IoT | Tomokatsu Yukishita | yre.jp

Build a Cloud-Connected Photo Frame with Raspberry Pi, Google Drive, rclone, fbi, and Python

Tue, 19 Jul 2022 00:00:00 +0000

I have been shooting with a digital SLR since 2003, and my total shot count is now at 388,524. I wanted a way to display those photos as a slideshow at home — without the hassle of manually copying files to an SD card every week.

The Problem with Commercial Photo Frames

Most digital photo frames read images from an SD card or USB drive, in the 7–10 inch range. They work fine, but constantly re-copying photos is tedious enough to make you stop using it.

What I wanted was a photo frame that pulls photos automatically from cloud storage like Google Drive. No such product seemed to exist, so I built one.

The concept: automatic photo sync from cloud storage to the photo frame

Hardware

Raspberry Pi 4

A photo frame runs 24/7, so low power consumption matters. Raspberry Pi 4 fits well.

Raspberry Pi 4 Model B 4GB (Japan authorized distributor)

Amazonで見る

Mobile Monitor

A 15.6-inch FullHD IPS mobile monitor. Choosing an IPS panel avoids the narrow viewing angles common in off-the-shelf photo frames. Mobile monitors are compact and flexible to position.

15.6-inch Mobile Monitor with Touch Panel

Amazonで見る

microSD Card and Power Supply

SanDisk High Endurance microSDHC 32GB for Dashcam and Action Cameras

Amazonで見る

Miuzei Raspberry Pi 4 USB-C Power Supply 5.1V 3A with Switch (PSE certified)

Amazonで見る

System Architecture

OS: Raspberry Pi OS

Write Raspberry Pi OS to the SD card using Raspberry Pi Imager. Since no desktop is needed for a photo frame, the Lite (CLI-only) version is sufficient.

Image Display: fbi

fbi renders images directly to the framebuffer from the command line. An earlier version used a GUI on Ubuntu but proved unstable, so I switched to this CLI-based approach.

Cloud Storage Sync: rclone

rclone downloads photos from Google Drive on a schedule. It supports over 70 cloud storage providers including Dropbox and S3.

The flow is: rclone downloads photos → fbi displays them as a slideshow.

Setup

Installing Raspberry Pi OS

Use Raspberry Pi Imager to write the image to the SD card.

Writing the OS image with Raspberry Pi Imager

Either 32-bit or 64-bit works for this use case.

Installing fbi

sudo apt-get -y install fbi

Granting Permissions

By default only root can use fbi to display images. Add the pi user to the video group:

pi@raspi3-photo:~ $ ls -Fla /dev/fb0
crw-rw---- 1 root video 29, 0 Apr 7 10:33 /dev/fb0

sudo usermod -aG video pi

Installing rclone

curl https://rclone.org/install.sh | sudo bash

After installing, run rclone config to authenticate with Google Drive.

In a headless SSH-only environment, rclone config will display a browser authentication URL during setup. Open that URL on another device to authorize access to your Google account.

Slideshow Program

Algorithm

Use rclone ls to list photos in the designated cloud storage folder
Start downloading images at random once an initial batch is available
Display downloaded images with fbi
Continue fetching more images in the background
Advance to a random new image at the configured interval
Use weighted random selection — recently shown images are deprioritized
Run rclone ls weekly to merge newly added photos into the database
Keep the database in memory (not a JSON file) for faster merging

Language: Python

The slideshow controller is written in Python. The refactored object-oriented version reduced startup time by 1700% compared to the original.

For Python learning, Python 3 Bootcamp by Silicon Valley Engineer (Udemy) was helpful.

Source Code

Published on GitHub:

GitHub - yukishita/photoView4: Cloud-connected slideshow program

Usage

Add this to crontab to start the slideshow at boot:

@reboot /home/pi/photoView4/photoView4.sh

Demonstration

The improved v4 demo. Startup to first image display is dramatically faster than the original:

Original Version (for reference)

The original downloads the full file list before starting, which makes startup slow:

Tips: Scheduled Display Power Off

Running the display 24/7 wastes electricity. Use vcgencmd in crontab to schedule power:

0 0 * * * /usr/bin/vcgencmd display_power 0
30 6 * * * /usr/bin/vcgencmd display_power 1

This turns the display off at midnight and back on at 6:30 AM. A motion or light sensor can make this even more automatic.

