ffmpeg linesize注意事项

原创 于 2021-12-27 19:59:18 发布 · 1.6k 阅读 · 0 ·
CC 4.0 BY-SA版权
版权声明：本文为博主原创文章，遵循 CC 4.0 BY-SA 版权协议，转载请附上原文出处链接和本声明。
文章标签：

#音视频

博主在尝试使用ffmpeg合并不同分辨率视频时遇到问题，当在高分辨率（2880x1800）笔记本上运行时，视频图像出现异常。原因是原始AVFrame的linesize非64的倍数，导致数据冗余。为解决此问题，博主调整了代码，正确处理每一行中多余的字节，并确保linesize对齐。最终成功修复了图像显示异常。

摘要生成于 C知道 ，由 DeepSeek-R1 满血版支持， 前往体验 >

最近在做视频混合，为此本人写了两篇博客，ffmpeg利用滤镜合并两个视频，一左一右
和ffmpeg利用滤镜合并四个视频，左一右三

在本人台式机上(19201080)上混合是ok的，但是在笔记本电脑上(28801800)上混合产生了问题，图像变成了下面这种样子。
在这里插入图片描述
我一度以为是滤镜那块出了问题，幸好不是。
在排查过程中，本人写了一个例子，通过ffmpeg读取本地视频文件，然后再经过编码，写入文件，结果发现，写入的文件播放时，就出现上面这个样子。

经过排查，发现，读取到的AVFrame的linesize为2944,1472,1472；而之前想的应该是2880,1440,1440。
所以读取到的数据其实多了一些冗余信息，拿Y分量而言，第一行到第1800行，每行中的第2881个字节数据到第2994个字节数据是多余的，需要去掉。

话说linesize需要是2944，1472,1472，有人说是为了对齐，分量需要是64的倍数，而1440不是64的倍数，在此先记录下。

在这里插入图片描述

下面罗列下处理，首先构建了三个变量pY，pU，pV，这里面的y_size是2880x1800

int y_size = m_pReadCodecCtx_VideoA->width * m_pReadCodecCtx_VideoA->height;

	char *pY = new char[y_size];
	char *pU = new char[y_size / 4];
	char *pV = new char[y_size / 4];

然后重新给这三个分量赋值，去掉每行中冗余的部分

 ///Y
int contY = 0;
for (int i = 0; i < pFrame->height; i++)
{
	memcpy(pY + contY, pFrame->data[0] + i * pFrame->linesize[0], pFrame->width);
	contY += pFrame->width;
}


///U
int contU = 0;
for (int i = 0; i < pFrame->height / 2; i++)
{
	memcpy(pU + contU, pFrame->data[1] + i * pFrame->linesize[1], pFrame->width / 2);
	contU += pFrame->width / 2;
}


///V
int contV = 0;
for (int i = 0; i < pFrame->height / 2; i++)
{
	memcpy(pV + contV, pFrame->data[2] + i * pFrame->linesize[2], pFrame->width / 2);
	contV += pFrame->width / 2;
}

最后是将处理后的分量送入队列

EnterCriticalSection(&m_csVideoASection);
av_fifo_generic_write(m_pVideoAFifo, pY, y_size, NULL);
av_fifo_generic_write(m_pVideoAFifo, pU, y_size / 4, NULL);
av_fifo_generic_write(m_pVideoAFifo, pV, y_size / 4, NULL);
LeaveCriticalSection(&m_csVideoASection);

工程目录如下：
在这里插入图片描述
其中main所在文件FfmpegCopyFileTest.cpp的代码如下：

#include <iostream>
#include "CopyFile.h"

#ifdef	__cplusplus
extern "C"
{
#endif

#pragma comment(lib, "avcodec.lib")
#pragma comment(lib, "avformat.lib")
#pragma comment(lib, "avutil.lib")
#pragma comment(lib, "avdevice.lib")
#pragma comment(lib, "avfilter.lib")
#pragma comment(lib, "postproc.lib")
#pragma comment(lib, "swresample.lib")
#pragma comment(lib, "swscale.lib")


