A Windows toolkit for what is really inside an image
StegoPixel is a cross-platform desktop tool for image metadata management, privacy cleanup, steganography and forensics — shipped as a Windows application, and built so that none of it needs a server.
What it is
An image file is two things at once: the picture you see, and a large amount of data you do not. StegoPixel exists for the second part. It reads the metadata other tools hide, hides data of its own in pixels and metadata, finds what somebody else hid, and estimates whether a photograph was ever a photograph.
It is one application rather than a collection of command-line utilities because the tasks share a working set. You load an image once, and the metadata inspector, the privacy scanner, the forensic panels and the steganography tools all operate on that same file, with results that stay available to each other. When you need to do the same thing to four hundred files, the workflow builder turns the sequence you just performed by hand into a saved pipeline.
Hidden data
Write payloads into pixels or metadata, and read payloads out of images that arrive from elsewhere.
Visible evidence
Twelve forensic instruments and eight enhancement modes for finding what an edit left behind.
Private by default
No account, no cloud, no telemetry. Your images are processed on your machine and stay there.
How it is designed
StegoPixel's interface is built on a design system the project calls “Data Clarity First”. It exists because forensic work involves long sessions of scanning subtle visual anomalies, and every unnecessary interface element costs attention that should be spent on the image.
Zero UI distraction. Panels and borders stay muted so your eyes go to the image. Floating toolbars that overlay the content are banned — controls sit in a docked bar above the canvas, never on top of it.
Contrast does the work. Critical readouts — hashes, dimensions, tag values — are separated by strong text contrast rather than by colour, so they remain legible in either theme and for colour-blind users.
The accent means something. The cyan accent is never decorative. It marks an active state, a primary action, or a status that needs attention — nothing else.
Data is structured, not dumped. Dense metadata becomes labelled cards with an uppercase title above a high-contrast value. Hashes are never wrapped mid-character.
Discovery surfaces look different from analysis surfaces. Dashboards, empty states and feature cards use a frosted-glass treatment with an icon-forward layout; forensic panels stay solid and quiet, so the two modes are never confused.
Motion is minimal on purpose. Only a handful of short opacity fades exist in the whole application. Everything else snaps, because animation in a tool you use for hours is noise.
This website is a translation of that same system — the same palette, type scale, corner radii and spacing values, carried over so the product and the page that describes it read as one thing.
Cryptography and algorithms
What protects a hidden payload, and what the forensic checks are built on:
AES-256-CBC with PBKDF2-HMAC-SHA256 at 100,000 iterations, a 6-byte salt and a 16-byte IV for payload encryption.
Reed–Solomon RS(255,223) over GF(256) for the optional error-correcting mode, with Berlekamp–Massey, Chien and Forney decoding.
RSA-2048/4096 and ECDSA P-256/384/521 for image signing, with PKCS#8 and PEM key import.
SHA-256 and SHA-512 for content digests, plus average, difference and perceptual hashes for similarity.
PCA via a closed-form trigonometric eigensolver, and a chi-square pairs-of-values LSB test reported with a Wilson–Hilferty p-value.
Block-hash copy-move detection with union-find clustering of matched regions.
Requirements
StegoPixel is a Windows desktop application. Everything it needs ships inside the package, so there is no separate runtime or framework to install first.
Product
StegoPixel
Version
1.0.0 pre-release
Status
Not yet published
Distribution
Microsoft Store
Operating system
Windows 10 / 11
Minimum version
10.0.17763 (1809)
Architecture
x64 · arm64
Runtime
Bundled — nothing to install separately
Disk footprint
Self-contained package
Administrator rights
Not required
Per-user install. Nothing is written to a system-wide location, and installation needs no elevation.
Self-contained. The runtime and native libraries are bundled, which is why the package is larger than a framework-dependent build and why it works on a clean machine.
Data written to your profile only, under %LocalAppData%\StegoPixel — settings, database, templates, logs.
No network requirement. The app is fully functional offline, permanently.
Uninstalls cleanly and appears in Settings → Apps like any other trusted application. Your data folder is left in place unless you remove it.
Project status
StegoPixel is in pre-release. It has not been published, and this site will not pretend otherwise.
Complete
What works today
All seven modules, sharing one loaded image
Pixel and metadata steganography with encryption and error correction
Metadata inspection, editing, templates and export
Privacy scanning and profile-based cleanup
Twelve forensic instruments, eight enhancement modes, six recipes
Heuristic AI, watermark and provenance detection
45 workflow node types and 54 presets, with execution history
HTML, PDF, JSON, CSV and hash-manifest reporting
Dark and light themes; twelve interface languages
MSIX and MSI packaging for Windows x64 and Arm64
In progress
What is still being finished
Store submission, listing and certification
Pricing and licensing terms, still undecided
Reconciling the internal version string (1.0.0) with the in-app display (1.0.0-alpha)
Populating the image index and collection tables, so library-wide search becomes real rather than schema-only
Broadening the format pipeline beyond the six raster formats currently decodable
Independent validation of the signing round-trip and the detection heuristics against known test material
Why we list unfinished things on a marketing page
Because the alternative is discovering them after you have trusted the tool with something that mattered. A forensics product whose own claims cannot be audited is not worth using, and the fastest way to demonstrate good faith is to say what is not done yet.
Editorial principles
How claims get made on this site. It seemed worth writing down.
No capability is described that the code does not implement. Where a feature is partially wired, the page says so — see the status section above.
Heuristics are labelled as heuristics. AI verdicts, watermark carrier scans and provenance parsing are described as the statistical methods they are, not as machine learning or as verification.
No accuracy figures are quoted. There is no labelled benchmark behind them, so quoting a percentage would be inventing one.
Limits are stated next to capabilities, not buried — the format matrix, the capacity caveat and the re-encoding warning are all on the features page rather than in a footnote.
Nothing is claimed about licensing that has not been decided. The FAQ says so plainly rather than implying “open source” because it sounds reassuring.
Contact
Support and feedback channels will be published alongside the Microsoft Store listing, since that is where the product will actually live and where the support link belongs. Until then:
Questions about the product
The FAQ answers most of what people ask first, and the privacy page covers the offline design in detail.