A video is a collection of images, and it's the images that move in quick succession that make us perceive movement and watch the video. Each successive image is called a "Frame," and we called "30fps" and "25fps" depending on how many frames pass in a second. Then, how many images are used per minute? If you take 30fps as an example, we’re using about 1,800 images per minute to play the video.

Why compression technology came about

That’s why the size of the video data is huge. If we assume that each FHD image is 500KB, 1800 images are consumed per minute, resulting in a capacity of 900MB. That's going to be very hard to handle, especially if the movie is about 2 hours long. So, humans use a technique called "Compression". Depending on the codec, the results may vary, but compression can reduce the size of a movie by a factor of 100 or more, allowing us to go from being able to store a single movie in 1TB of storage space to being able to store over 200 movies.

Compression Process

Video compression is accomplished in the following steps. It starts with inputting a huge amount of raw data, encoding it, storing or transmitting the data over the network, and then decoding it to convert it into output data. The reason video can be compressed during the encryption and decryption process is due to "Redundant Data Removal". When you have a bunch of images in a row, you're bound to end up with the same pixels in the same place, and it's hard for the human eye to tell the difference, which is why we rely on a technique called compression.

There are several compression methods that reduce the size of a video by removing redundant data. Typical methods include the following.

- Spatial Redundancy: Remove duplicate data within a frame

- Temporal Redundancy: Remove duplicate data between frames

- Statistical Redundancy: Remove duplicates found in the statistics of the data

- Based on cognitive abilities: Remove data that has no impact when viewed by a human eye.

And in this post, I’m going to cover the "Spatial Redundancy" and "Temporal Redundancy" compression methods for compressing frames.

Intra-frame coding

Spatial Redundancy is a technique for removing data within a single frame, called “Intra-frame Coding”. When compressing video in this way, it looks at the correlation of pixels within a single frame and removes pixels that have similar values.

The intra-frame compression method can maintain excellent image quality even when compressed because each frame has the same information, but it is difficult to reduce much data due to low compression efficiency. This is typically used by the 'ProRes' codec developed by Apple.

Inter-frame coding

Temporal Redundancy is a compression method that removes redundant data based on information between frames and is called Inter-frame Coding. Earlier, I mentioned that many images appear in succession to form a video. When compressing multiple frames at once, redundant data appears, and the data capacity is reduced by removing it. In other words, based on one frame, the next frame is created by detecting only the motion vector.

Inter-frame compression can achieve high quality for the space it uses, but it is slow to edit because the frames don't have all the information about the video, and if the reference frame is of poor quality, the overall video will look bad. This is the type of compression used by the H.264 codec.

The image above shows intra-frame compression and inter-frame compression briefly: intra-frame compression compresses each frame that contains all the data, while inter-frame compresses multiple frames at once, using the similarities between adjacent frames as a reference point.

Compression, and the rise of new codecs

To get a better understanding of the video, I've introduced two compression methods today. Understanding compression is crucial because as the video market grows, more and more technologies are required to enjoy video.

And recently, codecs have emerged that implement efficient compression. Among other high-performance codecs, BLUEDOT has developed a high-performance codec based on the industrial open-source codec 'AV1' developed by the Alliance for Open Media (AOMedia). BLUEDOT’s AV1 codec is highly compressible, requiring 50% less space than the h.264 codec for the same quality, and encodes at a faster rate to reduce network costs.

In an already huge video market, consumers are demanding higher quality video, and video providers want to deliver high-quality video that satisfies consumers' high standards while managing it efficiently. To meet their needs, BLUEDOT and many other tech companies will continue to work on developing codecs that achieve higher compression ratios in the future, so that everyone can enjoy high-quality media content.