Choosing a low light CCTV camera

Over the years I’ve setup and run a number of CCTV camera systems using a variety of brands and NVRs (Network Video Recorders). Hikvision, Dahua, Reolink to name a few more common ones. As far as NVRs are concerned I have tended to focus on DIY solutions using the likes of BlueIris (for Windows) and Frigate (for Linux).

In reality, there is a lot to consider with implications for the various choices, the full depth of which are beyond the scope of this blog. However, I will aim to share some key areas to consider when choosing a CCTV camera, in particular one for low light / night vision.

Mega pixels and sensor size

When considering megapixels it is important to consider the use case, especially as it relates to low light and subject identification. With megapixels more is not always better, especially with smaller sensors.

Cramming lots of pixels in to a small sensor can be fine for day-time viewing (within reason), but at night when you need to boost the sensor signal to see in low light, noise becomes a problem. Noise is essentially electrical interference that when amplified appears as grain or distortion in the image. This noise is often worse on lower quality sensors and/or when too many pixels are packed in to smaller space and they interfere with each other. When you have fewer pixels, you also have more light per pixel, which again helps. As such striking a balance between the sensor size and pixel count becomes important.

When it comes to cameras suitable for low light, the following is a useful guide :

• 1/2.5" (or smaller) sensor with 2 or 3MP
• 1/1.8" sensor with 4 or 5MP
• 1/1.2" sensor for 8MP (4k)

Typically a 1/1.8" sensor is the sweet spot for price and pixels.

Lens aperture and shutter speed

When capturing video, getting as much light as possible on to the camera sensor, especially at night, is usually a good thing.

• The more natural light that falls on the sensor the less you have to boost the signal to get a bright image which means less noise.
• At the same time, the more natural light there is, the faster the shutter speed can be which helps prevent motion blur.

The lens aperture is measured in f-stops. Usually you want as wide an aperture as possible F1.0 being the most you usually achieve. When you go from F1.0 to F1.4 you get half the light and to F2.0, you get half again.

If you can get a good image with a shutter speed at 1/100 with an F1.0 lens, then with an F1.4 lens you will either need to half the shutter speed (1/50) or boost the gain (ISO) of the sensor. While increasing the gain might seem like a good way to keep the shutter speed high, it can cause significant degradation in the image quality in lower light. Hence wanting to keep the aperture as wide as possible to allow you to keep a relatively fast shutter speed in lower light.

Good cameras will give you control over gain and shutter speed, with options for night and day, including auto. This is very useful if you want to ensure the shutter speed doesn’t drop too low, or the gain get too high. Without such settings you don’t really know what the camera is doing and they might drop the shutter speed to 1/25 or even slower which is detrimental to capturing a clear image of an object in motion.

One point to note is that with a wider lens aperture the depth of field gets smaller, i.e the depth of the image which is in focus gets shorter. When you get that nice blurry background in photos (called bokeh), that is because parts of the image go out of focus with a wider aperture. This tends to be less of a problem with CCTV cameras because the sensor sizes are small and the distances involved, but its worth being aware the larger the sensor size and the wider the aperture the narrower this depth of field gets.

Focal length

In simple terms the focal length determines the field of view or “zoom”. A common mistake is to push for a wide angle lens to capture a lot of the scene then hope that a higher megapixel count will allow you to retain detail. Unless going for very high end cameras this is likely a false assumption, especially in low light and even then not a wise strategy.

In most situations, a more focused view with lower pixel count covering key areas (such as entry and exit points) will prove much more valuable.

Another important consideration is as you increase the sensor size, the relative field of view changes for the same focal length. The aspect ratio of the sensor (4:3 or 16:9 for example) will also play a role.

As a general guide when mounting a camera at 3 or 4 meters from the subject :

• 30°-40° portrait or Head-and-Shoulders
• 50°-70° full-body shot such as a doorway
• 80°-110° Capturing a larger portion of the room, showing more context and surroundings.

That said if you are free to position the camera the focal length becomes less of an issue as you can move the camera closer or further away to suit your needs.

A typical solution would be to use a camera to capture subjects a key entry / exit points then use a wider angle camera (which could even be a 180 or 360 camera) to get an overview of everything going on in a space, but don’t expect to get detail from this wide angle, especially in low light

White and Infrared Light

While low light cameras with a smaller sensor can give you an acceptable picture under street lighting, you need to be careful with this as it’s possible to get an image to appear brighter by reducing the shutter speed which will lead to blurring.

If light drops, a supplemental white light can help you maintain your shutter speed at a relatively high speed to ensure you get a sharp image, but once the light falls away, you may need to consider infra-red light.

Cameras can come with no light, white light, infrared light, or dual lights depending on the make and model. While you might think you don’t need any sort of light, you should consider specific conditions, such as if the ambient light might be a problem or fail, such as due to a power outage or failed bulb.

If low light is proving a problem, infrared is typically the best choice for maintaining imagine clarity in very low light settings, this is particularly true of cheaper small sensor cameras.

wired or wireless

Wireless network CCTV cameras can be beneficial but will still need power, some will plug in, use batteries or even solar. While these can be useful for some applications, the reliability and way they work with your system can be suboptimal either due to placement or reliability issues.

