Skip to content

Pasting, Merging & Blending Images with Python (PIL)

You can combine, merge, blend or paste images for artistic or practical reasons and achieve ingenious results. You can also come up with creative backgrounds or different textures to add to your digital images.

In this tutorial, we will elaborate different methods for combining images, elaborate transparency settings for merging images and provide plenty of Python examples to communicate these ideas efficiently.

What is merging images?

Merging an image means combining parts of two or more images to achieve a composition. Pasting images can be useful for creating presentations, art, special effects and rich compositions. Image pasting is a powerful concept in Python’s PIL library.

Holy Python is reader-supported. When you buy through links on our site, we may earn an affiliate commission.

Image.paste method

Paste method is probably the most popular and convenient method to paste an image on another image.
It is applied to an existing image and its first argument is the image being pasted.
Its second argument is a box. This box can be a tuple of 2 elements or 4 elements. If you use 2 elements it will define where the upper left corner of the image being pasted starts. If you use 4 elements then you would be defining all 4 corners of the image being pasted relative to the base image.
img.paste(img2, (400,400))
img.paste(img2, (400,400,1200,1200))
We have a few examples below.

Example 1: Paste Method

from PIL import Image

img ="/home/zoom_call.jpg")
img2 ="/home/wind_turbines.png")

img.paste(img2, (400,400))
The code above should output something similar to Image #2 below.
We will use various techniques to achieve a cooler result than just pasting the image and introduce .resize() and .rotate() methods down below.
from PIL import Image

img ="/home/zoom_call.jpg")
img2 ="/home/wind_turbines.png").resize((2750,1920)).rotate(-2, expand=1)

img.paste(img2, (2450,450))
By only adjusting the image size with .resize() and changing the box coordinates of the image being pasted, we can achieve a result similar to Image #3, that’s close but still not perfect.
To overlay the wind turbine on the laptop screen almost perfectly, we can use the .rotate() method, -2 degrees or 363 degrees should be perfect.
Check out the evolution of our images and you can find the final code below.
This is the original image
This is minimal paste method
This is paste method with resizing but no rotation adjustment
This is the final image achieved by paste method with rotation and resizing
Rotate method has an argument called expand which is super useful. The default setting is expand=0, which means some parts of your image after rotation will be left outside the image canvas. You can amend it as expand=1 which will ensure the canvas of your image is expanded to fit all parts after rotation.
Expand by default fills the gaps around the rotated image with transparency but you can also fill them with a color using fillcolor parameter. For example, you can use fillcolor=”blue”.  In the Python example below we will stick to defaults and leave fillcolor as is. You can see the final Python code used to achieve the Image #4 below.
Here is the full code.
from PIL import Image, ImageDraw

img ="/home/zoom_call.jpg").convert("RGBA")
img2 ="/home/wind_turbines.png")
.resize((2750,1920)).rotate(-2, expand=1)

img.paste(img2, (2450,450), mask=img2)

So, that’s the final code that’s been used to create the final image. A few important points:

  • First image (base layer) is a jpg without transparency. We can convert it to have RGBA mode to be able to paste images and conserve the transparency when needed. So, the .convert(“RGBA”) method is often convenient. We will heavily use it in the examples below which involve transparency channels.
  • Second image. This is the image being pasted on top. To perfect the pasting operation so the image fits the laptop screen we need to:
    • resize the image (.resize() method, new dimensions passed as tuple)
    • rotate the image (.rotate() method, rotation degree passed)
There is also the mask argument. It’s extremely useful and often confused. You can read the comprehensive information about the mask argument below. Mask argument can be used crop images in cicular, triangular, hexagonal shapes with Python.

Final result achieved by a few adjustments is nearly perfect and it can all done from the  command line environment (cli). Another big advantage of computational photography or more specifically batch editing digital images with Python is that you can scale it to thousands or even millions of images.

Mask argument explained

Another important component of computational image pasting is the mask argument. Mask is necessary when the image being pasted requires some kind of transparency. (For example, image being pasted is an apple but since all images have a box shape the areas surrounding the apple might have no image data except transparency.)

