Convert RGB values to HSV

RGB describes a color by how much red, green and blue light a screen gives off in one spot. HSV describes the same color by three things that are easier to picture: hue (H), the position on the color wheel as an angle where red sits at 0°, green at 120° and blue at 240°, saturation (S), the amount of color, and value (V), how strongly the color shines[1].

Alvy Ray Smith described the model in 1978. He wanted people to pick screen colors the way a painter mixes paint: choose a pure hue, then lighten it with white or darken it with black. In his model, lower saturation means more white and lower value means more black[2].

The same model is often called HSB, with B for brightness[1]. That is the name Photoshop uses in its Color Picker, where hue is an angle from 0° to 360° and saturation and brightness are percentages[3]. If you need the same RGB color as a hex code, the RGB to HEX converter writes it out.

If your color is in another format, such as OKLCH, Lab or Android code, the color converter shows it in every format at once.

Divide each channel by 255. Value is the largest of the three results. Saturation is the gap between the largest and the smallest, divided by the largest. Hue depends on which channel is the strongest[4].

Formula: V = max, S = (max − min) ÷ max, and when blue is the strongest channel, H = 240° + 60° × (R − G) ÷ (max − min). With red on top the hue counts from 0°, and with green on top it counts from 120°[4].

Take rgb(59, 130, 246). Divided by 255, the channels are 0.231, 0.510 and 0.965. Blue is the highest, so value is 96.5%. The gap between the highest and lowest is 0.733, and 0.733 ÷ 0.965 gives a saturation of 76%.

Hue is 240° + 60° × (0.231 − 0.510) ÷ 0.733, which is 240° − 22.8° = 217.2°. The whole color is hsv(217.2, 76%, 96.5%). When all three channels are equal the color is a gray, saturation is 0 and hue has no meaning[2], so the converter puts 0 there.

Most often because OpenCV loads an image with its channels in blue, green, red order[9]. Convert an image from cv2.imread with cv2.COLOR_BGR2HSV, not COLOR_RGB2HSV. The mix-up changes the hue alone: the same three bytes 59, 130, 246 read as BGR give OpenCV a hue of 11 instead of 109, an orange instead of a blue.

The second trap is red. It sits at both ends of the hue scale, so pure red has an OpenCV hue of 0, while rgb(225, 0, 15), a red with a touch of blue, is already at 178. A range meant to pick red pixels has to cover both ends of the scale.