Summary

Component	Role
Raspberry Pi 4	Main computer (low power, always-on)
Mobile monitor	FullHD IPS display for high-quality photo viewing
fbi	Direct framebuffer image rendering from CLI
rclone	Automatically downloads photos from Google Drive
Python	Slideshow control with weighted random selection

Raspberry Pi 4 was used here, but Raspberry Pi Zero 2 W should also work. For an always-on device, choosing a lower-power board reduces running costs.

Continuous CO2 Monitoring with Raspberry Pi, CO2-mini, and Zabbix: A Complete Setup Guide

Tue, 19 Jul 2022 00:00:00 +0000

During the COVID-19 pandemic, I bought a CO2 monitor — the CUSTOM CO2-mini. It runs on USB power, and as it turns out, it can also communicate over USB. I connected it to a Raspberry Pi and set up Zabbix to continuously monitor indoor CO2 levels. Here is the full setup guide.

Hardware

Reading Data from the CO2-mini (Python)

When you connect the CO2-mini to a Raspberry Pi via USB, it is recognized without any special drivers. Data retrieval uses the heinemml/CO2Meter Python library.

Installing the Library

On Raspberry Pi OS Bookworm (2023 and later), installing packages directly into the system Python with pip is restricted. Use a virtual environment instead:

python3 -m venv /root/co2env
/root/co2env/bin/pip install git+https://github.com/heinemml/CO2Meter

On Raspberry Pi OS Bookworm and later, sudo pip3 install ... may fail with an “externally managed environment” error. Using a virtual environment as shown above is the safe approach.

Script: Writing CO2 and Temperature to Shared Memory

To make values accessible to the Zabbix agent, the script continuously writes them to files under /dev/shm/.

from CO2Meter import *
import time

sensor = CO2Meter("/dev/hidraw0") # may be hidraw1 or similar on your system

while True:
 time.sleep(1)
 data = sensor.get_data()
 if 'temperature' in data and 'co2' in data:
 with open('/dev/shm/co2', 'w') as f:
 f.write(str(data['co2']))
 with open('/dev/shm/temperature', 'w') as f:
 f.write(str(data['temperature']))

The device path /dev/hidraw0 may differ on your system. Run ls /dev/hidraw* to find the correct path after plugging in the CO2-mini.

Auto-start at Boot (crontab)

The device requires root access, so create a shell script that runs as root:

/root/co2.sh

#!/bin/bash
sleep 5
/root/co2env/bin/python3 /root/co2.py

Add to root’s crontab to run at boot:

@reboot /root/co2.sh

Zabbix Configuration

The approach is:

The Python script continuously writes CO2 and temperature values to /dev/shm/
The Zabbix agent reads those files via cat and returns the values

Zabbix Server: Creating the Host and Items

Install the zabbix-agent on the Raspberry Pi and add it as a host in Zabbix server.

Adding the Raspberry Pi as a Zabbix host

Create two items — one for CO2 concentration and one for temperature.

Creating CO2 and temperature items in Zabbix

Configuring item keys to match UserParameter names

Use the following item keys:

co2.co2 (CO2 concentration)
co2.temp (Temperature)

Zabbix Agent: Configuration File

Edit /etc/zabbix/zabbix_agentd.conf to set the server address:

Server=<zabbix-server address>
ServerActive=<zabbix-server address>
Include=/etc/zabbix/zabbix_agentd.conf.d/*.conf

UserParameter Configuration

Create /etc/zabbix/zabbix_agentd.conf.d/userparameter_co2.conf. It simply cats the shared memory files:

UserParameter=co2.co2,cat /dev/shm/co2
UserParameter=co2.temp,cat /dev/shm/temperature

Restart the agent:

sudo systemctl restart zabbix-agent

Results

CO2 concentration and temperature are now graphed in Zabbix.

