One toolkit, five guides

Finding a region of an image that has been copied from elsewhere in the same image — or a near-identical second file — is a copy-move problem, not a filename problem. StegoPixel compares image content, not file names or sizes.

• Copy-move detection — a block-level similarity engine with four matching algorithms, union-find clustering of matched regions, and displacement-vector filtering to discard coincidental texture repeats.
• Near-duplicate search — perceptual hashes (average, difference and DCT-based) find resized, recompressed and lightly edited copies that a byte comparison would miss.
• Progressive folder scanning — a folder loads in batches with live counters rather than freezing the window, and a thumbnail strip with a worker-pooled cache keeps it responsive.
• Recent files and folders — reopen what you were last working on.
• Persisted workflow state — saved workflows and their full execution history live in a local SQLite database, with configurable retention, auto-vacuum and a size warning threshold.

Content hashes are SHA-256 and SHA-512; similarity uses average, difference and perceptual hashes.

Sometimes the only way to be sure is to look at the bytes.

• Raw binary inspection of any file, with a monospaced offset/hex/ASCII layout built for reading structure rather than just scrolling.
• Structure parsing — the file's segment and chunk layout, so you can see where metadata and payload regions actually live.
• Format detection — confirms what a file really is, independent of its extension.
• Byte interpretation — read integers, floats and timestamps out of the bytes under the cursor.

A finding you cannot hand to anybody else is not much of a finding. Everything StegoPixel discovers can leave the app in a form somebody else can read.

Reports are produced by the same engine the workflow system uses, so a report generated by hand and one generated by an automated pipeline contain the same information.

• Dark and light themes — designed as a matched pair, not one theme with the colours inverted.
• 12 interface languages — English, German, Spanish, French, Italian, Japanese, Polish, Portuguese, Russian, Simplified Chinese, Hindi and Arabic, switchable at runtime without a restart. Arabic switches the layout to right-to-left.
• Configurable hashing — choose which digests are computed for integrity reports.
• Sensible defaults — backup before destructive operations, warning modes for overwrites, and a thumbnail size preference, all adjustable.
• Notification centre — long operations report back without modal interruptions.
• Local logs — written to your user profile for troubleshooting, never transmitted.

Capabilities are format-dependent, and knowing which is which saves a lot of confusion. This matrix reflects what the current build actually does.

HEIC/HEIF, AVIF and camera RAW formats (CR2, CR3, NEF, ARW, RAF, DNG, ORF, RW2, PEF, SRW) appear in the format catalogue, but they cannot be opened — the image pipeline decodes only the six raster formats above. StegoPixel will not quietly pretend otherwise: an unsupported file is reported as unsupported rather than partially analysed.

Two more format facts worth knowing before you start:

• Pixel-embedded output is always written as PNG. PNG is lossless, so it is the only output container that preserves an LSB payload. Saving an encoded result over a .jpg path raises a format-mismatch warning rather than silently destroying the payload.
• Conversion targets exclude GIF. You can read GIF, but not write it — output formats are JPEG, PNG, TIFF, WebP and BMP.

Capacity, honestly

A metadata channel targets 44,984 bytes in JPEG, PNG, TIFF and WebP — and that figure is explicitly a typical maximum, not a guarantee. The app's own documentation notes that external EXIF readers may stop honouring metadata profiles beyond 64 KB, which is why the ceiling sits below it. It also depends on the image, the tags already present and the payload's own overhead.

Pixel capacity scales with resolution, the number of channels and the bit depth you select. Rather than estimate, the app shows the real available capacity for the specific file and settings before you encode, and refuses rather than dropping a payload that does not fit.