文章目录
现在互联网上很多人都通过表情包来表达自己的情绪，常用的表情包很多都是视频文件的一部分。这里就介绍一下如何通过ffmpeg截取视频生成gif动图。其实原理很简单，首先我们seek到视频对应的位置，然后读取数据帧修改帧的数据格式并输出到gif文件当中，读取完毕之后我们就得到了一个视频动图。具体的操作步骤如下:
1.封装视频滤镜
首先封装一下视频滤镜，方便对数据帧进行变换处理
//video_filter.h
#ifndef VIDEOBOX_VIDEO_FILTER_H
#define VIDEOBOX_VIDEO_FILTER_H
extern "C" {
#include <libavcodec/avcodec.h>
#include <libavfilter/avfilter.h>
#include <libavfilter/buffersrc.h>
#include <libavfilter/buffersink.h>
#include <libavutil/channel_layout.h>
#include <libavutil/opt.h>
}
//滤镜输入和输出对应的配置
struct VideoConfig {
AVPixelFormat format;
int width;
int height;
AVRational timebase{
1, 30};
AVRational pixel_aspect{
1, 1};
VideoConfig(AVPixelFormat format, int width, int height, AVRational timebase = {
1, 30},
AVRational pixel_aspect = {
1, 1}) {
this->format = format; //像素格式
this->width = width; //视频宽
this->height = height; //视频高
}
};
class VideoFilter {
protected:
AVFilterContext *buffersink_ctx;
AVFilterContext *buffersrc_ctx;
AVFilterContext *buffersrc_ctx1;
AVFilterGraph *filter_graph;
const char *description = nullptr;
public:
VideoFilter() = default;
//构建对应的滤镜
int create(const char *filter_descr, VideoConfig *inConfig, VideoConfig *outConfig);
int create(const char *filter_descr, VideoConfig *inConfig1, VideoConfig *inConfig2, VideoConfig *outConfig);
//获取滤镜的输入和输出
int filter(AVFrame *source, AVFrame *dest);
int filter(AVFrame* source1, AVFrame* source2, AVFrame*dest);
//添加输入1
int addInput1(AVFrame * input);
//添加输入2
int addInput2(AVFrame* input);
//获取处理之后的结果
int getFrame(AVFrame* result);
void dumpGraph();
void destroy();
};
#endif
//video_filter.cpp
#include "video_filter.h"
int VideoFilter::create(const char *filter_descr, VideoConfig *inConfig, VideoConfig *outConfig) {
this->description = filter_descr;
char args[512];
int ret = 0;
const AVFilter *buffersrc = avfilter_get_by_name("buffer");
const AVFilter *buffersink = avfilter_get_by_name("buffersink");
AVFilterInOut *outputs = avfilter_inout_alloc();
AVFilterInOut *inputs = avfilter_inout_alloc();
enum AVPixelFormat pix_fmts[] = {
outConfig->format, AV_PIX_FMT_NONE};
filter_graph = avfilter_graph_alloc();
if (!outputs || !inputs || !filter_graph) {
ret = AVERROR(ENOMEM);
goto end;
}
snprintf(args, sizeof(args),
"video_size=%dx%d:pix_fmt=%d:time_base=%d/%d:pixel_aspect=%d/%d",
inConfig->width, inConfig->height, inConfig->format,
inConfig->timebase.num, inConfig->timebase.den,
inConfig->pixel_aspect.num, inConfig->pixel_aspect.num);
ret = avfilter_graph_create_filter(&buffersrc_ctx, buffersrc, "in",
args, nullptr, filter_graph);
if (ret < 0) {
goto end;
}
ret = avfilter_graph_create_filter(&buffersink_ctx, buffersink, "out",
nullptr, nullptr, filter_graph);
if (ret < 0) {
goto end;
}
ret = av_opt_set_int_list(buffersink_ctx, "pix_fmts", pix_fmts,
AV_PIX_FMT_NONE, AV_OPT_SEARCH_CHILDREN);
if (ret < 0) {
goto end;
}
outputs->name = av_strdup("in");
outputs->filter_ctx = buffersrc_ctx;
outputs->pad_idx = 0;
outputs->next = nullptr;
inputs->name = av_strdup("out");
