You bought a 4K security camera. In daylight the image is crisp enough to read a license plate across the street. At 10 PM that same camera shows a grainy grayscale blob walking up the driveway, and the face is unrecognizable. You check the spec sheet — it says “IR night vision up to 100 feet” — so what’s going wrong?
Night footage is where the gap between marketing numbers and real-world usable video gets widest. Here is why your camera looks terrible after dark, what the actual variables are, and the three or four fixes that make the biggest difference.
The core problem: there is no light
A camera doesn’t create an image; it records the light that reaches the sensor. In daylight that’s trivial — there are millions of photons per pixel per second hitting the glass. At night the sensor is starved. Everything that goes wrong with night footage is downstream of that fact.
Cameras handle low light in three ways:
- Switching to infrared (IR) mode— the IR-cut filter slides out of the way, IR LEDs on the camera turn on, and the sensor captures near-infrared light the eye can’t see. This is the classic black-and-white “night vision” image.
- Cranking up the gain (ISO) — amplifying the signal from each pixel. More signal, also more noise. This is where the grainy, dancing-speckle look comes from.
- Slowing the shutter — exposing each frame longer. More light per frame, but anything moving blurs into a smear.
Every camera does some combination of the three. The quality of the footage depends on how well the sensor, lens, and firmware balance them — and how much help the camera gets from the install.
Reason #1: the sensor is too small
Low-light performance is dominated by sensor size, not megapixels. A 1/1.8” sensor gathers roughly 2.5x the light per pixel of a 1/2.8” sensor at the same resolution — that’s the difference between a usable image at dusk and a noisy mess.
This is also why cramming more megapixels onto a small sensor makes night video worse, not better. An 8MP camera on a 1/2.8” sensor has smaller, noisier pixels than a 4MP camera on the same sensor. We went deeper on this tradeoff in the megapixels guide — the short version is: if you primarily care about night performance, pick the camera with the bigger sensor, even if the resolution is lower.
UniFi’s AI Pro, AI Theta, and the G5 Pro lineup all use larger sensors specifically because night footage is where cameras are judged, not daylight.
Reason #2: IR is blown out or falling off
IR illuminators on the camera body look impressive in the product photo but create two common problems:
The “IR hotspot” on nearby walls
If the camera is mounted under a porch eave six feet from the front door, the IR LEDs blast the nearby wall and porch ceiling. The auto-exposure meters off that bright surface, the sensor steps the exposure down, and anyone standing ten feet away in the darker driveway disappears into black. The close-in geometry is completely wrong for built-in IR.
Fall-off past 30–40 feet
That “100-foot IR range” spec is measured with nothing in the frame except a human-sized target at the exact distance being advertised. In real life, IR intensity falls off with the square of distance. Usable identification-quality footage from built-in IR is more like 25–40 feet, not the advertised number.
The fix: external IR illuminators or white-light cameras
For long driveways, large yards, and commercial parking areas, a separate IR illuminator mounted away from the camera (so the light comes from a different angle than the lens) transforms the image. Cameras with built-in white-light spotlights — UniFi G5 Flex with floodlight, UniFi AI Theta — solve this differently by turning the scene into full color and triggering the light on motion.
Reason #3: glare, reflections, and dirty glass
Three physical issues we see on almost every “my night video looks bad” service call:
- IR bouncing off glass or plastic— the camera is mounted behind a window, or inside a dome housing, and the IR LEDs reflect straight back at the lens. The result is a washed- out center and black edges. Cameras should never shoot through glass. Ever.
- Spider webs on the lens — spiders love the warm glass of an outdoor camera at night. A single web strand across the lens catches IR and creates a glowing streak that blinds the sensor. Wipe camera housings during each seasonal maintenance pass.
- Headlights and streetlights in the frame— auto-exposure meters off the bright light, stopping the sensor down, which plunges the rest of the frame into black. The fix is usually repositioning the camera or tilting it down a few degrees so the light source is just outside the frame.
Reason #4: the bitrate is too low
Camera manufacturers love to quote resolution numbers but quietly ship firmware that compresses night footage aggressively to keep storage costs down. Noise is the enemy of every video codec — grainy night footage compresses poorly, so the codec either spends more bits (large files) or smears the detail away (mushy, blocky image).
On most NVR platforms you can raise the bitrate manually. On UniFi Protect, the “Enhanced” quality setting nearly doubles the bits per second allocated to each stream, which makes night video noticeably sharper at the cost of storage. If you have disk space, use it — we covered the storage math in the UniFi Protect vs Ring vs Nest post.
Reason #5: the shutter is too slow for motion
If your camera shows ghostly, smeared people walking up the driveway at night, the shutter is slowed to 1/15th or 1/8th of a second to gather more light. Anything moving more than a few inches per frame blurs.
Most camera firmware lets you set a minimum shutter speed — typically 1/30th or 1/60th of a second. Faster shutter means less light per frame, so the image gets grainier, but moving subjects stay sharp. For security footage that’s the correct tradeoff — a clear noisy image of a face is always more useful than a clean blur.
Reason #6: white-light “Smart Detections” aren’t triggering
Modern cameras (UniFi AI series, Reolink TrackMix, Amcrest AD) can switch to full-color mode when they detect a person or vehicle, triggering an integrated spotlight. Full-color nighttime video — with a white LED filling in the scene — is dramatically better than IR for identifying faces, clothing colors, and vehicle paint.
The catch: the detection and spotlight have to fire before the subject has passed the camera. A spotlight that turns on three seconds too late only lights up the back of someone’s jacket as they walk away. That’s usually a motion-zone tuning problem or a detection-model issue — not a hardware limitation.
The short checklist to fix your night footage
If your existing cameras look bad at night, work through these in order:
- Clean the lens and housing (spider webs, dust, water spots on the dome).
- Aim the camera so it isn’t pointed at a streetlight, headlight path, or reflective surface.
- If shooting through glass, re-mount the camera outside. This is non-negotiable.
- In the NVR, raise the bitrate (to “Enhanced” or equivalent) and set a minimum shutter of at least 1/30s.
- If the subject is beyond 30 feet, add an external IR illuminator or upgrade to a camera with a white-light spotlight.
- If none of that works, the sensor is the bottleneck. Upgrade to a camera with a larger sensor (1/1.8” or larger).
What we actually install
For homes across the Wasatch Front, the cameras that give us consistently good night footage without hand-holding are UniFi G5 Pro, G5 Flex (with the floodlight accessory), and AI Theta. All three pair with UniFi Protect, record locally on PoE in your network rack, and don’t require a monthly cloud subscription to keep the footage.
For driveways longer than 40 feet — common on Alpine, Highland, and mountain-adjacent properties — we almost always pair a bullet camera with an external 850 nm IR illuminator mounted away from the camera body. It’s the single biggest upgrade you can make for long-distance night identification.
Bottom line
Cameras don’t see in the dark, they fake it. Good night footage is the result of a big sensor, a clean lens, thoughtful positioning, enough light (IR or white), a fast enough shutter to freeze motion, and a bitrate that doesn’t compress the image into mush. Miss any one of those and you get the grainy, blurry, unusable footage that makes homeowners replace perfectly good cameras thinking the hardware is broken.
Keystone Integration designs and installs security camera systems across Draper, Cottonwood Heights, Park City, and the rest of the Wasatch Front — with sensor sizing, illumination, and bitrate dialed in during a free site survey. See our full service list or reach out if your existing cameras look great in daylight and terrible at night.