The short definition
A fisheye camera packs one large sensor (typically 1/1.7-inch or larger) behind a hemispherical lens that captures roughly 180 degrees in every direction. From a ceiling mount, that's the entire room below. The raw image is a circle inside a square frame with severe edge distortion; dewarping software corrects it mathematically into normal-looking flat views.
Top-tier fisheyes from Axis, Hanwha Vision, Hikvision (NDAA-blocked for federal work), Bosch, and Avigilon ship at 6MP, 12MP, and 16MP. The higher-resolution variants make dewarping practical: more raw pixels means each virtual camera still has usable resolution.
How dewarping actually works
Dewarping is a coordinate transformation. Each pixel in the output maps to a pixel in the raw fisheye source via a lens-specific projection function. Three modes cover most use cases:
- Panoramic (180 or 360 degrees). Single horizontal strip showing the entire field of view. Useful for retail aisle ends, school corridors, and atriums.
- Four-quadrant (quad). Four virtual flat cameras facing four directions, like four PTZs from one mount. Useful for office bullpens and open-plan rooms.
- Virtual PTZ. One pannable, tiltable, zoomable view that the operator drives inside the captured 360. Live look uses a single virtual view; the full 360 is always recorded so historical PTZ is possible.
Dewarp can happen on the camera (saves VMS CPU; locks the dewarp mode at recording), in the VMS (more flexible; recording stays raw), or in the client viewer (most flexible, raw recording, slight client-side performance cost). Genetec, Milestone, Avigilon, and Verkada all support client-side fisheye dewarp on major brands.
Fisheye vs multi-sensor
Same coverage goal, different tradeoffs. Fisheye uses one sensor and one lens with software dewarp: lower hardware cost, but it loses pixels to the dewarp so per-face resolution drops at distance. Best indoors under 14-foot ceilings. Multi-sensor uses 3 or 4 sensors and lenses for true edge-to-edge resolution with no dewarp distortion, at typically 1.8 to 2.5x a fisheye from the same vendor. Best for outdoor corners, atriums, and high ceilings. See the multi-sensor entry.
Resolution and the per-face pixel question
Per-face pixel math separates a fisheye install that works from one that doesn't. Forensic capture typically requires 80 pixels between the subject's eyes. On a 12MP fisheye:
- 12MP fisheye, 12-foot ceiling. Subject at 8 feet from the mount: roughly 100 pixels between eyes. Forensic-grade.
- 12MP fisheye, 14-foot ceiling. Subject at 12 feet from the mount: roughly 70 pixels between eyes. Below forensic threshold; suitable for situational awareness only.
- 6MP fisheye. Halve those numbers. Use 6MP for indoor coverage where forensic capture isn't the goal.
For sites where the camera plan needs forensic capture at distance, scope multi-sensor or supplement the fisheye with a fixed bullet at the entry point.
Where fisheye is the right call
- Open-plan offices. One 12MP fisheye covers a 1500 sq ft bullpen. Replaces four traditional cameras at one-third the cabling cost.
- Retail mid-aisles. Center-aisle ceiling mount captures both sides of the aisle and the cross-aisle ends. Pairs well with vertical bullets at the registers.
- Conference rooms and classrooms. One fisheye at the ceiling center covers all participants. Good for incident review and room utilization analytics.
- Hospitality lobbies. Aesthetic dome form factor; one cable run; full 360 coverage. Pair with bullet at the entrance for face capture.
When to ask Tec-Tel about fisheye selection
Per-face pixel math is the part installers get wrong. We'll walk a building, mark up the ceiling plan, and tell you whether a 6MP, 12MP, or 16MP fisheye covers the zone or whether you need a multi-sensor instead. Free scoping call.