Image.fromarray just produces black image

Question

I’m trying to save a numpy matrix as a grayscale image using Image.fromarray. It seems to work on a random matrix, but not on a particular one (where there should appear a circle). Can anyone explain what I’m doing wrong?

from PIL import Image
import numpy as np
radius = 0.5
size = 10
x,y = np.meshgrid(np.linspace(-1,1,size),np.linspace(-1,1,size))
f = np.vectorize(lambda x,y: ( 1.0 if x*x + y*y < radius*radius else 0.0))
z = f(x,y)
print(z)
zz = np.random.random((size,size))
img = Image.fromarray(zz,mode='L') #replace z with zz and it will just produce a black image
img.save('my_pic.png')

Asked By: flawr

||

Source

Answer 1

Image.fromarray is poorly defined with floating-point input; it’s not well documented but the function assumes the input is laid-out as unsigned 8-bit integers.

To produce the output you’re trying to get, multiply by 255 and convert to uint8:

z = (z * 255).astype(np.uint8)

The reason it seems to work with the random array is that the bytes in this array, when interpreted as unsigned 8-bit integers, also look random. But the output is not the same random array as the input, which you can check by doing the above conversion on the random input:

np.random.seed(0)
zz = np.random.rand(size, size)
Image.fromarray(zz, mode='L').save('pic1.png')

Image.fromarray((zz * 255).astype('uint8'), mode='L').save('pic2.png')

Since the issue doesn’t seem to be reported anywhere, I reported it on github: https://github.com/python-pillow/Pillow/issues/2856

Answered By: jakevdp

Image.fromarray just produces black image

Question:

Answers: