Build a rehearsal, not a show. The first OBS production should be small enough that every signal can be checked and every failure can be repeated without an audience waiting.
This runbook stops before specialist configuration
The steps below create one scene sequence, one audio mix, one local recording, and one private destination output. Read what OBS Studio is if the roles of scenes, sources, encoding, and destinations are unfamiliar. Detailed bitrate presets, recording-only workflows, SRT, RTMP, plugins, multistreaming, and platform interfaces remain separate jobs.
Install OBS from the official channel and record the rehearsed version
Use the project’s current download route and the official OBS Quick Start Guide. Record operating system, OBS version, installer source, graphics driver, and whether this is a clean or existing profile. Keep that version fixed through the rehearsal; an update between setup and review makes failures harder to reproduce.
Launch OBS and let the Auto-Configuration Wizard create a starting point if appropriate. Treat its result as a proposal for this machine, not as a guarantee for the destination or a permanent quality setting.
Create a three-scene production
Make a holding scene, a programme scene, and an ending scene. Add only the sources required for the rehearsal: for example one camera or application capture, one graphic, and one microphone. Name them by job rather than by device number. Confirm source visibility, crop, scale, and ordering in every scene.
Switch through the sequence while watching the programme view. If a source takes time to wake, note it now and decide whether the holding scene needs to cover that delay.
Prove audio before adding polish
Speak at normal and excited levels, play any programme audio, and observe the mixer. Listen through the monitoring path used by the operator, then make a short recording and listen on another device. Check voice clarity, clipping, unwanted desktop sound, channel balance, and lip sync. Muting a channel should create an audible and visible result.
Do not copy a universal filter chain. Noise suppression, compression, limiting, and gain depend on the room, microphone, interface, speaker, and intended loudness. Add one change at a time and keep the before-and-after clip.
Make the local recording the first quality checkpoint
Choose a recording location with enough space and write permission. Capture several minutes that include every scene, transitions, motion, quiet and loud speech, and the most demanding source. Stop cleanly, reopen the file, seek across the full duration, and verify picture, audio, sync, and finalization.
If recording is the main deliverable, continue with OBS recording settings. For this rehearsal, the file is a controlled observation of the local production before the network and destination are introduced.
Choose one private destination output
Select the actual service or server intended for the first run, but create a private, unlisted, or otherwise non-public destination according to that service’s current controls. Enter the service and credentials through the supported method. Keep secrets outside screenshots and shared notes. Confirm the event, channel, privacy state, and destination-side preview before sending media.
This step proves one path. Route SRT work to the OBS SRT guide, RTMP-specific work to the OBS RTMP guide, and wider output planning to OBS streaming. Do not fold those specialist procedures into the first rehearsal.
Record, send privately, stop, inspect, and decide
Start a local recording, then start the private output. Run the planned scene sequence, speak and play programme audio, introduce a normal motion burst, and leave enough time to observe stability. Have another person watch through the destination rather than the OBS preview. Record first visible frame, audio presence, sync, interruption, and privacy result.
Stop the stream cleanly, then stop recording. Open the local file and the destination replay if available. From OBS, open the log for this exact session and save it with start and stop times. Mark each checkpoint pass or fail before any public stream: local picture, local audio, local sync, destination receipt, viewer picture, viewer audio, privacy, clean stop, recording playback, and log captured.
Change one cause after a failed checkpoint
Use the earliest failed boundary. If the local recording is wrong, remain in sources, render, audio, or encoding. If the local file is correct but the destination receives nothing, compare OBS connection status with destination ingest. If ingest is present but viewing fails, continue into access, processing, delivery, and player observations. Save the failed log before opening a new session.
Recovery rehearsal: a scene switch overloads the machine
Duplicate the demanding switch in private. Observe render lag, encoding lag, system load, and which source appears at the failure. Remove or simplify one suspected source, repeat the same sequence, and compare. Then rehearse the operator response: return to the holding scene, preserve audio if possible, announce the interruption through the approved channel, and stop only if the output cannot recover.
Reduce or isolate scene complexity; compare the same local recording path.
Confirm the selected encoder and load; adjust only after a matched retry.
Preserve the OBS log and check ingest, access, processing, and playback in order.
Test the start and stop sequence with a second operator
Give another trained person the run sheet without verbal hints. Ask them to open the correct profile and scene collection, check sources and audio, locate the recording folder, select the private destination, start in the documented order, and finish cleanly. Note every ambiguous label or hidden dependency. Improve the card, then repeat until the operator can identify a failed checkpoint without improvising a risky change.
Keep destination credentials out of the card. Verify who can retrieve or rotate them, how the operator confirms the account and event without exposing a secret, and what should happen if authorization expires during rehearsal. The safest recovery may be to remain on local recording while the authorized person repairs the downstream connection.
Turn the successful rehearsal into an operator card
Write the OBS version, profile and scene collection names, source checks, audio check, recording location, private destination, start order, stop order, log location, and first recovery action. Keep passwords elsewhere. Add the date and the exact file or session that passed.
When the generic checkpoints are repeatable, the tested output may be connected to Callaba or another destination required by the workflow. That later choice does not alter the steps above and does not imply an undocumented plugin or interface. Repeat the private run whenever the OBS version, driver, scene collection, main source, encoder, network, or destination changes.
Schedule the next rehearsal before the first public programme, then repeat a shortened form before every materially different production. Preserve a known-good local recording and session log as comparison points. If the operator cannot reproduce the good state, pause the public start rather than changing several settings under pressure.
