ELA Scanner

Error Level Analysis re-saves the image as JPEG at a known quality level, then computes the pixel-by-pixel difference between the original and re-saved version. Unedited regions that were compressed uniformly show similar error levels across the image. Manipulated areas — pasted objects, cloned regions, retouched faces — were compressed differently and produce noticeably higher or lower error values. The difference is amplified and displayed as a color heatmap where bright patches indicate potential edits. For double-compression detection that complements ELA, try the JPEG Ghost Scanner. For a metadata-based approach, see the Authenticity Checker.

The JPEG quality slider (85–99%) controls the re-save compression level. Lower values create more visible differences, making subtle edits easier to spot but also increasing noise from natural edges. The amplification scale (3–30×) magnifies the error signal — higher values reveal faint manipulation but may introduce false positives on textured or high-contrast areas. Start with the defaults (95% quality, 10× scale) and adjust based on what the image shows. For heavily compressed originals, try lowering quality to 90%. For high-quality originals, raise it to 98%.

Journalists and fact-checkers verify whether submitted photos have been manipulated before publication. Insurance investigators examine claim images for signs of doctoring — pasted damage, altered timestamps, or removed objects. Legal teams use ELA as part of forensic evidence review. OSINT researchers analyze images from social media to assess credibility. Content moderators flag potentially manipulated images on marketplaces and dating platforms. For a complete verification workflow, pair with the photo forensics hub.

The Authenticity Checker analyzes metadata — software tags, timestamp gaps, EXIF inconsistencies — to detect whether editing software touched the file. ELA analyzes pixels — compression patterns that reveal where visual content was altered. Metadata analysis is faster and catches software-level edits, while ELA catches pixel-level manipulation like clone stamping or object removal even when metadata has been stripped. For maximum forensic coverage, use both tools together. Add the AI Detector to check for fully synthetic images.