Skip to main content
العودة إلى المنتجات
python

DotScramble

أداة قوية لحماية خصوصية الصور

1 نجوم0 تفرُّع0 مشاكلالإصدار v1.6.0
DotScramble • README•16 min read
DotScramble icon: a dot dissolving into pixelsDotScramble icon: a dot dissolving into pixels

DotScramble: blur faces, license plates and text in photos on Linux

Hide what shouldn't be seen, before you share. A free, open-source photo privacy studio: blur, pixelate or black-bar faces, eyes, license plates and text, strip or fake EXIF and GPS metadata, and hide faces from AI face detection. Everything runs on your computer.

Version License: Apache 2.0 Python 3.10+ Linux CI 7 languages Part of DotSuite

Download · Website · Changelog · DotSuite · All my tools

DotScramble finding four faces in a group photo and pixelating themDotScramble finding four faces in a group photo and pixelating them


Why DotScramble?

A photo you are about to post often shows more than you meant to: the faces of people who never agreed to be online, a car's number plate, a name on a badge, an address on a parcel, and the exact GPS location hidden in the file. DotScramble removes all of that in one place:

  1. Open a photo, or drop it on the window.
  2. Pick what to hide (faces, eyes, plates, text, a word, or regions you draw) and how (blur, pixelate, black bar, mosaic…).
  3. Apply, compare and save. Choose whether the saved file keeps, strips or fakes its metadata.
  • Private: your photos are processed on your machine and never uploaded.
  • See before you commit: Detect shows every region it will hide, and you click one to skip it. Compare puts the original and the result side by side.
  • Honest: it tells you how many regions it hid, and it refuses to pretend: if detection can't run, you get an error, never a "done" with the face still visible.
  • AI-aware: blurring is enough for people, not always for software. AI Evasion targets face detectors too.
  • Free and open source: Apache 2.0, part of the DotSuite family. An optional Pro plan adds AI-powered detection and the advanced tools.

Features

🎯 What to hideFace, eyes, full body, license plate, any text (OCR), one specific word, regions you draw, or the whole image.
🎨 How to hide itBlur, pixelation, black bar, gradient, mosaic, frosted glass and oil paint, with strength, block size and opacity controls.
🛡️ AI EvasionHybrid blurs the face for people and hides it from AI face detection; Invisible leaves it looking unchanged while hiding it from the detector.
🔍 Detect previewEvery region outlined before anything changes. Click a region to skip it; with several faces you protect exactly the ones you choose.
↔️ CompareA before / after split view with a draggable divider.
🔎 Zoom and panMouse-wheel zoom around the cursor, fit or 100 %, and drag to move around a large photo.
⚡ Live previewChange a setting and the result updates on its own, without stacking the effect.
🧾 Metadata on saveKeep it, remove only the location, strip everything, or replace it with believable decoys (Blend in: a current phone, a spot a few km away, the date shifted by days; or Ghost, Troll, Artist), or set every field yourself. A report shows each field's value in the photo and in the saved file, and every save is read back and checked.
🕵️ What's in this photo?Free for everyone: a warning when a photo carries its exact location or tags people's faces, and a list of everything the file holds, with a Show on map button.
🧹 Deep cleaningRemoves what lives outside the usual EXIF fields too: the hidden preview of the unedited photo, XMP (edit history, face regions, names), IPTC, comments and PNG text (where AI tools store the prompt), and extra images hidden after the photo. The color profile is kept, so colors don't change.
✂️ Metadata onlySave metadata only writes the open photo back untouched with new metadata, and File → Clean Metadata Only (Ctrl+Shift+M) does it for many photos or a folder. JPEG image data is copied byte for byte, so there's no quality loss.
📦 Batch processingOne detection mode and effect for a whole folder, with progress, a summary, and per-file errors.
💾 Presets, undo and redoSave your favourite settings; step back and forward through every change.
🌍 7 languagesEnglish, العربية, فارسی, Español, Русский, Deutsch, Türkçe, with right-to-left layouts and correct plurals.
🎨 16 themesMidnight, Light, Nord, Dracula, Tokyo Night, Catppuccin, Rosé Pine, Gruvbox and more; every screen follows the theme.
🔐 Signed updatesReleases are built by GitHub Actions and signed; the app checks the checksum and the GPG signature before it installs anything.

