Search

The All-seeing Pi

About the project

In this resource, you will make a tweeting touchscreen photo booth using a Raspberry Pi.

Project info

Difficulty: Moderate

Platforms: Raspberry PiPython

Estimated time: 1 day

License: GNU General Public License, version 3 or later (GPL3+)

Items used in this project

Hardware components

Raspberry Pi 3 Model B Raspberry Pi 3 Model B x 1
Raspberry Pi Camera Module V2 Raspberry Pi Camera Module V2 x 1
Tactile Switch Buttons 10 Pack Tactile Switch Buttons 10 Pack x 1
Breadboard Mini Self Adhesive Red Breadboard Mini Self Adhesive Red x 1
Male-female jumper leads Male-female jumper leads x 4
Large Arcade Led Blue Button Large Arcade Led Blue Button (optional, to replace tactile push buttons) x 2

Software apps and online services

guizero guizero sudo pip3 install guizero
twython twython sudo pip3 install twython

Story

Software Installation

This resource requires a number of additional software libraries. You will need to be connected to the internet to install these extra libraries.

To install the software you need, run the following commands in a terminal window:

  1. sudo pip3 install guizero
  2. sudo pip3 install twython

This will install the necessary software to control the Camera Module, create a GUI, and tweet and manipulate images.

If you are using the Raspberry Pi touchscreen to make this resource, you will also need to enter the following commands:

  1. sudo apt-get update
  2. sudo apt-get upgrade
  3. sudo apt-get dist-upgrade
  4. sudo apt-get install -y raspberrypi-ui-mods
  5. sudo apt-get install -y raspberrypi-net-mods

Making the photo booth housing

You can create your All-Seeing Pi using any housing you like. Our first iteration used a humble cardboard box, but you might want to create a masterpiece of carpentry or laser-cutting wizardry to hold your All-Seeing Pi.

If you don’t have fancy buttons or a touchscreen, that’s OK too: you can still make the All-Seeing Pi! Here is a picture of the setup we used when creating this resource: the only extra hardware requirements are two tactile push buttons, four jumper leads, a breadboard, and a Camera Module. You can use your usual monitor, keyboard, and mouse.

Connecting the Raspberry Pi touchscreen

  • Set up your Raspberry Pi touchscreen. There is a good tutorial available to help you. Ensure that you power the touchscreen via the Raspberry Pi with jumper leads as shown in the tutorial, and that the power supply you use is a good quality one.

  • Situate your screen within the housing of your choice, ensuring you can still access the Raspberry Pi.

Connecting the Camera Module

  • Connect the Camera Module to the Raspberry Pi with the blue side of the connector facing the USB ports:

  • Situate the Camera Module in your chosen housing. If you are using glue to secure it in place, be careful not to get any glue onto the connectors or camera components or they may stop working.

Connecting the buttons

  • Using jumper leads, wire one button to GPIO 23 and any ground pin, and the other button to GPIO 25 and any ground pin.

  • Situate the buttons in your chosen housing. In the software, the button connected to GPIO 23 will select the next overlay, and the button connected to GPIO 25 will take the picture. (Don’t worry if you accidentally wire your buttons up the other way around: you can simply swap the pin numbers in the code!)

Finishing the setup

  • To write the software, you will also need to connect a keyboard and mouse to your Raspberry Pi, as well as a display if you are not using the touchscreen.

  • Power your Raspberry Pi on and, if you are using one, check that the touchscreen works.

Test the buttons

  • To begin, open the File Explorer, then right click on a blank space inside the File Explorer window.

  • Select Create New and then click Folder.

Type in the name of the folder where you will store the code and the photographs. We chose to call ours allseeingpi. Double click on the  allseeingpi  folder and make a note of the path to it (this is displayed in the bar at the top), which should be /home/pi/allseeingpi.

  • From the Programming menu, open up Python 3.

  • Create a new Python file by clicking on File > New File.

