What is 4: 4: 4, 4: 2: 2 and 4: 2: 0 or subsampling color

When the 4: 4: 4, 4: 2: 2 and 4: 2: 0 digit sets are read it means that through these notations a video formula related to chroma subsampling (also called chrominance subsampling) is being expressed.. These number combinations can be found in photos and videos, that is why it is necessary to know what they are for.

Before analyzing these notations, it must be considered that both the content in photos and in videos cause their distribution to slow down, related to the limits offered by broadband.

In this scenario, and to achieve greater compression and transfer speed in audiovisual content, chrominance subsampling is used, widely used in various content formats, such as Blu-ray discs and streaming services.

Index of contents

What is chroma subsampling or subsampling?

Chromatic subsampling (color subsampling) is a technique by which the color information contained in a signal is compressed to favor the information contained in the luminance. In this way, the bandwidth is reduced, but without affecting the quality of this compressed image.

Several years ago, with the introduction of digital video, videos weighed heavily, making it difficult to transmit and store them. Trying to find a solution to these size problems, the chrominance subsampling was arrived at.

If we investigate the composition of all digital video, we will find two main components that we call luminance and chrominance.

The first term, which we also know brightness or contrast, encompasses all the differences we see between the darkest and lightest areas in the video.

For its part, the chrominance is the component of the color saturation of the video. Because the vision of a human being has more sensitivity to contrast (luminance) than to color saturation (chrominance), it was decided that there was a part of the video that could be compressed without affecting its quality.

Therefore, to make digital video management easier, the compression technique was implemented. This means that a true color video signal (4: 4: 4) in which we find all the information of red, green and blue in each pixel, this will be compressed if chromatic subsampling is applied, making it its transfer is lighter and it requires less bandwidth when the color has already been removed.

Once the image is compressed, the quality of the black and white will not be less than the quality of the colors, since, as indicated, human vision has less capacity to assimilate chrominance. In this way, after subsampling, the video will have more luminance than chrominance information.

With this it is possible to maintain the quality of the image while making a significant reduction in its size by up to 50%. In some formats like the YUV, the amount of luminance only reaches a third of the total, so there is a wide margin to reduce chrominance and thus achieve greater compression.

Taking into account that there are certain limitations in the speeds that constitute the wide bands of internet and HDMI, for example, this compression achieves that a digital video can be transmitted with greater efficiency.

Both CRT monitors, LCDs, and charge coupled devices (CCDs) use components to capture red, green, and blue colors. However, in a digital video a distinction is made between luma and chroma just to be able to make a compression and make it lighter for transmission.

There are several chroma subsampling methods that use different notations that we will briefly explain, noting that the first number is for luma and the second and third numbers are for chroma.

Color subsampling / subsampling methods

4: 4: 4

This is the full and original resolution, in which there is no compression of any kind, with the first number indicating the luminance (4) and the following two numbers (4: 4) used for the Cb and Cr chroma components. This notation 4: 4: 4 is commonly used for RGB images, although it is also used for the YCbCr color space.

4: 2: 2

In the first issue we see a full resolution of the luma, while we see a half resolution for the chrominance. This notation is the standard in images and carries a compression that does not affect the image quality. It is used for DVCpro50 and Betacam Digital video formats, among others.

4: 1: 1

Again, we have a full resolution luma, while we now have even less chrominance - just a quarter. This is the subsampling scheme used by the NTSC DV and PAL DVCPro formats.

4: 2: 0

This notation indicates that the resolution of the luma is complete (4), while it has a half resolution in the vertical and horizontal direction for the chroma components. Actually 4: 2: 0 is a pretty difficult color sampling that includes a lot of variations considering if the video is interlaced or progressive, or if it is being used by MPEG2 or PAL DV.

With this 4: 2: 0 sampling, you get a 1/4 color resolution, just like 4: 1: 1 sampling. However, in the first case the color is compressed horizontally and vertically, while in the second notation the compression is horizontal.

1920 x 1080 color subsampling

Analog HDTV was followed by digital HDTV, a technology of higher quality and resolution. However, it also brought a great challenge for the engineers, since they had to create a form that made it possible for this new technology to be used in the systems present at that time, mainly PAL and NTSC.

