example/muxing/camm_muxing.cc - mp4v2 - Git at Google

 /*
  * This code sample muxes an artificially generated video stream with a
  * data stream for camera motion metadata (CAMM) into a single mp4 file.
  * The sample can be built with the "make" command.
  * After building the sample, it can be run as follows:
  *
  *    > ./camm_muxing output_file.mp4
  *
  * Note that this must be run in the sample directory as the "sample.jpg" file.
  *
  * The video stream is a single JPEG file for the entire video duration. It
  * is for illustration purposes only. The calling code is expected to
  * already have the encoded bytes of video frames to mux into the mp4 file.
  */

 #include <endian.h>
 #include <math.h>
 #include <mp4v2/mp4v2.h>
 #include <stdio.h>
 #include <stdlib.h>
 #include <sys/stat.h>
 #include <cstring>
 #include <fstream>
 #include <iostream>
 #include <vector>

 #define FRAME_WIDTH 2880
 #define FRAME_HEIGHT 1800
 #define VIDEO_TIME_SECONDS 10
 #define VIDEO_TIME_SCALE 90000
 #define NUM_CAMM_SAMPLES 200
 #define BYTE_BUFFER_SIZE 1000

 namespace {
 // Number of bytes for the packet type in a CAMM data packet.
 const int kCammPacketBytes = 4;
 // These sizes are fixed when building with IEEE floating point format.
 const int kMetadataTypeSizes[] = {
     // 3 float angle_axis values.
     3 * sizeof(float),
     // 2 int64 values for pixel_exposure_time and rolling_shutter_skew_time.
     2 * sizeof(uint64_t),
     // 3 float gyro values.
     3 * sizeof(float),
     // 3 float acceleration values.
     3 * sizeof(float),
     // 3 float position values.
     3 * sizeof(float),
     // 3 floats for latitude, longitude, and altitude.
     3 * sizeof(float),
     // 3 doubles for time_gps_epoch, latitude, and longitude
     // 1 int32 for gps_fix_type
     // 7 floats for altitude, horizontal_accuracy, vertical_accuracy,
     // vertical_east, vertical_north, vertical_up, and speed_accuracy.
     3 * sizeof(double) + sizeof(uint32_t) + 7 * sizeof(float),
     // 3 floats for magnetic field.
     3 * sizeof(float)};

 static uint64_t double_to_bytes(double d) {
   uint64_t result;
   memcpy(&result, &d, 8);
   return htole64(result);
 }

 static uint32_t float_to_bytes(float f) {
   uint32_t result;
   memcpy(&result, &f, 4);
   return htole32(result);
 }

