Video redaction removes or obscures information that the intended viewer does not need. The work has two parts: place the right mask over the sensitive area, then keep it there for every frame where the information appears.
Use manual masks when you need to cover faces, license plates, documents, or one-off interface details. Automatic detection can reduce the search work for supported categories, but the finished export still needs a full human review.
Decide what must be removed before you open an editor
List the information the final audience should not see. Include text, faces, plates, notifications, filenames, open tabs, reflections, and anything spoken in the audio. A clear scope makes the review measurable.
Avoid capturing the information when you can. Use sample data, close unrelated apps, disable notifications, and record only the necessary window or region. Redaction is a second line of defense, not a reason to record a crowded desktop.
Choose a manual or automatic method
| Method | Best fit | Main review risk |
|---|---|---|
| Manual masks | Faces, plates, documents, unusual layouts, and details an automatic detector does not support | The mask can drift, start late, end early, or miss another appearance |
| Automatic text detection | Repeated supported text patterns across a longer screen recording | Detection can miss a value, misclassify text, or cover the wrong region |
| Capture-time page blur | Known webpage elements you can hide before a Chrome recording starts | Navigation, dynamic content, embedded frames, or layout changes can expose new content |
Manual method: place and track masks through the video
- Import or open the highest-quality source in an approved editor. Keep an untouched restricted copy according to your organization’s policy.
- Move to the first frame where the sensitive detail appears and place a mask that fully covers it, including the edges.
- Track or keyframe the mask as the subject moves, the page scrolls, or the camera changes. Split the mask at cuts instead of stretching one track across unrelated scenes.
- Check the frame before the mask starts and the frame after it ends. Extend the timing if any part of the information flashes into view.
- Repeat the process for every separate appearance. Search the whole recording; the same name, face, plate, or account number may return later.
Automatic method: treat detection as a first pass
An automatic system can find supported categories and create draft masks. With Zight’s account-enabled post-recording text redaction, the owner starts redaction on an eligible completed Zight-hosted MP4 video. Zight detects supported text and covers it with permanent opaque masks rather than blur.
Supported categories are email addresses, phone numbers, payment card numbers, Social Security numbers, AWS access key IDs, and selected API keys or secrets. Automatic face and license-plate detection are not verified current Zight capabilities.
Review every proposed mask. Detection narrows the search; it does not transfer responsibility for the finished video to the software.
Inspect the actual export after the editor preview
- Watch the complete exported file at normal speed.
- Pause around every mask entrance, exit, scroll, zoom, scene cut, and occlusion.
- Zoom in to check that the original pixels do not remain visible around the edges.
- Listen for names, account numbers, addresses, or other sensitive details in the audio.
- Open the final share link with the same access the recipient will have.
- Confirm the unredacted original is not attached, linked, or available through ordinary version history by mistake.
Blur, pixelation, and opaque covers solve different problems
Blur and pixelation preserve some visual context, but weak settings can leave text or identifying features readable. An opaque cover removes more visual information and is easier to inspect, although it can hide nearby context.
Choose the treatment your policy requires, then test the rendered output. Zight automatic redaction uses opaque masks. Manual blur, pixelation, and black-box controls are available in the web editor through account enablement.
Frequently asked questions about video redaction
Is blur the same as redaction?
Blur is one visual treatment. Redaction is the broader process of removing or obscuring information and verifying the result. An opaque cover can be a stronger choice for readable on-screen text.
Can video redaction be fully automatic?
Automation can detect and track supported content, but a person should review the full export. Unsupported categories, motion, cuts, low contrast, and brief appearances can produce misses.
Does Zight redact every upload automatically?
No. The video owner starts redaction on an eligible completed Zight-hosted MP4 video. Contact Zight to discuss enablement for your account.