Choose by what moves
| Profile | Makes sense for | Main trade-off |
|---|---|---|
| 720p30 | talk, podcasts, drawing and limited setups | more stability, less detail and smoothness |
| 720p60 | fast games with tight bitrate or encoder | smooth motion, less crisp text |
| 1080p30 | camera, tutorials and text with little motion | more detail, less smooth motion |
| 1080p60 | fast games, fine text and a setup with headroom | highest load and need for bitrate |
Don't pick 1080p60 because your monitor is Full HD. The output has to survive the worst minute of your stream.
Bitrate changes the result
On Twitch, the official reference uses 3,000 kbps for 720p30, 4,500 for 720p60 and 6,000 for 1080p60. On YouTube, the current table recommends, in H.264, 8 Mbps for 720p30 and 720p60 and 17 Mbps for 1080p60, with minimums of 3, 3 and 6 Mbps. In AV1 or HEVC, the recommended values drop to 6, 6 and 12 Mbps.
With tight bitrate, more pixels can make the picture worse: the encoder splits the same amount of data across more detail and more frames. Blocky motion is a sign to lower resolution, FPS or scene complexity.
In multistream, there are three paths
- Lowest common denominator: set OBS to a profile every platform accepts and copy the signal.
- One exception: copy to the compatible destinations and transcode only the one that is different.
- All separate: transcode each platform when per-destination quality is worth the load.
In Corneta, these paths are the modes on the Quality screen: Straight up (1), Smart (2, which copies the landscape platforms and transcodes only the vertical ones) and All out (3). Set it up for me only points OBS at Corneta, with server and key; you set resolution, FPS and bitrate in OBS itself.
Start with the first one. Only open a new encoder session when the benefit is visible.
Test without fooling yourself
- use your heaviest scene;
- move the camera and the game like in a real stream;
- let the test run for ten minutes;
- check rendering, encoding and network separately;
- watch on a phone and on a regular connection;
- test every platform at the same time.
A smooth local recording doesn't prove that three network outputs will stay stable. A high speed test doesn't prove that your GPU can handle transcoding either.
If the profile you picked goes over your bandwidth, work out the total in the bitrate and upload guide.