The encoder is the box (or software) that takes the finished programme picture and sound and compresses it into a stream the internet can carry. Its settings decide how your event looks on a phone and how likely it is to buffer. Here's what each one does, and what we set.
Hardware or software?
Software encoders (vMix, OBS, Wirecast) run on a production PC and are flexible, cheap and integrated with the vision mix — the same machine can switch cameras, play graphics and encode. Hardware encoders (dedicated boxes from Teradek, LiveU, AJA, Blackmagic and others) do one job with great reliability and no operating system updates to worry about. Our standard on anything important: software encoding for the primary programme, a hardware encoder as the independent backup taking the same programme feed.
RTMP vs SRT
RTMP is the veteran protocol most platforms accept. It's simple, but it handles packet loss badly — a dropped packet becomes a stall. SRT was built for unreliable networks: it retransmits lost packets within a latency window you set, encrypts by default and copes with jitter. Use SRT wherever the destination supports it, especially for contribution links between sites and into cloud galleries; fall back to RTMP where you must (many social platforms).
Resolution: 1080p is the default, not 4K
1080p at 25 or 30 frames per second is the sweet spot for events: sharp on a laptop, efficient on a phone, achievable on ordinary venue lines. 4K quadruples the bitrate for a benefit almost no viewer sees on a stream, and most platforms will downscale it anyway. 720p is a sensible choice when the line is marginal — a stable 720p beats a stuttering 1080p every time. Frame rate: 25 (UK) or 30 fps for presentations; 50/60 only for fast motion like sport.
Bitrate: enough, with headroom
Typical settings we use: 1080p at 4.5–6 Mbps for talking heads and slides; up to 8 Mbps for busy, moving content; 720p at 2.5–4 Mbps. Higher isn't better if the line can't sustain it. Use a constant bitrate (CBR) for the live encode — it's predictable for the network — and keep the total of all your streams under half the tested upload.
Keyframes, profiles and audio
Set the keyframe interval to 2 seconds (most platforms require it), H.264 High profile, and AAC audio at 128–192 kbps, 48 kHz, stereo. Audio at too low a bitrate is far more noticeable than a slightly softer picture. If your platform supports HEVC (H.265) you gain quality per megabit, but check device compatibility before you rely on it.
Recording alongside the stream
Always record the programme locally at a higher quality than the stream — an ISO or programme recording at 20–50 Mbps — so the on-demand version and the highlight edit aren't built from the compressed stream. Record on two devices if the content matters.
The settings card we tape to the encoder
| Setting | Presentations / talking heads | Motion-heavy content |
|---|---|---|
| Resolution / frame rate | 1920×1080 @ 25 | 1920×1080 @ 50 |
| Video bitrate (CBR) | 5 Mbps | 8 Mbps |
| Keyframe interval | 2 s | 2 s |
| Profile | H.264 High | H.264 High |
| Audio | AAC 160 kbps 48 kHz | AAC 192 kbps 48 kHz |
| Protocol | SRT (RTMP fallback) | SRT (RTMP fallback) |
