[MUSIC PLAYING]

Welcome to the TI Precision Lab series on light sensing. My name is Rahland Gordon, and I am an applications engineer for Texas Instruments optical sensors. This video will explain some of the differences between RGB and XYZ color sensors and give insight on how to choose between the two for your application.

The two color sensors most typically used are RGB and XYZ color sensors. Because of their differences, choosing between the two can have a distinct variation and performance of your application. RGB color spaces use an additive process that combines red, green, and blue to represent any color within that space. What color the RGB value translates to is determined by the color space, such as DCI-P3 or SRGB, because each color space is different and just point to the primaries for each RGB channel.

This is true for all light producing or additive color processes, but in the case of sensing color, there is no universal standard. Each manufacturer has their own specifications for RGB [? spectrum ?] sensing curves, so the RGB value doesn't have a meaning in a sensing space. A mathematical transformation is needed to convert the manufacturer's specific RGB values sensed to a standard CIE-XY or UV color space.

This transformation introduces inaccuracy, and not all visible colors are represented. Because of this, along with other reasons, the CIE 1931 XYZ color space standard was created in order to represent all visible colors. This standard is absolute and independent of a device or manufacturer variation, making it more accurate and more sensitive to color variation than a sensor based on an RGB color space.

In addition, the y channel of an XYZ color sensor has a photopic response, making it simple to extract a value of lux or brightness. Based on the use case, using an RGB color sensor or an XYZ color sensor will produce different results and considerations. Ambient light color sensing is one of the most common use cases for color sensors.

Since, in most cases, the ambient light will be some form of white light, Correlated Color Temperature, or CCT, is most likely the desired output. The CCT can be used to adjust the display's color profile to ensure that the display's images seem natural or as they would be seen on a white piece of paper.

RGB color sensors, because there is no universal standard, often requires some sort of calibration to convert the RGB values to CCT. Some manufacturers provide a matrix or a simple linear equation to convert to CCT, but these vary in accuracy. In addition, non white light will often get mapped to the closest CCT and provide invalid results. In the RGB color space, there is no easy way to determine if the color point should be mapped to the CCT or not.

XYZ color sensors can easily and accurately be mapped to the nearest CCT value without calibration. To determine if the color points should be mapped to the CCT, it is common practice to convert the xyz coordinates to the UV color space. With a simple equation, you can determine the distance of the color point from the ideal blackbody curve, and invalid values can be ruled out. Nonetheless, if a cover material is used that shifts the color point, both RGB and XYZ color sensors will require calibration.

There are many applications, such as overhead lighting color control, where a color sensor is simply used as feedback to keep the color constant. These applications will be an example of relative color matching. For example, light sources such as LEDs change color under different conditions, such as temperature, driver current, PWM dimming, and aging.

Many lighting systems add blue and/or red LEDs to offset any changes in the color from the White LEDs. If the color sensor shows that the color point has shifted from the original factory calibration, the colored LEDs are added until the factory calibration and the color sensed match again.

In relative color matching applications, both RGB and XYZ color sensors can be used effectively without the need for calibration. Since only one color point is needed to match to, and it was measured in the same color space, an XYZ sensor's ability for brightness measurements does not offer a significant advantage.

There are also applications where measuring an exact color point is needed. In this case, the color is not subjective to a relative color point, but requires an objective match between two colors without comparing them directly. These applications require absolute color matching.

For example, paint will change in color over time due to aging, exposure to heat, and sunlight. It may be needed after some time to get a new paint that matches the current color. If the color sensor uses its own color space, such as with an RGB color sensor, there's no way to describe what the color of the paint is so that the machine can make a copy, making it oftentimes a non workable solution, and could require a complex calibration and be limited in accuracy.

If an XYZ color sensor is used, then the color is represented in a universally agreed upon standard. Even if the paint manufacturer works in a different color space, there should be an existing way to convert to the desired color space.

In many color sensing applications, the lux value of the light is also desired. This can be used in conjunction with the color value to set the display brightness and color to ensure the best possible user experience.

For RGB color sensors, sensing lux is more complicated and limited. Typically, no channel of the RGB color sensor provides a full photopic response, so a complicated calibration and matrix is required and will vary in accuracy across the color gamut.

With an XYZ color sensor, the y channel has a photopic response, and is linear to lux. The manufacture of the color sensor usually provides a ratio, or a simple single point calibration can be used to find the ratio between the y channel and lux.

In summary, both RGB color sensors, like TI's OPT 4060, and XYZ color sensors, like TI's 4048, can sense different colors and distinguish the differences between them. RGB color sensors do not represent the entire visible spectrum, and are dependent on light source and type unlike XYZ color sensors.

For most applications, either an RGB color sensor or an XYZ color sensor can be used. However, in some applications, there are advantages and disadvantages to using either color sensor type. For the most part, while usually more cost effective, RGB color sensors require additional calibration and transformations while being less accurate when compared to an XYZ color sensor.

To find more light sensor technical resources and search TI products, please visit the link shown. Thanks for taking the time to watch this video. Please try the following quiz.

What color sensor type would be best suited for measuring the brightness and correlated color temperature of an overhead light? The correct answer is a, an XYZ color sensor. While, in theory, an RGB sensor could be calibrated to do this task, the XYZ color sensor will provide both a simpler way of measuring the lux and converting the reading to a color temperature.