What is the encryption mechanism? 5 Points to Consider for HD Video Encoding [Part 1]
Encode
The image quality of Internet videos is almost proportional to the bit rate. However, if the bit rate increases unnecessarily, the file size will increase. “Keep bit rate low”, “Reduce file size” and “Reduce load time” are linked, and there is nothing to say if you can encode in high definition while keeping the bit rate low.
Also, at the beginning, I wrote that “image quality is almost proportional to bit rate”, but I think some of you may have experienced that “I increased the bit rate and encoded, but I am not satisfied with the quality of the image. “So, this time, I will explain five points that are often used to do high definition video encoding.
” Table of Contents ”
Encoding Mechanism
Point 1: learn a suitable bit rate guideline for resolution
Point 2: Increase the frame rate if motion is important and lower the frame rate if image quality is important.
Point 3: insert “keyframes” at short intervals for moving images
* You can read the second part (Point 4, Point 5) here.
Encoding Mechanism
First, I will briefly explain the encoding mechanism.
An image is a collection of continuous images (frames), and by changing this in a short time like a flip book, it appears that you are visually moving. Japanese television images are 29.97 frames per second (short for 29.97 fps / frame per second) and most movies and animations have a standard of 24 fps. Since a large number of frames are required for video, the amount of data is also huge. Therefore, data compression is indispensable for distributing videos on the Internet.
When coding
“Prediction in frame” that compresses data within a frame
“Prediction between frames” that compresses data into consecutive frames
Information is reduced and data is compressed within the range that does not affect the visual sense.
■ What is in-frame prediction?
There are various methodologies for data compression, so I will skip the details here, but the basic idea of within-frame prediction is to divide a frame into small blocks called cells and the colors adjacent to each other in the block. they are the same or similar, they are compressed together.
For example, if there is information “blue blue blue blue blue blue blue blue blue blue red red yellow yellow yellow” in the divided block, the amount of data can be reduced by combining this with “blue 11, red 2, yellow 3”. It’s an image.
An example of data compression in in-frame prediction (image)
■ What is cross-frame prediction?
However, in the case of video with a time axis, the number of frames is large, so there is a limit to the overall weight reduction based solely on the prediction within the frame. On the other hand, in the prediction between frames, based on the idea that “the contents are similar before and after the consecutive frames”, the cells that do not change from the previous frame reduce the amount of data by reusing information and the cells that change It becomes data as difference information.
Prediction between frames (image)
From here, I’ll explain five specific code points.