  • Click on File > Save and save your file into the allseeingpi folder you just created, with the filename allseeingpi.py.

  • We will need the gpiozero library. At the start of your Python file, add an import statement:

 from gpiozero import Button

  • Next we will set up the buttons. In a previous section, we wired our buttons to GPIO 23 and GPIO 25. Let’s go ahead and set both buttons up.

  1. next_overlay_btn = Button(23)
  2. take_pic_btn = Button(25)

  • Now we will use gpiozero to tell the buttons what to do when pressed. In the code below, next_overlay and take_picture are functions which will be called when the corresponding button is pressed:

  1. next_overlay_btn.when_pressed = next_overlay
  2. take_pic_btn.when_pressed = take_picture

  • We will write these two functions so that the buttons know what to do when they are pressed. Functions are usually written at the start of a program, immediately after the import statements. Add the following two functions immediately after the import statement, with some placeholder code to print a message when they are pressed, so we can test them.

  1. def next_overlay():
  2. print("Next overlay")
  3.  
  4. def take_picture():
  5. print("Take a picture")

  • Press F5 to run your program. Try pressing each button and check that a different message pops up for each in the Python shell.

If your buttons do not produce this result, check that they are wired up correctly, and that they are connected to GPIO pins 23 and 25 on the Raspberry Pi. The button pins should be in different rows of the breadboard, like this:

You may have buttons with two legs on each side. These should be placed across the gap on your breadboard with the jumper wires both attached into one side. Take care to ensure the jumper wires are in the same rows as the legs of the button.

Set up the camera

  • Now that we know the buttons work, let’s set up the code for the camera. First add an import statement to the existing ones at the top of the program: 

  1. from picamera import PiCamera

  • Locate the existing line take_pic_btn.when_pressed = take_picture and, below it, add the following code to set up the camera object:

  1. camera = PiCamera()
  2. camera.resolution = (800, 480)
  3. camera.hflip = True
  4. camera.start_preview(alpha=128)

This code creates a ‘PiCamera’ object with the resolution set to 800 × 480, which is the resolution of the Raspberry Pi touchscreen. We also tell the camera to flip the preview horizontally (hflip): if we don’t do this, the preview image will be mirrored, which makes it hard for people to align themselves with the overlays! We then start the preview with alpha set to 128 so that it is semi-transparent; this is in case we get an error and need to see what is happening underneath. When you are confident that your code works, you can remove the alpha=128 to make the preview fully opaque.

Take a picture when the button is pressed

  • Since we will probably take lots of pictures with the All-Seeing Pi, we will put the date and time at which the picture was taken within the filename to avoid a picture being overwritten each time a new one is taken. To do this, we will need the gmtime and strftime functions from the time library, so add this line to the other import statements: 

  1. from time import gmtime, strftime

  • Underneath the code to set up the camera, add the following line:

  1. output = strftime("/home/pi/allseeingpi/image-%d-%m %H:%M.png", gmtime())

This will create a variable called output which contains the location and filename of where the captured photo will be saved. The %d, %m (etc) characters are how we specify the time format:  %d means the day and  %m means the month, for example. If you would like the date format in your filename to be different, there is a full reference guide to strftime available. The current date and time is provided by calling the function gmtime().

  • Now let’s revisit the take_picture() function and add some new code so that it actually takes a picture instead of just printing a message. Locate the line def take_picture(). Delete the line print("Take a picture") and in its place, add the following lines, making sure they are indented: 

  1. def take_picture():
  2. camera.capture(output)
  3. camera.stop_preview()

This code captures a picture, saving it to the location we just defined in the variable output.  It then stops the camera preview.

  •  Press F5 to run your program, then press the button to take a picture.

  • Navigate to the folder /home/pi/allseeingpi and check that the picture you just took has saved correctly.

