Color depth (computer graphics)
The color depth determines an essential property of raster and vector graphics : the differentiation of all brightness and color values.
Basics
The color and brightness values of digital images are stored within the smallest unit of each image: for raster graphics within each pixel , for vector graphics within each color-defined vector. Each image unit contains a specified number of the maximum possible gradations (for example, for an average digital photo: 256 gradations per color channel of a pixel) as well as the specific color and brightness information (on the scale of these specified gradations).
The number of possible gradations is not necessarily synonymous with the number of possible colors. A distinction is made here between the number of color channels and the scope of the color table. Only the relationship between the type of color definition (color channels and their number, tables, ...) and the specification of the gradations (in bits) results in the maximum possible color depth.
Color differentiation
The maximum possible amount of (color) gradations is specified in bits and thus designates the color depth of an image. These gradations represent a scale on which the actual color information is stored. The color depth is the mathematical basis of the actual color information. In practice, an image never has the amount of colors that the range of this scale (color depth) makes available.
A color depth of 1 bit would mean that exactly two states would be possible in each color channel (usually red, green and blue on the computer screen). As an example, this would be black and red for the red color channel . With a color depth of 2 bits, 4 states would be possible, for example black, dark red, medium red and light red . With the usual color depth of 8 bits, 2 8 = 256 states and thus just as many individual red tones are possible.
The most common is the RGB color space with 8 bits per channel , corresponding to (2 8 ) 3 = 16,777,216 (approx. 16.8 million) theoretically possible colors. With 16 bits (per channel) this results in 281,474,976,710,656 (281 trillion) color possibilities.
Color tables
Images with indexed colors are a special form: in these, the data structure of a pixel does not contain the colors themselves, but an index to an entry in the color table. Instead of a color table, one often speaks of a color palette. The color depth indicates the maximum number of usable entries in the color table. In practice, color tables with 1 to 8 bpp (= bits per pixel) are used, corresponding to 2 1 = 2 to 2 8 = 256 simultaneously coded or representable colors. 1 bpp is common for images that contain only black and white.
Examples:
- The GIF format allows 1 to 8 bpp.
- PNGs with a color table allow 1, 2, 4 or 8 bpp.
- The EGA - graphics standard allows 4 bpp.
application
Digital photos usually have a color depth of 24 bits. In practice, of course, there is no photo that really has all the individual colors - the average is well below that. The advantage of the 24-bit color depth is clearly evident in almost every photo, as can be seen from a comparison with a 16-bit version of the same photo. The photo with 16-bit color depth shows recognizable, often very annoying podium patterns with color transitions that are no longer visible with 24-bit color depth. The advantage of the higher color depth is less in the maximum number of possible colors, but rather in the greater color differentiation.
Most computer monitors can only display 8 bits per channel. In professional photography and medical applications, 16 bits per channel are also required. Extreme areas of brightness (deep black shadow and glaring light) cannot be saved with 8 bits. For this a drastic reduction of the contrast range and the contrast differentiation is necessary. In order to make this change in the dynamic range visually appealing, high dynamic range images (high-contrast images) are used, which are scaled down to 8 bits for display using tone mapping . This process is a special form of image optimization .
With the ITU-R recommendation BT.2020 (Rec.2020) for Ultra High Definition Television (UHDTV) with up to 8K image resolution , a color depth of 10 or 12 bits per channel was specified, i.e. an application of 30 or 36 bits on the screen. These 30 and 36 bits are included in the HDMI specification 1.3, as well as 48 bits. With Displayport , 30 bit to 5K image resolution from version 1.3 are possible. The use of the color space of Rec. 2020 for high dynamic range video is given with HDMI 2.0 and Displayport 1.4. Dolby Vision introduced 10 or 12 bits per channel (in the format war with HDR10 +), and HDMI version 2.1 knows a transmission with 14 bit RGB, which supports this extended color space.
Color depth | Name / use | Coding | Number of displayable colors |
---|---|---|---|
1 bit | Monochrome | No clear assignment | 2 1 = 2 |
4 bit | Used with EGA graphics cards | No clear assignment | 2 4 = 16 |
6 bit | Used by the Amiga computers for HAM and Halfbright mode |
No clear assignment | 2 6 = 64 (thanks to the special HAM mechanism, however, up to 4096) |
8 bit | Used by the MSX 2 computers | Red: 3 bits Green: 3 bits Blue: 2 bits |
2 8 = 256 (due to the special HAM8 mechanism on the Amiga, however, up to approx. 2 million) |
12 bit | Used in several NeXT workstations | Red: 4 bit Green: 4 bit Blue: 4 bit |
2 12 = 4096 |
15 bit | Real Color | Red: 5 bits Green: 5 bits Blue: 5 bits |
2 15 = 32,768 |
16 bit | High color | Red: 5 bits Green: 6 bits Blue: 5 bits |
2 16 = 65,536 |
24 bit | True color | One byte each (8 bits) for R, G and B | 2 24 = 16,777,216 |
24 bit color + 8 bit alpha |
True color with 8-bit alpha channel | One byte each (8 bits) for R, G and B and α | 2 24 = 16,777,216 |
30 bits | Deep Color , HDR Video , e.g. B. Internal color depth in flatbed scanners | 10 bits each for Y, U and V | 2 30 = 1,073,741,824 |
36 bits | Deep Color, HDR10 +, Dolby Vision, for example high quality photography | 12 bits each for R, G and B | 2 36 = 68,719,476,736 |
42 bits | Deep color, such as high quality flat screen televisions | 14 bits each for R, G and B | 2 42 = 4,398,045,511,104 |
48 bit | Deep Color, for example high quality flatbed scanners | 16 bits each for R, G and B | 2 48 = 281,474,976,710,656 |
In Scan -, cinema, TV and print area 40 and 48-bit color depths are also more at 30, 32, 36, frequently such. Sometimes only in the internal processing (i.e., for example, the hardware of a scanner works internally with 30 bits to read in an original, but the scanner then only outputs the scanned image to the computer with a color depth of 24 bits).