Base Layer · File-Level
File-level mutations that modify the binary without affecting pixels. Zero visual quality loss.
Pixel Layer · pHash Evasion
Spatial-domain mutations that evade perceptual hashing (pHash) algorithms used by ad platforms.
Trims image edges and stretches back to original size. Shifts every pixel position to break pHash fingerprint.
Injects subtle random pixel-level color variations. Nearly invisible but disrupts AI pixel-sample comparison.
Flips image horizontally. Highly effective for background/scenery images without text or logos.
Color & Lighting · CNN Evasion
Color-space mutations that evade CNN (convolutional neural network) based duplicate detection.
Shifts R,G,B channels uniformly. Changes color histogram without visible difference within ±2% range.
Adjusts color intensity relative to grayscale. Subtle shifts alter CNN color-space features.
Changes the difference between light and dark areas. Alters edge-intensity features used by CNN models.
Non-linear brightness curve. Adjusts dark/bright region transitions to disrupt edge-modeling algorithms.
Overlays a near-invisible color veil. Forces full canvas color-space rewrite to disrupt CNN analysis.
Click or drag & drop ad images here
Supports PNG, JPG, WEBP · Multi-file batch · ZIP input
🔒 Local offline processing. 100% privacy. No files ever uploaded.