#ifdef __cplusplus
};
#endif





int main()
{
	CCopyFile cVideoCopy;
	const char *pFileA = "E:\\learn\\ffmpeg\\FfmpegFilterTest\\x64\\Release\\in-desktop-2880x1800.mp4";

	const char *pFileOut = "E:\\learn\\ffmpeg\\FfmpegFilterTest\\x64\\Release\\out-copy.mp4";
	cVideoCopy.StartCopy(pFileA, pFileOut);
	cVideoCopy.WaitFinish();
	return 0;
}

CopyFile.h的代码如下：

#pragma once

#include <Windows.h>

#ifdef	__cplusplus
extern "C"
{
#endif
#include "libavcodec/avcodec.h"
#include "libavformat/avformat.h"
#include "libswscale/swscale.h"
#include "libswresample/swresample.h"
#include "libavdevice/avdevice.h"
#include "libavutil/audio_fifo.h"
#include "libavutil/avutil.h"
#include "libavutil/fifo.h"
#include "libavutil/frame.h"
#include "libavutil/imgutils.h"

#include "libavfilter/avfilter.h"
#include "libavfilter/buffersink.h"
#include "libavfilter/buffersrc.h"


#ifdef __cplusplus
};
#endif

class CCopyFile
{
public:
	CCopyFile();
	~CCopyFile();
public:
	int StartCopy(const char *pFileA, const char *pFileOut);
	int WaitFinish();
private:
	int OpenFileA(const char *pFileA);
	int OpenOutPut(const char *pFileOut);
private:
	static DWORD WINAPI VideoAReadProc(LPVOID lpParam);
	void VideoARead();


	static DWORD WINAPI VideoCopyProc(LPVOID lpParam);
	void VideoCopy();
private:
	AVFormatContext *m_pFormatCtx_FileA = NULL;

	AVCodecContext *m_pReadCodecCtx_VideoA = NULL;
	AVCodec *m_pReadCodec_VideoA = NULL;


	AVCodecContext	*m_pCodecEncodeCtx_Video = NULL;
	AVFormatContext *m_pFormatCtx_Out = NULL;

	AVFifoBuffer *m_pVideoAFifo = NULL;

	int m_iMergeWidth = 1920;
	int m_iMergeHeight = 1080;
	int m_iYuv420FrameSize = 0;
private:
	CRITICAL_SECTION m_csVideoASection;
	HANDLE m_hVideoAReadThread = NULL;
	HANDLE m_hVideoCopyhread = NULL;
};

CopyFile.cpp的代码如下：


#include "CopyFile.h"
//#include "log/log.h"





CCopyFile::CCopyFile()
{
	InitializeCriticalSection(&m_csVideoASection);
}

CCopyFile::~CCopyFile()
{
	DeleteCriticalSection(&m_csVideoASection);
}

int CCopyFile::StartCopy(const char *pFileA, const char *pFileOut)
{
	int ret = -1;
	do 
	{
		ret = OpenFileA(pFileA);
		if (ret != 0)
		{
			break;
		}

		ret = OpenOutPut(pFileOut);
		if (ret != 0)
		{
			break;
		}

		m_iYuv420FrameSize = av_image_get_buffer_size(AV_PIX_FMT_YUV420P, m_pReadCodecCtx_VideoA->width, m_pReadCodecCtx_VideoA->height, 1);
		//申请30帧缓存
		m_pVideoAFifo = av_fifo_alloc(30 * m_iYuv420FrameSize);

		m_hVideoAReadThread = CreateThread(NULL, 0, VideoAReadProc, this, 0, NULL);

		m_hVideoCopyhread = CreateThread(NULL, 0, VideoCopyProc, this, 0, NULL);

	} while (0);

	return ret;
}

int CCopyFile::WaitFinish()
{
	int ret = 0;
	do 
	{
		if (NULL == m_hVideoAReadThread)
		{
			break;
		}
		WaitForSingleObject(m_hVideoAReadThread, INFINITE);

