This demo shows how ImgFact Compare Mode presents a high-risk image discrepancy: two versions of the same insurance claim photo, one consistent with original photography and one showing strong synthetic or manipulated-image indicators.
Comparison preview
Discrepancy found
claim_photo_original.jpg
Authentic
claim_photo_revised.jpg
AI Generated
Comparison Verdict
Significant Discrepancy Detected
Image A shows strong indicators of authentic photography. Image B exhibits hallmark signs of AI generation including atypical noise distribution, near-zero EXIF data, and anomalous power-law frequency behavior.
→ Do not rely on Image B without obtaining the original source file and conducting further forensic review.
Derivative risk
The revised image appears visually related to the original but carries stronger manipulation indicators.
Metadata gap
Image B has missing camera metadata and lower native resolution, suggesting re-export or regeneration.
Escalation path
Request the original source file and preserve both versions before relying on either in a matter.
Image Aclaim_photo_original.jpg
Authentic
88
% confidence
Original photo shows consistent metadata, natural noise behavior, and low manipulation indicators.
Forensic Signals
ELA Score
8%
Noise Score
22%
Power-Law Freq
7%
Complexity Var
9%
Metadata
EXIF Camera Data
present
Dimensions
3840×2160
Image Bclaim_photo_revised.jpg
AI Generated
66
% confidence
Revised image shows missing EXIF data, abnormal texture behavior, and elevated AI-style signal deltas.
Forensic Signals
ELA Score
71%
Noise Score
84%
Power-Law Freq
91%
Complexity Var
88%
Metadata
EXIF Camera Data
missing
Dimensions
1024×1024
Signal delta review
Material differences between submitted image versions
This is a simulated comparison using synthetic data designed to illustrate how Compare Mode presents results. Actual scan verdicts and signal values will vary based on the specific images analyzed.
Limitation notice
ImgFact is a first-pass forensic screening tool. For formal legal proceedings, consult a qualified forensic examiner before relying on any digital forensic output.