inputs->filter_ctx = buffersink_ctx;
inputs->pad_idx = 0;
inputs->next = nullptr;
if ((ret = avfilter_graph_parse_ptr(filter_graph, filter_descr,
&inputs, &outputs, nullptr)) < 0) {
goto end;
}
if ((ret = avfilter_graph_config(filter_graph, nullptr)) < 0) {
goto end;
}
end:
avfilter_inout_free(&inputs);
avfilter_inout_free(&outputs);
return ret;
}
int VideoFilter::create(const char *filter_descr, VideoConfig *inConfig1, VideoConfig *inConfig2, VideoConfig *outConfig) {
this->description = filter_descr;
char args1[512];
char args2[512];
int ret = 0;
const AVFilter *buffersrc1 = avfilter_get_by_name("buffer");
const AVFilter *buffersrc2 = avfilter_get_by_name("buffer");
const AVFilter *buffersink = avfilter_get_by_name("buffersink");
AVFilterInOut *outputs = avfilter_inout_alloc();
AVFilterInOut *inputs = avfilter_inout_alloc();
AVFilterInOut *full_output = avfilter_inout_alloc();
enum AVPixelFormat pix_fmts[] = {
outConfig->format, AV_PIX_FMT_NONE };
filter_graph = avfilter_graph_alloc();
if (!outputs || !inputs || !filter_graph || !full_output) {
ret = AVERROR(ENOMEM);
goto end;
}
snprintf(args1, sizeof(args1),
"video_size=%dx%d:pix_fmt=%d:time_base=%d/%d:pixel_aspect=%d/%d",
inConfig1->width, inConfig1->height, inConfig1->format,
inConfig1->timebase.num, inConfig1->timebase.den,
inConfig1->pixel_aspect.num, inConfig1->pixel_aspect.num);
snprintf(args2, sizeof(args2),
"video_size=%dx%d:pix_fmt=%d:time_base=%d/%d:pixel_aspect=%d/%d",
inConfig2->width, inConfig2->height, inConfig2->format,
inConfig2->timebase.num, inConfig2->timebase.den,
inConfig2->pixel_aspect.num, inConfig2->pixel_aspect.num);
ret = avfilter_graph_create_filter(&buffersrc_ctx, buffersrc1, "in",
args1, nullptr, filter_graph);
if (ret < 0) {
goto end;
}
ret = avfilter_graph_create_filter(&buffersrc_ctx1, buffersrc2, "in1",
args2, nullptr, filter_graph);
if (ret < 0) {
goto end;
}
ret = avfilter_graph_create_filter(&buffersink_ctx, buffersink, "out",
nullptr, nullptr, filter_graph);
if (ret < 0) {
goto end;
}
ret = av_opt_set_int_list(buffersink_ctx, "pix_fmts", pix_fmts,
AV_PIX_FMT_NONE, AV_OPT_SEARCH_CHILDREN);
if (ret < 0) {
goto end;
}
outputs->name = av_strdup("in");
outputs->filter_ctx = buffersrc_ctx;
outputs->pad_idx = 0;
outputs->next = full_output;
full_output->name = av_strdup("in1");
full_output->pad_idx = 0;
full_output->filter_ctx = buffersrc_ctx1;
full_output->next = NULL;
inputs->name = av_strdup("out");
inputs->filter_ctx = buffersink_ctx;
inputs->pad_idx = 0;
inputs->next = nullptr;
if ((ret = avfilter_graph_parse_ptr(filter_graph, filter_descr,
&inputs, &outputs, nullptr)) < 0) {
goto end;
}
if ((ret = avfilter_graph_config(filter_graph, nullptr)) < 0) {
goto end;
}
end:
avfilter_inout_free(&inputs);
avfilter_inout_free(&outputs);
return ret;
}
int VideoFilter::filter(AVFrame *source, AVFrame *dest) {
int ret = av_buffersrc_add_frame_flags(buffersrc_ctx, source, AV_BUFFERSRC_FLAG_KEEP_REF);
if (ret < 0) {
