Overview

Smelter performance is highly dependent on the runtime, hardware, and the feature set you need. It is a great choice in some scenarios, but suboptimal in others.

When to use Smelter?

Smelter is very efficient if you need a solution that will decode, modify raw frames and encode it back. However, if you only need some of those features there might be a better solution for you out there.

You should use Smelter:

• If you need to modify video content (add overlay, combine multiple sources, add some text or effects)
• You need to transcode the video to a different codec.
• You need to generate output in multiple resolutions.

You should not use Smelter:

• If you want to convert your stream into different protocol or format. For example, convert incoming RTMP stream into HLS playlist.
• If you want to re-stream something to more viewers. For example, broadcast incoming WHIP stream over WHEP consider a solution like Fishjam instead.

Key performance factors

The processing pipeline in Smelter consists of many components. However, only a few specific factors have deciding impact on the performance. We can divide or heaviest workloads into following categories:

• decoding - Converting encoded video e.g. H264 into raw frames (2D array of bytes representing specific pixels).
• rendering - Process that takes set of frames (one for each input) and produces set of frames (one for each output).
• encoding - Converting stream of raw frames (2D array of bytes representing specific pixels) into encoded video e.g. H264

Runtimes

You can run Smelter on the server or in the browser. Each environment has its own performance bottlenecks.

See runtime specific pages to learn more:

• Media Server
• Browser WASM module