Check Images & Videos for Unusual Technical Signals
This tool scans for hidden technical patterns — things the eye cannot see — in any image or video file. Everything runs inside your browser. No upload. No login. Your file never leaves your device.
This tool highlights unusual technical patterns. It does not determine truth.
How to Use This Tool
Drop or select any image or video. It opens locally — nothing is sent to a server.
See which technical layers show unusual patterns. Read the plain-English summary at the top.
Use the OSINT next steps section if something looks unusual. A single tool result is never a verdict.
Drop an image or video here
or click to browse — Images up to 50 MB · Videos up to 200 MB
Who This Tool Is For
Scascan's Media Signal Analyser is designed as a professional forensic workflow aid, not a consumer novelty.
Verify viral images and videos before publication. Surface technical anomalies to flag for further investigation.
Analyse media provenance as part of open-source intelligence workflows alongside geolocation and reverse search.
Study visual disinformation patterns. Understand how synthetic and manipulated media presents technically.
Add technical signal analysis as one layer of a content review process for suspected manipulated media.
Audit scraped or reused media assets for signs of manipulation or misrepresentation.
Better understand what to look for in viral media before sharing — not to get a verdict, but to ask better questions.
How This Analysis Works
Seven independent signal layers run entirely inside your browser tab — no file leaves your device. Each layer examines a different forensic dimension. Weights are dynamically adjusted per file type: JPEG images emphasise Error Level Analysis, PNG files emphasise fractal continuity, videos add temporal coherence, and screenshot contexts reduce unreliable metadata layers.
Reads the invisible data embedded in the file — which software saved it, GPS location if present, timestamps. Files shared on social media usually have this stripped, so the weight is reduced automatically for those cases.
Looks at small regions of the image for areas that are unnaturally smooth or uniform. AI-generated images often look perfectly smooth in ways that real-world photos — with natural noise and imperfections — do not.
Looks for invisible repeating patterns in the image that can be left by older AI generation pipelines. Less reliable on modern tools, but still useful as a supporting signal.
Uses machine learning to examine the statistical distribution of pixel values across the image. This is the broadest and most reliable signal layer — it works across all file types and survives recompression better than others.
Examines the file size and structure for signs of how the image was saved — looking at common AI output dimensions, and how the file has been re-saved.
Re-saves the image at a lower quality and measures the difference. Real photos show uneven differences between textures and smooth areas. Synthetic images often show a suspiciously even difference map across the entire image.
Checks whether fine detail and broad detail in the image are statistically consistent — as they are in natural photography. AI tools often produce images where fine detail looks rich but broader texture is oddly uniform.
Why This Tool Exists in 2026
In 2026, AI image and video generators have become powerful enough that no tool — ours included — can reliably tell you whether a file was made by AI or a human. The generators have become too good, and they actively minimise the very patterns that detection tools look for.
Rather than make claims we cannot support, we built something honest: a tool that surfaces the technical signals a trained investigator would look at — metadata, compression patterns, texture consistency, and frame variation — and presents them clearly. The interpretation is always yours to make.
Our philosophy
We believe transparency is better than certainty. A tool that shows you the evidence is more useful than a tool that gives you a verdict it cannot back up. Verification always requires multiple methods — this tool is one input, not a conclusion.
Why videos are also analysed across frames
Real video has small natural differences between frames — camera shake, changing light, background motion. AI video generators tend to produce frames that are too similar to each other. The tool measures this consistency and flags when it falls outside the normal range for natural video. This is a signal, not a verdict.
Privacy Architecture
Frequently Asked Questions
Related Resources
Extend your media verification workflow with these tools and guides.