Free and Pro

Everything you need to hide faces and plates is free, and so is seeing what a photo gives away. Pro (the DotScramble Max plan on DotSuite) adds AI-powered tools and full metadata control:

FreePro
All 7 effects, Detect preview, Compare, zoom, live preview✅✅
Face and eye detectionHaar cascadesAI face ovals and tilted eye bars (MediaPipe)
Manual regions and full image✅✅
Full body, license plates, text (OCR), one specific word✅
AI Evasion (Hybrid and Invisible)✅
What's in this photo? and the location warning✅✅
Remove the location only4 photos a dayUnlimited
Strip everything, decoys, per-field metadata editor✅
Clean metadata only (many photos or a folder)Location only, within the daily 4Every choice
Batch processing3 images per batch, metadata always strippedUnlimited, any metadata choice

Activate Pro from Help → Upgrade to Pro: you sign in on dotsuite.dev in your browser and the app is activated automatically. Pro works on 2 computers per account. Help → License & devices shows this computer's Machine ID and the devices in use; you can remove 2 devices every 30 days yourself (in the app or on dotsuite.dev), and support can always move Pro for you.

AI Evasion

A blurred face is unreadable to people, yet face detectors can often still find it, and a photo that software can find a face in can be indexed, matched and tracked. AI Evasion adds a small, optimized pattern of noise to the face region until a face detector no longer finds a face there.

Original → face oval found → black-box optimisation (SPSA) against the detector → protected face
  • Hybrid: blur for people plus the adversarial pattern for software.
  • Invisible: the face looks the same to people; only the detector is fooled.
  • It runs on the CPU and takes about 30–90 seconds per face, with a progress dialog you can cancel.
  • The result is reported honestly: complete, partial (with the evasion rate), or failed (the face is still blurred, but evasion was not verified).
  • Save the result as PNG: JPEG compression can weaken the pattern.

The pattern is tuned against the face detector DotScramble itself uses (MediaPipe). It is not a guarantee against every face recognition system, so treat it as an extra layer on top of a visible effect, not a replacement for one.

Install

Every package installs the same app. All of them are on the Releases page.

MethodHowBest for
AppImagechmod +x DotScramble-Linux-x86_64.AppImage && ./DotScramble-Linux-x86_64.AppImageany distribution, no install
Debian / Ubuntu / Kali / Mintsudo apt install ./DotScramble-Linux-x86_64.deb, then run dotscrambleapp-menu integration
Arch / Manjarosudo pacman -U DotScramble-Linux-x86_64.pkg.tar.zst, then run dotscrambleArch-based systems
Portabletar xzf DotScramble-Linux-x86_64.tar.gz && ./DotScrambleone binary, anywhere
From sourcesee belowdevelopment

Optional, for the text and word modes:

BASH
sudo apt install tesseract-ocr        # Fedora: sudo dnf install tesseract · Arch: sudo pacman -S tesseract
  • AppImage: if it won't start because of FUSE, install libfuse2 (libfuse2t64 on Ubuntu 24.04+) or run it with --appimage-extract-and-run.
  • AI model: the first time you use a Pro AI feature, DotScramble asks to download Google's MediaPipe face landmark model (about 8 MB).

Supported systems

The packages are built on Ubuntu 22.04 and need glibc 2.35 or newer on x86_64:

DistributionWorks
Ubuntu 22.04 LTS and newer (Linux Mint 21+, Pop!_OS, Zorin, elementary OS 7+)✅
Debian 12 (bookworm) and newer✅
Kali Linux (rolling)✅
Fedora 36 and newer✅
Arch, Manjaro, EndeavourOS✅
openSUSE Tumbleweed✅
Ubuntu 20.04, Debian 11, RHEL / Rocky / Alma 9 and older❌ glibc too old: run from source
Windows, macOSnot yet

Works on X11 and Wayland (GNOME, KDE and others), and snaps to half the screen like any window.

Verify your download

Every release has a SHA256SUMS file, a detached GPG signature (.asc) for every package, and the public key (DotScramble-signing-key.asc):

BASH
sha256sum -c SHA256SUMS --ignore-missing          # checks the files you downloaded
gpg --import DotScramble-signing-key.asc
gpg --verify DotScramble-Linux-x86_64.AppImage.asc DotScramble-Linux-x86_64.AppImage

Releases are signed by FreeRave <kareem209907@gmail.com>, with this key fingerprint:

0F04 6781 3146 DE68 272E DC82 145D 3459 56DE 18D2

Check it after importing with gpg --fingerprint 145D345956DE18D2. A Good signature from that key means the file is exactly what the release workflow built. The built-in updater checks the same signature against the same pinned fingerprint before it installs an update.