		CloseHandle(m_hVideoAReadThread);
		m_hVideoAReadThread = NULL;

		WaitForSingleObject(m_hVideoCopyhread, INFINITE);
		CloseHandle(m_hVideoCopyhread);
		m_hVideoCopyhread = NULL;
	} while (0);

	return ret;
}

int CCopyFile::OpenFileA(const char *pFileA)
{
	int ret = -1;

	do
	{
		if ((ret = avformat_open_input(&m_pFormatCtx_FileA, pFileA, 0, 0)) < 0) {
			printf("Could not open input file.");
			break;
		}
		if ((ret = avformat_find_stream_info(m_pFormatCtx_FileA, 0)) < 0) {
			printf("Failed to retrieve input stream information");
			break;
		}

		if (m_pFormatCtx_FileA->streams[0]->codecpar->codec_type != AVMEDIA_TYPE_VIDEO)
		{
			break;
		}
		m_pReadCodec_VideoA = (AVCodec *)avcodec_find_decoder(m_pFormatCtx_FileA->streams[0]->codecpar->codec_id);

		m_pReadCodecCtx_VideoA = avcodec_alloc_context3(m_pReadCodec_VideoA);

		if (m_pReadCodecCtx_VideoA == NULL)
		{
			break;
		}
		avcodec_parameters_to_context(m_pReadCodecCtx_VideoA, m_pFormatCtx_FileA->streams[0]->codecpar);

		m_iMergeWidth = m_pReadCodecCtx_VideoA->width;
		m_iMergeHeight = m_pReadCodecCtx_VideoA->height;

		m_pReadCodecCtx_VideoA->framerate = m_pFormatCtx_FileA->streams[0]->r_frame_rate;

		if (avcodec_open2(m_pReadCodecCtx_VideoA, m_pReadCodec_VideoA, NULL) < 0)
		{
			break;
		}

		ret = 0;
	} while (0);


	return ret;
}


int CCopyFile::OpenOutPut(const char *pFileOut)
{
	int iRet = -1;

	AVStream *pAudioStream = NULL;
	AVStream *pVideoStream = NULL;

	do
	{
		avformat_alloc_output_context2(&m_pFormatCtx_Out, NULL, NULL, pFileOut);

		{
			AVCodec* pCodecEncode_Video = (AVCodec *)avcodec_find_encoder(m_pFormatCtx_Out->oformat->video_codec);

			m_pCodecEncodeCtx_Video = avcodec_alloc_context3(pCodecEncode_Video);
			if (!m_pCodecEncodeCtx_Video)
			{
				break;
			}

			pVideoStream = avformat_new_stream(m_pFormatCtx_Out, pCodecEncode_Video);
			if (!pVideoStream)
			{
				break;
			}

			int frameRate = 10;
			m_pCodecEncodeCtx_Video->flags |= AV_CODEC_FLAG_QSCALE;
			m_pCodecEncodeCtx_Video->bit_rate = 4000000;
			m_pCodecEncodeCtx_Video->rc_min_rate = 4000000;
			m_pCodecEncodeCtx_Video->rc_max_rate = 4000000;
			m_pCodecEncodeCtx_Video->bit_rate_tolerance = 4000000;
			m_pCodecEncodeCtx_Video->time_base.den = frameRate;
			m_pCodecEncodeCtx_Video->time_base.num = 1;

			m_pCodecEncodeCtx_Video->width = m_iMergeWidth;
			m_pCodecEncodeCtx_Video->height = m_iMergeHeight;
			//pH264Encoder->pCodecCtx->frame_number = 1;
			m_pCodecEncodeCtx_Video->gop_size = 12;
			m_pCodecEncodeCtx_Video->max_b_frames = 0;
			m_pCodecEncodeCtx_Video->thread_count = 4;
			m_pCodecEncodeCtx_Video->pix_fmt = AV_PIX_FMT_YUV420P;
			m_pCodecEncodeCtx_Video->codec_id = AV_CODEC_ID_H264;
			m_pCodecEncodeCtx_Video->codec_type = AVMEDIA_TYPE_VIDEO;

