What is an IP camera? A plain English 2026 guide to IP vs analogue, resolution, PoE, ONVIF, bandwidth, security, and how one platform brings every camera together.
An IP camera captures, compresses and sends video as network data, so it can be managed and analysed with software.
The specifications that matter most: resolution, low light performance, lens and field of view, compression, and the weather rating of the housing.
Power over Ethernet carries data and power on one cable, the biggest installation advantage of IP over analogue.
ONVIF lets cameras and software from different brands work together, but always test the exact camera and firmware against your platform.
IP cameras are networked computers, so security is essential. Change default passwords, update firmware, isolate the network and keep video on premise where you can.
An IP camera is a networked digital camera that sends its video over the same kind of network your computers and phones use. IP stands for Internet Protocol, the addressing system that moves data across networks.
An analogue camera sends a continuous signal down a dedicated coaxial cable to a recorder. An IP camera does three things itself before anything leaves the device. It captures light through a lens onto a sensor. It compresses the picture into a digital video stream, usually H.264 or H.265. And it sends that stream across the network to a recorder, a video platform or a viewer.
Because the output is data rather than a signal, an IP camera can be reached from anywhere on the network, given its own address, and managed with software. That shift, from signal to data, is why IP cameras have become the default for new installations.
Analogue CCTV is not dead, but IP is now the standard for new systems, and the differences are practical.
An analogue camera sends a signal over coax to a DVR (digital video recorder), usually at up to about two megapixels, and needs a separate power cable. An IP camera sends a compressed stream over a network cable to an NVR (network video recorder) or software platform, commonly at two to eight megapixels or more, and can take its power from the same cable.
Interoperability changed just as much. Analogue systems were often tied to one vendor, while IP cameras that follow open standards such as ONVIF and RTSP can be mixed across brands on one platform. AI moved too: IP video can be analysed on the camera and across every feed in the video platform. Our NVR versus DVR guide covers the recorder side in depth, and our SD, HD and 4K guide covers resolution.
Light passes through the lens onto an image sensor. The sensor's size, pixel count and light sensitivity set the ceiling on picture quality; a larger sensor with larger pixels usually sees better in low light.
A processor in the camera handles exposure, colour balance, wide dynamic range for high contrast scenes, noise reduction and switching between day and night modes.
Raw video is far too large to send as it is, so the camera compresses it. The main formats are H.264 and the newer, more efficient H.265 (also called HEVC), both standardised by the ITU-T and ISO/IEC. H.265 can roughly halve the data needed at similar quality, which directly cuts bandwidth and storage.
The compressed stream is sent using a streaming protocol. RTSP (Real Time Streaming Protocol) is the classic choice in surveillance. A recorder or video platform receives it, stores it, and makes it available for live view, playback, search and AI.
Datasheets are long, but a few lines deserve most of your attention.
Resolution. Two megapixels (1080p) is the practical minimum. Four and eight megapixels give more detail and more room to zoom in, at the cost of more bandwidth and storage.
Low light performance. Stated as minimum illumination in lux; lower means the camera sees in darker conditions. For true darkness, move to infrared or thermal.
Lens and field of view. A wider lens covers more with less detail; a narrower one covers less with more detail. A varifocal lens lets you adjust, and a PTZ camera changes its view on demand.
Frame rate. Fifteen to twenty five frames per second is typical for surveillance.
Compression. H.265 with smart bitrate, which spends data only when something moves, is the biggest lever on storage.
Wide dynamic range. Balances bright and dark areas in the same picture; essential at entrances and windows.
Housing. A weather rating such as IP66 or IP67 (under IEC 60529) shows dust and water resistance; vandal resistance is a separate IK rating.
One of the biggest practical advantages of IP cameras is Power over Ethernet (PoE), standardised by the IEEE 802.3 working group. One network cable carries both the video and the power, so there is no separate power cable to each camera: the single largest installation saving of IP over analogue.
The common standards are IEEE 802.3af (PoE), delivering about 13 watts to the camera, which suits fixed cameras; IEEE 802.3at (PoE+), about 25 watts, for PTZ cameras, heaters and infrared lights; and IEEE 802.3bt (PoE++), roughly 50 watts or more for high power and multi sensor cameras.
A PoE switch or injector supplies the power. A standard network cable reaches about 100 metres before you need a switch or fibre, so you plan distance around switch locations rather than around each camera. Our PoE guide covers power budgeting in detail.
Cameras and recorders from different vendors once refused to talk to each other. ONVIF (the Open Network Video Interface Forum) fixed that with a common standard so cameras and video software work together regardless of brand.
You will see three profiles most often: Profile S for video streaming and PTZ control, Profile G for recording on the camera and retrieving it, and Profile T for advanced streaming including H.265 and analytics data. When a camera lists Profile S or T, a standards based platform can usually find it, pull its stream and control it without custom work.
One practical point: ONVIF support is a strong signal, not a guarantee every advanced feature works, because firmware quality varies between vendors and versions. Test the exact camera and firmware against your platform before standardising on it.
Choosing a camera type is about coverage and lighting, not brand.
Bullet cameras are visible and directional: ideal for perimeters and long corridors. Dome and turret cameras are discreet and harder to tamper with: ideal indoors. PTZ cameras pan, tilt and zoom to actively watch large areas. Fisheye and multi sensor cameras cover a very wide or full 360 degree area from one device. Thermal cameras see heat rather than light, for total darkness, smoke or long range perimeters.