CO2 concentration graph in Zabbix. Values drop at ventilation events

Long-term trend. In winter, windows tend to stay closed, allowing CO2 to build up

Since setting up the monitor, I ventilate much more frequently. Seeing the numbers makes it obvious exactly when ventilation is needed.

Summary

Component	Role
CO2-mini	Measures CO2 concentration and temperature via USB
Raspberry Pi	Reads sensor data and writes it to shared memory
Python (CO2Meter library)	Communicates with CO2-mini and parses data
Zabbix (agent + server)	Collects data, graphs it, and stores long-term history

The complete scripts are available on GitHub:

GitHub - yukishita/co2-mini: Program to log CUSTOM CO2-mini data with Zabbix

Raspberry Pi 3 + Official 7-inch Touchscreen Photo Frame with Ambient Light Sensor Auto-Brightness (BH1750FVI + I2C + Python)

Tue, 19 Jul 2022 00:00:00 +0000

After building a cloud-connected photo frame with Raspberry Pi 4, I had a spare Raspberry Pi 3 and official 7-inch touchscreen sitting around. I turned them into another photo frame — and added a GY-30 BH1750FVI ambient light sensor to automatically adjust display brightness based on the room’s lighting.

Hardware

Raspberry Pi 3

Raspberry Pi 3 Model B Single-board Computer

Amazonで見る

Official 7-inch Touchscreen

800×480px, 24-bit RGB color, 60 fps, up to 10-point multitouch.

Raspberry Pi Official 7-inch Touchscreen LCD

Amazonで見る

Case for the Official Touchscreen

Designed to house both the Raspberry Pi 3 Model B and the 7-inch display.

Raspberry Pi & 7-inch LCD Touchscreen Case ABS (RS Components, Black)

Amazonで見る

Connecting the Display

Follow the Raspberry Pi Touch Display official documentation to assemble the display.

When connecting, pay attention to the ribbon cable orientation. For the power wires, connect red to the 5V pin and black to the GND pin.

Display Rotation Setting

By default the display is rotated 180°. The fix depends on your OS version.

The lcd_rotate=2 option below is for Raspberry Pi OS Bullseye and earlier (legacy/FKMS mode). On Bookworm (2023 and later), lcd_rotate is not officially recommended — use the Screen Configuration GUI tool instead.

The config file path also differs by OS version:

Bullseye and earlier: /boot/config.txt
Bookworm and later: /boot/firmware/config.txt

For Bullseye and earlier (legacy/FKMS mode):

sudo vi /boot/config.txt

Add the following line:

lcd_rotate=2

Slideshow Program

The slideshow uses the Python script from the cloud-connected photo frame project (photoView4).

Display Brightness Control

Running the display at full brightness all the time is uncomfortable in a dark room. The official 7-inch display allows brightness control by writing a value (0–255) to a sysfs file:

echo "100" | sudo tee /sys/class/backlight/rpi_backlight/brightness

Adding an Ambient Light Sensor (BH1750FVI)

Rather than setting brightness manually, I wanted it to adjust automatically based on ambient light.

Parts

WINGONEER GY-30 BH1750FVI Digital Light Intensity Sensor Module I2C for Raspberry Pi

Amazonで見る

ELEGOO 120pcs Dupont Jumper Wires Male-Female / Male-Male / Female-Female Set

Amazonで見る

Connecting to Raspberry Pi via I2C

The BH1750FVI communicates over I2C. Refer to the Raspberry Pi GPIO pin layout for wiring.

