A production-grade, fully standalone Raspberry Pi timelapse system with live preview and optional RTMP restreaming. No external server required.

Designed, built, and tested by Brett Chapin / MieTech LLC.

• Fully standalone — captures, stores, and generates videos on-device (no server required)
• Scheduled capture — interval, fixed times, or solar (sunrise/solar noon/sunset)
• Live preview — MJPEG (Pi Zero W) or HLS (Pi Zero 2 W+) via built-in web server
• RTMP restreaming — stream live to YouTube / Twitch / any RTMP server (Pi Zero 2 W+)
• Automatic video generation — ffmpeg stitches frames when your threshold is hit
• Disk space guard — configurable free-space floor with automatic oldest-first cleanup
• Multi-profile — run multiple independent timelapse setups simultaneously
• Systemd service — runs on boot, restarts on failure, survives reboots
• Secure by default — preview binds to localhost; built-in Basic Auth option

Pi Zero W: single-core ARMv6 @ 1 GHz, 512 MB RAM. MJPEG preview at 640×360 / 1 fps uses ~15–25% CPU. Run ffmpeg video generation with nice -n 19 to avoid camera stalls.

Pi Zero 2 W: quad-core Cortex-A53 @ 1 GHz, 512 MB RAM. HLS at 1280×720 / 15 fps is feasible using the V4L2 hardware H.264 encoder. Avoid running HLS + RTMP simultaneously.

# libcamera + picamera2 (camera access)
sudo apt install -y python3-picamera2 libcamera-apps

# ffmpeg (video generation + HLS + RTMP)
sudo apt install -y ffmpeg

# Python build tools
sudo apt install -y python3-pip python3-venv git

# From local clone:
pip install -e ".[dev]"

Minimum required packages (installed automatically):

Note: picamera2 is installed via apt, not pip.

timelapse setup
# or for a named profile:
timelapse --profile garden setup

timelapse config validate

timelapse run

sudo cp systemd/[email protected] /etc/systemd/system/
sudo systemctl daemon-reload
sudo systemctl enable --now timelapse@default
journalctl -u timelapse@default -f   # live logs

timelapse [--profile NAME] COMMAND

Commands:
  setup               Interactive setup wizard
  run                 Start capturing (foreground)
  status              Show capture statistics and disk status
  snapshot            Take one frame immediately (outside normal schedule)
  gen-video           Generate video from current photo batch now
  config validate     Check config for errors
  config show         Print resolved config as YAML

~/.config/timelapse/
  config.yaml              ← global settings (timezone, log level)
  profiles/
    default.yaml           ← default profile
    garden.yaml            ← custom profile
~/.local/share/timelapse/
  default/state.json       ← schedule state (auto-managed)

See profiles/default.yaml and config.example.yaml for fully commented examples.

Interval — every N seconds/minutes/hours:

schedule:
  mode: interval
  interval: "5m"   # 30s, 5m, 2h, 1d

Fixed times — one or more HH:MM times per day:

schedule:
  mode: times
  times: ["08:00", "12:00", "17:00"]

Solar — sunrise / solar noon / sunset with optional offsets:

schedule:
  mode: solar
  latitude: 51.5074
  longitude: -0.1278
  owm_api_key_env: "TIMELAPSE_OWM_KEY"
  events:
    - type: sunrise
      offset_minutes: 10
    - type: sunset
      offset_minutes: -15

The preview server binds to 127.0.0.1 by default.

Option A — Tailscale (recommended):

curl -fsSL https://tailscale.com/install.sh | sh && sudo tailscale up
# Access preview at http://<tailscale-ip>:8080/

Option B — Cloudflare Tunnel (public HTTPS URL):

cloudflared tunnel --url http://127.0.0.1:8080

Option C — nginx + Let's Encrypt (see README for full config)

Stream keys and passwords are read from environment variables only — never stored in config files.

# /etc/timelapse/default.env (chmod 0600, loaded by systemd)
TIMELAPSE_RTMP_KEY_DEFAULT=your-stream-key
TIMELAPSE_WEB_PASSWORD=your-preview-password
TIMELAPSE_OWM_KEY=your-owm-api-key

• Auto-upload after video: hook in timelapse/video/generator.py
• Webhook notifications: hook in timelapse/capture/pipeline.py
• Additional RTMP destinations: extend timelapse/streaming/rtmp.py
• ML trigger: intercept in timelapse/capture/pipeline.py before record_capture()

pip install -e ".[dev]"
pytest

MIT — This program is brought to you by MieTech LLC.