How Well do IP Cameras Work in Low Light?

JH
John Honovich
•Published Jun 06, 2009 04:00 AM

[2016 Update: Camera technology has changed substantially since 2009. The same pattern occurs but cameras are generally much better in low light. For newer tests on this, see:

IP cameras have a bad reputation of working poorly in low light conditions. It's part of a greater problem with all surveillance cameras - as critical as low light performance is to video surveillance, high quality low light performance can be difficult to achieve.

There's no way to get this from manufacturers as they provide cryptic (and often overstated) illumination statistics. Plus, they usually provide still images that mislead more than inform.

Over the last two months of IP camera testing, the importance of measuring and understanding tradeoffs of camera settings has become clear.

As we examined in our report on lux meters, measuring light levels is crucial. Image quality degrades over a range of light levels (almost always higher than what the manufacturer states). You need to determine what those light levels are and how bad the image quality becomes at each light level.

Secondly, significant video quality tradeoffs exist in manufacturer's low light optimizations. These functions can distort moving objects, introduce significant artifacts and change the appearances of objects.

This premium report digs into these elements, explaining and sharing actual video samples that demonstrate these issues over a wide range of light levels.

Image Quality Degrades

Image quality in low light conditions does not suddenly go from good to bad. It is not as if at 2 lux, the image quality is perfect and at 1 lux, the image is entirely black.

Usually, image quality degrades over a wide range of lux levels. Unfortunately, a lot of surveillance video needs to occur within the degradation range where capturing video is neither ideal nor impossible.

For example, let's say at .5 lux, the camera generates an all black image - it cannot see anything useful. By contrast, at 50 lux, the camera generates its best image (an image that's as good as it is at 500 lux or greater). Between 50 and .5 lux, the video quality continuously degrades.

The degradation ranges from problems in seeing fine details (at the high end) to only being able to see the outlines of large objects (at the low end).

Determine the Camera's Objectives

How low a light level a camera can handle depends on the surveillance objective of the camera:

  • If all you need is to see if something or someone is there, you will be able to use a light level at the low end of the range. The picture won't be pretty but it will be good enough.
  • If you want to see the details of someone's face or specific actions that a person is making, you will need a light level at the higher end. Lower levels will have noise or blurring effects that can distort those details.
  • If you want to do video analytics, you will likely need a higher level. The noise and distortions found at lower levels and with low light enhancements used can significantly increase false alerts.

Tradeoffs of Low-Light Optimizations

Manufacturers offer a number of functions to increase low-light performance. However, these increases usually have negative side effects that users need to carefully consider. Here are the 3 general categories of low-light optimizations:

  • Black and White mode: Cameras provide better low light quality when switched to B&W mode. This is accomplished by the use of a mechanical cut filter and are usually labeled as being Day/Night cameras. [Note: a few cameras provide 2 imagers - one B&W and another for color]. This is the one feature that works the best with the least tradeoffs. Black and white mode almost always provides a substantial better image than the color mode at the same light level. The obvious tradeoff is capturing color.
  • Slowing Shutter speed/ lengthening exposure: If there's very little light, one way to get more light is to take a longer time to capture the image. Instead of using a 1/30 second exposure, it could be lengthed to 1/5 second or even 1/2 second. The two significant drawbacks of this is: (1) reduces frame rate and (2) causes issues with motion blur. If a car or person is moving quickly across the camera, the long exposure can distort details - rendering the image potentially useless. Manufacturers obscure this problem by showing still images of stationary objects.
  • Increasing Gain / enhancing images: Manufacturers can artificially enhance images to increase details in the image. The problem is that these techniques routinely introduce noise (e.g., the appearance of 'snow' or lines in the image). These artifacts can create new problems, especially if you are attempting to see fine details or perform video analytics. In our videos below, these issues can be seen very clearly.

Analyzing Video Samples

In the screencast below, we present and comment on the series of sample videos recorded:

Note: The camera tested a day/night model with a mechanical cut filter that was rated to .05 lux minimum illumination (that's all the details provided on the data sheet). Obviously, looking at the test results, there's no usable video at any lux level close to .05 lux. However, this camera was one of the better ones tested.

Download Video Samples

All of the video samples from the low light test may be downloaded: