The resolution is right, the lens is right, and the recording is still useless: the plate can’t be read, the face is blurred, the night image is snowy. The cause is often exposure. The shutter (exposure time), gain and iris trio is the group most often left at factory settings at commissioning, because in live view the image looks “good enough.” Below we work through how to calculate exposure for a moving target, the effect of gain on noise and bitrate, and starting values by scenario.
The exposure trio: shutter, iris and gain
Two physical parameters determine the light reaching the sensor: shutter (the time the sensor collects light for one frame) and iris (the lens aperture opening, the F-number). The third parameter, gain, doesn’t add light. It amplifies the signal from the sensor electronically.
| Parameter | When increased | Side effect |
|---|---|---|
| Shutter time (e.g. 1/250 → 1/30) | More light | Motion blur |
| Iris opening (e.g. F2.0 → F1.4) | More light | Depth of field narrows |
| Gain (dB) | Brighter image | Increased noise and bitrate |
Think of the amount of light in “stops”: one stop is light doubling or halving. Shortening the shutter from 1/30 to 1/250 is a loss of about 3 stops (8 times) of light. On the gain side, 6 dB means roughly doubling the signal; making up the same loss with gain alone requires about 18 dB of extra gain. Those 18 dB come back to the image as noise.
In auto exposure, the camera changes these parameters in its own order of priority to reach the target brightness. The order varies by manufacturer and exposure mode. On cameras with DC-iris or P-iris lenses, the iris usually opens first, then the shutter lengthens up to the defined upper limit, and gain rises last. With a fixed-iris lens, the first step is directly the shutter. This is where the night problem comes from: in the dark, the camera stretches the shutter to 1/25, or to 1/12 or longer if “slow shutter” is enabled. The image is bright, and everything that moves is blurred.
Why motion blur happens and how to calculate it
An object moving during the exposure leaves a trail on the sensor. The size of the blur in pixels is found with a simple multiplication:
Blur (m) = target speed (m/s) × shutter time (s)
Blur (px) = blur (m) × pixel density (px/m)
Walking person, 1.5 m/s, 250 px/m:
1/30 s → 5 cm → ≈ 12.5 px
1/250 s → 0.6 cm → ≈ 1.5 px
Vehicle, 30 km/h (≈ 8.3 m/s), 250 px/m:
1/30 s → ≈ 28 cm → ≈ 69 px
1/250 s → ≈ 3.3 cm → ≈ 8 px
1/1000 s → ≈ 0.8 cm → ≈ 2 px
In an image where identification is expected, the aim is for blur not to exceed a few pixels; for fine detail such as letters and digits, the limit is even lower. The calculation is based on the component of motion perpendicular to the image plane. A vehicle coming toward the camera moves less in the image than one passing horizontally in front of it. Positioning the camera to look along the direction of motion at a narrow angle gives a sharper image with the same shutter. Geometry, not settings. Determining pixel density is explained in the IP camera system design guide, and its relationship to the lens in the camera lens and focal length calculation article.
The relationship between gain, noise and bitrate
Gain amplifies the sensor’s read noise and photon noise along with the signal. At high gain the image turns “snowy”; random brightness fluctuations appear on flat surfaces. This has two consequences on site:
- Bitrate increase: H.264 and H.265 encoders encode the difference between frames. Noise is different in every frame, and the encoder treats it like motion. In VBR mode, bitrate rises; in CBR mode, bitrate stays constant but blocking and loss of detail increase.
- Side effect of noise reduction (DNR): 3D DNR suppresses noise by averaging across frames. At aggressive settings, it leaves a ghost trail behind moving objects and softens facial detail.
That’s why storage and network calculations should use the night bitrate, not the daytime one. On an NVR sized with the daytime value, retention time shrinks in the first winter. Stream profiles and bitrate modes are covered separately in the CBR, VBR and smart codecs article.
Settings table by scenario
The values aren’t a recipe; they’re reasonable starting points for testing on site. In interfaces, the same setting appears under different names such as “Exposure time,” “Max shutter,” “Shutter limit,” “Gain limit” or “AGC.”
| Scenario | Starting shutter | Gain approach | Note |
|---|---|---|---|
| Indoor corridor, office | 1/100 | Medium limit | Prevents flicker on 50 Hz mains |
| Pedestrian door, turnstile | 1/250 | Medium limit, additional light | Check the lighting on the face area |
| Loading dock, forklift area | 1/250–1/500 | Low–medium limit | Inside the dock may be dark and outside bright |
| Vehicle entrance with barrier (slow) | 1/500 | Low limit | IR or white light essential for plates |
| Road, fast vehicle flow | 1/1000 and shorter | Low limit | Usually requires a dedicated LPR camera |
Indoors, fluorescent and some LED luminaires flicker at twice the mains frequency; on Türkiye’s 50 Hz grid, that means 100 Hz. If the shutter time isn’t a multiple of this period, horizontal bands or brightness fluctuations between frames appear in the image. The camera’s “anti-flicker 50 Hz” setting or values such as 1/100 and 1/50 solve the problem. Outdoors in daylight, this constraint doesn’t apply.
