As a newcomer to Python and its libraries, I am exploring the conversion of HDR images to RGBM as outlined in WebGL Insights Chapter 16.
import argparse
import numpy
import imageio
import math
# Handling arguments
parser = argparse.ArgumentParser(description='Convert a HDR image to a 32bit RGBM image.')
parser.add_argument('file', metavar='FILE', type=str, help='Image file to convert')
args = parser.parse_args()
# Loading the image
image = imageio.imread(args.file)
height = image.shape[0]
width = image.shape[1]
output = numpy.zeros((height, width, 4))
# Image conversion process
for i in numpy.ndindex(image.shape[:2]):
rgb = image[i]
rgba = numpy.zeros(4)
rgba[0:3] = (1.0 / 7.0) * numpy.sqrt(rgb)
rgba[3] = max(max(rgba[0], rgba[1]), rgba[2])
rgba[3] = numpy.clip(rgba[3], 1.0 / 255.0, 1.0)
rgba[3] = math.ceil(rgba[3] * 255.0) / 255.0
output[i] = rgba
# Saving the resulting image as PNG
imageio.imsave(args.file.split('.')[0] + '_rgbm.png', output)
The code accomplishes the task accurately but operates at a slow pace due to individual pixel iteration within Python, leading to longer processing times for larger images (approximately 4 minutes and 30 seconds for a 3200x1600 image).
I am seeking advice on optimizing this process. While exploring vectorization and broadcasting in numpy, I have not yet found an effective way to implement them in my scenario.
Edit:
With guidance from Mateen Ulhaq, I managed to enhance the efficiency:
# Image conversion optimization
rgb = (1.0 / 7.0) * numpy.sqrt(image)
alpha = numpy.amax(rgb, axis=2)
alpha = numpy.clip(alpha, 1.0 / 255.0, 1.0)
alpha = numpy.ceil(alpha * 255.0) / 255.0
alpha = numpy.reshape(alpha, (height, width, 1))
output = numpy.concatenate((rgb, alpha), axis=2)
This revised approach completes the process within seconds.