Trying to merge gif and image file but getting unexpected output... Plz help

[Pratham31]

What did you do?

Actually I have two files. One is a Transparent gif and the second one is a JPG image and my aim is to use image as a background and gif as a foreground. In short, I want to merge them.

What did you expect to happen?

I want an image as a background and gif as a foreground. In short, I want to merge them.

What actually happened?

But using the pillow library is giving me output like this.

What are your OS, Python, and Pillow versions?

• OS: Windows
• Python: Python 3.8.5
• Pillow:

from PIL import Image, ImageSequence
transparent_foreground = Image.open("C:\\Users\\PRATHAMESH\\Desktop\\Python try\\me.jpg")
animated_gif = Image.open("C:\\Users\\PRATHAMESH\\Desktop\\Python try\\celeb (2).gif")

x, y = transparent_foreground.size
frames = []
for frame in ImageSequence.Iterator(animated_gif):
    frame = frame.copy()
    frame.paste(transparent_foreground,(x,y),mask=transparent_foreground)
    frames.append(frame)
    
frames[0].save('output.gif', save_all=True, append_images=frames[1:])

[radarhere]

This is similar to #4917 - except that issue has disappeared.
Please provide copies of me.jpg and 'celeb (2).jpg' so that we have everything necessary to run your script like you do.

[Pratham31]

Yeah Sir, sure.
me.jpg -

celeb (2).gif -

I want output like this(made this for reference) -

Note - celeb (2) is a gif not jpg.

[radarhere]

A few comments.

• As per https://pillow.readthedocs.io/en/stable/reference/Image.html#PIL.Image.Image.paste, a 2-tuple for the box argument is the upper left corner of the paste. You want it to be 0, 0, to paste one image exactly on top of the other. If it is the width and height of one of the images, then it will be pasted off down and to the right.
• You've labelled your JPG image as 'transparent_foreground'. This is clearly not correct, as you want the JPG to be the background image.

Given that, and working around a transparency mask error, I get

from PIL import Image, ImageSequence
background = Image.open("me.jpg")
animated_gif = Image.open("celeb (2).gif")

frames = []
for frame in ImageSequence.Iterator(animated_gif):
	output = background.copy()
	transparent_foreground = frame.convert('RGBA')
	output.paste(transparent_foreground, (0, 0), mask=transparent_foreground)
	frames.append(output)

frames[0].save('output.gif', save_all=True, append_images=frames[1:])

But there is an issue with the background there.

Instead, see what you think of this - it loops through the pixels and copies them individually, ignoring the transparent or background pixels.

from PIL import Image, ImageSequence
background = Image.open("me.jpg")
animated_gif = Image.open("celeb (2).gif")

frames = []
for frame in ImageSequence.Iterator(animated_gif):
	output = background.copy()
	frame_px = frame.load()
	output_px = output.load()
	transparent_foreground = frame.convert('RGBA')
	transparent_foreground_px = transparent_foreground.load()
	for x in range(frame.width):
		for y in range(frame.height):
			if frame_px[x, y] in (frame.info["background"], frame.info["transparency"]):
				continue
			output_px[x, y] = transparent_foreground_px[x, y]
	frames.append(output)

frames[0].save('output.gif', save_all=True, append_images=frames[1:])

[Pratham31]

Thank you, sir. It works 👍🏻

[Pratham31]

But I have one question that It is taking time to save the GIF, right. Why is it so?

[Pratham31]

And after some time it is again showing that weird image.

[radarhere]

As is noted in the docs, manipulating individual pixels like I am doing in my second example is slow. The first code example I provided is much faster, but has a problem with the background.

With the 'weird image', I presume you're referring to the very last frame of the output GIF?

[radarhere]

Not the best way, but the easiest way to fix that would be to just remove the last frame.

from PIL import Image, ImageSequence
background = Image.open("me.jpg")
animated_gif = Image.open("celeb (2).gif")

frames = []
for frame in ImageSequence.Iterator(animated_gif):
	output = background.copy()
	frame_px = frame.load()
	output_px = output.load()
	transparent_foreground = frame.convert('RGBA')
	transparent_foreground_px = transparent_foreground.load()
	for x in range(frame.width):
		for y in range(frame.height):
			if frame_px[x, y] in (frame.info["background"], frame.info["transparency"]):
				continue
			output_px[x, y] = transparent_foreground_px[x, y]
	frames.append(output)

frames[0].save('output.gif', save_all=True, append_images=frames[1:-1])

[Pratham31]

Can we reduce the file size of GIF?

[nulano]

Can we reduce the file size of GIF?

Not with Pillow, see #617 (comment) for that.

[Pratham31]

And what if I want a gif at the left below corner?

[Pratham31]

Any modification in the above removed last frame program ?

[radarhere]

Here is a modified version where I first scale down your GIF to half the size, and then locate it in the lower left corner.

from PIL import Image, ImageSequence
background = Image.open("me.jpg")
animated_gif = Image.open("celeb (2).gif")

frames = []
for frame in ImageSequence.Iterator(animated_gif):
	output = background.copy()
	
	# Reduce the GIF to half the size
	frame = frame.resize((frame.width // 2, frame.height // 2))
	
	frame_px = frame.load()
	output_px = output.load()
	transparent_foreground = frame.convert('RGBA')
	transparent_foreground_px = transparent_foreground.load()
	for x in range(frame.width):
		for y in range(frame.height):
			if frame_px[x, y] in (frame.info["background"], frame.info["transparency"]):
				continue
			
			# Offset the y value when setting the pixel
			# since this is now the lower left, rather than the upper left
			output_px[x, y+output.width/2] = transparent_foreground_px[x, y]
	frames.append(output)

frames[0].save('output.gif', save_all=True, append_images=frames[1:-1])

There are some bugs in the way that Pillow handles GIFs. The code I have given you in this issue is not ideal, but works because I presume you are personally more interested in an immediate solution than in a permanent one.