			av_opt_set(m_pCodecEncodeCtx_Video->priv_data, "b-pyramid", "none", 0);
			av_opt_set(m_pCodecEncodeCtx_Video->priv_data, "preset", "superfast", 0);
			av_opt_set(m_pCodecEncodeCtx_Video->priv_data, "tune", "zerolatency", 0);

			if (m_pFormatCtx_Out->oformat->flags & AVFMT_GLOBALHEADER)
				m_pCodecEncodeCtx_Video->flags |= AV_CODEC_FLAG_GLOBAL_HEADER;

			if (avcodec_open2(m_pCodecEncodeCtx_Video, pCodecEncode_Video, 0) < 0)
			{
				//编码器打开失败，退出程序
				break;
			}
		}

		if (!(m_pFormatCtx_Out->oformat->flags & AVFMT_NOFILE))
		{
			if (avio_open(&m_pFormatCtx_Out->pb, pFileOut, AVIO_FLAG_WRITE) < 0)
			{
				break;
			}
		}

		avcodec_parameters_from_context(pVideoStream->codecpar, m_pCodecEncodeCtx_Video);

		if (avformat_write_header(m_pFormatCtx_Out, NULL) < 0)
		{
			break;
		}

		iRet = 0;
	} while (0);


	if (iRet != 0)
	{
		if (m_pCodecEncodeCtx_Video != NULL)
		{
			avcodec_free_context(&m_pCodecEncodeCtx_Video);
			m_pCodecEncodeCtx_Video = NULL;
		}

		if (m_pFormatCtx_Out != NULL)
		{
			avformat_free_context(m_pFormatCtx_Out);
			m_pFormatCtx_Out = NULL;
		}
	}

	return iRet;
}


DWORD WINAPI CCopyFile::VideoAReadProc(LPVOID lpParam)
{
	CCopyFile *pVideoMerge = (CCopyFile *)lpParam;
	if (pVideoMerge != NULL)
	{
		pVideoMerge->VideoARead();
	}
	return 0;
}

void CCopyFile::VideoARead()
{
	AVFrame *pFrame;
	pFrame = av_frame_alloc();

	int y_size = m_pReadCodecCtx_VideoA->width * m_pReadCodecCtx_VideoA->height;

	char *pY = new char[y_size];
	char *pU = new char[y_size / 4];
	char *pV = new char[y_size / 4];

	AVPacket packet = { 0 };
	int ret = 0;
	while (1)
	{
		av_packet_unref(&packet);

		ret = av_read_frame(m_pFormatCtx_FileA, &packet);
		if (ret == AVERROR(EAGAIN))
		{
			continue;
		}
		else if (ret == AVERROR_EOF)
		{
			break;
		}
		else if (ret < 0) 
		{
			break;
		}

		ret = avcodec_send_packet(m_pReadCodecCtx_VideoA, &packet);

		if (ret >= 0)
		{
			ret = avcodec_receive_frame(m_pReadCodecCtx_VideoA, pFrame);
			if (ret == AVERROR(EAGAIN))
			{
				continue;
			}
			else if (ret == AVERROR_EOF)
			{
				break;
			}
			else if (ret < 0) {
				break;
			}
			while (1)
			{
				if (av_fifo_space(m_pVideoAFifo) >= m_iYuv420FrameSize)
				{
					///Y
					int contY = 0;
					for (int i = 0; i < pFrame->height; i++)
					{
						memcpy(pY + contY, pFrame->data[0] + i * pFrame->linesize[0], pFrame->width);
						contY += pFrame->width;
					}


					///U
					int contU = 0;
					for (int i = 0; i < pFrame->height / 2; i++)
					{
						memcpy(pU + contU, pFrame->data[1] + i * pFrame->linesize[1], pFrame->width / 2);
						contU += pFrame->width / 2;
					}


