HOW SERIOUS A FAILURE IS depends on which priority it threatens, not how loud it is. A dead encoder costs the live feed. A recorder silently not rolling costs the footage — the one thing we cannot get back.
01 BEFORE LEAVING — PRE-FLIGHT
Encoder configured, burned in, and tested — single-encoder rig: this box has no backup, treat it accordinglyWindows Update disabled on both CCUs and the encoder; no pending restartsFull gear check against the gear list — every case opened, every item verified working All quad-link BNC sets labeled (camera, link number, direction) Spares packed: DA, LC patch cables, BNC, 12V-2×6, batteries — fibre runs come from the DAZN truck Fiber borrow confirmed with the DAZN truck — runs are borrowed from truck fiber stock: count (2 per camera + spare), lengths, connector type, and who hands them over at load-inAll recording media formatted and verified; capacities match the continuous-roll plan Audio track count locked at 16 and written down — the number that gets taped to the encoderCircuit (Tata) tested end-to-end to real ingest before event day; night-of support number saved Alternate camera position confirmed with venue — "refused on the day" has an answer Abort criteria agreed (thermal, network, attention) and the person who calls it named
02 ABELCINE REHEARSAL — WED 19 AUG, ALL GEAR In hand: 2× V-Raptor, 2× Connect, 2× CCU, encode computer. The rest (DAs, switcher, HyperDecks, sync, monitoring) arrives from AbelCine at the rehearsal — check it in, then full rig-and-run. Nothing goes to the arena untested.
AbelCine gear checked in on arrival against the gear list — DAs, switcher, HyperDecks, sync, monitoring; every item powered and verified before the rig starts. Anything missing or dead is known by noon, not on SaturdayFull signal chain assembled exactly per the plan: camera → Connect → CCU → DA → switcher → encoder → cloud Genlock/PTP proven here — clap/strobe test, ISOs compared frame-by-frameBoth record layers exercised: HyperDecks off the DAs, cloud VOD — each played back. CCUs have no record space; the HyperDecks are the only ISODebayer + encode on the same computer, tested under full load — R3D debayer and 8K HEVC encode running together; watch GPU utilization, temps, and dropped frames over a sustained runSwitcher 8K-mode behavior confirmed on the bench: aux outputs and multiview — the router question answered for real Pull-Connect-and-record-internal drill performed once — timed, so the fallback is muscle memory Audio path run at the locked track count; a channel-check pass end to end Cesar walked through the encoder and the console — the second Prism operator exists Every 3 August landmine re-checked against the new topology; findings written down same day Repack after rehearsal is the arena pack — cases packed as they will travel, labels verified
03 LOAD-IN & RIG
Second fibre run to each camera position while the floor is open — not carried in a bagFibre runs armored/tactical, matting at every crossing, cable paths per the approved routes DA-bypass cables staged — CCU direct to HyperDeck and switcher, labeled, coiled at positionSwitcher-bypass path staged — one camera direct to encoder UPS on the encode + record path — nothing else on that batteryPower load on agreed circuits only; nothing critical shares with unknown loads Comms up and tested between camera positions, tech world, and audio Cellular backup on site, powered, tested at reduced bitrate to real ingest
04 SYNC
PTP (SMPTE 2059-2) reaching both cameras over Connect — settings per the RED conversation Tri-level/blackburst to both CCU DeckLink cards, switcher, and HyperDecks — one generator, locked Genlock proven: clap or strobe to both cameras, ISOs compared frame-by-frame
05 RECORD LAYERS — PRIORITY 1 CCUs have no record space — the HyperDecks are the only per-camera ISO. Every layer verified by playback, never by tally light.
HyperDeck A verified — recording off the DA, played back. The only record of CAM AHyperDeck B verified — recording off the DA, played back. The only record of CAM BThird HyperDeck standing by — the record path's only spare: configured, media loaded, reachable fastCloud VOD test clip — contributed, recorded to S3, played backCAM C internal record verified by playback DIT in position, clip log started (camera, bout, time), offload drives connected Two-copies-before-reuse rule briefed to everyone touching media Crew briefed: if a Connect module fails, pull it and record on the camera
06 COLOR
Log-to-picture conversion location confirmed (one place, every camera) and written down SDR confirmed end-to-end; encoder, converter and 8K display settings recorded Cameras A and B matched by the V1; check again after arena lighting changes 8K display in PC/Game mode, every processing feature off, settings photographed
07 AUDIO
16-channel external audio interface connected to the encoder — levels verified on every inputTrack count locked and taped to the encoder — it does not change mid-showInterface clock locked (sample rate confirmed at the encoder) — no dropouts in a 10-min soak Channel-check run: each of the 16 channels soloed and confirmed on the decoded return
08 MEASUREMENT
QR timecode slate visible in frame at the agreed position and size Return display up next to program — decoded CDN output, A/B against source Console session recording; per-stage latency capture confirmed writing If both circuit paths exist: same content run over each, results labeled
09 DURING THE SHOW
One named owner for signal-path changes — everyone else reportsExperiments run on undercards only; nothing untested in the path during main bouts Debayer + encode load watched through the first undercard — GPU utilization, temps, dropped frames logged before committing to main bouts Human notes written during quiet periods — someone owns the log Media swaps per plan; DIT log current at every swap Encoder temps checked each hour; airflow unobstructed
10 STRIKE & POST
Every card and drive accounted for against the DIT log before anything is packedTwo verified copies of everything before any media is reused or leaves the building separately Cloud VODs confirmed complete in S3; transfer plan for masters agreed All positions struck clean; venue walk-through with no trace left This checklist exported/screenshotted for the test report; debrief scheduled
Reset all
Expand all
Collapse all