Working with overlays

  • The All-Seeing Pi is no ordinary photo booth! The second button we set up, next_overlay_btn,  is used to change between ‘overlays’: these are fun pictures such as hats, beards, and glasses which appear on the screen as if you are wearing them. Here is an example of a picture taken with an overlay:

You can make your own overlays, or use the ready-made ones we have provided for you to download. If you are creating your own overlays, make sure that they are saved at 800 × 480 resolution as PNG files, with the background set to transparent.

  • Create a subfolder called overlays within your allseeingpi folder, and place your overlay images inside it.

  • Navigate to the overlays folder of the GitHub repo for this project. Click on the filename of the overlay you would like to use, then right-click on the download link and save the image into the overlays folder you just created. Repeat this process until you have saved all of the overlays you would like to use.

  • Now right-click here and save this file as overlay_functions.py. Make sure you save this file in your allseeingpi directory (where the allseeingpi.py script is also saved).

  • In the overlay_functions.py file, find this comment:

  1. # EDIT THESE VALUES ------------------------

You will need to change this code to specify two things:

  1. - Set the `overlays_dir` to the directory where your overlays are stored. If you are following this tutorial exactly, you will **not** need to change this directory location.
  2. - Set the `overlays` to be a list of the filenames of the overlays (without extension), surrounded by quotes and separated by commas. For example, if you had overlay images called `rock.png`, `paper.png`, and `scissors.png`, your line of code would look like this:

  1. overlays = ['rock', 'paper', 'scissors']

  • Now go back to your allseeingpi.py program. Underneath the other import statements in your program, add another one to import this file:

  1. from overlay_functions import *

This will allow us to use all of the overlay functions defined in the overlay_functions.py file from within our allseeingpi.py file.

Change overlays with a button

  • The other button you wired up to your All-Seeing Pi (called next_overlay_btn) will be the one we use to switch between the various overlays. Locate the function def next_overlay() : and delete the indented line print ("Next overlay") . In its place, add the following code, making sure the lines are indented to show that they are part of the function:

  1. def next_overlay():
  2. global overlay
  3. overlay = next(all_overlays)
  4. preview_overlay(camera, overlay)

First, we have to declare that we want to use the global variable, overlay.  This means that when we change the overlay, that value is saved so that we can access it and use it from anywhere, and the change isn’t lost when we exit this function.

The second line gets the next overlay from the list of all_overlays  (defined within the overlay_functions.py file), and sets this as the current overlay.  Then, the function preview_overlay() is called to display the new overlay.

  • Save your program, and run it by pressing  F5. Check that when you press the button to change between overlays, the overlays change. Ensure you have at least one overlay image in your overlays folder to be able to change between them!

Here is the program so far if you want to check your progress.

  • You will notice that, when you take a picture, two things happen. Firstly, the overlay does not disappear and probably makes it quite difficult to see what you are doing: close the Python shell window to get rid of the overlay. Secondly, people can see a camera preview and can choose a silly hat from the overlays, but, when they take the photograph, the overlay disappears. We need to add code to remove the overlay from the screen once the picture is taken, and superimpose it onto the saved photograph.

Save an overlay on your picture

  • Locate the function def take_picture() :  and add two lines of code at the end of the function:

  1. def take_picture():
  2. camera.capture(output)
  3. camera.stop_preview()
  4. remove_overlays(camera) # Add this line
  5. output_overlay(output, overlay) # Add this line

Here we are using two more functions from the overlay_functions  file. The function remove_overlays does exactly what it says, and removes all of the overlays so they don’t hang around after we take a photograph. The output_overlay function takes the photograph and the overlay and glues them together so the resulting final output is a photograph with the chosen overlay superimposed.

  • Once again, save your file and run it using F5 to check that you can now change between overlays, and that, when you take a photograph, your chosen overlay is saved as part of the picture. 