One benefit of wired ethernet cameras you can power them over the network cable using PoE (Power over ethernet). This means a single cable can run from your camera to a network switch which supplies power to your device. Your network switch will need to support PoE but there are basic models that are very affordable.

Wired ethernet cameras are likely the better choice for more serious systems and they will be more reliable than wifi cameras or those that rely on batteries.

On top of the benefit of a single cable and reliability, if you wanted to battery back-up your CCTV system, a relatively modest UPS (Uninterruptible Power Supply) can keep your cameras going during brief power outages. With a low power NVR (Network video recorder) and small number of cameras, even an affordable UPS can provide an hour or two of run-time should the power fail.

Storage / recording

For the purposes of this section we’re talking about network CCTV cameras where you own and control your footage locally. We are ignoring cloud based systems as they tend come with subscriptions, privacy issues, and limits on storage. Local systems will give you maximum flexibility and privacy, but that can come with a bit more setup overhead and initial cost, but ultimately it gives you greater flexibility and lower cost over time.

Typically many cameras come with SD card storage allowing clips to be recorded directly on the device. It’s important you use a high endurance SD card as they wear out over time due to the intensity of write operations.

CCTV cameras can usually copy footage to a central location using NFS or FTP protocols. If you have a NAS (Network attached storage) device this may be a quick and simple solution for you to centralize / backup your footage, but wont be as useful as when using some sort of NVR (Network video recorder).

RTSP (Real Time Streaming Protocol) is supported by many cameras and is likely the preferred way for recording streams centrally. Your NVR can connect to this stream and record the footage centrally as well as provide a friendlier interface for you to search and view your footage.

Recording footage centrally using an NVR is usually the preferred option, with the SD card providing a handy backup should the centralized storage be lost or interrupted. You may want to consider wider backup and resilience plans beyond this for more mission critical applications.

There are dedicated NVR units which support common standards such as RTSP and ONVIF, or for the DIY enthusiast you can use fairly moderate computer hardware to create your own and use popular software such as BlueIris (Windows) and Frigate (Linux). The key thing is to try and enure it is not propriety so that you can mix and match hardware.

Privacy masks and region of interest (ROI)

Privacy masks are used to obscure specific areas of a video frame from being recorded or viewed which is important for respecting privacy and regulations. When a privacy mask is applied, the selected area is either blurred, pixelated, or completely blacked out in the recorded footage. This ensures that certain parts of the scene, like private property, personal information, or sensitive areas, are not visible. It can also be helpful when it comes to filtering out areas where you might not want to detect motion and events.

Region of interest (or ROI) refers to an area within the camera image that is of particular interest. This area can be configured to allocate more resources, ensuring better clarity and detail in that part of the image. This is useful for focusing on critical areas while potentially reducing data usage and storage requirements for the rest of the frame.

Privacy masks and ROI may be features of interest when choosing a camera depending on your particular use case.

Motion and object detection

Most modern IP CCTV cameras come with some basic motion detection. This can be useful for triggering alerts and making more effective use of storage, by limiting recording to specific events. It can be a bit hit and miss, with false alarms or missed events. More advanced object detection uses image recognition algorithms to identify specific objects, such as people or cars, or events such as line crossing, items being left, or removed.

With NVRs such as using software like BlueIris and Frigate you will likely want to centralize your motion and object detection, but it might still be advantageous to use it on the cameras too to help with local on device storage.

It is still possible that motion and object detection can miss some key recordings so you should consider recording everything for a period of days and using motion/events to help with identifying and triggering events. You can retain footage for longer that do have events, but keep everything for the immediate period to ensure if something significant like a burglary happens you have everything.

More complex setups and retention schedules will likely require an NVR as cameras can only typically be configured in a fairly basic manner, but this may be suitable for many use cases.

Smart features and apps

Interoperability and compatibility for individual cameras as it relates to smart homes and apps can be very hit and miss. When it comes to the more professional cameras they are likely more robust and better for general CCTV applications but lack smart home features. At the same time, many cheaper cameras have variable user experiences.

Getting the right mix of features in this area and a high quality camera can be particular tricky. In reality, most people wont be looking at their CCTV regularly and many will likely lose interest in the events once the novelty wears off. In such cases you should focus on quality cameras and robust recording mechanisms.

If smart home features are a feature for you, it might be that you rely on your smart home or NVR to handle motion and detect events, rather than trying to get every camera to play nicely. In such cases, cameras that support RTSP streaming can likely be used with your NVR / smart home to detect motion and raise events.

Conclusion

Choosing a low-light camera can be hit and miss, especially with the volume of cheap cameras with false marketing claims, but if you stick to a few simple rules you will more likely have better success :

• Choose a camera with a moderate pixel count and larger sensor
• Go for a wide F1.0 aperture if possible
• Consider if you might need supplemental lighting and what sort

Then think about what further features and options you might need :

• Wireless or wired network camera
• What storage method works best for you
• Look for privacy masks and ROI features if needed
• Consider if motion and object detection is required