Base Percentile

The percentile threshold for detecting noise pattern peaks in the Fourier residuals.

Base Percentile sets the sensitivity for detecting structural noise patterns in the frequency domain analysis.

How it works

• After Stage 1, residuals are analyzed in Fourier space

• Peaks above this percentile threshold are identified as structured noise

• Lower percentile = more sensitive (detects weaker patterns)

• Higher percentile = less sensitive (only strong patterns)

Values (30-70)

• 30-40%: Very sensitive, may detect subtle patterns but risk false positives

• 50% (Default): Balanced detection

• 60-70%: Conservative, only detects prominent patterns

Recommendations

• Start with default (50%)

• Decrease if the mask seems to miss visible noise patterns

• Increase if the mask includes too many false detections

Note: When Adaptive Thresholding is enabled, this serves as the starting point that gets adjusted based on pattern strength.