Raspberry Pi GPIO pinout (source: Raspberry Pi official documentation)

Wiring Table

Raspberry Pi	GPIO Pin	BH1750FVI
GPIO2 (SDA)	3	SDA
GPIO3 (SCL)	5	SCL
5V Power	2	VCC
Ground	14	ADO
Ground	20	GND

BH1750FVI wired to Raspberry Pi 3 via I2C

Verifying the Sensor (Python)

After wiring, confirm data is coming through. Based on a reference implementation on GitHub:

#!/usr/bin/python3

import smbus

Bus = smbus.SMBus(1)
Addr = 0x23
LxRead = Bus.read_i2c_block_data(Addr, 0x11)
print("Illuminance: " + str(LxRead[1] * 10) + " lx")
LxRead2 = Bus.read_i2c_block_data(Addr, 0x10)
print("Brightness: " + str((LxRead2[0] * 256 + LxRead2[1]) / 1.2))

Running it confirms the sensor is working:

root@raspi3-photo:~# python br.py
Illuminance: 1650 lx
Brightness: 990.8333333333334

Auto-brightness Control Program

The final program adjusts backlight brightness smoothly based on ambient light readings:

Target brightness levels are defined per illuminance range
The display brightness is incremented or decremented one step at a time toward the target
The I2C sensor is polled every 300ms — polling faster than this results in incorrect readings with SMBus

Sensor Placement

The sensor is mounted on the protruding back section of the case, where it receives ambient light from the room.

Sensor mounted on the case back where it can see ambient light

Source Code

The complete program (object-oriented version) is on GitHub:

GitHub - yukishita/lcdBrightness2: Auto-brightness control for Raspberry Pi official display using BH1750FVI light sensor

Demonstration

The brightness adjusts smoothly as ambient light changes:

Summary

Component	Role
Raspberry Pi 3	Main computer
Official 7-inch Touchscreen	Photo frame display (800×480px)
GY-30 BH1750FVI	I2C ambient light sensor
Python (smbus)	I2C communication and brightness control logic

Adding the light sensor makes the photo frame genuinely pleasant to use — bright during the day, dimmed at night, automatically.

DIY JJY Simulator with ATOM Lite: Sync a Radio-Controlled Clock Without a Signal

Thu, 12 May 2022 00:00:00 +0000

After moving to a new home where my radio-controlled clock could no longer receive the JJY time signal, I built a low-cost JJY simulator using an M5Stack ATOM Lite and a JJY antenna board. This is a record of how I solved the problem for well under JPY 5,000.

The Problem: Radio-Controlled Clock Stopped Syncing After Moving

Radio-controlled clocks are convenient precisely because they set themselves automatically. After moving, however, my clock stopped receiving the JJY signal indoors, and the displayed time slowly drifted out of sync.

The clock could still pick up the signal near a window, but carrying it there and back every time it needed syncing was not a practical long-term solution.

What Is JJY?

JJY is the call sign for Japan’s standard time radio signal, broadcast by the National Institute of Information and Communications Technology (NICT). Two transmitters cover the country: 40 kHz from Fukushima (Ōtakadoya-yama) and 60 kHz from Saga (Haganeyama). Radio-controlled clocks receive this signal to maintain accurate time automatically.

The device described in this article simulates the JJY signal and rebroadcasts it locally, allowing a clock to sync even when it cannot receive the real transmission.

Commercial Retransmitters Are Expensive

There are commercial products designed to solve exactly this problem — they receive accurate time from NTP servers or GPS and retransmit a JJY-compatible signal indoors.

Kyohritsu Denshi AC-Synchronized Time Transmitter P18-NTPAC

Amazonで見る

KEISEEDS GPS Radio Clock Repeater P18-NTPGR

Amazonで見る

Radio Clock Signal Transmitter with Built-in Clock P18-NTPLR

Amazonで見る

These products work well, but they typically cost JPY 10,000–30,000. Hard to justify for a single clock at home.

The DIY Solution: ATOM Lite + JJY Antenna Board

Searching Switch Science (a Japanese electronics parts retailer), I found a JJY antenna board designed to work with M5Stack devices — exactly what I needed.