					///V
					int contV = 0;
					for (int i = 0; i < pFrame->height / 2; i++)
					{
						memcpy(pV + contV, pFrame->data[2] + i * pFrame->linesize[2], pFrame->width / 2);
						contV += pFrame->width / 2;
					}


					EnterCriticalSection(&m_csVideoASection);
					av_fifo_generic_write(m_pVideoAFifo, pY, y_size, NULL);
					av_fifo_generic_write(m_pVideoAFifo, pU, y_size / 4, NULL);
					av_fifo_generic_write(m_pVideoAFifo, pV, y_size / 4, NULL);
					LeaveCriticalSection(&m_csVideoASection);

					break;
				}
				else
				{
					Sleep(100);
				}
			}

		}


		if (ret == AVERROR(EAGAIN))
		{
			continue;
		}
	}

	av_frame_free(&pFrame);
	delete[] pY;
	delete[] pU;
	delete[] pV;
}

DWORD WINAPI CCopyFile::VideoCopyProc(LPVOID lpParam)
{
	CCopyFile *pVideoMerge = (CCopyFile *)lpParam;
	if (pVideoMerge != NULL)
	{
		pVideoMerge->VideoCopy();
	}
	return 0;
}


void CCopyFile::VideoCopy()
{
	int ret = 0;

	AVFrame *pFrameVideoA = av_frame_alloc();
	uint8_t *videoA_buffer_yuv420 = (uint8_t *)av_malloc(m_iYuv420FrameSize);
	av_image_fill_arrays(pFrameVideoA->data, pFrameVideoA->linesize, videoA_buffer_yuv420, AV_PIX_FMT_YUV420P, m_pReadCodecCtx_VideoA->width, m_pReadCodecCtx_VideoA->height, 1);

	int iOutVideoWidth = m_pReadCodecCtx_VideoA->width;
	int iOutVideoHeight = m_pReadCodecCtx_VideoA->height;

	AVPacket packet = { 0 };
	int iPicCount = 0;

	while (1)
	{
		if (NULL == m_pVideoAFifo)
		{
			break;
		}

		int iVideoASize = av_fifo_size(m_pVideoAFifo);

		if (iVideoASize >= m_iYuv420FrameSize)
		{
			EnterCriticalSection(&m_csVideoASection);
			av_fifo_generic_read(m_pVideoAFifo, videoA_buffer_yuv420, m_iYuv420FrameSize, NULL);
			LeaveCriticalSection(&m_csVideoASection);


			pFrameVideoA->pkt_dts = pFrameVideoA->pts = av_rescale_q_rnd(iPicCount, m_pCodecEncodeCtx_Video->time_base, m_pFormatCtx_Out->streams[0]->time_base, (AVRounding)(AV_ROUND_NEAR_INF | AV_ROUND_PASS_MINMAX));
			pFrameVideoA->pkt_duration = 0;
			pFrameVideoA->pkt_pos = -1;

			pFrameVideoA->width = iOutVideoWidth;
			pFrameVideoA->height = iOutVideoHeight;
			pFrameVideoA->format = AV_PIX_FMT_YUV420P;

			ret = avcodec_send_frame(m_pCodecEncodeCtx_Video, pFrameVideoA);

			ret = avcodec_receive_packet(m_pCodecEncodeCtx_Video, &packet);

			av_write_frame(m_pFormatCtx_Out, &packet);

			iPicCount++;
		}
		else
		{
			if (m_hVideoAReadThread == NULL)
			{
				break;
			}
			Sleep(1);
		}
	}

	av_write_trailer(m_pFormatCtx_Out);
	avio_close(m_pFormatCtx_Out->pb);

	av_frame_free(&pFrameVideoA);
}
tusong86
关注
评论
添加红包

请填写红包祝福语或标题

个

红包个数最小为10个

元

红包金额最低5元

当前余额3.43元 前往充值 >
需支付：10.00元
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付元
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明：

1.余额是钱包充值的虚拟货币，按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载，可以购买VIP、付费专栏及课程。

余额充值