Live-Streaming Arendalsuka 2024 to Facebook & LinkedIn Simultaneously

Challenge

Arendalsuka (Norway's largest political week) 2024 needed:

• 4 cameras (Logitech Brio 4K)
• Simultaneous stream to Facebook & LinkedIn
• 6 hours/day streaming for 5 days
• Automatic scene switching based on speaker
• Zero budget for commercial streaming platforms

Solution

Repository: Arendalsuka-Multicam-Stream documents the full setup.

Stack

Architecture

[4× Logitech Brio] → [USB 3.0 Hub] → [OBS Studio]
                                                      ↓
                         ┌────────────────────────────┴────────────────────────────┐
                         ▼                                                         ▼
                  [nginx-rtmp]                                                 [nginx-rtmp]
                    (Facebook)                                                   (LinkedIn)
                         │                                                         │
                         ▼                                                         ▼
                   rtmp://live-api-s.fb...                                  rtmp://linkedin...

Key Components

1. nginx-rtmp for Multi-Output

# nginx.conf
rtmp {
    server {
        listen 1935;
        chunk_size 4096;

        application live {
            live on;
            record off;
            
            # Facebook
            push rtmp://live-api-s.facebook.com/rtmp/STREAM_KEY;
            
            # LinkedIn
            push rtmp://rtmp-api.linkedin.com/rtmp/STREAM_KEY;
        }
    }
}

Run in Docker:

docker run -d \
  -p 1935:1935 \
  -v $(pwd)/nginx.conf:/etc/nginx/nginx.conf \
  tiangolo/nginx-rtmp

2. Auto Scene Switching (Python)

# auto_switch.py
# Switch scene to loudest camera based on audio levels

import obsws_python as obs
import time

client = obs.ReqClient(host='localhost', port=4455, password='secret')

SCENES = ['cam1', 'cam2', 'cam3', 'cam4']

def get_audio_levels():
    # Query OBS for input volume levels
    levels = {}
    for scene in SCENES:
        resp = client.get_input_volume(scene)
        levels[scene] = resp.input_volume_mul
    return levels

while True:
    levels = get_audio_levels()
    loudest = max(levels, key=lambda x: levels[x])
    if levels[loudest] > 0.1:  # threshold
        client.set_current_program_scene(loudest)
    time.sleep(0.5)

3. OBS Scene Collection

Results

Lessons Learned

1. nginx-rtmp is rock-solid — handled 5 days of continuous streaming without restart
2. Audio-based switching works — no manual operator needed for panel discussions
3. USB 3.0 bandwidth — 4× 4K@30fps needs a quality powered hub
4. Test stream keys early — LinkedIn keys expire, Facebook keys persist
5. Backup recording — OBS local recording saved us when LinkedIn had ingest issues

Links

• Repository: https://github.com/turbolego/Arendalsuka-Multicam-Stream
• Full Documentation: README in repo (hardware list, OBS config, nginx config, Python scripts)
• Arendalsuka 2024: https://arendalsuka.no/

Open source streaming infrastructure — no vendor lock-in, no monthly fees.