Most estates mix several. Map each viewpoint to the coverage, detail and lighting it needs, then choose the type that delivers it.
Every IP camera uses network capacity and storage, and a quick calculation avoids expensive surprises.
Start with each camera's bitrate. A 4MP H.265 camera at 15 frames per second might average 2 to 4 megabits per second, depending on scene activity and smart bitrate settings. Multiply by the number of cameras: fifty cameras at 3 megabits per second is about 150 megabits per second of constant traffic, which is why surveillance usually gets its own switch or network segment.
For storage, a rough rule: gigabytes needed equals the bitrate in megabits per second divided by eight, times the seconds of retention, times the number of cameras. Retention, often 30 to 90 days, dominates the total. H.265, smart bitrate and motion based recording are the biggest levers on both network and storage.
IP cameras are networked computers, and therefore a target. The Mirai botnet of 2016, which hijacked hundreds of thousands of poorly secured cameras and recorders, remains the standing lesson.
Change default passwords and use strong, unique ones: defaults remain the most exploited weakness. Update firmware on a schedule. Put cameras on their own network segment with no direct internet access. Switch off services you do not use, such as universal plug and play, unneeded open ports and cloud relays you never asked for.
Check procurement rules too. US federal buyers and many enterprises restrict certain named camera vendors under NDAA Section 889, so check the current list before standardising. Above all, keep video on premise where you can. Keeping video and AI inside your own network removes a whole class of cloud exposure, which is why on premise and air gapped deployments remain the strongest choice for sensitive sites.
Cameras are only as useful as the platform that unites them. A video management system (VMS) finds cameras, records them, provides live view and playback, and, in the best platforms, watches every feed with AI.
Visylix is an automated AI video surveillance platform built in India. It watches every camera with 22 AI analytics built in house, decides what matters with Sudarshan automation, and acts: alerting the right people, raising incidents, talking down intruders through camera speakers and feeding your SIEM, PSIM and access control.
Every brand, one platform. Visylix connects standards based cameras over ONVIF for control and RTSP for video, with device packs for Hikvision, Dahua, Axis, Hanwha and Uniview that smooth out each brand's quirks. Its native streaming engine supports 13+ protocols, and bulk administration makes large estates quick to set up. Visylix also reads the archives of existing NVRs from Hikvision, Dahua, Uniview, I8H, Bosch and Panasonic, plus CP Plus, Honeywell, Sparsh and Godrej models, so moving to IP never means losing your history.
Fast and secure. Live view arrives in under a second over WebRTC, down to under 200 ms. Visylix ships as a Docker image on your own hardware, so video never leaves the building, and it runs on premise, at the edge, hybrid or fully air gapped. Active Directory and LDAP sign in, 46 permissions across 4 roles and a tamper evident audit trail come as standard.
Simple licensing. Starter and Pro include unlimited cameras on the VMS licence, from ₹4,999 or $49 per month. AI analytics are a separate subscription, billed per camera per month, on the cameras you choose.
The wider point: the camera you buy and the platform you run are one system. Choose cameras with genuine ONVIF support and open streaming, and you keep your options open.
Match each camera honestly to its job. Choose the right resolution for the scene: do not overspend on 4K where two megapixels will do, or underspend where you must read a plate or a face. Match the lens to the coverage you need, and check low light or infrared performance if the site is dark.
Prefer H.265 with smart bitrate to control storage. Confirm ONVIF Profile S or T support and test it against your platform. Check the weather and vandal ratings suit the environment. Make sure the PoE class fits both the camera and your switch budget, and choose a vendor that ships firmware updates and meets your procurement rules.
Finally, choose a platform that connects cameras over open standards and does not charge you more for every camera you add. Get those choices right and an IP camera estate grows cleanly from a handful of cameras to thousands.
IP stands for Internet Protocol, the standard way data travels across networks. An IP camera sends its video as network data rather than as an analogue signal. The same letters mean something else in a weather rating such as IP66 or IP67: there, IP means Ingress Protection, a dust and water resistance scale defined by IEC 60529.
CCTV traditionally means analogue cameras that send a continuous signal over coaxial cable to a DVR. An IP camera is digital: it compresses video itself and sends it over a network to an NVR or a software video platform. IP cameras usually offer higher resolution, power and data on one cable, and the freedom to mix brands.
No. IP cameras need a network, not the internet. Many secure sites keep cameras completely offline on an isolated network, with the video platform running on premise. You only need internet access if you choose remote viewing over the internet.
ONVIF is an open standard that lets cameras and video software from different manufacturers work together. A camera supporting ONVIF Profile S or T can usually be added to a standards based platform without custom integration, which is what lets you mix camera brands in one system.
It depends on resolution, frame rate, compression and how busy the scene is. A 4MP camera using H.265 at 15 frames per second commonly averages 2 to 4 Mbps. Multiply by the number of cameras to size the network, and give surveillance its own switch or network segment.
Yes, if they support open standards. ONVIF and RTSP let a mixed brand estate be recorded and viewed in one platform. Visylix connects standards based cameras this way, with device packs for the top five brands, and Starter and Pro include unlimited cameras on the VMS licence.
You need something to record and manage video, but it does not have to be a hardware NVR. A software platform running on your own servers can replace a fixed NVR and add AI across every camera at once. Visylix can even read the archives of existing Hikvision, Dahua, Uniview, I8H, Bosch and Panasonic NVRs, so you keep your history.