A two-storey waterfront events venue needed a camera plan that holds on a banquet night, in the rain and in an evacuation, in a hall whose layout changes for every event. Render rebuilt the venue as a twin from its drawing and its photographs, placed each camera only where it can really be mounted, and tested the plan the way a security consultant would: by simulation.
Two event floors, a ground storey over the water, a beach and a street.
Seating is never fixed. Banquets, theatre rows, standing receptions and an empty hall each bring their own crowd, and drapes dress the ceiling on event nights.
Where do the cameras go, and how far can the plan be trusted?
Not a coverage picture, but a decision: which mount, which aim, what it sees by day and by night, and what happens when something goes wrong.
The building rebuilt from the CAD drawing, floor by floor, and checked against photographs of the site.
doneCameras placed only on real ceilings, walls and poles, and aimed against each zone's requirement.
doneThousands of simulated intrusions per scenario, with an operator watching and a steward responding.
doneNoon, golden hour and night, rain, low sun and the compressed recording, measured on each camera's simulated view.
doneSix seating layouts with their crowds and their dressing, and an evacuation for each.
doneEvery input with no published value pushed to both ends of its range, to find where the decision would change.
doneThe steward arrives in time in about 97 runs in 100.
In the weakest scenario, across 2,000 simulated intrusions. With no cameras, none was interrupted.
Five degrees off costs less than one percent.
An installer can miss the aim on any camera by five degrees and the plan barely moves.
No camera loses its view to the dressing.
Mounting the indoor cameras just below the lowest fold of the drapes kept every view clear in all six layouts.
Response time, not the cameras.
The plan holds unless calling and sending a steward takes longer than about fifteen seconds. That is a staffing rule, not more hardware.
One gap, found on paper.
The far end of the main footbridge was barely seen. The twin showed it before installation, where moving a camera costs nothing.
Every camera on its own sheet.
Installers receive a drawing for each camera, its mount measured from the building's own columns, its bearing and its tilt, and a CAD file with every camera placed.
Each figure in the decision record is tagged: the drawing, a photograph, a published source, a measurement on the twin, or a simulation. What could not be measured is marked as assumed, tested across its range, and listed to confirm on site.
Nobody is identified. The plan works at the level of observing a scene, and it gives no compliance verdict.
Render is in early access. Tell us about the place you want to plan.