Color Versus Day/Night ShootoutBy: Antony Look, Published on Apr 16, 2011
Most people prefer color over black and white but when it gets dark out, almost all surveillance experts agree that black and white video provides superior performance. To that end, professional surveillance users generally choose Day / Night cameras over Color only cameras for night time outdoor applications.
A number of important questions, though, remain unanswered:
- How dark does it need to be before you should switch to black and white mode?
- How much better of an image does black and white mode provide over color mode?
- How much better is a Day / Night camera over a Color only camera?
- What is the impact on visual quality for people and vehicles?
To better understand these tradeoffs, we conducted a series of 4 tests using 6 IP cameras.
- 6 IP Cameras: We selected 2 color only cameras (an SD from Axis and a MP from IQinVision) and 4 day / night cameras (2 Bosch's - one at SD and HD, and MP Sony and Pelco Sarix cameras). We wanted a broad range of cameras to see how performance varied across camera types and camera manufacturers.
- 4 Test Scenes: We tested (1) in an unlit field at under 1 lux to understand challenges in poorly lit scenes; (2 and 3) in a parking lot with typical street lighting for vehicles and people; (4) Finally, we tested in an outdoors athletic complex at ~50 lux to see the impact of very well lit night time scenes.
The images below provide a visual overview of some of the tests we conducted.
The video below shows you the tradeoffs in action:
Video Clips Download (ZIP File)
Review our Color vs. D/N videos zip file [link no longer available] (~500 MB) containing all the video clips and images generated during our testing. The clips are embedded inside an ExacqVision player and simply needs to be double clicked to view. In the Exacq player's Options menu we recommend enabling 'Show Camera Names' to show the names of the cameras on-screen.
~1.0 Lux Color vs. BW Analysis
In this video we examine the results for tests performed in the ~1.0 lux open field. At this particularly low light level (<1.0 lux) BW mode is shown to clearly outperform color mode. Digitally zooming in to both human subject and the ensemble of license plates demonstrates that a better level of detail and overall clearer image is produced by the BW modes of TDN cameras. These results are generally expected, and most TDN cameras will automatically switch to monochrome under these lighting conditions.
In color mode testing all cameras (both color-only and D/Ns set to color-mode) appear to 'struggle' to produce a strong picture quality especially in terms of details on the human subject. Somewhat unexpectedly, D/Ns set to color mode, tend to outperform the color-only cameras at this ~1.0 lux level.
3 to 5 Lux Color vs. BW Analysis (Person)
In the video below, we examine the results for tests performed in the 3 to 5 lux parking lot. Interestingly, results at this light level generally show BW mode outperforming color mode. Thus, despite many TDN camera specifications citing color mode minimum illuminations much lower (e.g. 1.0 lux) than this 3 to 5 lux scene, our real-world tests show material benefits using BW mode. When TDN cameras are tested in BW mode, both human and license plate details improve versus color mode. Applying this finding in the real-world can prove challenging since most D/Ns will tend to operate in color-mode automatically at this light level. Thus, an ability to fine-tune the cut-over between color and monochrome mode must be provided by the camera and properly configured by the user.
In the color mode tests, no significant additional value is provided by color capability. In fact, in the darker scenes typical to this light level it is not so obvious that the D/Ns display any additional 'coloring'. While in color mode the Sony CH140 and Bosch NBN-921 (both MP) appear to provide best overall image quality compared to the rest of the field (color only Axis 1011 and IQEye 042SI; and the Pelco HD and Bosch SD D/Ns). At such a light level where scenes are dark, the additional color details (e.g., color of clothing or vehicles) are less likely to come through clearly enough to make a material difference.
3 to 5 Lux Color vs. BW Analysis (Vehicle)
In this video we revisit the 3 to 5 lux parking lot scene, but instead focus on a vehicle subject. Our first observation is that headlight glare or blooming effects appear more problematic in BW mode. This is evident across all TDN cameras from our study. However, both the Bosch NBN-921 (720p) and NBN-498 (D1) were able to mitigate the blooming phenomenon relatively better than the other D/Ns (Sony CH140 and Pelco Sarix). In BW mode, only the HD version of the Bosch provided a legible reading of the license plate through the glare of headlights. With headlights off, all the HD/MP D/Ns provide readable license plates.
~50 Lux Color vs. BW Analysis
In this video we analyze the differences and trade-offs of color vs. BW in a very well lighted (~50 lux) athletic complex. It is no surprise that in such a well lit environment TDN cameras will almost always switch to color-mode or stay in color-mode. It is also not surprising that our tests suggest that this is the 'right' behavior as monochrome mode did not produce any clear advantages in image quality. In fact, color mode brought our greater overall 'richness' to the scene - e.g. identifying color of clothing and other objects of interest.
A total of six (6) cameras are included in the Color vs. Monochrome Low-light Shootout. Four (4) cameras feature true D/N (IR cut-filter). Two (2) cameras are color only.
True Day/Night Cameras:
- Bosch NBN-921 [link no longer available] (online $750) - 1/3" CCD; F1.3 Bosch; 720p; 0.4 Lux (30 IRE, B/W); 1 Lux (Color)
- Bosch NBN-498 [link no longer available] (online $700) - 1/3" CCD; F1.3 Bosch, 4CIF; 0.06 Lux (30 IRE, B/W)
- Sony CH140 (online $800) - 1/3" CMOS; F1.2 Fujinon; 720p; 0.1 Lux (50 IRE, B/W); 0.2 Lux (Color)
- Pelco Sarix IXE20DN (online $1,300) - 1/3" CMOS; F1.2 Pelco; 720p; 0.25 Lux (Mono, 33 ms)
Color Only Cameras:
- Axis M1011 (online $160) - 1/4" CMOS; Integrated F2.0; 4CIF; 1.0 Lux (Color)
- IQEye 042SI (online $400) - 1/3" CMOS; F1.6 IQinvision; 2MP; 0.2 Lux (Color)
All six (6) cameras were simultaneously recorded to an Exacqvision VMS throughout three (3) low-light environments. Each of the three (3) environments were designed to simulate real-world low-light surveillance applications:
- ~1.0 lux open field with no direct lighting
- 3 to 5 lux parking lot with street lamps
- ~50 lux recreation area with liberal degree of lighting present
For each light level both an all-color mode test (all TDN cameras forced into color mode) and a monochrome test (TDN cameras forced into monochrome mode) are conducted. Each test involves a human subject carrying an ensemble of license plates at an approximate 20ft FoV. The subject moves across horizontally to demonstrate motion and momentarily stops in the center of the scene to present static facial and object details. Note that color-only cameras (Axis M1011 and IQEye 042SI) were run again in color mode during the monochrome portions.
The 3 to 5 lux parking lot scene features two (2) additional tests. The tests involve a vehicle rather than human subject. The vehicle is driven from background to foreground with head-lights turned on directly facing the cameras.
Here is a summary of all eight (8) tests:
- ~1.0 lux open field with human/license plate subjects - cameras all set to color mode
- ~1.0 lux open field with human/license plate subjects - D/N cameras set to monochrome mode (2 color mode cameras repeated)
- 3 to 5 lux parking lot with human/license plate subjects - cameras all set to color mode
- 3 to 5 lux parking lot with human/license plate subjects - D/N cameras set to monochrome mode (2 color mode cameras repeated)
- 3 to 5 lux parking lot with vehicle headlights - cameras all set to color mode
- 3 to 5 lux parking lot with vehicle headlights - D/N cameras set to monochrome mode (2 color mode cameras repeated)
- ~50 lux recreation area with human/license plate subjects - cameras all set to color mode
- ~50 lux recreation area with human/license plate subjects - D/N cameras set to monochrome mode (2 color mode cameras repeated)
Throughout all tests, all cameras are maintained at default settings.
Here are the key default settings relevant to low light performance for each camera:
- Bosch NBN-921 - 1/60s shutter; SensUp 4x; AGC on '15'
- Bosch NBN-498 - 1/60s shutter; SensUp 4x; AGC on '15'
- Sony CH140 - 1/30s shutter; View-DR/VE Off; AGC 'Middle'
- Pelco Sarix IXE20DN - 1/8s shutter
- Axis M1011 -Exposure Value '50'; Exposure Control 'Automatic'
- IQEye 042SI -1/30s shutter