Forgery Detection
Upload an image or single-page PDF to check if it was photoshopped or AI-edited
Upload File
Upload an image or 1-page PDF
PNG, JPG, JPEG, BMP, TIFF, WEBP, PDF
Detection Method
Position-independent ELA with LLM vision labeling. Best for detecting edited text fields in documents.
AI agent iteratively adjusts ELA parameters like a forensic analyst, exploring different quality levels until it finds evidence of editing.
Runs ELA, noise analysis, clone detection, and EXIF metadata inspection.
Ready to Analyze
Upload an image or PDF and select your detection method to check for signs of manipulation or AI editing.
Analyzing Image...
Running forensic analysis. This may take a moment.
Overall Verdict
Enhanced ELA Heatmap (blue→red→yellow = increasing difference):
Standard ELA (brighter = more likely edited):
Mean Difference
Suspicious Pixels
Noise Variation
Anomalous Blocks
Clone Pairs
Metadata Fields
Findings:
Editing Software Detected:
Agent Iteration Log
Click to expand/collapseELA Heatmap
Brighter areas in the heatmap indicate regions with different compression levels, which may suggest editing.
Mean Difference
Detected Regions Overlay
Bounding boxes highlight areas where the ELA heatmap shows rectangular bright regions. Red = high confidence, Orange = medium, Yellow = low.
Detected Regions:
Detailed LLM Analysis
Click to expand/collapseAPI Example