Create a GUI

We have an almost-working All-Seeing Pi. However, when a picture is taken, the camera preview disappears and the user is left staring at the Python shell and the Raspbian desktop. You probably don’t want your selfie-takers to have to restart the Python program every time someone takes a picture. We will create a very simple GUI to display the picture that was taken and allow them to take another picture.

  • To create the GUI we will use a library called guizero,  which you should have already installed in the software installation step. Add another import line with the others at the start of your program to bring in the guizero functions we need:

  1. from guizero import App, PushButton, Text, Picture

  • At the bottom of your current program, create the beginning of your GUI. 

  1. app = App("The All-Seeing Pi", 800, 480)
  2. message = Text(app, "I spotted you!")
  3. app.display()

First, we create an  app, which is the basic container for the GUI.  The dimensions are 800 × 480 because that is the resolution of the touchscreen, and the title bar will contain the text “The All-Seeing Pi”. It is possible to make the GUI full-screen, but we will not do this for now  because it can be difficult for testing. We also create a message, "I spotted you!", and add it to the app before displaying everything.

  • Save and run your program again. Check that, when you press the button to take the photo, the camera preview exits and you see a mostly blank GUI with a message saying “I spotted you!”.

  • Now, between the message line and the app.display()  line, add another line of code to create a button.

  1. new_pic = PushButton(app, new_picture, text="New picture")

Examining the arguments passed to this PushButton object, we have three parts:

  • app: tells the button to add itself to the app

  • new_picture: this is the command. When the button is pushed, it will call the function new_picture() (which we haven’t written yet!)

  • text="New picture": this is the text which will appear on the button

  • Now write the new_picture  function so that the button knows what to do when it is pressed. Write this code after the take_picture()  function, but before the code where we set up the buttons. Ensure that your cursor is not indented, otherwise the code you write now will become part of the take_picture() function, which we do not want.

  1. def new_picture():
  2. camera.start_preview(alpha=128)
  3. preview_overlay(camera, overlay)

This function is very straightforward: it simply tells the camera to restart the preview, and to display the overlay (which will be the last overlay we used).

  • Save your program, and run it using F5 once again. Check that you can press your physical button to take a picture, and that the GUI displays once the camera preview disappears. Check that you can press the on-screen button to restart the camera preview and take another picture.

Stop the picture overwriting

Now that we have introduced the ability to run the program only once but take multiple pictures, we have a problem. The filename for the picture is generated by this existing line of code:

  1. output = strftime("/home/pi/allseeingpi/image-%d-%m %H:%M.png", gmtime())

However, we only execute this line of code once during the program. This means that, every time the button is pressed to take a picture, it is saved to the same location, with the same filename. To fix this, we need to regenerate the filename every time we take a picture.

  • Locate this line of code and copy it so you can paste it somewhere else shortly. Then, change the output to be equal to an empty string:

  1. output = ""

  • Now find your take_picture() function. At the start of the code within the function,  add the line global output  and then paste in the line you copied. The altered function should look like this:

  1. def take_picture():
  2. global output
  3. output = strftime("/home/pi/allseeingpi/image-%d-%m %H:%M.png", gmtime())
  4. camera.capture(output)
  5. camera.stop_preview()
  6. # .... code continues...

We are dealing with scoping here: this is an important concept for programmers to understand. Why did we bother to create the variable output in the main part of the program, and initialise it as a blank string, when we could have just created it within the take_picture()  function? The answer is that if we only created it within the take_picture()  function, once the function finished executing, the variable would no longer exist. By declaring that we are talking about the global version of the output variable, we are telling the program that we want to use the variable output which we created in the main part of the program. This means that once the function exits, the variable output with the location of the saved picture will still exist. We need to have a permanent record where the picture was saved because it is used in other places within the program.

Display the picture

You probably don’t want your photo booth participants to have to go digging through the Raspbian filesystem to see the picture they took either, so let’s display the picture they took on the GUI.

  • Locate the line of code where you intialise the output variable:

  1. output = ""

Immediately underneath it, add a new line of code to define the location where we will store the latest-photo, i.e. the photo most recently taken using the booth.

  1. latest_photo = '/home/pi/allseeingpi/latest.gif'

  • Now locate the line of code where you added the PushButton to your GUI. Immediately before that line, insert a line of code to display an image on the GUI:

  1. your_pic = Picture(app, latest_photo)

  • The file we are referring to, latest.gif, does not yet exist, so if you run your program now you will not see a photograph displayed on the GUI. We must add code inside the take_picture() function to generate this image so that it can be displayed. Locate the  take_picture() function and, underneath the other code in the function, add the following lines (remembering to ensure that the new lines of code are also indented):

  1. size = 400, 400
  2. gif_img = Image.open(output)
  3. gif_img.thumbnail(size, Image.ANTIALIAS)
  4. gif_img.save(latest_photo, 'gif')
  5.  
  6. your_pic.set(latest_photo)

This code opens the output image (the image containing the photo combined with the overlay), creates a smaller thumbnail of that image in gif format, and saves it at the location set up in latest_photo. It then sets the image on the GUI (your_pic) to be that latest photo image using the set() function which is part of the guizero library.

  • Save your code and test whether, when you take a photograph, it is displayed on the GUI. You may find that there is a short delay between the camera preview exiting and the image displaying on the GUI while it is saving. 

You may notice that the picture quality of the image displayed on screen is not optimal. This is because the picture has been converted to gif format to be displayed on the GUI. The full-quality png version of the photograph will still be saved in the allseeingpi folder.

Tweet picture

If you just want a fun photo booth to take and save pictures, you could stop there. Alternatively, you could go one step further and make your All-Seeing Pi tweet the photo that was taken.

  • You will need to set up a Twitter account and create an app for your All-Seeing Pi. Follow steps 1-6 on the Getting started with the Twitter API resource in a separate file, and check that you can successfully send a textual tweet from Python.

  • Save a copy of the auth.py file containing your Twitter API keys (which you created  during the ‘Getting started’ tutorial) inside your /home/pi/allseeingpi folder.

  • Go back to your allseeingpi.py file and, after the other import statements, import Twython:

  1. from twython import Twython

  • Immediately after importing Twython, add the following code to import your Twitter API credentials from your auth.py file:

  1. from auth import (
  2. consumer_key,
  3. consumer_secret,
  4. access_token,
  5. access_token_secret
  6. )

  • Create a new function after the new_picture()  function, called send_tweet():

  1. def send_tweet():

  • Inside the function, instantiate a Twitter object:

  1. def send_tweet():
  2. twitter = Twython(
  3. consumer_key,
  4. consumer_secret,
  5. access_token,
  6. access_token_secret
  7. )

  • Add some more code inside the send_tweet() function to tweet the output picture.  You can change the text in your message  code if you want your tweet to say something different:

  1. message = "The All-Seeing Pi saw you!"
  2. with open(output, 'rb') as photo:
  3. twitter.update_status_with_media(status=message, media=photo)

  • Now, find the code for the GUI where you create the PushButton for a new picture, and add another PushButton underneath it which will call the send_tweet()  function when it is pressed:

  1. tweet_pic = PushButton(app, send_tweet, text="Tweet picture")

  • Save and run your program. Test whether, when you take a picture and press the Tweet picture button on the GUI, the picture is tweeted from your Twitter account.

The finished code is here: you can check it against your code if you need to.

Once you are happy that your All-Seeing Pi works, you may wish to remove the alpha=128  command from the camera preview to make it fully opaque. You can also make the GUI full- screen: locate the line app = APP("The All-Seeing Pi", 800, 480)  and, immediately after it, add the line app.tk.attributes("-fullscreen", True).

