Every year, somewhere around the second week of November, the questions I get change. There are not more of them. They are different. A property manager pulls a clip from the night before and tells me the camera looks worse than it did in September. A warehouse manager asks why the truck sitting in the lot is a pair of headlights and a shape in the dark. Nobody touched the system. It is the same equipment on the same settings it was running in August.
What moved was the sun. On September 1 in Atlanta the sun sets at 8:03 in the evening, and there is still usable light in the sky until about 8:28. On November 2, the Monday after the clocks change, the sun sets at 5:44 and the light is gone by 6:10. That is two hours and nineteen minutes taken off the end of your day across nine weeks. Those times come from the United States Naval Observatory for downtown Atlanta.
Your building does not move with them. The last shift still ends when it ends. The cleaning crew still arrives when it arrives. The dumpster still gets rolled out at a quarter past six. Those events begin the fall in full daylight and finish it in the dark, and the only thing that changes in between is how much your cameras can tell you about them.
I founded Verified Security in 2007 and have spent nearly two decades designing, installing, and servicing commercial security systems across metro Atlanta. September is the month to look at this, because every fix below is cheaper and less urgent now than it is in December, and because you still have the better part of two months to make it.
The short version. Night vision rarely fails because it is dark. It fails because the scene is lit unevenly, because the camera quietly slows its shutter to compensate and smears anything that moves, and because infrared gives back a flat monochrome picture that blows out the reflective surfaces you most want to read. All three show up in about twenty minutes, with two people and a flashlight. None of them show up on the live view of an empty lot, because each one only costs you something once there is a face, a plate, or a moving person in the frame.
What Actually Changes Between Now and Thanksgiving
Three things move at once, and none of them touch your equipment.
Sunset walks earlier by a little over a minute a day through September and October. Gradual changes are the ones nobody notices. Then daylight saving time ends on Sunday, November 1, and an hour of evening light disappears in a single night. A door people were using at dusk on Friday is a door people are using in the dark on Monday.
The second change is the trees. Leaf cover comes off through late October and November, and a camera that has been looking at a wall of green since April suddenly has a sightline it did not have. That works in both directions. You may pick up the view of the back fence you have wanted for two years. You may also pick up a streetlight, a neighboring wall pack, or a road full of headlights pointed straight into the lens.
The third change is your own calendar. From Thanksgiving into the first week of January, most commercial buildings run a different pattern: reduced staff, contractors working off hours, deliveries at odd times, and stretches of days when nobody is on site at all. More of what happens at your building happens after dark, and more of it happens with no one there to see it.
There is a fourth, and it does touch your equipment, which is why it is the one to check first. It takes two minutes. Your recorder has to move its own clock on November 1. If it is set to the wrong time zone, or its daylight saving option is switched off, or it has no time source to check itself against, every timestamp it writes after that Sunday is an hour off from the one on the police report, the badge reader, and the phone in somebody’s pocket. Pull up any camera on the morning of November 2 and confirm the time in the corner is the time on your watch. That single check protects the timestamp on everything the recording is later asked to prove.
The Problem Is Almost Never Darkness
When somebody tells me a camera is bad at night, I have learned to go look at the lighting before I look at the camera. Buildings that struggle after dark are usually not dark. They are patchy.
A camera has to settle on one exposure, and the brightest thing in the frame pulls that exposure toward itself. Put a wall pack directly above a door and the camera sets itself for the pool of light under that fixture. A person standing four feet outside the pool is now a silhouette, and a silhouette has no face, no lettering on the shirt, and no color. The camera did exactly what it was designed to do. The lighting was simply laid out before anyone considered that a camera would be reading it.
The same thing happens when a vehicle turns into the lot with its headlights toward the camera, and again in a lobby that is bright inside and watched at night through a glass storefront. In each of those cases the picture would improve more from moving or shielding one light fixture than from replacing the camera underneath it. Newer cameras with wide dynamic range handle that contrast better, and they are worth having. They narrow the gap without closing it, and the fixture is still the cheaper fix.
The test costs nothing. Stand where the frame is brightest, walk to where it is dimmest, and have somebody watch the recorded video while you do it. If you vanish somewhere in the middle, you have a contrast problem, and no amount of resolution is going to solve it.
The Trade Your Camera Makes Without Telling You
This is the one that surprises people, including people who have watched their own cameras for years.
When light drops, a camera has three ways to keep the picture from going black. It can open the iris, the opening in the lens that lets light through, which it has almost certainly already done. It can raise gain, which brightens the image and adds noise with it. Or it can hold the shutter open longer, gathering more light into every frame.
On default settings, most cameras lean on the third one, and on an empty scene it works beautifully. That is exactly why the live view looks fine. You pull the lot up on your phone at nine at night, the pavement is sharp, the striping is crisp, everything looks healthy. Nothing in the picture is moving.
Now put a person in it, walking at a normal pace. With the shutter held open, that person is a smear across the frame. The face is gone. The lettering on the jacket is gone. Whether they were carrying anything is a matter of opinion. Nothing is broken and nothing is corrupted. You are looking at motion blur, and it appears only in the frames you actually care about, which is how it sits on a system for years without anybody catching it.
