Continuing the case study I started in the previous post.
To understand why "just shoot a video" wasn't an option, you need to see what actually happens when a partner shoots it wrong.
What the product does
Our client is AVT — they build seamless, interactive 3D tours of any vehicle, built from real photography instead of 3D models. The core feature: you can shoot anywhere — a parking lot, a street, any room. No studio, no special lighting. The result is a full interactive tour: walk around the vehicle, open the doors, look inside, even hear the engine run.
That only works if the shoot follows a precise, patented methodology. The camera goes in specific points, at specific levels, in a specific sequence. Deviate, and the tour doesn't stitch together.
Who's actually doing the shooting
The client has no in-house operators. Partners sign up through a form on the website, arrange access to a vehicle with a dealer, and shoot it themselves — no training, no supervisor on site.
If a partner shoots it wrong, the tour fails. The partner's time is gone. The dealer's vehicle and slot are gone. The client doesn't get a sellable tour — and doesn't get paid.
A mistake in a repeatable process doesn't cost what it takes to fix once. It costs however many times it repeats before anyone notices.
Why we ruled out live-action video immediately
Real footage depends on too many variables to control at scale. To make every tutorial look and feel like one consistent product, you'd need:
- Identical lighting on every shoot
- Real vehicles — meaning manufacturer agreements to use them
- An actor to demonstrate each action
- A location, equipment, and a full production setup
- Contingencies for everything that can go wrong: the actor's sick, the location falls through, and so on
CGI removes every one of those variables. Building a tutorial takes a handful of calls and iteration — no rentals, no logistics, no production days. If the methodology changes, or an error turns up, nothing gets reshot: we open the project, make the fix, and by the next morning every partner is working from the updated version.
What "correct" actually meant here
The bar wasn't "watch it and get the general idea." It was "watch it and know exactly what to do at every step" — for someone with zero training, zero supervision, and one shot to get it right.
Once we started breaking the methodology down in detail, we found something we hadn't expected: dozens of variations, each depending on the vehicle's configuration. That's next.