Security cameras have become commoditized — $30 buys a camera that would have cost hundreds ten years ago. But the spec sheets that cover these products use terms that look meaningful without explaining what they mean in practice. Understanding how cameras actually work helps you evaluate specs that matter versus marketing language that doesn’t.
Resolution: What the Numbers Mean
Camera resolution is measured in pixels. Common resolutions:
- 1080p (Full HD): 1920×1080 pixels — the current standard for entry-level cameras
- 2K (QHD): 2560×1440 pixels — 78% more pixels than 1080p
- 4K (Ultra HD): 3840×2160 pixels — 4× the pixels of 1080p
More pixels mean more detail — theoretically. In practice, resolution interacts with several other factors:
Lens quality and sensor size. A sharp lens on a larger image sensor outperforms a soft lens on a high-resolution sensor. Pixel count only matters if the optical system can resolve that detail. Many budget cameras advertise 4K sensors but pair them with mediocre lenses.
Compression. All security cameras compress video for storage and transmission. Aggressive compression (lower bitrate) destroys detail regardless of the sensor resolution. A 1080p camera at 4Mbps bitrate will look better than a 4K camera at 2Mbps.
Practical viewing distance. 4K provides meaningful advantages when you need to zoom into footage — identifying a face from across a driveway, or reading a license plate at distance. For a front door camera viewed at normal distances, 1080p is typically sufficient.
Storage implications. 4K video requires approximately 4× the storage of 1080p at equivalent quality. This significantly affects how many days of footage you can retain locally.
How Night Vision Works
Most security cameras advertise night vision, but there are three fundamentally different technologies behind that claim:
Infrared (IR) illumination is the most common type. The camera has IR LEDs around the lens that emit near-infrared light (typically 850nm or 940nm wavelength). This light is invisible to the human eye but detectable by the camera’s image sensor. The result is black-and-white footage in darkness. The 850nm wavelength produces a visible red glow from the LEDs at night; 940nm is invisible but typically provides a shorter effective range.
Performance metrics that matter: IR range (how far the LEDs illuminate), number of IR LEDs, and whether the camera uses “smart IR” that adjusts LED intensity to avoid overexposure at close range. Many cheap cameras wash out the image completely when a person walks directly in front of the camera at close range.
Color night vision uses a different approach: a large image sensor that’s highly sensitive to light, combined with a very wide aperture lens. These “starlight” sensors can produce color footage in conditions that appear dark to the human eye, by collecting and amplifying ambient light (streetlights, indoor light spillover). The result is color footage in near-darkness rather than black-and-white IR footage. Color footage provides significantly better identification capability — a red jacket looks very different from a blue one. Color night vision requires at least some ambient light; it doesn’t work in true darkness.
Integrated spotlights are a third approach. The camera activates a white LED floodlight when motion is detected, illuminating the scene with visible light and enabling full-color footage. This deters intruders and captures higher-quality footage than IR. The tradeoff: it alerts the subject that they’ve been detected, which isn’t always desirable.
Motion Detection: How It Distinguishes Events from Noise
Older security cameras used pixel-change detection: if a certain percentage of pixels changed between frames, an alert was triggered. This method generates enormous numbers of false alerts from moving shadows, lighting changes, trees in wind, and rain.
Modern cameras use more sophisticated approaches:
PIR (Passive Infrared) sensors detect changes in infrared radiation — essentially, heat signatures. A human walking into frame produces a different heat signature than a cloud shadow. PIR dramatically reduces false alerts compared to pixel-change detection, but it can miss very slow-moving objects or fail to detect people behind glass.
AI-based object detection uses machine learning models to classify what’s in the frame. The camera sends frames to an on-device or cloud-based model that identifies whether the motion is a person, vehicle, animal, or package. This enables notifications like “a person is at your front door” rather than generic motion alerts. The quality varies significantly — some systems are excellent at distinguishing people from pets at various distances and angles; others frequently misclassify.
Zones and sensitivity allow you to exclude specific areas (a busy sidewalk, a street with passing cars) from triggering alerts. This is often more effective than algorithm improvements for reducing nuisance notifications in specific environments.
Local Storage vs Cloud Storage
Where footage is stored has implications for cost, privacy, reliability, and access.
Cloud storage sends footage to the camera manufacturer’s servers. Advantages: footage is preserved even if the camera is stolen or damaged, accessible from anywhere, typically involves automatic retention management. Disadvantages: ongoing subscription fees (often $3–10/month per camera, or more for multi-camera plans), footage is controlled by a third party, dependent on upload bandwidth and internet connectivity, potential privacy concerns about footage leaving your premises.
Local storage keeps footage on-device (SD card) or on a local recorder (NVR — Network Video Recorder). Advantages: no ongoing fees after hardware purchase, footage stays on your property, works during internet outages (for motion detection and recording, if not remote viewing). Disadvantages: footage is lost if the camera is stolen (for SD card storage), manual management of storage capacity, no off-site backup.
Hybrid systems store motion clips locally while optionally backing up to cloud, or store continuous footage locally while sending alerts to cloud. This approach is increasingly common in prosumer cameras.
NVR systems connect multiple cameras to a central recorder that manages storage, provides a single interface for all cameras, and often supports local-only operation without any cloud dependency. They require more setup but provide better long-term value for multi-camera installations.
Encryption and Privacy
The camera’s video feed is a high-value privacy target. End-to-end encryption means footage is encrypted from the camera to the viewing app, and the camera manufacturer cannot access the contents even if they wanted to. Many consumer cameras encrypt the transport layer (HTTPS/TLS) but retain the ability to access footage on their servers. True end-to-end encryption is less common but increasingly offered by privacy-focused products.
For local-storage cameras, the main concern is whether the camera transmits any data externally and whether the local network interface is secured. Default credentials, unpatched firmware, and unnecessary cloud features that can’t be disabled are common security concerns with budget cameras.