Consequently, all efforts had to be directed towards making compatibility between PAL and NTSC possible. The new HDTV standard had to be compatible for both PAL and NTSC, among its main features.

The variations that this standard suffered over the years were many, until it was finally set at 1125 vertical lines, with 1080 of these dedicated exclusively to the image. At that time, the maximum rate for 1080 was 29.97 fps (NTSC), while for 720 it was 59.94 fps (NTSC).

These are some of the most widely used chromatic subsampling values in the different popular digital video formats:

HDCAM: 3: 1: 1NTSC: 4: 1: 1PAL, DV, DVCAM, HDTV: 4: 2: 0Internet Video: 4: 2: 0HDTV Transmission Quality: 4: 2: 2 Uncompressed (full information): 4: 4: 4: 4

Is a 3: 1: 1 subsampling better than 4: 2: 2?

In the old 1080p HDCAM format, 3: 1: 1 was used, while 720p resolution had and still has 4: 2: 2 subsampling. But which of these was the best?

If we only based on the data, it is a simple answer: 4: 2: 2 is twice 3: 1: 1 in terms of color sampling, so we could clearly state that the best in this case is 4: 2:two.

However, this cannot be an absolute answer, since the size of the image is not considered in the 4 × 4 notations of the color sampling.

So which of these notations is better? An image that contains a lot of color information or another with less information but with a better sample color? There is no clear answer.

The intention of this analysis was for us to see that an image has much more information and complexity as a background than what is superficially seen.

Of course, always bearing in mind that we use a sample of an image at 4: 4: 4, since this is a complete notation in which the best sampling frequency is obtained.

Subsampling 4: 4: 4 vs 4: 2: 2 vs 4: 2: 0

The number 4, which is the first number from the left, indicates the size of the sample.

As for the two numbers that precede this, they are related to the chroma information. These depend on the first number (4) and are responsible for defining the horizontal and vertical sampling, respectively.

An image with a 4: 4: 4: 4 color component is not compressed at all, which means that it was not sub-sampled and therefore fully contains the luminance and color data.

Analyzing a four by two pixel matrix, we see that 4: 2: 2 contains half the chroma that we find in a 4: 4: 4 signal, while analyzing a 4: 2: 0 matrix we see that it contains even less: only a color information room.

The horizontal sampling rate on a 4: 2: 2 signal will be only half (2), while its vertical sampling will be full (4). In contrast, in a 4: 2: 0 signal, there is only color sampling in half of the pixels in the first row, completely ignoring the pixels in the second row of the signal.

Calculating the size of the subsampling data

There is a fairly simple calculation with which we can know exactly how much information is lost after having sub-sampled color. The calculation is as follows:

As we have already indicated, the maximum quality for a sample is 4 + 4 + 4 = 12

This means that an image with full color is 4: 4: 4 = 4 + 4 + 4 = 12, where we find 100% quality, without any compression. From this point on, the quality of a sample can vary as follows:

4: 2: 2 = 4 + 2 + 2 = 8, which is 66.7% of 4: 4: 4 (12) 4: 2: 0 = 4 + 2 + 0 = 6, which is 50% of 4: 4: 4 (12) 4: 1: 1 = 4 + 1 + 1 = 6, which is 50% of 4: 4: 4 (12) 3: 1: 1 = 3 + 1 + 1 = 5, which is 42% of 4: 4: 4 (12)

Therefore, if a 4: 4: 4 full color signal is 24 MB in size, it means that a 4: 2: 2 signal will be about 16 MB in size, while a 4: 2: 0 signal It will be 12 MB in size and a 3: 1: 1 signal will be 10 MB.

With this we can already understand why chromatic subsampling is so important and continues to exist. For sectors such as the internet and television it is essential because it reduces the size of the files and therefore requires less bandwidth resources.

Conclusion about subsampling

With chromatic subsampling we can compress an image file to reduce its size in this way. With this, it is achieved that less bandwidth is required to transmit it, without losing the quality of the image with the naked eye. This means that after color subsampling or subsampling, no major imperfections are visually noticeable.

Currently, the 4: 2: 0 sample is essential for audiovisual content platforms, so without this compression technique, it would surely have been much more difficult and expensive to access services such as 4K content from Amazon and Netflix.

Wikipedia source