Strivr Labs · Case Study

“Project Airline”

A high-fidelity R&D environment built to intentionally break and rebuild our CG pipeline.

The Challenge

Strivr's reputation was built on live-action VR. For some clients that quality was a constraint — filming meant flying a crew out and shutting a working facility down. CG was the answer, and we'd done it a couple of times — never as a repeatable pipeline. A CG project for a large account was kicking off, and it needed the pipeline working from day one.

RoleDirection · Procedure research · Pipeline integration
ClientStrivr, and a fictional airline invented to stress-test the pipeline
FormatInteractive CG VR · MetaHuman characters · Motion capture
TimelineScript to deployment in two weeks

The Strategy

Rather than risk a live engagement, we manufactured a customer: Project Airline, a fictional airline simulation, directed with the same rigor as paid work.

Two weeks script to deployment
  1. Script
  2. Produce
  3. QA
  4. Deploy

The goal: find every place the pipeline would break, before a paying client did.

The Playthrough

Three minutes as cabin crew

You meet your instructor, who introduces the goal: complete the pre-flight safety inspection checklist. From there, you're on your own.

01

Find it

Green icons mark the bins holding emergency equipment, the same convention as a real cabin. The learner has to find them under time pressure.

Scene huntLocate the equipment before the time runs out.

02

Inspect it

Three checks, in order: valve off, dust caps secured, pressure gauge. The gauge has to read 1500 PSI to pass.

The gaugeThis cylinder reads below 1500 PSI. It fails.

03

Make the call

Using environmental cues, make a decision about whether the equipment passes inspection.

The decisionThree options, including "I'm not sure."

The Pipeline

What broke

Environment generation, MetaHumans, automated lip sync, motion capture — we'd used each of these on its own. The sprint forced us to run them together, on a deadline, with render time budgeted against office hours. Two things broke.

Can you spot the render issue?

This went into the build with a mistake in it. It's obvious once you see it.

The aircraft cabin interior, looking down the aisle at the emergency oxygen bottle

The Craft

The CG crew member standing in the cabin with his arms open, his face covered by a question mark

Casting someone who doesn't exist

The second break wasn't technical: it was a communication problem. CG characters are assembled by different teams in separate rooms — someone picks the body, someone else picks the voice, someone else animates. The result is only cohesive when direction is clear and effective.

The brief called for a veteran crew member — someone with enough seniority that a rookie would follow their instructions without hesitation. I built the character's voice around that natural authority.

Audition takessame line, different voices

When the voice assets landed with CG, we realized a problem. They had built the character without checking the brief — decades younger than the man I'd imagined. The voice I generated wouldn't work.

On a live-action shoot, recasting voiceover means going out to talent agencies, booking a studio, re-recording. With our new AI-powered voice pipeline, recasting for CG took an hour.

Voice, art, and animation working together: meet James.

Impact

The demo did more than test our tech; it became our production engine.

Pipeline Validation
We fixed the breaking points on our own time. Real client work kicked off straight after, on the same pipeline.
Process Established
The demo's milestones became our operational plan, timelines and quality bar.
Sales Utility
Sales adopted Project Airline as a demo tool — proof to skeptical clients that our CG matched our live-action standard.
Tooling Adopted
The AI voice pipeline used here went on to be adopted across the studio, eliminating the hard cost of dialogue for fully synthetic characters.