Video Surveillance

How video surveillance work is scoped and estimated: what drives the hours, what makes jobs harder than they look, and the effort patterns experienced estimators rely on.

How It Works & Technology Overview

An IP surveillance system: cameras (PoE-powered data producers) → network transport → recording (on-prem NVR or cloud) → viewing/analytics. Design order is objectives-first: define each camera's forensic objective (situational awareness ~20 px/ft, human recognition ~40–60 px/ft, positive identification/plate capture ~80+ px/ft at the target distance), then the environment (day/night, backlight, weather), and only then camera type, lens, resolution, and storage. Camera families: fixed domes (default indoor), bullets (outdoor reach, visible deterrence), turrets (dome optics without IR bounce-back), PTZ (operator-dependent — wasted without tours or active monitoring), multi-sensor panoramics (one drop covering four views — license and bandwidth math differ), and specialty (LPR with dedicated lens/positioning, thermal, explosion-rated). Recording architectures: on-prem NVR/VMS (client-owned hardware, local retention, IT-managed, ages) vs cloud VMS (subscription per camera, remote access, auto-firmware, no server — higher recurring, lighter head-end labor). Storage math baseline: ~1 TB per camera per 30 days of continuous 1080p; 4K roughly doubles it; motion-based recording and smart codecs (H.265) cut it substantially but verify against retention requirements, which may mandate continuous.

Common Deployment Scenarios

  • Small commercial (4–16 cameras): cloud VMS or compact NVR, existing network or a dedicated PoE switch, mobile viewing.
  • Warehouse/industrial: high mounting (lift logistics — measure aisle clearance and floor loading before renting), long exterior runs (conduit, media converters past 328 ft), harsh-environment housings.
  • Multi-site retail: standardized per-site kits, centralized cloud management, bandwidth-constrained uploads.
  • Enterprise/campus: VMS servers or clusters, dedicated surveillance VLANs/switches, integration with access control (video verification), retention compliance.
  • Analog-to-IP migration: existing coax (evaluate condition — encoders or IP-over-coax adapters vs re-pull), phased cutover keeping coverage live.

Products, Vendors & Characteristics

  • Cameras/on-prem VMS: Axis (premium, open ecosystem), Hanwha Vision, Bosch, i-PRO, Avigilon (Motorola), Uniview; Hikvision/Dahua carry NDAA Section 889 procurement bans — disqualifying for federal/government-adjacent clients and many insurers; check before quoting.
  • VMS software: Milestone, Genetec (enterprise standards, per-camera licensing tiers), Exacq, Avigilon; NVR appliances bundle licensing per channel.
  • Cloud VMS: Verkada, Rhombus, Eagle Eye, Avigilon Alta — per-camera-per-year subscription, direct-ship hardware, fast enrollment, tight IT/firewall coordination requirements instead of head-end builds.
  • Transport hardware: PoE budgets per camera class (IR heaters and PTZs draw multiples of a fixed dome — size on peak with headroom; keep switch PoE under ~80% loaded), industrial media converters and PoE extenders for long runs.
  • Analytics: on-camera (line-cross, object classification) vs server/cloud AI search; analytics promises need nighttime tuning to survive headlights and shadow play.

What Makes It Hard (Gotchas & Failure Modes)

  • Coverage promised without pixel math: "good coverage" in a proposal is a dispute after the first incident — put per-camera objectives (awareness/recognition/identification at distance) in the contract.
  • Retention discovered late: "we need 30 days" after a 2-week NVR shipped; regulated clients (cannabis, banking, education) carry mandated retention/redundancy — ask first, size storage second.
  • Network reality: 20 4K streams saturate uplinks that "have a fiber backbone"; multicast/VLAN isolation for camera traffic needs the network expert; cloud VMS upload bandwidth per camera matters on thin circuits.
  • No power/no network at the "perfect spot": the wireless compromise underdelivers; exterior conduit and trenching are the honest answer and the real cost.
  • Environmental failures: direct sun on housings, IR bounce off eaves, spider webs on IR domes, thermal throttle in sealed enclosures, cold-weather heater draw exceeding the PoE class.
  • Bargain-tier hardware efficiency penalty: unstable firmware and missing documentation turn 10-minute configurations into multi-hour struggles — the cheap camera is rarely cheap by commissioning.
  • Privacy/legal boundaries: audio recording is legally restricted in many states; cameras over neighboring property or sensitive areas create liability — flag during design.
  • Handshake scope creep: "just move this camera while you're here" changes pathing and lens work — change order, always.

Effort & Estimation Patterns

  • Standard interior IP dome (existing network port, standard mount): 1–1.5 hours installed, aimed, enrolled.
  • Outdoor bullet with new conduit run: 3–4 hours plus conduit sub-scope.
  • PTZ on pole/pan-tilt mount: 3–4 hours plus ~1 hour tour programming.
  • Multi-sensor panoramic: ~2 hours (one drop, four aims); LPR: half-day per lane including positioning trials and validation.
  • NVR/VMS head-end: 2–3 hours for appliance-class install and configuration; server-class VMS builds 1–2 days plus per-camera enrollment.
  • Cloud VMS per camera: 45–60 minutes (no head-end), plus a per-project IT-coordination block (firewall rules, VLAN) of 2–4 hours.
  • Focus/verification pass: 10–15 minutes per camera at night for IR/analytics-sensitive positions — budget the evening tuning visit explicitly.
  • Recurring: cloud licensing resale; maintenance agreements at ~10–15 minutes per camera annually (clean, re-focus, verify recording/retention) plus head-end updates; monitoring integration (video verification with the alarm category) priced per site.

These figures are planning baselines drawn from common industry practice, not a quote. Real hours vary with site conditions, crew, and scope. DarkBird exists because that variance is measurable: it calibrates your estimates against your own completed-project actuals, so your numbers stop being someone else's averages.

Stop estimating from averages

DarkBird turns this kind of knowledge into Install Cards calibrated against your own completed projects. 45-day free trial, full Pro access, no credit card.