  • Mask layer: Mask argument must be used when pasting images with transparency. Mask argument takes any image and communicates to python that only pixels with information should be pasted (fully transparent pixels of the mask layer will be neglected in paste operation)

One thing that’s commonly confused is the usage of mask layer. Mask layer only determines the coordinates of the pixels that are being passed, the actual image being pasted is still the image that’s passed to .paste() method as the first argument.

In summary, coordinates of pasting comes from the mask argument (or mask layer provided by it) and the actual visual that’s being pasted comes from a separate image.

So it’s convenient that in many examples, mask argument and the image being pasted will be the same image. It’s useful when pasting an image to assign the coordinates and the visual of the non-transparent part to be the same image if that’s the goal.

But if you are looking to paste, say an orange in an apple shape this can also be possible in which case mask image will be different than the image being pasted.

Two important properties of the image being passed to the mask argument are as below.

  • Mask image must always be the same size as the image being pasted. It can be a different image with different transparency data but the images sizes must be the same.
  • Mask image must be in RGBA mode for color images.
  • Mask image can also be in L (whitegray) mode, in that case black pixels will be left out (pixels equaling 0) and only pixels with a value other than 0 will be pasted.

Example 2: Mask argument

Here we have a Python example that might be even more useful to understand how mask argument works.

from PIL import Image, ImageDraw

img ="RGBA", (2000,2000))
img2 ="/home/usa/Downloads/orange_2.jpg").convert("RGBA").convert("RGBA")
x, y = img2.size
img3 ="/home/usa/Downloads/transparent-apple-21.png").resize((x,y))

img.paste(img2, (0,0), mask=img3)

So, here we are using three images.
  • A base layer being used for everything to be pasted on it.
  • An orange image that’s being pasted on the base.
  • An apple image that’s being used as the mask argument.

You can use the same technique to create infinite cool combination. A city photo in a car shape, puppy in a heart shape, an oil refinery in a wind turbine shape are just a few other ideas.

Image.composite method

Composite method works very similarly to the .paste() method. The main difference is the way they are applied. While .paste() method is applied on the image as an image method, composite works as a function and it needs both images to be passed as arguments inside it. Check out the mini examples below.

  • img1.paste(img2, (100,100))
  • Image.composite(img1, img2)

Composite is an ancient word and it signifies physical or digital things which are composed of multiple parts.

In science and technology, composite is a popular and commonly used term such as composite material, composite signal, composite numbers and composite images. For example, space crafts have heat shield made of composite material (such as fibrous refractory which can withstand unimaginable orbit entry conditions. Google Patent)

Composite comes from Latin roots of com (with, together) + posite (position).

Anyway, back to composite method for pasting images. .composite() method has a couple of drawbacks to the previously demonstrated .paste() method.

Disadvantages of composite method:

  • Both images need to be the same size for composite method
  • Both images need to be the same mode for composite method
  • Paste method has a box argument which allows assigning the coordinates of the top left corners of the image being pasted in relation to the base image.
Although we can paste a smaller size image on a larger image, this isn’t immediately possible with the composite method. Additionally, both images need to be in the same color mode such as RGB, RGBA, RGBa, L, LA, 1 etc. This situation is tolerated in paste mode so that if the image being pasted is in a different mode, it will be converted to the base image’s mode automatically.

Example 3: Composite Method

Composite method for pasting images

Python code below creates an image mask in ellipse mode. Check out how to draw geometric shapes on images with Python tutorial to learn more about drawing ellipses, circles and other shapes with Python Image Library (PIL).

from PIL import Image, ImageDraw

img ="/home/fox.jpg").convert("RGBA")
mask ="L", img.size, 0)
draw = ImageDraw.Draw(mask)
draw.ellipse((300, 1500, 1220, 2250), fill=185)

x,y = img.size
img2 ="/home/crow.jpg").convert("RGBA").resize((x,y))

img3 = Image.composite(img2, img, mask=mask)

Above you can see the code used to create a new image with the Image module’s composite method.
Mask image is created in whitegray mode (L) and it’s only channel is filled with 185 out of 255. This will create an approximate 72.5% opacity (or 27.5% transparency).

