Remove transparency/alpha from any image using PIL

Question

How do I replace the alpha channel of any image (png, jpg, rgb, rbga) with specified background color? It must also work with images that do not have an alpha channel.

Answer 1

This can be done by checking if the image is transparent

def remove_transparency(im, bg_colour=(255, 255, 255)):

    # Only process if image has transparency (http://stackoverflow.com/a/1963146)
    if im.mode in ('RGBA', 'LA') or (im.mode == 'P' and 'transparency' in im.info):

        # Need to convert to RGBA if LA format due to a bug in PIL (http://stackoverflow.com/a/1963146)
        alpha = im.convert('RGBA').split()[-1]

        # Create a new background image of our matt color.
        # Must be RGBA because paste requires both images have the same format
        # (http://stackoverflow.com/a/8720632  and  http://stackoverflow.com/a/9459208)
        bg = Image.new("RGBA", im.size, bg_colour + (255,))
        bg.paste(im, mask=alpha)
        return bg

    else:
        return im

Answer 2

I'd suggest using Image.alpha_composite.
This code can avoid a tuple index out of range error if png has no alpha channel.

from PIL import Image

png = Image.open(img_path).convert('RGBA')
background = Image.new('RGBA', png.size, (255,255,255))

alpha_composite = Image.alpha_composite(background, png)
alpha_composite.save('foo.jpg', 'JPEG', quality=80)

I also recommend you inspect both results with a image.show().

Credits for this answer goes to shuuji3 and others who helped build a vast answers repertoire in this other question.

Answer 3

I updated Vincius M's code with the user, tutuDajuju, suggestion to make [height, width, 3] matrix (not [height, width, 4]):

from PIL import Image
import numpy as np

png = Image.open(img_path).convert('RGBA')
background = Image.new('RGBA', png.size, (255,255,255))

alpha_composite = Image.alpha_composite(background, png)
# if you check the matrix dimension, channel, it would be still 4.  
h, w, channel = np.asarray(alpha_composite).shape

alpha_composite_3 = alpha_composite.convert('RGB')
# if you check the matrix dimension, channel, it would be 3.  
h, w, channel = np.asarray(alpha_composite_3).shape