Starvis full-color camera technology

With the development of sensor technology, the low-light performance of security cameras has been greatly enhanced. In a dark environment, with the help of weak light, or even starlight, the camera can still shoot clear and colorful surveillance pictures. Haikang has a full-color series (ColorVu Series) cameras, and Dahua and Univision have launched full-color cameras. Security video surveillance has entered the starlight full-color era.

Compared with traditional infrared night vision, the advantages of starlight full-color cameras are:

  1. The color of the image is more than that of the infrared black and white image. Real-time monitoring, more information can be obtained after investigation and evidence collection;
  2. No need for additional infrared lights, no light pollution, better concealment;
  3. Without additional fill light, the power at night is relatively smaller, which is more conducive to POE power supply, and at the same time, the heat generation is smaller, and the camera and system are more stable and reliable;
  4. There will be no infrared overexposure, flashlight effect and other defects of traditional infrared cameras, so you can see more clearly.
  5. Some special scene applications. For example, in a parking lot at night, after the headlights are turned on, the infrared camera may not be able to see the license plate and the front of the car at all. The full-color camera perfectly solves this problem, and the license plate and the front of the car are clearly visible.

Before discussing the starlight full-color technology, let’s first understand a few basic concepts.


Illuminance is the luminous flux received per unit area. The unit is lux (lx=lux).

A candle illuminates an area of 1m2, and the brightness presented is about 1lux.

The following table is the illuminance value of some common scenes

Illuminance(LUX)Light conditions
100 000Scorching sun
50 000Operating room
10 000Clear sky
5Street lamp
0.2Full moon
0.02Moon night

Of course, the above illuminance value is only a reference value, not so accurate. Similarly, in the field of video surveillance, there is no clear standard for what illumination camera is considered to be starlight, which will be discussed in detail later. Generally, the sensitivity of the camera mainly depends on the lens and image sensor. The lower the specified lux value, the better the sensitivity of the camera.

Generally, manufacturers will specify the minimum level of illumination required for the camera to produce acceptable images. Although such specifications are helpful for comparing photosensitivity of cameras made by the same manufacturer, they are not so useful for comparing cameras from different manufacturers. This is because different manufacturers use different measurement methods and have different standards for generating acceptable images.


The concept of starlight camera has always been in security, but the name of a more professional system should be derived from SONY.

SONY named some of its own back-illuminated CMOS sensors with excellent low-light performance starvis series. This should be the source of the naming of the Starlight full-color camera.

For Starvis, Sony’s official interpretation is visibility under the starlight, which means visibility under the starlight. Of course, there is an explanation in the industry as starvision. Sony is registered with the starvis logo, and it is generally not allowed to be used freely without authorization. Therefore, many manufacturers have named them separately, such as the above-mentioned colorVU of Hikvision, Full-color of Dahua, LightHunter of Uniview, and Lightfinder of Axis.

Regarding Starvis technology, Sony defines it as follows: The STARVIS is back-illuminated pixel technology used in CMOS image sensors for security camera applications. It features a sensitivity of 2000 mV or more per 1 µm2 (color product, when imaging with a 706 cd/m2 light source, F5.6 in 1 s accumulation equivalent), and realizes high picture quality in the visible-light and near infrared light regions.

This technology is mainly used in back-illuminated CMOS sensors, because it can receive light from the backside of the sensor silicon substrate, increasing the amount of light entering, reducing light loss, and improving sensitivity.

A sensor with a sensitivity of 2000mV and above per 1µm2 area is called a starvis sensor. How to measure the performance difference of different starvis sensors? Sony has put forward the SNR1s concept.


In 2021, SONY updates starvis to starvis2. Starvis2 has a wide dynamic range (AD12bit) of more than 8dB compared to STARVIS for the same pixel size in a single exposure.

Full definition of STARVIS 2 is back-illuminated pixel technology used in CMOS image sensors for security camera applications. It features a sensitivity of 2000 mV or more per 1 µm2 (color product, when imaging with a 706 cd/m2 light source, F5.6 in 1 s accumulation equivalent). It also has a wide dynamic range (AD12bit) of more than 8dB compared to STARVIS for the same pixel size in a single exposure, and achieves high picture quality in the visible-light and near infrared light regions.


Sony is introducing SNR1s [lx] as an index used to quantitatively evaluate picture quality at low illumination. SNR1s [lx] is a proprietary index advocated by Sony, and is limited to CMOS image sensors for security camera applications.

A smaller value indicates better picture quality at low illumination.

SNR1s [lx] is an acronym consisting of “SNR” (Signal-to-Noise Ratio), “1” (represents that the signal level when noise = 1 is 1), and “s” (for Security).

SNR1s [lx] Prescribed Conditions

ApplicationSecurity Camera
Light Source3200 [K]
Target Object18% Grey
F number1.4
Exposure time1/60 [s]
Linear Matrixwithout
SignalG [e-] (Sensor Raw Output)
Noise√Shot Noise[e-] 2 + Dark Noise[e-] 2
Signal : Noise1 : 1
Note:SNR1s is a new index proposed by Sony for picture quality at low illumination, and is limited to CMOS image sensors for security applications.

Schematic of SNR1s Measurement Method

Two 3200 [K] light sources illuminate an 18% gray chart from different directions, and dimming is performed to 100 [lx]. A camera is placed so that the distance from the gray chart to the imaging surface of the image sensor is 1 m, the sensitivity [e-] and dark noise [e-] are measured, and SNR1s [lx] is calculated using the relational equation. The luminance when equation equals 1 is SNR1s [lx].

$SNR = \frac{G[-e]}{\sqrt{Shot Noise[-e]^2+Dark Noise[-e]^2}}= \frac{G[-e]}{\sqrt{G[-e]+Dark Noise[-e]^2}}$

SNR1s value of some common sensors of Sony

Four factors affecting the low illumination of the camera

  1. Image sensor. The photosensitive performance of the image sensor is one of the core factors that determine the low-light effect of the camera. In addition to the Sony Starvis series CMOS back-illuminated sensor described above, domestic manufacturers have also introduced some sensors with better photosensitive performance. The overall performance is still different from Sony. Relatively large, but cost-effective (cheap), and in line with the current domestic chip industry’s independent and controllable concept, so the market share is still very high, and the future can be expected!
  2. Large aperture lens. The larger the aperture, the more light is taken into the sensor through the lens, and the better the effect. The square ratio of the two adjacent aperture numbers in the full-level aperture value is 1:2, so the illuminance of the lens of the adjacent two aperture values ​​is doubled. That is to say, the light input of F1.0 aperture lens is twice that of F1.4 lens. For details, please refer to the article Lens in the Security Camera. Another thing to note is that the larger the lens aperture, the smaller the depth of field.
  3. Processing chip. The noise reduction and automatic gain capabilities of the ISP in the video encoding chip will affect the final imaging effect.
  4. Algorithms such as low illumination enhancement. When the performance of all hardware is the same, the final decision of the low-light performance of the product is the algorithm of each manufacturer and the optimization capability. For example, through a certain AI algorithm, the input incomplete information is restored. Some even use dual sensor mode, a black and white one can provide more details, and a color one can provide color. Then the two images are combined to enhance the camera’s low-light capability.
0 replies

Leave a Reply

Want to join the discussion?
Feel free to contribute!

Leave a Reply

Your email address will not be published. Required fields are marked *