return -1;
}
ret = av_buffersink_get_frame(buffersink_ctx, dest);
if (ret < 0) {
return -1;
}
return 0;
}
int VideoFilter::filter(AVFrame * source1, AVFrame * source2, AVFrame * dest)
{
int ret = av_buffersrc_add_frame_flags(buffersrc_ctx, source1, AV_BUFFERSRC_FLAG_KEEP_REF);
if (ret < 0) {
return -1;
}
ret = av_buffersrc_add_frame_flags(buffersrc_ctx1, source2, AV_BUFFERSRC_FLAG_KEEP_REF);
if (ret < 0) {
return -1;
}
ret = av_buffersink_get_frame(buffersink_ctx, dest);
if (ret < 0) {
return -1;
}
return 0;
}
int VideoFilter::addInput1(AVFrame * input)
{
return av_buffersrc_add_frame_flags(buffersrc_ctx, input, AV_BUFFERSRC_FLAG_KEEP_REF);
}
int VideoFilter::addInput2(AVFrame * input)
{
return av_buffersrc_add_frame_flags(buffersrc_ctx1, input, AV_BUFFERSRC_FLAG_KEEP_REF);
}
int VideoFilter::getFrame(AVFrame * result)
{
return av_buffersink_get_frame(buffersink_ctx, result);
}
void VideoFilter::dumpGraph() {
printf("%s:%s", description, avfilter_graph_dump(filter_graph, nullptr));
}
void VideoFilter::destroy() {
if (filter_graph)
avfilter_graph_free(&filter_graph);
}
2.截取视频生成gif
封装完毕视频滤镜之后，我们就可以读取视频的数据帧，修改数据结构输出gif动图了，对应的实现如下:
#pragma execution_character_set("utf-8")
#define _CRT_SECURE_NO_WARNINGS
#include <string>
#include <iostream>
#include <thread>
#include <memory>
#include <iostream>
#include <fstream>
extern "C"
{
#include "libavutil/opt.h"
#include <libavcodec/avcodec.h>
#include <libavfilter/avfilter.h>
#include <libavformat/avformat.h>
#include <libavformat/avio.h>
#include <libavutil/avutil.h>
#include <libswresample/swresample.h>
#include <libswscale/swscale.h>
#include <libavutil/frame.h>
#include <libavutil/time.h>
#include <libavfilter/buffersink.h>
#include <libavfilter/buffersrc.h>
#include <libavutil/imgutils.h>
}
#include "video_filter.h"
//@1输入视频文件的地址
//@2输出gif文件的地址
//@3起始时间(s)
//@4结束时间(s)
int convert_video_to_gif(const char *output_filename, const char *input_filename, float from, float to)
{
if (from < 0 || from >= to) {
return -1;
}
AVFormatContext *inFmtCtx = nullptr;
AVFormatContext *outFmtCtx = nullptr;
AVCodecContext *videoCodecCtx = nullptr;
AVCodecContext *gifCodecCtx = nullptr;
int ret = 0;
//打开输入文件
ret = avformat_open_input(&inFmtCtx, input_filename, nullptr, nullptr);
ret = avformat_find_stream_info(inFmtCtx, nullptr);
ret = avformat_alloc_output_context2(&outFmtCtx, nullptr, nullptr, output_filename);
int video_idx = 0;
for (int i = 0; i < inFmtCtx->nb_streams; ++i)
{
if (inFmtCtx->streams[i]->codecpar->codec_type == AVMEDIA_TYPE_VIDEO)
{
video_idx = i;
AVStream *inVideoStream = inFmtCtx->streams[i];
AVStream *outVideoStream = avformat_new_stream(outFmtCtx, nullptr);
//创建输出流和对应的编码器
av_dict_copy(&outVideoStream->metadata, inVideoStream->metadata, 0);
const AVCodec *inCodec = avcodec_find_decoder(inVideoStream->codecpar->codec_id);
videoCodecCtx = avcodec_alloc_context3(inCodec);
ret = avcodec_parameters_to_context(videoCodecCtx, inVideoStream->codecpar);
