Last time I ended on this: behind one request — "make the tutorials" — sit dozens of variations. Here's what that actually means.
What a "level" is
The shoot for AVT is broken into levels. A level is a walkaround of the vehicle built around one specific task: the lower ring around the wheels, the walkaround at mirror height, the walkaround with doors open, the interior walkaround, the upper ring. Nine levels per body type.
Why one level isn't one video
A level isn't a single video, because the shooting method itself depends on the vehicle's configuration: how many doors, where they're placed, how many rows of seats. A two-door car isn't shot the same way as a four-door. A cabin with one row isn't shot the same way as one with four.
Multiply levels by configurations, and that's how many videos one body type needs.
On the first body type, that math gave us 112 fragments. We shot 38.
How we cut 112 down to 38
Not every level depends on configuration.
On some levels, the difference between two doors and four doors is critical — those need separate videos. On others, that difference changes nothing about how the shoot goes. Those we merged into one shared fragment.
Getting there meant taking apart every step of the methodology and asking one question at each step: does anything here actually change with configuration — and if so, what?
From there we built the structure: which model is the baseline, how many variants branch off it, and where exactly they diverge. That structure is what we used to estimate scope, timeline, and cost.
What it would have cost without this
Timeline and cost would have scaled up proportionally — three times the videos, three times the work and the price.
But that's not the real problem. The real problem is maintenance.
A video doesn't just live once. The methodology gets refined, the interface changes, a step gets new on-screen text. Every time that happens, the fix has to go into every fragment it touches.
Every extra piece of content isn't just production. It's maintenance — change one detail, and you may have to redo everything.
At 112 fragments, any change means reworking the entire set. At 38, a chunk of the fragments never has to be touched at all, because they're shared across configurations.
Whose idea was this
Not ours. The optimization idea came from the client.
On the first body type, AVT handed us a finished breakdown — the full merge logic already worked out. We understood it and built it.
That didn't happen again. On later body types, the client gave us the configurations but not a map for how to collapse them. We worked that logic out ourselves: what exists, what actually affects the shoot, where things can be merged.
Every new body type, from scratch. Optimization doesn't carry over — vans have one logic, pickups another, trucks a third (with five cab subtypes), buses their own set of variants. RVs are the hardest case: every manufacturer builds the interior differently, so it's almost impossible to predict.
Same kind of work, every time starting from zero.
The system now runs to over two hundred video files — all in one consistent style, as a single product. It could have been several times that many, and the client would have paid for it.
Next time: how we built a visual language for the whole system — and why the vehicles in the tutorials had to be designed from scratch.