From source

BASH
git clone https://github.com/kareem2099/DotScramble && cd DotScramble
python3 -m venv venv && source venv/bin/activate
pip install -r requirements.txt
python src/main.py

Python 3.10 or newer. requirements.txt pins OpenCV to the 4.x contrib build on purpose: OpenCV 5 no longer ships the Haar cascade files the free face, eye and body detection uses.

Usage

Before / after split view: drag the divider to compare the original with the blurred faceBefore / after split view: drag the divider to compare the original with the blurred face

The basic flow: open a photo (Ctrl+O, the image browser, or drag and drop) → pick what to hide in step 1 → pick the effect in step 2 → fine-tune in step 3 → Apply effect (Ctrl+Enter) → Save (Ctrl+S).

  • Detect (Ctrl+F): outlines what the current mode finds. Click a region to skip it; Apply then hides only the regions still outlined.
  • Manual: choose Manual and drag on the photo to draw as many regions as you need.
  • Targeted text: type a word (a name, a phone number) and only that word is hidden.
  • Compare (Ctrl+Shift+C): drag the divider to see the original and the result.
  • Live preview: turn it on in the toolbar and every change re-applies by itself.
  • Metadata on save: in step 4 choose Keep, No location, Strip or Decoy. When you save, a report shows each field's value in the photo and what the saved file will hold. To change only the metadata, without any effect, press Save metadata only.
  • What's in this photo?: the link in step 4 lists everything the open photo carries.
  • Opening photos: the image browser opens where you were last, with your last photo selected, and Recent lists the photos you opened from any folder.
  • Batch (Ctrl+B): add files or a whole folder, pick a mode and an effect, choose an output folder, and start.
  • Presets: Presets → Save Current Settings, then load them any time.

Keyboard shortcuts

ShortcutActionShortcutAction
Ctrl+OOpen an imageCtrl+SSave the result
Ctrl+EnterApply the effectCtrl+DClear selections
Ctrl+ZUndoCtrl+YRedo
Ctrl+FDetect previewCtrl+Shift+CCompare
Ctrl++ / Ctrl+-Zoom in / outCtrl+0 / Ctrl+1Fit / actual size
Ctrl+BBatch processing

Languages and themes

DotScramble in Arabic, right to left, on the light themeDotScramble in Arabic, right to left, on the light theme

View → Language switches between English, العربية, فارسی, Español, Русский, Deutsch and Türkçe on the fly; Arabic and Persian use a right-to-left layout. DotScramble starts in your system language when it is one of them. View → Themes has 16 themes, and every dialog follows the one you pick.

Updates

DotScramble checks GitHub for a new release in the background. When there is one, it downloads the file that matches how you installed it, verifies the checksum and the signature, and offers to restart. If a version is ever retired, an older copy shows a clear Update required screen with a download button. Running from source never updates itself. The AppImage also carries update information, so AppImageUpdate, Gear Lever and AppImageLauncher can update it too.

What goes online

Your photos never leave your computer. DotScramble connects only to:

WhenWhereWhy
Startup and Check for Updatesapi.github.com, raw.githubusercontent.comnew releases and the supported-version policy
Downloading an updategithub.com release filesthe new version
Activating or re-checking Pro (about once a day)dotsuite.dev, core.dotsuite.devyour licence: an anonymous computer ID, a computer name, the app version and daily counts of Pro use, numbers only, never photos, file names or metadata
First use of a Pro AI feature, if you agreestorage.googleapis.comthe MediaPipe face model (about 8 MB)

Settings, presets and history stay in ~/.local/share/DotScramble.

FAQ

How do I blur faces in a photo on Linux?

Open the photo, keep Face selected in step 1, pick Blur (or Pixelation, Black Bar, Mosaic…) in step 2, and press Apply effect. Press Detect first if you want to see the faces it found and skip some of them.

How do I hide a license plate or a number plate?

Choose License Plate in step 1 (Pro), or Manual and draw a rectangle over the plate (free). Pixelation or Black Bar work best for plates.

How do I remove EXIF and GPS location from a photo?