ret = avcodec_open2(videoCodecCtx, inCodec, nullptr);
const AVCodec *codec = avcodec_find_encoder(AV_CODEC_ID_GIF);
gifCodecCtx = avcodec_alloc_context3(codec);
gifCodecCtx->codec_id = AV_CODEC_ID_GIF;
gifCodecCtx->time_base = {
1, 30 };
gifCodecCtx->bit_rate = 100000;
gifCodecCtx->pix_fmt = AV_PIX_FMT_RGB8;
gifCodecCtx->width = 600;
gifCodecCtx->height = 400;
ret = avcodec_open2(gifCodecCtx, codec, nullptr);
ret = avcodec_parameters_from_context(outVideoStream->codecpar, gifCodecCtx);
}
}
av_dict_copy(&outFmtCtx->metadata, inFmtCtx->metadata, 0);
if (!(outFmtCtx->flags & AVFMT_NOFILE)) {
ret = avio_open(&outFmtCtx->pb, output_filename, AVIO_FLAG_WRITE);
}
//写gif文件头
ret = avformat_write_header(outFmtCtx, nullptr);
int64_t first_pts = av_rescale_q_rnd((int64_t)from, AVRational{
1,1 },
inFmtCtx->streams[video_idx]->time_base,
AV_ROUND_DOWN);
av_seek_frame(inFmtCtx, video_idx, first_pts, AVSEEK_FLAG_BACKWARD);
//通过滤镜缩放视频,修改图片像素格式
VideoFilter *filter = nullptr;
filter = new VideoFilter();
char filter_descr[128];
snprintf(filter_descr, sizeof(filter_descr), "scale=600:400,format=pix_fmts=%s",
av_get_pix_fmt_name(gifCodecCtx->pix_fmt));
VideoConfig in(videoCodecCtx->pix_fmt, videoCodecCtx->width, videoCodecCtx->height);
VideoConfig out(gifCodecCtx->pix_fmt, gifCodecCtx->width, gifCodecCtx->height);
filter->create(filter_descr, &in, &out);
int gif_pts = 0;
int index = 0;
while (true) {
AVPacket packet{
0 };
av_init_packet(&packet);
ret = av_read_frame(inFmtCtx, &packet);
if (ret < 0)
{
break;
}
if (av_compare_ts(packet.pts, inFmtCtx->streams[packet.stream_index]->time_base,
(int64_t)(to * 10), AVRational{
1, 10 }) >= 0) {
break;
}
//读取数据帧并进行输出
if (packet.stream_index == video_idx)
{
AVFrame *frame = av_frame_alloc();
ret = avcodec_send_packet(videoCodecCtx, &packet);
if (ret < 0) {
continue;
}
ret = avcodec_receive_frame(videoCodecCtx, frame);
if (ret < 0) {
continue;
}
filter->filter(frame, frame);
frame->pts = gif_pts++;
ret = avcodec_send_frame(gifCodecCtx, frame);
if (ret < 0) {
continue;
}
av_frame_free(&frame);
AVPacket gifPkt{
0 };
av_init_packet(&gifPkt);
ret = avcodec_receive_packet(gifCodecCtx, &gifPkt);
if (ret < 0)
{
continue;
}
gifPkt.stream_index = 0;
av_packet_rescale_ts(&gifPkt, gifCodecCtx->time_base,
outFmtCtx->streams[0]->time_base);
ret = av_interleaved_write_frame(outFmtCtx, &gifPkt);
}
}
if (filter != nullptr) {
filter->destroy();
delete filter;
}
avformat_close_input(&inFmtCtx);
avformat_free_context(inFmtCtx);
avformat_free_context(outFmtCtx);
return 0;
}
int main(int argc, char* argv[])
{
if (argc != 3)
{
printf("usage:%1 input filepath %2 outputfilepath");
return -1;
}
//输入视频文件地址,输出gif的地址
std::string fileInput = std::string(argv[1]);
std::string fileOutput = std::string(argv[2]);
avformat_network_init();
convert_video_to_gif(fileOutput.c_str(), fileInput.c_str(), 30, 40);
}
3.gif优化
由于原始视频的像素比较高，帧率也比较高，这样截取出来的gif可能比较大。对于gif尺寸比较大，优化策略主要包括以下三点:
1.缩短截取的时间长度
2.对图片尺寸进行缩放处理
3.通过抽帧来降低视频的帧率
抽帧可能降低动图的流畅度，需要提起注意。
4.示例效果