 // Writes sample camm data in a byte stream format for a single packet.
 // Returns the number of bytes written.
 // In real code, you are not required to write all packet types like in this
 // example. The pts of each packet also does not need to be equally spaced.
 static int WriteSampleCammBytes(void *bytes, int packet_number) {
   uint32_t angle_x, angle_y, angle_z, gyro_x, gyro_y, gyro_z, acceleration_x,
       acceleration_y, acceleration_z, position_x, position_y, position_z,
       latitude32, longitude32, gps_fix_type, horizontal_accuracy_meters,
       vertical_accuracy_meters, vertical_east_velocity_mps,
       vertical_north_velocity_mps, speed_accuracy_mps, altitude,
       magnetic_field_x, magnetic_field_y, magnetic_field_z,
       pixel_exposure_nanoseconds, rolling_shutter_skew_time,
       vertical_up_velocity_mps;
   uint64_t latitude64, longitude64, time_gps_epoch;
   memset(bytes, 0, BYTE_BUFFER_SIZE);
   uint16_t *camm_data = reinterpret_cast<uint16_t *>(bytes);
   uint16_t packet_type = packet_number % 8;
   // The packet type comes after the first 2 reserved bytes.
   camm_data[1] = htole16(packet_type);
   camm_data += 2;
   switch (packet_type) {
     case 0:
       angle_x = float_to_bytes(M_PI / 2);
       angle_y = float_to_bytes(-M_PI / 2);
       angle_z =
           float_to_bytes(fmod(packet_number * M_PI / 20, 2 * M_PI) - M_PI);
       memcpy(camm_data, &angle_x, 4);
       memcpy(camm_data + 2, &angle_y, 4);
       memcpy(camm_data + 4, &angle_z, 4);
       break;
     case 1:
       pixel_exposure_nanoseconds = double_to_bytes(500);
       rolling_shutter_skew_time = double_to_bytes(300);
       memcpy(camm_data, &pixel_exposure_nanoseconds, 8);
       memcpy(camm_data + 4, &rolling_shutter_skew_time, 8);
       break;
     case 2:
       gyro_x = float_to_bytes(M_PI / 20);
       gyro_y = float_to_bytes(2 * M_PI / 20);
       gyro_z = float_to_bytes(3 * M_PI / 20);
       memcpy(camm_data, &gyro_x, 4);
       memcpy(camm_data + 2, &gyro_y, 4);
       memcpy(camm_data + 4, &gyro_z, 4);
       break;
     case 3:
       acceleration_x = float_to_bytes(0.1);
       acceleration_y = float_to_bytes(0.2);
       acceleration_z = float_to_bytes(0.3);
       memcpy(camm_data, &acceleration_x, 4);
       memcpy(camm_data + 2, &acceleration_y, 4);
       memcpy(camm_data + 4, &acceleration_z, 4);
       break;
     case 4:
       position_x = 0;
       position_y = 0;
       position_z = 0;
       memcpy(camm_data, &position_x, 4);
       memcpy(camm_data + 2, &position_y, 4);
       memcpy(camm_data + 4, &position_z, 4);
       break;
     case 5:
       latitude32 = float_to_bytes(37.454356 + .001 * packet_number);
       longitude32 = float_to_bytes(-122.167477 + 0.001 * packet_number);
       altitude = 0;
       memcpy(camm_data, &latitude32, 4);
       memcpy(camm_data + 2, &longitude32, 4);
       memcpy(camm_data + 4, &altitude, 4);
       break;
     case 6:
       time_gps_epoch = double_to_bytes(1500507374.825 + packet_number);
       memcpy(camm_data, &time_gps_epoch, 8);
       camm_data += 4;
       gps_fix_type = 0;
       memcpy(camm_data, &gps_fix_type, 4);
       camm_data += 2;
       latitude64 = double_to_bytes(37.454356 + .001 * packet_number);
       memcpy(camm_data, &latitude64, 8);
       camm_data += 4;
       longitude64 = double_to_bytes(-122.167477 + .001 * packet_number);
       memcpy(camm_data, &longitude64, 8);
       camm_data += 4;
       altitude = 0;
       memcpy(camm_data, &altitude, 4);
       camm_data += 2;
       horizontal_accuracy_meters = float_to_bytes(7.5);
       memcpy(camm_data, &horizontal_accuracy_meters, 4);
       camm_data += 2;
       vertical_accuracy_meters = float_to_bytes(10.5);
       memcpy(camm_data, &vertical_accuracy_meters, 4);
       camm_data += 2;
       vertical_east_velocity_mps = float_to_bytes(1.1);
       memcpy(camm_data, &vertical_east_velocity_mps, 4);
       camm_data += 2;
       vertical_north_velocity_mps = float_to_bytes(1.1);
       memcpy(camm_data, &vertical_north_velocity_mps, 4);
       camm_data += 2;
       vertical_up_velocity_mps = 0;
       memcpy(camm_data, &vertical_up_velocity_mps, 4);
       camm_data += 2;
       speed_accuracy_mps = float_to_bytes(2.5);
       memcpy(camm_data, &speed_accuracy_mps, 4);
       break;
     case 7:
       magnetic_field_x = float_to_bytes(0.01);
       magnetic_field_y = float_to_bytes(0.01);
       magnetic_field_z = float_to_bytes(0.01);
       memcpy(camm_data, &magnetic_field_x, 4);
       memcpy(camm_data + 2, &magnetic_field_y, 4);
       memcpy(camm_data + 4, &magnetic_field_z, 4);
       break;
     default:
       break;
   }
   return kCammPacketBytes + kMetadataTypeSizes[packet_type];
 }