In step 4, Metadata on save, choose No location to remove just the GPS position (free for 4 photos a day), Strip to remove everything, or Decoy to replace it with believable fake values (Pro). When you save, a report lists every field before the file is written, and the saved file is checked afterwards. Save metadata only does this without changing the photo, and File → Clean Metadata Only does it for a whole folder.

Can DotScramble hide my face from AI face recognition?

Its AI Evasion mode (Pro) adds an optimized noise pattern until a face detector no longer finds the face, and reports whether that fully worked. It is tuned against the MediaPipe detector the app uses and is not a guarantee against every system, so use it together with a visible effect (the Hybrid mode does both).

Can I blur one specific word, like a name, in a screenshot?

Yes: choose Targeted Text (Pro) and type the word. It needs Tesseract OCR (sudo apt install tesseract-ocr). Text hides every piece of text it finds instead.

Can I blur faces in many photos at once?

Yes, with Batch processing (Ctrl+B): add a folder, pick the mode and the effect, and DotScramble processes every image and reports any that failed. Free batches hold three images.

Does DotScramble upload my photos?

No. All detection and effects run locally. See What goes online for the few connections it does make, none of which send your images.

How do I move Pro to another computer?

On the old computer choose Help → License & devices → Deactivate this computer, then activate on the new one. If the old computer is gone, remove it under Devices on your DotScramble dashboard at dotsuite.dev. You can remove 2 devices every 30 days yourself; for more, email support@dotsuite.dev.

Is DotScramble free?

Yes. It is open source under Apache 2.0, and blurring faces, plates you draw and whole images is free forever, and so are the metadata warning, the inspector and removing the location from 4 photos a day. Pro adds the AI-powered detection, AI Evasion, full metadata control and unlimited batches.

Does it support Arabic and other languages?

Yes: English, العربية, فارسی, Español, Русский, Deutsch and Türkçe, with a right-to-left layout for Arabic and Persian and correct plural forms in every language. Translations other than English and Arabic are new: corrections are very welcome.

Which Linux distributions does it run on? What about Windows and macOS?

Any x86_64 distribution with glibc 2.35 or newer: Ubuntu 22.04+, Debian 12+, Fedora 36+, Kali, Arch and others (see Supported systems). On older systems, run it from source. Windows and macOS builds are planned once they can be tested properly.

Part of DotSuite

DotScramble is one of the developer and desktop tools in DotSuite by FreeRave:

Develop

BASH
pip install -r requirements.txt pytest
QT_QPA_PLATFORM=offscreen pytest -q tests
  • Tests run offscreen and on every push in CI. The licence module (src/managers/license_manager.py) is not in the repository: tests that need it run in CI, where it comes from a secret.
  • Translations live in src/assets/locales/<lang>.json; counted strings use CLDR plural forms ({"one": …, "other": …}). Adding a language is one JSON file plus an entry in languages.json.
  • Styling lives in one stylesheet, assets/themes/theme_template.qss, filled from the 16 themes in assets/themes/themes.json. Widgets only set object names and variant / role properties.
CODE
src/
  main.py               entry point
  config.py             settings, version, update source
  views/                main window, canvas (zoom / compare / detect overlay), sidebar widgets, icons
  models/               detection engine (Haar, MediaPipe, OCR), effects, AI evasion (SPSA)
  managers/             themes, localization (plurals, RTL), licence, local auth, database
  assets/locales/       en, ar, fa, es, ru, de, tr
core/                   image picker, batch processor, auto-updater, version gate, metadata writer
                        (plan, write, verify) and spoofer, daily quota, Pro usage counts
gui/                    batch window, metadata editor, save report, inspector, cleaner, license window
assets/                 themes (QSS + JSON), icons
tests/
.github/workflows/      ci.yml (tests), release.yml (build, smoke-test, sign, publish)
version_policy.json     the oldest version still supported

Issues and pull requests are welcome on GitHub. Start with CONTRIBUTING.md.

If DotScramble keeps your photos private, you can support it through GitHub Sponsors, Ko-fi or Buy Me a Coffee.

License

Apache 2.0 © FreeRave (Kareem). PySide6 (Qt for Python) is used under the LGPL v3. Part of DotSuite.

منتجات مشابهة

‌
‌
‌
‌
‌
‌
‌
‌
‌
‌
‌
‌
‌
‌
‌
‌
‌
‌
‌
‌
‌