Sample Report

The example file

Summary & provenance reading

Provenance reading: "Consistent with a camera original" — level: neutral-positive. The evidence chips: Rich context · Natural image range · Lossy (DCT) · 5568×3712 · GPS present · Baseline JPEG.

What the report says: EXIF identifies a Nikon D750 with the lens record matching the exposure data; the JPEG structure shows baseline encoding with a single compression pass; no editor strings, no platform traces, no appended data. Together these signals are consistent with a file that came straight from a camera. The report words it exactly that way — consistency, not certification.

Camera & EXIF card

Location card

GPS records present: 48.858370° N · 2.294482° E — approximate place resolved as Paris, Île-de-France, France. The report states coordinates as recorded facts and the place name as an approximation. On real reports, the page also discloses that coordinates may be sent to a third-party geocoding service (OpenStreetMap's Nominatim) to produce the place name — the image itself is never sent.

Identity, structure & signals

• SHA-256: 9f2c07a4e1b8d5f3a6c0e9d2b4f7a1c8e3d6b9f2a5c8e1d4b7f0a3c6e9d2b5f8 — the file's exact identity; any byte-level change alters it completely.
• dHash (perceptual): f0e1c3870f1e3c78 — visual similarity fingerprint, robust to resize/re-encode.
• Structure map: APP1(Exif) · APP1(XMP) · DQT · SOF0 (baseline) · DHT · SOS — a clean, camera-typical JPEG layout. No COM comments, no unknown segments, no trailing data.
• Entropy profile: average 79.4 / 100 — "Natural image range". Slices dip at the header (expected) and run flat across the compressed payload. No anomalous regions.
• JPEG quality estimate: ~96/100, single quantization table set — consistent with a camera save, not a web re-encode.
• Provenance markers: none found — no AI generator strings, no C2PA markers, no IPTC source declarations, no editor signatures.

Journey of the file

The Journey section reconstructs a timeline from the signals — presented as a consistency assessment, never an authenticity verdict:

1. Captured by a camera — EXIF identifies Nikon D750, taken 2026:08:14 17:42:19.
2. No naming or platform patterns — the filename follows a camera default (DSC_…), consistent with a direct transfer from a memory card.
3. No re-encode evidence — baseline JPEG, single-pass quantization, no editor or platform strings.
4. Metadata fully survived — EXIF, XMP and GPS all intact, consistent with a file that was never processed by a stripping platform.

Classification: "Consistent with a camera original." The report adds the standing caveat: metadata can be forged; a crafted file could reproduce every signal above. That is why the wording is consistency, never confirmation.

Limitations, exactly as the report shows them

Every report closes with the limitations block: metadata can be modified or removed; missing metadata does not necessarily mean manipulation; forensic indicators are not always conclusive; results should be interpreted in context. This sample is no exception — it cannot prove the file came from a camera. It can tell you that every signal the file carries is consistent with that story, and it shows you exactly which signals were weighed.

What a "dirty" case looks like

The sample above is the clean case. On a WhatsApp-forwarded screenshot, the same cards read very differently: camera EXIF absent, GPS absent, dimensions downgraded, progressive re-encode detected, filename matching a WhatsApp export pattern, metadata survival graded "Metadata-light", and Journey classified "Passed through a platform or app". Nothing is flagged as "fake" — the report explains that this pattern is exactly what normal platform processing produces. That calibration between "suspicious" and "normal stripping" is the product's core skill.