Conversion of player's actual 10 pixel format and size

SwsContext* vctx = NULL;

Create a context for format and size conversion , The assignment is NULL,

for (;;)// Every time I receive packet Go through it receive
		{
    
			// once send May correspond to multiple times receive, Because once send Of packet It may contain multiple frames of information 
			// Therefore, it is necessary to receive multiple times , Until you can't get it 
			re= avcodec_receive_frame(av, frame);//frame Application method and packet Agreement , But more than packet The space is much larger 
			if (re != 0)break;
			cout << "recv frame" << frame->format << " " << frame->linesize[0] << endl;// The value of the video is one line ,why？

			// The size and format are processed after decoding 
			if (av ==vc)
			{
    
				vctx = sws_getCachedContext(
					vctx,// Pass on NULL Will be newly created 
					frame->width,frame->height, (AVPixelFormat)frame->format,// Input width, height and format 
					frame->width, frame->height, AV_PIX_FMT_RGBA,// Output width, height and format 
					SWS_BILINEAR,0,0,0
				);// If there is a difference between the original and the target vctx Empty and recreate , Same handle vctx out 
				if (vctx)
					cout << " Format and size conversion succeeded " << endl;
				else
					cout << " Format and size conversion failed " << endl;
			}	
		}

In traversing each packet Of receive Used in the cycle of sws_getCachedContext
First determine whether the decoder context is video , Only video frames can use format and size conversion , This function is for context variables vctx The initialization , It stores the information of pixel format and size conversion , stay sws_scale（） In order to carry out the real transformation ,

To do a cast

Because I gave it to vctx The assignment is null, Here we pass null to the function vctx Will create a new vctx, The following is the input width, height and format , Output width, height and format , If there is a difference between the original and the target vctx Empty and recreate , Same handle vctx out

give the result as follows , You can see only video frames vctx If it is not empty, it will be printed successfully

if (vctx)
				{
    
					if (!rgb)
						rgb = new unsigned char[frame->width * frame->height * 4];//rgba 32 position ,4 Bytes 
					uint8_t* data[2] = {
     0 };// Namely unsigned char
					data[0] = rgb;
					int lines[2] = {
     0 };
					lines[0] = frame->width * 4;
					cout << " Format and size conversion succeeded " << endl;
					re=sws_scale(
						vctx,
						frame->data,// input data 
						frame->linesize, 0,// Enter row size , To align 
						frame->height,// Enter the height , Traverse all of linesize
						data,// Output data and size 
						lines
					);
					cout << "sws_scale=" << re << endl;
				}
				else
					cout << " Format and size conversion failed " << endl;

			}

int sws_scale(struct SwsContext *c, const uint8_t *const srcSlice[],
const int srcStride[], int srcSliceY, int srcSliceH,
uint8_t *const dst[], const int dstStride[]);

The function parameters are described in detail below ：

1. Parameters SwsContext *c, Context of conversion format . That is to say sws_getContext The result returned by the function .
2. Parameters const uint8_t *const srcSlice[], Enter the data pointer for each color channel of the image . It's actually decoded AVFrame Medium data[] Array . Because different pixels have different storage formats , therefore srcSlice[] dimension
It could be different .
With YUV420P For example , It is planar Format , Its memory is arranged as follows ：
YYYYYYYY UUUU VVVV
Use FFmpeg Decoded and stored in AVFrame Of data[] In the array ：
data[0]——-Y component , Y1, Y2, Y3, Y4, Y5, Y6, Y7, Y8……
data[1]——-U component , U1, U2, U3, U4……
data[2]——-V component , V1, V2, V3, V4……
linesize[] The data width of the corresponding channel is saved in the array ,
linesize[0]——-Y The width of the component
linesize[1]——-U The width of the component
linesize[2]——-V The width of the component

and RGB24, It is packed Format , It's in data[] There is only one dimension in the array , It is stored as follows ：
data[0]: R1, G1, B1, R2, G2, B2, R3, G3, B3, R4, G4, B4……
Special attention should be paid here ,linesize[0] The value of is not necessarily equal to the width of the picture , Sometimes in order to align the decoders CPU, The actual size will be larger than the width of the picture , This is when we program （ such as OpengGL Hardware conversion / Rendering ） Pay special attention to , Otherwise, the decoded image will be abnormal .

3. Parameters const int srcStride[], Enter the span of each color channel of the image .. That is, the number of bytes per channel , Corresponding to the decoded AVFrame Medium linesize[] Array . According to it, you can determine the starting position of the next line , however stride and width Not necessarily the same , This is because ：
a. Due to the alignment of data frame storage , It is possible to add some padding bytes after each line, so stride = width + N;
b.packet In color space , Several channels of data per pixel are mixed together , for example RGBA, Every pixel 3 Bytes are stored continuously , Therefore, the position of the next line needs to be skipped 3*width byte .

4. Parameters int srcSliceY, int srcSliceH, Define the processing area on the input image ,srcSliceY It's the starting position ,srcSliceH How many lines are processed . If srcSliceY=0,srcSliceH=height, Indicates that a complete image is processed at one time . This setting is for multithreading parallelism , For example, you can create two threads , The first thread handles [0, h/2-1] That's ok , The second thread handles [h/2, h-1] That's ok . Parallel processing speeds up .
5. Parameters uint8_t *const dst[], const int dstStride[] Define the output image information （ Data pointer of each color channel output , Number of bytes per color channel row ）

Reference resources ：https://www.cnblogs.com/yongdaimi/p/10715830.html