There is a setting for it. A minimum shutter speed tells the camera it may not slow past a certain point, and moving subjects stay sharp. It has a cost: the picture gets darker, because you took away the camera’s easiest way to brighten it. That is why the honest order is light first, settings second. Give the camera more light to work with and the trade stops hurting.
What Infrared Does Well, and What It Does Not
Most exterior cameras sold today have infrared illuminators built into the housing. That is what security camera night vision means on a spec sheet, and it is genuinely useful. The illuminators are all but invisible to the eye, so they do not bother a tenant or a neighbor, and in real darkness they will hand you a picture where there was none. They also carry four properties worth understanding before you lean on them.
Infrared is monochrome
The camera drops color and gives you gray. That matters more than it sounds like it should. “Male, dark jacket, jeans” describes a very large number of people. “Male, red jacket, jeans” narrows it considerably. Color after dark requires visible light, and there is no setting that recovers it later.
Infrared falls off quickly
The distance printed on a specification sheet describes how far the illuminator throws under favorable conditions. A person in dark clothing, in a drizzle, at the far end of a lot is a much harder subject than that. Treat the rated distance as the outer bound of an ideal case, not as a working range.
The plate you most want to read
License plates are built to be retroreflective. That is the entire point of them: a plate throws headlight light back toward the driver behind it so it can be read at night. A camera with infrared illuminators sitting an inch from its own lens is, from the plate’s point of view, a set of headlights. The plate returns that light into the camera far more efficiently than the car around it does, the camera meters for the scene as a whole, and the plate overexposes into a white rectangle with the characters washed out of it. This is a well-documented problem in plate recognition work, where it goes by the name halation. It is why plate capture belongs to a dedicated camera with its own optics, mounting angle, and exposure settings, rather than to the general-view camera that happens to point that way. Illuminators that dim themselves for close subjects help with a face at the door. They do not make a plate readable, because the plate is still returning far more light than the car around it.
The same physics catches high-visibility safety wear. If your yard crew wears reflective striping, expect that striping to flare under infrared the way the plate does, and the person inside it to go dark.
Reflections that are already in your building
Infrared bounces. Put a camera behind glass and its own illuminators reflect off the inside of that glass back into the lens, and the whole frame turns to white haze the instant the camera switches to night mode. The same thing happens under a tight soffit, or when a wall, a column, or a downspout sits inside the first few feet of the view. The near object lights up, the camera meters for it, and everything past it drops to black. Those are mounting problems, and they are cheaper to correct than they look. Get the camera outside the glass, stand it off the wall by a few inches, and turn it so nothing close sits in the frame.
The Small Things That Only Show Up After Dark
- Spiders and insects. Bugs collect around a warm housing, spiders follow the bugs, and a web strung across an infrared illuminator lights up like a snowstorm. In my experience it is the most common single cause of unusable night video, and of motion recording that fires all night for no reason anybody can find in the morning.
- A dome that looks clean at noon. Pollen, road film, and hard water spray leave a haze that daylight punches straight through. Infrared does not. It scatters inside that film and softens the entire image. Cleaning a dome is a five-minute job with the best ratio of improvement to cost of anything in this article.
- Focus set in daylight. A camera focused at noon was focused with the iris closed down. At night the iris opens, depth of field shrinks, and the camera can go soft on exactly the plane you cared about. If a camera has only ever been serviced during business hours, its focus has never been checked in the condition it does half its work in.
- Headlights and motion recording. A camera pointed anywhere near a road starts recording every passing car once headlights are on, which they are not at four in the afternoon in September and very much are at six in December. If your recorder keeps video until its drives fill and then writes over the oldest, one camera that triples what it writes shortens the window for every camera sharing that recorder. That is worth a look alongside how many days your recorder actually holds.
Everything Else You Bought Sits on Top of This Picture
Video verification depends on a person at a monitoring center being able to look at what your camera sent and say out loud what is happening in it. I have written before about why police may not respond to an unverified burglar alarm in metro Atlanta. That whole approach rests on the clip being legible. A smear crossing a dark lot verifies nothing.
Analytics are the same story. What an AI camera can actually do for a business comes down to whether the software can tell a person from a shadow and a vehicle from a reflection. Feed it a blown out, blurred, monochrome frame and detections get worse while false alerts climb, and people stop trusting the alerts. That loss of trust is the expensive part, and it is harder to reverse than the picture that caused it.
Access control has its own version. When a door is forced open at 6:40 on a December evening, the value of tying that door event to the camera above it is that somebody can see who was standing there. And the season when this matters most is the same season when contractors and cleaning crews start arriving in full dark instead of at dusk.
None of that is an equipment problem. It is a picture problem, and a picture problem is fixable.
The Twenty-Minute Walk
Do this once, this month, before the clocks change. It takes two people, a phone, and a flashlight you do not switch on until the very end.
- Pick three cameras, not thirty. Your main entrance, your primary vehicle approach, and whatever door your people actually use after hours. If those three are right, the rest of the system is a much smaller conversation.
