Open 1mike12 opened 2 months ago
Curious why the new_avg_r starts off as the correct avg red of the input image, but then gets updated as the sum of all colors?
new_avg_r
new_avg_r = np.sum(shifted)
IE why isn't the code this:
while(new_avg_r < MIN_AVG_RED): shifted = hue_shift_red(avg_mat, hue_shift) new_avg_r = np.mean(shifted[..., 0]) # <----- hue_shift += 1 if hue_shift > MAX_HUE_SHIFT: new_avg_r = MIN_AVG_RED
Sorry it's been a while since I edited this repo. Give me a couple of days to refresh my knowledge of the code and get back.
Curious why the
new_avg_r
starts off as the correct avg red of the input image, but then gets updated as the sum of all colors?new_avg_r = np.sum(shifted)
IE why isn't the code this: