This article will show how to encapsulate H.264 into FLV format.
Between this article, it is recommended to take a look at the following two articles:
H.264 code stream structure
FLV format detailed
First, what do you need to build a FLV file?
FLV HEADER + FLV Script Tag + FLV Video Tag + FLV Audio Tag
1
Since it is just the encapsulation H.264, AUDIO TAG is not included.
The three parts of the FLV Header + FLV Script Tag + FLV Video Tag will be described below.
1, FLV HEADER
By reading the above article "Detailed FLV Format" should be very clear about the content of FLV Header.
Below to analyze an example:
Signature: 0x46 ('f')
Signature: 0x4c ('L')
Signature: 0x56 ('V')
Version: 01
Flags: 01
TypeflagSReServed [bit 3-7]: 0
Typeflagsaudio [Bit 2]: 0
TypeflagSReServed [bit 1]: 0
Typeflagsvideo [bit 0]: 1
Dataoffset: 00 00 00 00 00
Pretagsize: 00 00 00 00 00 // The first TAG default is 0
1234567891011
2, FLV Script Tag
Also known as Metadata Tag, the first TAG behind FLV Header
Below to analyze an example
Tag header:
TagType: 12 (indicating this is a script tag)
DataSize: 00 00 9f (the size of the Tag Data section)
TimeSTAMP: 00 00 00 (Timestamp of the TAG)
TimeStamp_EX: 00 (Extended part of timestamp)
Streamid :: 00 00 00 (always 0)
12345
Tag Data:
Because tagType is 0x12, this part of the data is two AMF packages
The first AMF package:
TYPE: 02 (Representing strings)
Stringlen: 00 0A (string length is 10)
String: 6F 6e 4D 65 74 61 44 61 74 61 (Onmetadata)
123
Second AMF package:
TYPE: 08 (represents array)
ArrayNum: 00 00 00 07 (number of array elements)
Stringlen: 00 08 (First array element string length)
String: 64 75 72 61 74 69 6F 6E (DURATION)
ValType: 00 (Data Type, Double)
VAL: 40 03 70 A3 D7 0A 3D 71 (Double)
Stringlen: 00 05
String: 77 69 64 74 68 (WIDTH)
...
End: 00 00 00 09
12345678910
pretagsize: 00 00 00 AA
1
3, FLV Video TAG
This partially encapsulates image data, that is, the data package of H.264 is here.
H.264 is composed of a NALU, NALU types (SPS, PPS, I frame SLICE, non-IF SLICE)
The encapsulation H.264 is taken out of these NALUs and then packages into TAG.
It should be noted that SPS and PPS must be packaged in a TAG.
The following is made to analyze how various NALUs (SPS, PPS, I frame slice, SEI, and non-I frame slice) are encapsulated.
1, SPS and PPS
Tag header:
Tagtype: 09 (TAG type, including audio (0x08), video (0x09), Script data (0x12))
DataSize: 00 00 2E (the size of the Tag Data section)
TimeStamp: 00 00 00 (Timestamp, 19MS)
TimeStamp_EX: 00 (Extended part of timestamp)
StreamID: 00 00 00 (always 0)
12345
Tag Data:
Frametype | CODECID: 17 (KeyFrame | AVC) (parameter of video tag)
1
Avcvideopaket: (because codeciD == 7, thus videodata is avcvideopaket)
Avcpakettype: 00 (Type of Acvpacket, 0: AVC Sequence Header; 1: AVC NALU; 2: AVC End of Sequence)
CompositionTime: 00 00 00 00
12
(Because acvpakettype == 0, Data = AvcDecoderConfigurationRecord) AvcDecoderConfigurationRecord:
CONFIGURATIONVERSION: 01
AVCProfileindication: 64
PROFILE_COMPATIBILITY: 00
Avclevelindication: 1e
Lengthsizeminusone: FF
Numofsequences: E1 (low five is the number of SPS, the calculation method is: NumofsequenceParameterSets & 0x1f = 1)
SequenceParametersetLength: 00 18 (length, 24)
SequenceParametersetnalunits: 67 64 00 1E AC D9 40 A0 33 B0 11 00 00 03 02 47 00 6d 34 0F 16 2D 96 (SPS)
NumofPictureParameterSets: 01 (number of pps)
PictureParametersetLength: 00 06 (length of PPS)
PictureParameterSetNalUnalthS: 68 EB E3 CB 22 C0 (PPS)
1234567891011
Previoustagsize: 00 00 00 00 00
1
2, I frame of SLICE
Tag header:
Tagtype: 09 (TAG type, including audio (0x08), video (0x09), Script data (0x12))
DataSize: 00 35 78 (Size of Tag Data)
TimeStamp: 00 00 00 (Timestamp, 19MS)
TimeStamp_EX: 00 (Extended part of timestamp)
StreamID: 00 00 00 (always 0)
12345
Tag Data:
Frametype | CodeCID: 17 (Key Frame | AVC) (Videos TAG parameters)
1
Because codeciD == 7, thw so videodata is AVCVIDEOPAKET
Avcvideopaket:
AVCPAKETTYPE: 01 (ACVPACKET type, 0: AVC Sequence Header; 1: AVC NALU; 2: AVC End of Sequence)
CompositionTime: 00 00 00 00
12
Nalulen: 00 00 35 6F
Naludata: ...
12
pretagsize: 00 00 00 00 0035 83
1
The package is the same for the package of the SEI.
3, non IF SLICE
The unique package format of non-I frame is basically the same, the only difference is FrameType | CodeCID: section, should be set to:
Frametype | CodeCID: 27 (Inter Frame | AVC) (Videos TAG parameters)
1
Summary: The head information of the FLV is explained above, how to encapsulate the Script Tag, and the H.264 different types of NALUs, and release the source code below.
GitHub: https://github.com/imsjt/h.264toflv
CSDN Download: https://download.csdn.net/download/weixin_42462202/11037715
Our other product:
