I initially made the ☀️ Sol Mate GPT, but it didn't take too long until I wondered what it would look like on an e-Paper display.
The script to update the screen takes a location (maybe your own, maybe your best friend's, or even a fictional place!) and get the weather for that location, then generate an illustration of the location including the current weather and lighting conditions. So if it's raining, there will be people with umbrellas in the picture, and so on. If you do generate for a fictional location, the weather service will be given the latitude/longitude of the most similar place on Earth (according to GPT). The illustrations are rendered using Dall-E 3. This is why an OpenAI API token is necessary to run it.
This repository contains all the code that was needed to generate and display a weather report for any specified location on a Raspberry Pi with an attached Waveshare e-Paper display.
- Raspberry Pi 5
- Waveshare e-Paper 7.3" display (code needs to be updated for other sizes)
- 3 more nice examples
- 3D-printing the case
- 4 examples, some with more color
- The setup, without the box
- Video of the thinness of the display
This should all run on your Raspberry Pi.
I recommend setting up a virtual environment for Python, such as uv, first. Here are the instructions for if you're using uv
(to be run inside the clone of this repo):
uv venv
source .venv/bin/activate
uv pip install -r requirements.txt
Without uv
:
python -m venv .
source .venv/bin/activate
pip install -r requirements.txt
You will need to specify an OPENAI_API_KEY
environment variable. For your convenience, the code will load environment variables from a .env
file in the current working directory.
Now you can use the control.py
script to generate an image and show it on the screen:
python control.py show Barcelona
Waveshare recommends you don't leave the same image on the display for too long.
Use the clear
command to clear it:
python control.py clear
I set up a cron job (crontab -e
) to update the image two times per day, but keep in mind this can end up costing a non-trivial amount:
0 8 * * * cd ~/src/sol-mate-eink && .venv/bin/python control.py show Barcelona
0 18 * * * cd ~/src/sol-mate-eink && .venv/bin/python control.py show Barcelona
0 2 * * * cd ~/src/sol-mate-eink && .venv/bin/python control.py clear
(You'll need to tweak the paths for your setup, of course.)
Feel free to use the private API I hosted to get the weather if you're not going to hammer it. I also use this backend for my GPT. However, if you have high volume ideas in mind, please self-host it! The source code is here:
https://github.com/blixt/sol-mate
I'd love to help if I can – reach out on Twitter or create an issue in this repo!