What Is Chroma Subsampling?

Chroma subsampling is a way of saving space in an image or video. The file keeps the brightness detail of every single pixel but records color detail for only some of them, because the human eye barely notices the difference.

Where you find it and where it comes from

Chroma subsampling is used in video, in digital images and in color television, and it affects the visual information portrayed in them. In practice you meet it in JPEG photos and in almost every video file, whatever the container: MP4, MOV, AVI, MKV.

It only works if the file keeps brightness and color apart. Instead of storing red, green and blue for each pixel, these formats store luma, the brightness, separately from chroma, the color, as two components usually written Cb and Cr. Subsampling then targets the chroma alone and leaves the luma untouched. The choice is made by whatever encoded the file: the camera or phone, the export preset of a video editor, a screen recorder, or the program that saved the JPEG. A report may present it as YCbCr4:2:0 (2 2) for a photo, or as a pixel format such as yuv420p for a video stream. Those are two ways of writing the same thing.

Reading the notation

The value is typically displayed as YCbCr followed by three numbers, and those numbers describe a block of pixels four wide and two tall. The first number is the width of that block, the luma reference, which is not affected by subsampling and is why it is almost always 4. The second says how many color samples are taken across the top row of the block, and the third how many across the bottom row, where 0 means the bottom row simply reuses the color of the row above it.

• 4:4:4 - one color sample per pixel, no subsampling at all
• 4:2:2 - color at half the horizontal resolution
• 4:2:0 - color at half the horizontal and half the vertical resolution, a quarter of the color samples, and by far the most common setting for consumer photos and video
• 4:1:1 - color at a quarter of the horizontal resolution

Why it matters for your file

The gain is size and bandwidth. Fewer color samples mean fewer bytes for the same frame, which saves storage space and makes uploading, streaming and broadcasting cheaper. That is why cameras and streaming services settle on 4:2:0.

The cost is usually invisible, because the eye is far less sensitive to fine color detail than to fine brightness contrast. Where it does show is on hard, saturated color edges: red or blue text on a dark background, thin colored line art, and screen recordings of user interfaces, all of which can pick up a faint colored fringe. Color-critical work has the same problem for a different reason. Green-screen keying and heavy color grading cut into the color information directly, so they want 4:2:2 or 4:4:4 to work with.

For archiving, remember that subsampling is destructive and one-way. Converting a 4:2:0 file to 4:4:4 later gives you a bigger file, not the discarded color samples back. The field describes the encoding only and holds nothing personal.