About this tool
Redact images in your browser, black bars, blur, brush
Hide faces, license plates, signatures, addresses, credit card numbers, or any sensitive region of a photo before posting or sharing. The redaction is permanent, once exported, the hidden pixels are gone. Nothing uploads.
Why use OmegaPix
-
Permanent redaction : Pixels under a black bar or blur are physically overwritten in the exported file. There is no hidden layer a viewer can peel back. Unlike PDF "highlight" redaction that some tools produce, this is real.
-
Three quick tools : Rectangle black-bar for blocky redaction (the news-style hard cover), rectangle blur for softer obscuration, and a freehand brush for irregular regions (curved signatures, handwriting strokes).
-
Nothing leaves your device : Sensitive photos are the worst case for upload-based tools. The server sees exactly the thing you wanted to hide. This redactor runs entirely in your browser via Canvas. Verify with DevTools → Network during the operation.
How it works
Drop your image
JPG, PNG, WebP, or AVIF. Image loads into a full-resolution canvas in your browser.
Paint over what to hide
Pick black-bar, blur, or brush. Click and drag to cover the region. Repeat for multiple areas. One-step undo if you slip.
Export the redacted copy
Save as JPG, PNG, or WebP. The exported file has the redaction baked into the pixels, viewers see only the redacted version.
Frequently asked questions
Are the redacted pixels truly gone?
Yes. Black-bar redaction overwrites the pixels with solid color in the canvas before export, the original image data under the bar is not present in the exported file. Blur redaction overwrites with a blurred sample of the surrounding pixels; the original sharp pixels are also gone. You cannot recover the redacted content from the exported file.
Is blur safer than a black bar?
Black bar is strictly safer than blur. A heavy blur scrambles fine detail but can sometimes be partially reversed by clever image-processing techniques on simple content (e.g. license plates with limited character sets). For maximum safety against motivated attackers, use black bars. Blur is for casual social-share contexts where readers should know "something was here" without it being a hard visual interruption.
Does redaction work on a layered PSD or similar?
No. The redactor works on flattened raster images (JPG, PNG, WebP, AVIF). If you start from a layered file, export it as PNG first, then redact the export. Layered editing is outside the scope of this tool.
Can I undo a redaction stroke?
V1 supports single-level undo (undo the most recent stroke). Multi-step undo/redo and a history panel are on the roadmap. If you make multiple bad strokes, use "Clear all redactions" to start over.
Will the redaction survive social-media compression?
Yes. The redacted pixels are part of the exported image file. Instagram, Twitter/X, Facebook, and others recompress images on upload but cannot recover content that was never in the file in the first place.
Can I redact metadata at the same time?
V1 redacts pixels only. EXIF, GPS, and other metadata in the source file are preserved in the export. If you also want to strip metadata (likely yes, sensitive photos usually have sensitive EXIF), run the redacted export through the EXIF Remover after.
Does it support batch processing?
V1 is single-image only. Batch mode is on the roadmap. For now, repeat the workflow per image.
Can it auto-detect faces or license plates?
Not in V1. Auto-detection requires a machine-learning model that would add significant bundle weight and license complexity. The browser-only constraint also limits which models we can ship. Manual painting is the v1 approach, fast in practice for small numbers of regions.
When to use this tool
Photos with bystanders
Black-bar over faces of strangers before posting a street photo, restaurant shot, or event recap.
Screenshots with personal info
Blur the address bar, email column, or order numbers before sharing a bug report or how-to walkthrough.
Documents and IDs
Black-bar SSNs, account numbers, signatures, and addresses on photographed forms before sending to clients or posting to support forums.
Journalism and research
Obscure source identities in field photos, license plates of vehicles, identifying tattoos or clothing.
When not to use this tool
The redaction must be reversible
There is no reversible redaction in a flat image file. If the original needs to be recoverable, keep the original on a secure system and create a separate redacted copy.
You need exact-text redaction in a PDF
PDF redaction is a different operation, covering or removing specific text glyphs in a vector PDF. Use a dedicated PDF redactor for that. This tool is for images.
You need professional forensic redaction
For legal/forensic contexts where chain-of-custody and certified workflows matter, use a tool that produces an audit trail. This is a consumer-grade redactor for everyday sharing.
Technical details
How black-bar redaction works
The image loads into an off-screen canvas at full natural resolution. A black-bar rectangle is drawn with fillRect() over the selected region. The image is then exported via toBlob(). The pixels under the rectangle are physically overwritten in the export. There is no transparency, no layer, no metadata flag.
How blur redaction works
The selected region is extracted into a temporary canvas, blurred with the canvas filter property (filter: blur(Npx)) where N scales with image size, then composited back over the source region. The original sharp pixels in that region are overwritten by the blurred sample. The blur uses Gaussian convolution implemented natively by the browser.
How the freehand brush works
Mouse/pointer events accumulate into a path that is stroked onto the canvas with the chosen tool (black or blur). Brush size scales the stroke width. For large redacted areas of irregular shape, brush is faster than building up rectangles.
Why no auto-detect in v1
Browser-only face detection requires shipping a model file (typically 1-5 MB). Reasonable v2 candidates include MediaPipe Face Detection (~2 MB compressed) or a small custom YOLO variant. Both add real load time and licensing surface that v1 deliberately avoids. Manual paint is fast for the 1-3 regions most users need to redact.
Pixel-data integrity check
You can verify redactions are real by saving the export and opening it in an image editor. The pixels in the redacted region will be solid color (black-bar) or blurred (blur tool). No trace of the original. This is fundamentally different from "highlight" annotations in some PDF tools, which can be removed by a viewer.
Your files stay on your device
The image is read into your browser via the File API, redacted on a Canvas element, and exported via canvas.toBlob, no upload, no server processing, no third party sees the original or the redacted output. The Canvas API is supported in every modern browser; nothing else is loaded.