Required Parts

Product	Link	Notes
JJY antenna board for M5StickC	Switch Science	M5StickC specific
JJY antenna board for M5Atom	Switch Science	For ATOM Lite/Matrix
M5StickC	Switch Science	Discontinued
ATOM Lite	Switch Science	Available now — recommended
ATOM Matrix	Switch Science	Available now

Note: M5StickC has been discontinued. For new builds, the recommended combination is ATOM Lite with the JJY antenna board for M5Atom. The original article used M5StickC, which is what appears in the photos below.

At the time of purchase, the antenna board and M5 device together came to under JPY 5,000 (prices may vary — check current listings before ordering). That is roughly one-third to one-sixth the cost of a commercial retransmitter.

Setup

Setup follows the instructions in the GitHub repository linked from the Switch Science product page:

Flash the firmware following the instructions and there are no unusual steps. It worked on the first attempt without any issues.

Testing

With the firmware flashed, connect the M5 device to the JJY antenna board and place it next to the radio-controlled clock.

Place the device right next to the clock to receive the simulated signal

The transmitted signal is weak and short-range. The device needs to be within a few centimeters to a few tens of centimeters of the clock to sync reliably. In practice this is not a problem — once the clock has synced, it holds accurate time on its own until the next sync cycle.

Known Limitation

One issue worth noting: if the device is used near a microwave oven, the Wi-Fi connection drops and does not automatically reconnect. A manual restart is required. Either keep the device away from the microwave, or add a periodic auto-reset to the firmware.

Summary

Item	Detail
Total parts cost	Under JPY 5,000 (roughly 1/3–1/6 of commercial alternatives)
Soldering required	None
Setup difficulty	Low — flash firmware from GitHub and place next to clock
Signal range	Short (a few cm to a few tens of cm)
Known issue	Wi-Fi drops near microwave, requires manual restart

A practical and affordable way to keep a radio-controlled clock synced in a room with poor JJY reception. If you are in a similar situation, this combination is well worth trying.

IoT | Tomokatsu Yukishita | yre.jp

Build a Cloud-Connected Photo Frame with Raspberry Pi, Google Drive, rclone, fbi, and Python

The Problem with Commercial Photo Frames

Hardware

Raspberry Pi 4

Mobile Monitor

microSD Card and Power Supply

System Architecture

OS: Raspberry Pi OS

Image Display: fbi

Cloud Storage Sync: rclone

Setup

Installing Raspberry Pi OS

Installing fbi

Granting Permissions

Installing rclone

Slideshow Program

Algorithm

Language: Python

Source Code

Usage

Demonstration

Original Version (for reference)

Tips: Scheduled Display Power Off

Summary

Continuous CO2 Monitoring with Raspberry Pi, CO2-mini, and Zabbix: A Complete Setup Guide

Hardware

Raspberry Pi

CO2-mini (CUSTOM CO2 Monitor)

Power Supply and microSD Card

Reading Data from the CO2-mini (Python)

Installing the Library

Script: Writing CO2 and Temperature to Shared Memory

Auto-start at Boot (crontab)

Zabbix Configuration

Zabbix Server: Creating the Host and Items

Zabbix Agent: Configuration File

UserParameter Configuration

Results

Summary

Raspberry Pi 3 + Official 7-inch Touchscreen Photo Frame with Ambient Light Sensor Auto-Brightness (BH1750FVI + I2C + Python)

Hardware

Raspberry Pi 3

Official 7-inch Touchscreen

Case for the Official Touchscreen

Connecting the Display

Display Rotation Setting

Slideshow Program

Display Brightness Control

Adding an Ambient Light Sensor (BH1750FVI)

Parts

Connecting to Raspberry Pi via I2C

Wiring Table

Verifying the Sensor (Python)

Auto-brightness Control Program

Sensor Placement

Source Code

Demonstration

Summary

DIY JJY Simulator with ATOM Lite: Sync a Radio-Controlled Clock Without a Signal

The Problem: Radio-Controlled Clock Stopped Syncing After Moving

What Is JJY?

Commercial Retransmitters Are Expensive

The DIY Solution: ATOM Lite + JJY Antenna Board

Required Parts

Setup

Testing

Known Limitation

Summary