- Go out after the light is fully gone. Through September that means after 8:30 in the evening. By the first week of November it means after 6:15. Do not run it at dusk. Dusk is the easy case, and it will tell you everything is fine.
- Watch the recording, not the live view. This is the step people skip and it is the step that matters. Have the second person play back what was just recorded, at export quality, on an ordinary screen. The live view is a different stream and it flatters the system.
- Walk it normally. Do not stand still and pose. Walk toward the camera, across it, and past it at the pace a person actually walks. Motion blur only appears in motion.
- Wear what your people wear. If the crew wears dark work clothes, wear dark work clothes. A white polo shirt will pass a test that a navy hoodie fails.
- Park a car where a car parks and try to read the plate off the playback, then off a single still frame pulled out of it, because a still frame is what actually gets handed to somebody.
- Write one line per camera. Not a paragraph. Just what you could not see. “Cannot tell hair color at the front door.” “Plate unreadable past the second row.” “Person disappears between the light and the fence.” That list is your whole scope of work, and you built it in twenty minutes.
Now switch the flashlight on and point it at the spot where you disappeared. If that alone fixes the picture, you have a lighting job rather than a camera job, and lighting jobs are the cheaper of the two by a wide margin.
What to Fix, in the Order I Would Fix It
- Clean the domes and clear the webs. Free, immediate, and it improves more night images than any other single action on this list.
- Move light before you move cameras. Re-angling one wall pack so it washes an area instead of pooling underneath itself will often do more for a night image than a full generation of camera upgrade. Adding a fixture is the next cheapest step.
- Put white light where you need a face or a color. At entrances, at the door where cash or keys change hands, at the spot where somebody would stand and wait. Infrared is fine for coverage. It is not fine for description.
- Bring the camera closer and lower. A camera at nine feet covering one doorway will give you a usable face. That same camera at twenty feet covering an entire elevation gives you a person-shaped object. Height buys coverage and spends identification, and after dark that trade gets considerably worse.
- Set a shutter floor, then look again. That is the minimum shutter speed from earlier. Once the light is right, this is the setting that converts good light into a sharp moving subject.
- Then talk about equipment. If you have done the first five and a camera still cannot do the job, it is the wrong camera for that position, and you now know exactly which position and exactly why. That is a far better place to buy from than a catalog.
One more thing that only matters after dark. Exterior lighting runs on building power, while the security system runs on its own batteries. In an outage the cameras can stay up while the area they watch goes completely black, which is a very different experience in December than it is in July. It is part of what a power outage actually does to a business security system.
Frequently Asked Questions
Why does my security camera look fine at night on the live view but blurry in the recording?
To hold brightness in low light, most cameras slow the shutter. On an empty scene that works, which is why the live view looks sharp. Put a person in the frame walking at a normal pace and that person becomes a smear, so the blur appears only in the frames you actually care about. Setting a minimum shutter speed corrects it, at the cost of a darker picture.
Can infrared security cameras read license plates at night?
Usually not. Plates are retroreflective by design, so they throw a camera’s own infrared light straight back into the lens far more efficiently than the vehicle around them does. The plate overexposes into a white rectangle and the characters wash out, an effect known in plate recognition work as halation. Reliable plate capture needs a dedicated camera with its own optics, mounting angle, and exposure settings.
What is the cheapest way to improve night security camera footage?
Clean the domes and clear any spider webs first, because a web strung across an infrared illuminator whites out the entire image. Then re-angle the exterior lighting you already have so it washes an area instead of pooling under the fixture. Those two steps cost almost nothing and repair more night images than replacing a camera does.
Does a video recorder change its own clock for daylight saving time?
Only if it is configured to. If the recorder sits on the wrong time zone, has its daylight saving option switched off, or has no time source to check itself against, every timestamp it writes after Sunday, November 1 will be an hour away from the police report, the badge reader, and everyone’s phone. Confirm the on-screen time on the morning of November 2.
How do I test my security cameras at night?
Go out after the light is fully gone, which means after 8:30 in the evening through September and after 6:15 by early November. Have a second person play back the recording rather than watch the live view. Walk toward and across each camera at a normal pace, wearing what your staff wear, then write one line per camera describing what you could not see.
Where I Would Start
Put forty minutes on a calendar this month, after dark, with one other person. The walk itself takes twenty. The other twenty is finding the playback and getting everybody outside. Walk the three cameras that matter. Write the one-line notes. If the list comes back clean, you have bought a quiet winter for the price of one evening. If it does not, you have a specific written scope of work in September, when a lighting change is a simple job instead of an emergency.
If you are not sure whether what you wrote down is a lighting problem, a settings problem, or a camera problem, call us at 678-924-7480 or get in touch here.
Current as of September 2026. We review this post annually.
Scott Hightower founded Verified Security in 2007 and has spent nearly two decades designing, installing, and servicing commercial security systems across metro Atlanta. Verified Security is an authorized Honeywell security provider and specializes in access control, video surveillance, intrusion, and fire alarm systems. Reach Scott’s team at 678-924-7480.