Blend method

Blend mode is another mode that can be used to merge images with Python. Blend mode only requires image1, image2 and a ratio constant named alpha which can be between 0 and 1 inclusive.
Image.blend(img1, img2, alpha)
Alpha value determines the ratio of second image being pasted. A value of 0 will include 0% if the second image while a value of 1 will blend 100% of the second image.
from PIL import Image

img ="/home/orange.jpg").convert("RGBA")
x, y = img.size
img2 ="RGBA", (x,y), color="red")

img3 = Image.blend(img,img2, 0.7)

Example 4: Blend Method

Below we have a Python example regarding the blend method which shows blending of a new image with plain red color and an orange image using the blend method and various alpha values.

Blend method with alpha=0.3 and blue image
Blend method used with alpha=0.7 and red image

So blend method can be convenient for overlaying images when no transparency is involved. There are pros and cons of this method as following.

Pros of blend:

  • Very simple to use
  • Unique constant alpha to decide blending ratio

Cons of blend:

  • No mask means no pasting in specific shape
  • Images must be the same size always
  • Images must be the same color mode.

If when using .blend() method, image sizes don’t match you will get an error similar to below.

 return im1._new(core.blend(,, alpha))
ValueError: images do not match

This error is easy to fix, you just need to resize or crop the images so that they have the same height and width.

Alpha Composite method

Alpha_composite method can be useful for merging two images with transparency channels. What alpha_composite does is it only takes two arguments image 1 and image 2.
Image.alpha_composite(img1, img2)

Important thing is since there is no alpha adjustment parameter or
mask argument, images will need to have appropriate transparency values

This can easily be achieved by using the .putalpha() method.
Check out the mini sections regarding RGBA images and .putalpha method below.

RGBA Image Mode

RGBA is a useful color mode that can be applied to images. Most colorful digital images you come across in daily life will either have RGB or RGBA color modes.
Both of these image modes consist of Red, Green and Blue channels but RGBA color mode also has an Alpha channel in addition which can be used to modify the transparency levels of pixels.
So, when we convert and RGB image to RGBA, we get an additional alpha value for each pixel normally for a fully opaque image alpha value will be 255 for each pixel meaning each pixel of the image is fully visible (no transparency).
We can then use .putalpha() to manipulate the transparency of an image.

.putalpha() method for filling the transparency channel

.putalpha is a very convenient method that can be applied to images with Python’s PIL library.

Important thing is since there is no alpha adjustment parameter or
mask argument, images will need to have appropriate transparency values

Example 5: Alpha Composite Method

Below you can see the usage of the alpha_composite method in combination with the .putalpha method which is also explained down below.

from PIL import Image, ImageDraw

img ="/home/usa/Downloads/leio-mclaren-FwdZYz0yc9g-unsplash.jpg").convert("RGBA")
x,y = img.size
img2 ="/home/usa/Downloads/arthur-chauvineau-Dn7P1U26ZkE-unsplash.jpg").convert("RGBA").resize((x,y))


img3 = Image.alpha_composite(img, img2)
Airplane image
Fireworks image
Alpha composite method for merging airplane and fireworks
Alpha composite image with different putalpha() values
Alpha_composite method works slightly different than the composite method.
  1. Alpha_composite only accepts 2 arguments: image1 and image2.
  2. Alpha_composite doesn’t have mask argument.
  3. Alpha_composite needs transparency channel (A of RGBA) to work
  4. Transparency channel can be filled using putalpha() method. putalpha() accepts any value between 0 and 255. The higher the value the more opaque the image will be (You can see different image examples above).
We have a well-researched article explaining RGB and RGBA image modes.


In this Python Tutorial we learned how to paste, blend, mix and overlay images using different techniques from the Image module of PIL library.

To achieve various image merging results we used Python methods such as .paste(), .composite(), .alpha_composite() and .blend().

We also supportive classes and methods such as RGBA images, .convert() method, .resize() method, .putalpha() method, .size attribute and we made use of the mask argument when available.


  • Example 1: How to paste wind turbine image on laptop screen (paste)
  • Example 2: How to paste an orange in apple shape (paste with mask)
  • Example 3: How to paste crow image on fox image (composite)
  • Example 4: How to blend orange with red background (blend)
  • Example 5: How to merge airplane with fireworks (alpha_composite)

You can consider revisiting our basic image editing tutorials such as, resizing images, creating new images, saving images and cropping images.