public/kcc
1
0
Fork 0
mirror of https://github.com/ciromattia/kcc synced 2025-12-13 09:46:25 +00:00
Code Issues Releases Wiki Activity

Fine-tuned the settings and added the perpendicular direction for a better final rendering

Browse Source 
The finer settings allow for more information to be retained in the final image, while still effectively removing the rainbow effect.

Adding the perpendicular direction results in a better rendering of the final image (avoiding visual artifacts related to suppression at the main angle).
This commit is contained in:
Its-my-right
2025-07-13 14:36:38 +02:00
committed by Alex Xu
parent f6d10337d8
commit 2da5b11858
1 changed files with 7 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
kindlecomicconverter/rainbow_artifacts_eraser.py
View File

@@ -15,7 +15,7 @@ def fourier_transform_image(img):
return fft_result return fft_result
def attenuate_diagonal_frequencies(fft_spectrum, freq_threshold=0.3, target_angle=135, def attenuate_diagonal_frequencies(fft_spectrum, freq_threshold=0.3, target_angle=135,
angle_tolerance=15, attenuation_factor=0.1): angle_tolerance=10, attenuation_factor=0.25):
""" """
Attenuates specific frequencies in the Fourier domain (optimized version for rfft2). Attenuates specific frequencies in the Fourier domain (optimized version for rfft2).
@@ -61,16 +61,18 @@ def attenuate_diagonal_frequencies(fft_spectrum, freq_threshold=0.3, target_angl
# Convert to degrees and normalize in a single operation # Convert to degrees and normalize in a single operation
angles_deg = np.rad2deg(angles_rad) % 360 angles_deg = np.rad2deg(angles_rad) % 360
# Optimize angular condition # Calculation of complementary angle
# For rfft2, we only process angles in the positive half-plane of X
# So we only calculate the main angle, not its opposite
target_angle_2 = (target_angle + 180) % 360 target_angle_2 = (target_angle + 180) % 360
# Calulation of perpendicular angles (for better final rendering)
target_angle_3 = (target_angle + 90) % 360
target_angle_4 = (target_angle_3 + 180) % 360
# Create angular conditions in a vectorized way # Create angular conditions in a vectorized way
angle_condition = np.zeros_like(angles_deg, dtype=bool) angle_condition = np.zeros_like(angles_deg, dtype=bool)
# Process both angles simultaneously # Process both angles simultaneously
for angle in [target_angle, target_angle_2]: for angle in [target_angle, target_angle_2, target_angle_3, target_angle_4]:
min_angle = (angle - angle_tolerance) % 360 min_angle = (angle - angle_tolerance) % 360
max_angle = (angle + angle_tolerance) % 360 max_angle = (angle + angle_tolerance) % 360