Skip to content

Disclaimers and Gotchas

Jump to bottom
Ryan Laseter edited this page Nov 2, 2023 · 2 revisions

Why results from one machine might not apply to another

It can be difficult to determine what specific encoder settings would work best on your system without a lot of trial and error. A few of the factors that can affect this:

  • GPU's encoder generation/hardware
  • overall system's performance (combination of motherboard, CPU, and other factors such as storage/RAM)
  • driver versions being used to encode the video files

With this in mind, you may find that what works really well on your system, may not work the same on another's (even with the same GPU hardware). This is why it's important to run the benchmark on your own system for specific settings/capabilities of your entire setup.

Higher bitrate always increases vmaf score

This is expected. By convention, using a low encoder preset/tune combination but allowing more data to be sent per second, will mean a higher quality stream.

This tool will help you find at what bitrate you reach your max achievable quality at given encoder settings, but if you can afford it you can always increase the bitrate above that (with diminishing quality returns). This is almost always the case with at-home game streaming where you're less bandwidth limited.

Game streaming outside your network or over cellular is where you'll truly become bandwidth limited and where this tool can be useful.

Disclaimers

  • It is possible that when you go to apply these settings in OBS Studio, or Sunshine's game stream hosting software, that the encoder/ffmpeg version being used there may perform different, i.e. most likely worse than this benchmarking tool. With that in mind, the author has personal success with applying suggested encoder settings to sunshine and seeing a 1 to 1 performance. However your results may vary.

  • This tool only tells you whether your host system can encode the video at the given parameters; it does not tell you whether the client you intend to stream TO can decode at the same speed. You may find that your client device cannot decode the incoming video as fast as it's sent. The Moonlight streaming app has a nice statistics overlay that can help you identify if your client is the bottleneck.

  • You may find that what works on your machine, does not work on another with similar hardware. This is expected, and is why you should run the tool on your machine to get specific results to your setup.

  • Your encoder performance in real-life might vary if you play different game genres, or if you have overlays in your OBS studio setup. It is difficult/impossible to cover all possible inputs when benchmarking video encoding, however this tool does try it's best to get you 90% of the way there.

Clone this wiki locally