12
CAPTURE / 12 — DIGITAL CROWDS
Digital Crowds
Background actors scanned on your set, and a perfectly synchronized 4K–6K volumetric array — gear and personnel supplied — producing sprites, volumetric assets, 4D mesh sequences, and 4D Gaussian splats.
[ ON SET / ON LOCATION / GEAR + CREW SUPPLIED ]
SRC / CROWD SESSION / ARRAY ENTRY
WHAT IT IS
A crowd is a throughput problem.
Stadiums, streets, ballrooms, battlefields — the shot needs hundreds or thousands of people, and the day only has room for a fraction of them. Crowd systems fill the rest. What they fill it with is the whole question: generic library agents read as generic library agents, and an audience clocks it even when they cannot name why.
Feed the same system real, scanned people — the actual background cast, in the actual costume, from the actual day — and the crowd stops looking borrowed. We supply that data two ways: traditional 3D scans of background performers captured on your set, and moving capture from a synchronized volumetric array we bring with crew.
ON SET WITH THE BACKGROUND CAST
We come to the holding area.
Our Mobile CyberSCAN™ Studios are trucked to the production and stand up in about 15 minutes — a stage, a location, a backlot, a parking lot. Each performer is in the rig for roughly 60 seconds, and we take two scans as standard so a blink or a moved hand never becomes a callback. The rigs refire back to back with no offload gap, so a background unit moves through in the gaps that already exist in the schedule instead of needing a day of its own.
Costume, hair, and hand props are captured as shot, which is what keeps a scanned background performer usable in the sequence they were actually hired for. We work in Hero and Background tiers — Hero for anyone who may end up in a full-frame shot, Background for the fast, efficient throughput a crowd system demands — and the data is retopologized and textured for the crowd pipeline it is headed into. For close-up principal work, that same capture is a digital double.
[ ~15 MIN SETUP / ~60s PER PERFORMER / 2 SCANS / BACK-TO-BACK ]
THE VOLUMETRIC ARRAY
Gear and personnel for a synced 4K–6K array.
When the crowd has to move rather than stand, a static scan is not enough. We supply the rig and the people to run it: a genlocked 4K–6K camera array built in house on the C.A.L.V.I.N.N.™ platform, plus the crew who designed it. Genlock is the part that is not negotiable — if the cameras are not firing on the same instant, every reconstruction of a moving performer inherits that error, and no amount of cleanup puts it back.
The array is modular and travels on its own, so performance capture happens on your production instead of routing a background unit to a fixed stage. Output is scoped to the pipeline receiving it: sprites and imposter cards for deep-background density, volumetric assets and 4D mesh sequences for mid-ground action, and Gaussian splat sequences where the crowd has to render in real time. When the brief is a single moving performance rather than a populated frame, that is 4D Capture.
[ 4K–6K / GENLOCKED / MODULAR + MOBILE / GEAR + CREW SUPPLIED ]
FIG.12 / EVERY ANGLE, ONE INSTANT
WHAT YOU RECEIVE
Crowds your pipeline can actually run.
- Scanned background castReal performers in costume, captured on your shoot day.
- Crowd-ready assetsRetopologized and textured for the crowd system receiving them.
- 4D mesh sequencesPer-frame geometry of a performance, not a simulated cycle.
- 4D Gaussian splatsMoving performers as real-time splats, engine-ready.
- Volumetric assets & spritesFrom mid-ground volumes down to deep-background cards.
- Pipeline handoffScoped with your VFX vendor before the cameras ever fire.
SELECTED WORK
Crowd and volumetric capture, recently.
RELATED SERVICES
START HERE
Tell us how big the crowd has to be.
Send us the background count, the shoot dates, and the crowd system it feeds. We’ll scope the scanning day or the volumetric array around it.
FAQ
Common questions
Is there a wait between performers?
No. Raw frames are available and reviewable the moment they are shot, and the rigs refire back to back. There is no offload gap, so a scan fits inside a lighting setup instead of needing a slot of its own.
How long does an actor have to stand in the rig?
About 60 seconds door to door — that covers a hero scan plus a backup safety scan. The capture itself is 1/60th of a second, so nothing is asked of the performer: no held pose, no held breath, no repeat takes. Their time belongs to their character, not to our equipment.
How long do you need on set before you can shoot?
About 15 minutes from arrival to capture-ready — on a stage, on location, or in a parking lot. The studio travels to production, not the other way around.
Do you need a dedicated scan day?
Usually not. Because we are mobile and set up in about 15 minutes, most productions fold scanning into a day that is already on the schedule — no travel days, no flights, and no per diem for cast.
Does genlock actually matter for Gaussian splatting?
Yes. If the cameras are not genlocked, each one is sampling a slightly different moment, and any reconstruction of a moving subject inherits that error. Every array we build is genlocked for that reason.
What is the difference between 3D scanning and 4D capture?
A 3D scan is a single instant — one frame of geometry and texture. 4D capture is performance over time: the same volumetric capture repeated across a sequence, so motion and deformation are captured, not simulated after the fact.
Do you handle NDAs and secure delivery?
Yes. Almost everything we shoot is under NDA — unreleased titles, talent likenesses, and production designs. We work under production NDAs as standard practice and treat likeness data accordingly.