Color subsampling or chroma subsampling is a term used in video shooting. You may have heard about terms like 4:4:4 and 4:4:2, and so on. Ever wondered what they might be referring to and how those affect the quality of your video? Let’s dive into this beginner tutorial on color subsampling.
Color subsampling is a term associated with reducing the file size of your clips by cutting down on specific information, and that information is color information. It has been noticed in research that the human eye is more susceptible to the brightness levels in a clip rather than the color information or the chroma information. And that’s why a color subsampling routine is used when files are written to eliminate color information while retaining brightness information in the clips. The file size can be drastically reduced by just cutting down on the color.
Now coming back to the color subsampling part. How does it work?
When color information is recorded, the camera does not write all the available information in the image. Instead, they copy and paste color information from surrounding pixels, known as color subsampling or chroma subsampling.
The 4:4:4 ratio suggests that the sample area of two blocks, with each block containing four individual pixels, is considered for sampling. That’s denoted by the first 4. The following 4 suggest how many pixels from the first row have retained color information. In this case, all four of the pixels retain color information. The following 4 suggest how many pixels from the second or last row retain color information. In this case, it’s again all four. That means in this case, where 4:4:4 is used, none of the color information is overlooked.
In the case of 4:2:2 color subsampling, the sampling area is 2 rows of 4 pixels each. In this case, however, only two pixels in the first row and 2 pixels in the second row retain color information. What happens to the other 2 pixels? Well, the color information of the other 2 pixels is copied onto them.
The same happens in the bottom row. The color information in the other 2 pixels is copied onto the remaining 2 pixels. Using the 4:2:2 sampling systems, the overall color information is cut by 50%, significantly affecting the overall file size.
In the same way, in the 4:2:0 color subsampling system, only two pixels in the first row retain color information. While the other two copy the color information from them. None of the pixels in the bottom row retain color information. They copy the color information from the first row.
In the final scheme of things, since each of the pixels still retain the Luma or brightness value, we’re going to see different tones of the same color, which is, in the overall scheme of things, not as bad as it sounds. But on the bright side, we reduce the file size by a lot and therefore reduce the stress on the image processor.