The weather station runs on a Raspberry Pi Zero with a Pimoroni Inky Impression, displaying weather forecasts and sensor data. It retrieves weather data from OpenWeather and communicates with other IoT devices by exchanging sensor data through a local MQTT broker. The display shows a four-day weather forecast alongside a 12-hour forecast represented on a clock dial.
The system uses an AI image generator from getimg.ai to create an art image based on the current weather summary. A button press saves the image to Google Drive, and a slideshow feature allows viewing of previously stored images.
Starting at 07:00, the display updates every 15 minutes with weather and sensor data until 09:00, when a weather art image is generated and shown. The image remains on display for 30 minutes unless a button press shortens the duration. The artwork is created in the style of a randomly selected artist, with optional keywords to influence the artistic style. After midnight, updates pause for the display to rest.
New features include a birthday countdown list retrieved from Pastebin and a radial chart displaying the UV index forecast.¹
APIs are used to fetch the weather forecast, generate the weather art image, and upload it. The weather station receives local sensor data from a MQTT broker, which also relays data to other IoT devices and databases.
This image was generated in the style of Katsushika Hokusai using the prompt:
Painting in the style of artist Katsushika Hokusai depicting Misty and light winds in January, featuring Mount Fuji
However, not all generated images are worth keeping. More examples of generated images can be found here.
getimg.ai generates the weather art image using a text-to-image model based on the chosen settings and supplied prompt.
{
"model": "absolute-reality-v1-6", # Replace with your desired model
"prompt": f"Painting in the style of artist {artist_name} depicting {summary} in {month}, {keywords}", # Your prompt for generating the image
"negative_prompt": "Disfigured, cartoon, blurry, frame",
"output_format": "jpeg", # Specify JPEG format
"width": 576, # Your desired width x 64
"height": 448, # Your desired height
"steps": 25,
"guidance": 6 # 0-20 Higher guidance forces the model to better follow the prompt
} More information on image generation is available in the getimg.ai Text to Image API Reference.
Sending sensor data to Adafruit IO is handled by another device running an MQTT broker.
This image indicates an error occurred while fetching OpenWeather data. It displays the Inky icon, -1° temperatures, and replaces the weather summary with part of the error message.
Examples of other errors: local file error, getimg.ai response code 404, slideshow google drive error
As part of another project logs are sent to cloud logging services such as cloudwatch.
The screen updates every 15 minutes while displaying the weather but stops updating between 00:00 and 07:00. However, this refresh rate may be considered fast for this type of screen. Pimoroni suggested the expected refresh rate was once per day or once every couple of days.
⬜ Experiment with additional colour options on the e-ink display
⬜ Automatically remove old images from the generated images folder
⬜ Integrate a Xiaomi Mi Temperature/Humidity sensor via Bluetooth
⬜ Publish the current birthday event to the MQTT broker for other IoT devices
✅ Refactor the code to be more object-oriented, modular, and maintainable
✅ Display the UV index similarly to the PoP - resolving any overlap issues
✅ Expand the set of available weather icons
✅ Add a birthday event countdown
Enjoy the current weather with Picasso's painting
