Never really understood the concept of GOP .
The video sequence is composed of several time continuous images, which are divided into several small image groups and GOP when compressing them. There are two types of GOP in video encoding
Closed GOP, each GOP starts with an IDR image, and each GOP is encoded independently of each other
The first GOP in the development GOP is the IDR image, the first of the subsequent GOP is the NON-IDR image, that is, the image of the GOP after the inter-frame prediction can refer to the information of a GOP image.
SS(slice segment)
The SS is a fragment, each piece consists of one or more fragments, and a SS is divided into the same size as the CTU at the time of encoding.
SS from tablet Slice
A picture can be divided into one or more slices (note that the slice here, the boundary does not need to be horizontal, it can be some asymmetrical parts), each piece of data is independent
The header information of different pieces of information is obtained independently, so each slice needs to initialize and update its decoding environment before entropy decoding.
At the same time, intra-frame prediction cannot be done across the edge of the slice
However, when the loop filter is allowed to cross the edge of the chip (here recall that the loop filter in AVS is the end of the image reconstruction, the whole frame of the image loop filter)
A separate slice can be divided into several SS, including a separate SS, and several dependent SS, and begin decoding with independent SS
The independent SS means that all syntactic elements are determined by themselves.
Depending on the SS, the syntactic elements involved are inferred from the independent SS that have been decoded and can share information about some independent SS
The prediction process cannot span the boundaries of the independent SS, but can span the bounds of the dependent SS
The SS in the same frame picture is reference to the same PPS
Tile:
The newly introduced concept is a rectangular area that splits horizontally and vertically. The main purpose is to increase parallel computing without introducing new error diffusion.
Slice including SS, SS including CTU
Tile is directly including CTU
The division of slice and the partitioning of tiles are subject to one of the following two conditions:
(1) All CTU in a slice/ss belongs to the same tile.
(2) All CTU in a tile belongs to the same slice/ss.
hevc-parameter Gop,slice,tile Troubleshoot