Fix reliability of the optimization / quality

[alexjc]

Mashing up grass and dirt sometimes produces low-quality or even buggy results.

mashup_dirt5_grass6 (left: bug on first-run, right: expected on second-run)

mashup_dirt8_grass7 (left: quality=2 lower quality than others, right: quality=4 as expected)

EDIT: It may be that the second row (low quality matches) is a result of recovering from the first row (gradients exploded).

[sharoseali]

@alexjc i have more then 2 images. I want to mashup and create texture. But the code is not writhing the resultant file. I even try with 2 image files but it didn't work (May be I m wrong). Just guide me how to use mashup command in order to merge input images. Thanks. The code below is given on git-hub but its not working.

from texturize import api, commands, io
from PIL import Image
# newsize = (512, 512)
# # Opens a image in RGB mode

# The input could be any PIL Image in RGB mode.

# io.load_image_from_file("/home/casper/Desktop/point_clouds/pre_tex.png")
sources = [
    io.load_image_from_file("/home/casper/Desktop/point_clouds/texturize/tex1.png"),
    io.load_image_from_file("/home/casper/Desktop/point_clouds/texturize/tex2.png")
    # io.load_image_from_file("/home/casper/Desktop/point_clouds/texturize/tex3.png"),
    # io.load_image_from_file("/home/casper/Desktop/point_clouds/texturize/tex4.png")
]

# Only process one octave to retain photo-realistic output.
mashup = commands.Mashup(sources)

for result in api.process_octaves(mashup, octaves=5):
    pass

# The output can be saved in any PIL-supported format.
result.save("./output.png")