I can generate a list of pixel values from an existing image file using a procedure like the following:
from PIL import Image
image = Image.open("test.png")
pixels = list(image.getdata())
width, height = image.size
pixels = [pixels[i * width:(i + 1) * width] for i in xrange(height)]
How could I convert this list of pixel values back to an image file?
First, you need to have your pixel tuples in a single un-nested list:
pixels_out = []
for row in pixels:
for tup in row:
pixels_out.append(tup)
Next, make a new image object, using properties of the input image, and put the data into it:
image_out = Image.new(image.mode,image.size)
image_out.putdata(pixels_out)
Finally, save it:
image_out.save('test_out.png')
Your list comprehension generates a list of lists, the latter being generated by the slicing (i*width:(i+1)*width). Your comprehension can be much easier: pixels = [pixel for pixel in pixels]. Obviously this outputs the same list, pixels, but you can use the idea to perform an operation on the pixels, e.g. pixels = [operation(pixel) for pixel in pixels].
Really, you overthought it. You don't have to manage the image dimensions. Getting the pixels in a list, and then putting them into an equal-sized image with putdata keeps the in order because they are linearized the same way by PIL.
In short, this is what your original snippet should have been:
from PIL import Image
image = Image.open("test.png")
image_out = Image.new(image.mode,image.size)
pixels = list(image.getdata())
image_out.putdata(pixels)
image_out.save('test_out.png')
You can you cairo. Pretty easy.
#!/usr/bin/python
# Extracting pixels from an image ------
from PIL import Image
image = Image.open("test.png")
pixels = list(image.getdata())
width, height = image.size
pixels = [pixels[i * width:(i + 1) * width] for i in xrange(height)]
# Putting pixels back to an image ------
import cairo
Width=len(pixels[0])
Height=len(pixels)
surface = cairo.ImageSurface (cairo.FORMAT_ARGB32, Width, Height)
context = cairo.Context (surface)
y=0
for row in pixels:
x=0
for rgb in row:
r=rgb[0] /255.0
g=rgb[1] /255.0
b=rgb[2] /255.0
context.set_source_rgb(r, g, b)
context.rectangle(x, y, 1, 1)
context.fill()
x+=1
y+=1
surface.write_to_png ("out.png") # Output to PNG
