60 and 59.94 are not the same number
The stutter arrives every fifteen or twenty seconds, all night, and no amount of hardware fixes it. Two devices agreed on 60 and meant different things.
The judder is regular. That is the clue.
Random stutter is load. Stutter on a metronome is arithmetic.
Where 59.94 came from
American colour television needed to stay compatible with black and white sets, so the frame rate was nudged down by a factor of 1000/1001. Sixty became 59.94. Thirty became 29.97.
That decision is seventy years old and still in your capture chain. Cameras, capture cards and consoles all still offer both.
What the mismatch does
Run a 59.94 source into a 60 pipeline and you are 0.06 frames short every second. After about 16.7 seconds you are a whole frame behind.
Something has to give. The system repeats a frame or drops one. You see a single hitch, and then you see it again seventeen seconds later, all the way through the session.
Time the gap between hitches. Around 17 seconds at 60 fps means a 60 against 59.94 mismatch. Around 33 seconds means 30 against 29.97.
The other cadence problem
Regions differ. 25 and 50 are standard where the mains runs at 50 Hz. A 50 fps camera into a 60 fps canvas is a far bigger mismatch. It judders all the time instead of on a schedule.
Mixing 24 fps film content into a 60 fps stream is the same class of problem. It is handled by repeating frames in a pattern, and that pattern is visible on slow camera moves.
Setting it straight
- Pick one cadence for the whole chain. Either the 60 family or the 59.94 family, never both.
- Set the camera, the capture card, the OBS canvas and the output to match.
- Check the capture card properties. Cards often default to 59.94 while listing it as 60.
- Consoles usually output 59.94 even when the menu says 60. Match the pipeline to the console, not to the label.
- Where a source cannot be changed, set the whole pipeline to its cadence instead.
The one that catches people
Your monitor refresh rate is a third number. A 144 Hz display showing a 60 fps preview is fine, because the preview is not what you ship.
Chasing a stutter you only see in the preview window wastes an evening. Judge the recording, not the preview.
Sources
The Rig Wire reports on streaming technology from public documentation, source repositories and vendor releases. We did not bench-test what is described here. When we have tested something, we say so.