenshot, a cloud copy, a backup, a message that was already sent, or a copy held by another person.

If the original contains sensitive information, handle those other copies separately according to the sensitivity of the information and the procedures that apply to your situation.

Privacy review should include the filename as well as the visible image. A filename can contain a person's name, project, date, ticket number, customer reference, or case identifier.

The tool defaults to a generic redacted-screenshot filename and creates a newly encoded PNG instead of copying the source file container. However, the browser may add new encoder-generated information such as standard resolution metadata, so the output should not be described as containing no metadata of any kind.

Cropping can be an excellent privacy measure when the sensitive information is outside the part of the screenshot that actually needs to be shared. Removing the entire region is often simpler than covering it.

Redaction is useful when the sensitive detail sits inside the content that must remain visible. In practice, a careful workflow can combine both: crop unnecessary surroundings first, then redact information that remains inside the useful frame.

Screenshots are useful evidence for reproducing software problems, but they often capture much more than the bug itself. Before attaching one to a support ticket or issue report, check for customer names, email addresses, internal hostnames, tokens, URLs, account IDs, browser tabs, and notification previews.

Keep the technical evidence that helps explain the problem while removing information the recipient does not need. This is particularly important when a screenshot will be copied into a public issue, forum post, documentation page, or third-party support system.

This tool is designed for manual image editing, not automatic discovery of every sensitive field. It does not perform OCR, face detection, automatic PII detection, document classification, or compliance verification.

Legal disclosure, litigation, regulated investigations, formal records requests, classified material, medical information, or evidence handling may require approved software, documented procedures, trained review, and independent verification. A browser redaction tool should not be treated as proof that a release is legally or organisationally compliant.

The optional manifest records the source-image dimensions, normalized redaction rectangles, opaque replacement colour, region count, and the number and percentage of covered pixels.

It does not contain OCR text or copies of the covered pixels, but its coordinates can reveal where information appeared and how much of the image was hidden. Treat the manifest as potentially sensitive and normally keep it private unless a reviewer specifically needs it.

A browser-based redaction workflow can be useful when the screenshot itself is sensitive. This tool keeps the source image and redaction state in the browser instead of sending the image to a remote redaction service.

That privacy property should be stated precisely: it means this tool does not upload the screenshot for processing. It does not guarantee that every copy of the source image has disappeared from the device or from systems where the screenshot was previously stored or shared.