Chroma subsampling is the practice of encoding a video’s color information at a lower spatial resolution than its brightness information. It relies on the fact that human vision resolves fine detail mainly through luminance, so reducing chroma resolution saves a large share of data with little visible effect on video quality in most natural images.
Subsampling is described with a three-part ratio such as 4:4:4, 4:2:2 or 4:2:0. It is applied in cameras, codecs, broadcast interfaces and streaming formats, and it is one of the oldest and most effective forms of data reduction in digital video.
How chroma subsampling works
Subsampling requires the image to be in a luma and color difference format, written Y’CbCr. The RGB signal is converted with a matrix defined by the relevant standard, such as Rec. 709 or Rec. 2020, into one luma channel (Y’) and two chroma channels (Cb and Cr) that describe the blue and red difference from luma.
The luma channel keeps every sample. The chroma channels are then filtered and decimated: neighboring chroma values are averaged and only one value is stored for a group of pixels. On playback, the decoder upsamples the chroma back to full resolution by interpolation before converting the picture to RGB for the display.
Because the Y’ channel carries most of the perceived sharpness, edges and textures survive. The cost appears only where color changes sharply on a small scale, such as thin colored text or the edge of a saturated object against a contrasting color.
Reading the J:a:b notation
The ratio describes a reference block four pixels wide and two rows high. The first number is the width of the block, the second is the number of chroma samples in the first row and the third is the number of chroma samples in the second row.
| Scheme | Chroma resolution | Data relative to 4:4:4 | Typical use |
|---|---|---|---|
| 4:4:4 | Full horizontal and vertical | 100% | Graphics, VFX plates, screen content |
| 4:2:2 | Half horizontal, full vertical | About 67% | Broadcast production, professional cameras |
| 4:2:0 | Half horizontal, half vertical | 50% | Streaming, Blu-ray, consumer cameras |
| 4:1:1 | Quarter horizontal, full vertical | 50% | Older DV formats |
The data percentages assume uncompressed samples of equal bit depth. With 4:2:0, a 1920 x 1080 frame stores luma at 1920 x 1080 but each chroma plane at only 960 x 540 samples, so the frame needs half the samples of a 4:4:4 frame.
The schemes also differ in chroma siting, which defines where the subsampled value sits relative to the luma samples. MPEG-2 and most HD formats place 4:2:0 chroma between two rows but aligned horizontally with the left luma sample, while some other formats center it. A decoder that assumes the wrong siting produces a slight color shift of half a pixel.
Chroma subsampling and other quality factors
Chroma subsampling is one of several separate losses in a video chain, and it is often confused with bit depth or compression. Each factor removes information in a different dimension.
| Factor | What it reduces | Typical visible defect |
|---|---|---|
| Chroma subsampling | Spatial resolution of color | Color bleeding at sharp color edges |
| Bit depth | Number of tonal steps | Banding in smooth gradients |
| Lossy compression | Detail judged least visible | Blocking artifacts and mosquito noise |
| Spatial resolution | Total pixel count | Overall softness |
Standard objective metrics treat subsampling unevenly. PSNR, SSIM and VMAF are usually computed on the luma channel only, so they can report high scores for material whose chroma has been reduced. Color-sensitive evaluation requires metrics on all three channels or careful visual inspection.
Choosing a subsampling scheme
The right scheme depends on what will happen to the picture later. Material that will be keyed, graded heavily or composited needs more chroma resolution than footage that goes straight to viewers.
| Situation | Recommended scheme | Reason |
|---|---|---|
| Green screen acquisition | 4:2:2 or 4:4:4 | Clean color edges for the key |
| Final streaming delivery | 4:2:0 | Supported by all players and hardware decoders |
| Screen recording with text | 4:4:4 | Prevents fringes on colored text |
| Broadcast contribution | 4:2:2 10-bit | Headroom for further processing |
| Archive or mezzanine master | 4:2:2 or 4:4:4 | Avoids repeated losses in later encodes |
The general rule is to subsample once, as late as possible. Every conversion between schemes involves filtering and interpolation, and repeated conversions add softness and color shifts that cannot be reversed.
Typical artifacts and mistakes
Subsampling artifacts are most visible on synthetic content, such as titles, user interfaces and saturated graphics, because natural images rarely contain hard single-pixel color edges.
- Colored text fringes: red or blue text on a dark background looks soft or blurred in 4:2:0.
- Keying edges: a chroma key on 4:2:0 footage produces blocky, stair-stepped matte edges.
- Chroma upsampling error: some players use poor interpolation, which adds jagged color edges.
- Repeated conversion: alternating between 4:2:0 and 4:2:2 in a pipeline adds cumulative blur.
- Assuming 4:4:4 means RGB: 4:4:4 can be Y’CbCr or RGB, and the two behave differently in processing.
Explained without jargon
Our eyes notice fine detail in brightness much more than in color. Chroma subsampling uses this. The video keeps every detail of light and dark, but it stores color more coarsely, for example one color value for every group of four pixels.
It is a little like a coloring book. The black outlines stay sharp, and the colors are filled in more loosely. For most pictures nobody notices, and the file becomes much smaller.
A typical use case
A production company shoots an interview series against a green screen. Its first test used a consumer camera recording 8-bit 4:2:0, and the keyed edges around the presenter’s hair looked blocky and showed green fringes. The team switched to a camera recording 10-bit 4:2:2, which doubles the vertical chroma detail compared with 4:2:0 and gives the keyer cleaner edges.
After compositing and grading in a 4:4:4 timeline, the series is exported as a 4:2:0 file for online platforms. Because the key was pulled from the 4:2:2 originals, the edges stay smooth in the final delivery, even though the viewer receives only a quarter of the original chroma samples per frame.