Detailed design criteria for vehicle entrances and plate reading are the subject of the license plate recognition camera selection article; here we cover only the exposure side.
Day/night profiles and scheduling
A single exposure setting doesn’t work correctly both in the midday sun and under IR at night. Most enterprise camera systems allow separate image profiles to be defined for day and night. The profile switches in two ways:
- By light level: The camera switches to the night profile the moment it removes the IR-cut filter. If the threshold and delay (hysteresis) are wrong, it keeps switching back and forth (flapping) at dusk. In the recording, this shows up as color–black-and-white switching several times a minute.
- By schedule: The profile changes at set times. If the schedule isn’t updated during the year to follow sunrise and sunset, the night profile kicks in too early on summer evenings.
The typical approach for the night profile: maximum shutter is limited to the value in the scenario table, the gain limit is pulled down to a level where noise remains acceptable, slow shutter is turned off, and 3D DNR is lowered to a medium level. If the result is a dark image, what’s needed isn’t a setting but lighting. The IR range and external lighting decision is in the IR range and night image article. WDR mode also affects exposure time; with multi-exposure WDR enabled, additional artifacts may appear on moving objects.
If you’ll distribute settings with a VMS or automation tool, the ONVIF Imaging service defines exposure mode, minimum/maximum exposure time, minimum/maximum gain and iris parameters as standard. Manufacturer-specific features (slow shutter, smart IR, DNR levels), however, are often set only from the manufacturer’s own interface or API. For details, see the Imaging Service specification on the ONVIF website.
Testing and recording at commissioning
Whether exposure is correct is measured not in live view but on a single frame taken from the recording. In live view, the eye merges moving images and hides blur. A paused frame doesn’t lie.
- For each critical point, write down the target speed and the expected image level (observation, recognition, identification).
- In daylight, have a person walk past the target point at normal walking speed and record; at vehicle entrances, have a real vehicle pass at the expected speed.
- Step through the recording frame by frame, export the best frame at the target point and assess face/plate readability.
- Repeat the same test at night, under real lighting conditions. Full darkness, not dusk.
- Record the average and peak bitrate values for day and night from the VMS or camera statistics.
- Write the approved settings (shutter, gain limit, DNR, WDR, profile switch threshold) into the camera inventory and take a configuration backup.
This record makes it easier to restore settings after a firmware update or camera replacement; it’s also evidence of the test result at project acceptance. Firmware updates resetting image settings to defaults do happen on site; a setting without a backup has to be found all over again.
Common mistakes
- Leaving factory settings: Auto exposure chooses a long shutter at night; it produces blur at every point with moving targets.
- Forgetting slow shutter turned on: The image looks bright, but the frame rate actually drops and motion trails appear.
- Approving settings in live view: Blur only shows up on a paused frame.
- Leaving the gain limit at maximum and calculating storage with the daytime bitrate: Night recordings take more space than planned, and retention shrinks.
- Skipping the flicker setting indoors: Banding under LED and fluorescent lighting lowers recording quality.
- Copying the same profile to all cameras: Doors, corridors and docks have different speed and light conditions. Bulk copying is fast; it isn’t right.
If exposure is addressed at the project stage together with camera positioning and lighting decisions, commissioning gets shorter. To do this work across a facility, see our IP camera system design service.
Frequently asked questions
Why do faces come out blurred in night footage?
In the dark, auto exposure first lengthens the shutter time. At times such as 1/25 or 1/12, even a walking person moves a few centimeters during the exposure and facial detail is lost. The solution is to limit the maximum shutter time and make up for missing light with lighting.
If I fix the shutter at 1/1000, will the image go dark at night?
Without additional light, yes. Going from 1/30 to 1/1000 is a loss of about five stops (32 times) of light; the camera tries to make up for it with gain, that is, with noise. A short shutter gives a clean image only if there’s enough IR or white light on site.
Why does recording capacity shrink as gain increases?
Gain amplifies noise along with the signal. The encoder treats random noise like motion; in VBR mode the bitrate rises, and in CBR mode quality drops within the same bitrate. That’s why you should do the storage calculation with the night bitrate value.
Should exposure settings be redone when the camera changes?
Yes. Because the sensor, lens aperture and image processing differ, values can’t be carried from model to model. Even the same model needs different values at a point with different light conditions; run day and night tests for each point.