Other ideas

  • Can you add a text box or perhaps a touchscreen keyboard to your GUI to allow someone to enter their Twitter handle?
  •   Can you use this Twitter handle to add an @username  mention to the tweet text?
  • Could you make a more imaginative housing for your All-Seeing Pi?

The All Seeing Pi (Overlay Functions)

The following section is for more advanced learners as it explains in detail what the code and functions inside overlay_functions.py do. It is possible to make the All Seeing Pi without understanding what these functions do - simply save a copy of the file overlay_functions.py into the folder with your code and they will be available.

Importing necessary libraries

These statements import functions from the PIL library to process and save the images and the itertools library so that we can cycle through the overlays. 

  1. from PIL import Image
  2. from itertools import cycle

Setting up the variables

This part sets up the directory where the overlays are saved, and the names of the various overlays. The overlay variable is initialised with the first value in the list.

  1. # EDIT THESE VALUES ------------------------
  2. overlays_dir = "/home/pi/allseeingpi/overlays"
  3. overlays = ['girl', 'cowboy', 'top', 'pink', 'glassesnose', 'moustache', 'sunglasses', 'elvis', 'emo', 'blackhat', 'emo2', 'baseball', 'flowers', 'santa', 'alps', 'mop', 'glasses']
  4. # ------------------------------------------
  5. overlay = overlays[0] # Starting value

Get the overlay as a PIL Image

This function is only used within other functions in this file. Given the name of an overlay as a string, it creates a PIL Image object of that overlay and returns it.

  1. def _get_overlay_image(overlay):
  2.  
  3. # Open the overlay as an Image object
  4. return Image.open(overlays_dir + "/" + overlay + ".png")

Pad the overlay

This function ensures that the overlay is padded correctly so it can be displayed on the preview.

  1. def _pad(resolution, width=32, height=16):
  2. # Pads the specified resolution
  3. # up to the nearest multiple of *width* and *height*; this is
  4. # needed because overlays require padding to the camera's
  5. # block size (32x16)
  6. return (
  7. ((resolution[0] + (width - 1)) // width) * width,
  8. ((resolution[1] + (height - 1)) // height) * height,
  9. )

Remove all overlays

This function iterates over all overlays attached to the camera object, and removes them.

  1. def remove_overlays(camera):
  2.  
  3. # Remove all overlays from the camera preview
  4. for o in camera.overlays:
  5. camera.remove_overlay(o)

Put the overlay on the camera preview

This function is passed a PiCamera object (camera) and a string (overlay).  It removes all overlays currently associated with the camera object, creates a PIL Image object of the chosen overlay called overlay_img, pads that image to display correctly and then adds it to the camera preview. The alpha of the preview is set to 128 so that the overlay is semi transparent. If the overlay was made fully opaque it would obscure the camera preview.

  1. def preview_overlay(camera=None, overlay=None):
  2.  
  3. # Remove all overlays
  4. remove_overlays(camera)
  5.  
  6. # Get an Image object of the chosen overlay
  7. overlay_img = _get_overlay_image(overlay)
  8.  
  9. # Pad it to the right resolution
  10. pad = Image.new('RGB', _pad(camera.resolution))
  11. pad.paste(overlay_img, (0, 0))
  12.  
  13. # Add the overlay
  14. camera.add_overlay(pad.tobytes(), alpha=128, layer=3)

Save picture with overlay

This function takes the location of the photograph (output) and the given overlay (overlay), both as strings. It then creates a PIL Image object of the specified overlay, also creates a blank PIL Image to save the finished output to, and then combines the photograph with the overlay, re-saving the finished photograph at the output location. 

  1. def output_overlay(output=None, overlay=None):
  2.  
  3. # Take an overlay Image
  4. overlay_img = _get_overlay_image(overlay)
  5.  
  6. # ...and a captured photo
  7. output_img = Image.open(output).convert('RGBA')
  8.  
  9. # Combine the two and save the image as output
  10. new_output = Image.alpha_composite(output_img, overlay_img