We will not quote a vision program before visiting the site. This is what those two days consist of, and why they change the plan more often than not.
Optics and lighting
Working distance, angle, field of view, and what the light does across a full day. We photograph the station under every condition it runs in, including the third shift, which is where most surprises live.
Mounting and environment
Where a camera can physically go, what vibration reaches it, and how hot, dusty or wet that location gets. This drives the hardware enclosure and often the whole approach.
The process itself
How fast parts move, where the decision has to happen, and what mechanism already exists to act on it. If there is no reject mechanism, that is a different project than the one being discussed.
Network and controls
What controllers are installed, how the network is segmented, and what the controls team will and will not permit. Their constraints shape the design more than any preference of ours.
The people
We spend time with the inspectors doing the work today. They know which defects matter, which are cosmetic, and which the paperwork does not capture — and their judgement becomes the annotation schema.
What changes most often
Two findings recur. Lighting needs to be controlled and enclosed, which adds cost nobody budgeted. And the defect classes as written on the quality sheet do not match what inspectors actually reject on, which has to be resolved before annotation rather than after.