The Mechanics of JPEG Quantization
JPEG is a lossy compression format that divides images into 8x8 pixel blocks and applies the Discrete Cosine Transform (DCT). High-frequency details are rounded off based on a standardized quantization table.
When an image is saved repeatedly at a specific quality (e.g. 75%), pixel errors converge toward an equilibrium. If an investigator inserts a foreign object into that image, the spliced region has not undergone identical generational loss, producing a sharp divergence in compression error.
In authentic photos, areas with similar textures exhibit uniform error brightness. Spliced objects glow brighter or appear distinctly darker.
Interpreting White Noise vs Dark Residuals
In an ELA error map, high-contrast edges and freshly added elements compress at much higher error rates, causing them to glow bright white or cyan.
Conversely, heavily airbrushed areas or smooth digital paint often appear as flat black or deep violet patches because they lack the natural sensor noise present across camera capture frames.
Step-by-Step ELA Verification Playbook
1. Upload the suspect image to our in-browser ELA module.
2. Adjust the JPEG re-compression slider to 75% (the forensic baseline).
3. Increase the error amplification slider to 20x to reveal subtle block differences.
4. Drag the split-screen loupe over text, borders, and subject boundaries to compare original pixels with their error profiles.
Always cross-reference ELA findings with Laplacian edge analysis to eliminate false positives caused by natural high-contrast borders.