Pi Inky Weather Display

A beautiful weather dashboard for Raspberry Pi with 7.3" e-paper display

Features • Quick Start • Hardware • Configuration • Development

The generation of the image is abstracted away from the hardware, so it can be used on any hardware stack.

Hardware

• Raspberry Pi (Zero model requires soldering the GPIO header)
• Inky Impression 7.3" E-Paper display
• 3D printed case (optional)

Setup on Raspberry Pi

1. Install the Inky library:

   curl https://get.pimoroni.com/inky | bash

   For detailed installation steps, refer to the official documentation.

2. Download the latest release:

   Download the latest release for your architecture from the releases page and extract it:

   Architecture Guide

   Raspberry Pi Model	Architecture	Download
   Pi 1, Zero, Zero W	arm-unknown-linux-gnueabihf	ARMv6
   Pi 2, 3, 4, Zero 2 W (32-bit OS)	armv7-unknown-linux-gnueabihf	ARMv7
   Pi 3, 4, 5 (64-bit OS)	aarch64-unknown-linux-gnu	ARMv8
   x86 Linux	x86_64-unknown-linux-gnu	x64

    # Download and extract
    wget https://github.com/mt-empty/pi-inky-weather-epd/releases/latest/download/pi-inky-weather-epd-<architecture>.zip
    unzip pi-inky-weather-epd-<architecture>.zip
    chmod +x pi-inky-weather-epd

3. Configure your weather data provider and location:

   Get your latitude and longitude from https://www.latlong.net/ and create a configuration file:

   mkdir -p ~/.config
   cat > ~/.config/pi-inky-weather-epd.toml << EOF
   [api]
   latitude = YOUR_LATITUDE   # e.g., -33.8727 # Sydney
   longitude = YOUR_LONGITUDE # e.g., 151.2057
   provider = "open_meteo"    # "open_meteo" (worldwide) or "bom" (Australia only)
   EOF

   See ./config/default.toml for more configuration examples.

4. Set up an hourly cron job to update the display:

   (crontab -l 2>/dev/null; echo "0 * * * * cd /path/to/extracted/files && ./pi-inky-weather-epd && sudo <PYTHON_PATH> <IMAGE_SCRIPT_PATH> --file dashboard.png --saturation <SATURATION>") | crontab -

   Replace:

   • /path/to/extracted/files with your installation directory
   • <PYTHON_PATH> with path to Python (e.g., /usr/bin/python3)
   • <IMAGE_SCRIPT_PATH> with path to Inky's image.py (e.g., /home/pi/Pimoroni/inky/examples/7color/image.py)
   • <SATURATION> with the desired saturation level depending on your display (e.g., 1.0). If using the Inky Impression 7 colours, it is not recommended to change this for current icons

   Example of complete cron command:

   0 * * * * cd /home/pi/pi-inky-weather-epd && ./pi-inky-weather-epd && sudo /home/dietpi/env/bin/python3 /home/dietpi/Pimoroni/inky/examples/7color/image.py --file dashboard.png --saturation 1.0

Configuration

You can override the default configs located at ./config/ by creating a file at:

~/.config/pi-inky-weather-epd.toml

Configuration Examples

Here are example configurations.

Default Configuration

Imperial Units

[render_options]
temp_unit = "F"
wind_speed_unit = "mph"

Date Format

You can customise the date format using chrono strftime specifiers. The default is "%A, %d %B" (e.g., "Saturday, 06 December").

[render_options]
# Example formats:
# date_format = "%B %-d, %Y"     # December 6, 2025 (US style)
# date_format = "%d/%m/%Y"       # 06/12/2025 (Australia/UK)
# date_format = "%m/%d/%Y"       # 12/06/2025 (USA)
# date_format = "%Y-%m-%d"       # 2025-12-06 (ISO 8601)
# date_format = "%a, %-d %b"     # Sat, 6 Dec
# date_format = "%d.%m.%Y"       # 06.12.2025 (Germany)

date_format = "%A, %d %B"

See chrono strftime documentation for all available format specifiers.

Use Clear night Icon instead of Moon Phase icon when Time=night and Weather=clear

When the sky is clear, the moon phase icon is used instead of the clear night icon, you can disable with:

[render_options]
use_moon_phase_instead_of_clear_night = false

Set X-Axis Placement to be always at y=0

The x-axis is no longer at the bottom of the graph when the temperature is below zero, it is now always positioned at x = 0.

[render_options]
x_axis_always_at_min = false

Dark Theme

[colours]
background_colour   = "black"
text_colour         = "white"

x_axis_colour       = "white"
y_left_axis_colour  = "red"
y_right_axis_colour = "blue"

actual_temp_colour  = "red"
feels_like_colour   = "green"
rain_colour         = "blue"

Auto-Update Interval

Enable auto-update when a new release is available. This is enabled by default.

[release]
# Set to 0 to disable auto-updating
update_interval_days = 7
# Opt in to pre-release versions (requires update_interval_days > 0)
allow_pre_release_version = false

Degraded Operation

The dashboard can still work using cached data for a while if the API is unreachable. A diagnostic icon and message appears on the display when issues occur.

Diagnostic Type	Priority	Icon
API Error	High
No Internet	Medium
Incomplete Data	Low
Update Failed	Low

When multiple diagnostics occur, the highest priority diagnostic is displayed, lower priority ones are cascaded.

Inky Impression 7.3

Supported Colours at 1.0 Saturation (Without Dithering)

[0, 0, 0],        # Black
[255, 255, 255],  # White
[0, 255, 0],      # Green
[0, 0, 255],      # Blue
[255, 0, 0],      # Red
[255, 255, 0],    # Yellow
[255, 140, 0],    # Orange

Documentation and Resources

• EPD used: Inky Impression 7.3
• Actual Panel: Waveshare 7.3" E-Paper HAT
• Panel documentation: Waveshare Wiki
• Open-Meteo API: Open-Meteo Weather Forecast API (default provider)
• BOM API: Bureau of Meteorology API Documentation (Australia only)
• Icons: Custom SVG icons complete overhauled of Meteocons

Design Decisions

Image generation is hardware-agnostic. Weather data is rendered to an SVG via TinyTemplate, then rasterized to PNG with resvg. The same pipeline runs on a Pi or a dev laptop, so cargo run on your machine shows exactly what ships to the panel — no on-device-only rendering path to debug blind.

Degrades instead of failing. If the API is unreachable, the fetcher falls back to the last cached response and keeps rendering, surfacing a priority-ordered diagnostic icon instead of a blank display or a crash log nobody on a headless Pi will ever see.

Config, not code, for personalization. Colours, units, date format, and axis behaviour are all TOML overrides layered on config/default.toml — no fork needed to reskin the dashboard.

Backward-compatible by default. The binary can self-update in place on unattended Pi devices, so a newer binary routinely reads state (cached JSON, TOML config) written by an older one. New fields must tolerate absence rather than hard-fail a fetch.

Time is injected, not read. Time-dependent logic goes through a Clock abstraction rather than calling the system clock directly, so tests and the 24-hour simulation script can drive arbitrary times deterministically.

Contributing

Contributions are welcome — see docs/CONTRIBUTING.md for dev setup, running tests, simulating dashboards, and troubleshooting.