 static void WriteCammPacket(MP4FileHandle handle, int camm_track_id, int pts,
                             int packet_num, char *buffer) {
   int packet_size = WriteSampleCammBytes(buffer, packet_num);
   if (!MP4WriteSample(handle, camm_track_id,
                       reinterpret_cast<uint8_t *>(buffer), packet_size,
                       /* duration */ 0,
                       /* rendering_offset */ pts,
                       /* is_sync_sample */ true)) {
     printf("Failed to write camm track sample\n");
     exit(1);
   }
 }

 static void WriteVideoFrame(MP4FileHandle handle, int video_track_id,
                             const std::vector<char> &jpg_bytes, int pts,
                             int duration) {
   // This code sample doesn't illustrate how to encode video. An encoder
   // for h264 or another format should be used to do that, and then call
   // MP4WriteSample with the bytes returned from the encoder.
   if (!MP4WriteSample(handle, video_track_id,
                       reinterpret_cast<const uint8_t *>(jpg_bytes.data()),
                       /* num_bytes */ jpg_bytes.size(),
                       /* duration */ VIDEO_TIME_SCALE,
                       /* rendering_offset */ pts,
                       /* is_sync_sample */ true)) {
     printf("Failed to write mp4 sample to video track\n");
     exit(1);
   }
 }

 static void WriteStreams(MP4FileHandle handle, int video_track_id,
                          const std::vector<char> &jpg_bytes,
                          int camm_track_id) {
   char camm_buffer[BYTE_BUFFER_SIZE];
   int video_pts = 0;
   int camm_pts = 0;
   int camm_packet_num = 0;
   int max_pts = VIDEO_TIME_SECONDS * VIDEO_TIME_SCALE;
   while (video_pts < max_pts && camm_pts < max_pts) {
     if (video_pts <= camm_pts) {
       WriteVideoFrame(handle, video_track_id, jpg_bytes, video_pts,
                       VIDEO_TIME_SCALE);
       video_pts += VIDEO_TIME_SCALE;
     } else {
       WriteCammPacket(handle, camm_track_id, camm_pts, camm_packet_num,
                       camm_buffer);
       camm_pts += max_pts / NUM_CAMM_SAMPLES;
       camm_packet_num++;
     }
   }
 }

 static int AddVideoTrack(MP4FileHandle handle) {
   printf("Adding video track\n");
   int video_track_id = MP4AddVideoTrack(
       handle, VIDEO_TIME_SCALE, VIDEO_TIME_SCALE * VIDEO_TIME_SECONDS,
       FRAME_WIDTH, FRAME_HEIGHT, MP4_JPEG_VIDEO_TYPE);
   if (video_track_id == MP4_INVALID_TRACK_ID) {
     printf("Can't add video track\n");
     exit(1);
   }
   return video_track_id;
 }

 static int AddCammTrack(MP4FileHandle handle) {
   printf("Adding camm track\n");
   int data_track_id = MP4AddTrack(handle, "camm", VIDEO_TIME_SCALE);
   if (data_track_id == MP4_INVALID_TRACK_ID) {
     printf("Can't add camm data track\n");
     exit(1);
   }
   if (!MP4SetTrackName(handle, data_track_id, "camm")) {
     printf("Couldn't set the track name of the CAMM track.\n");
     exit(1);
   }
   return data_track_id;
 }

 // Reads the contents of the sample jpeg file into memory.
 static std::vector<char> ReadSampleFileBytes() {
   struct stat buffer;
   const char *filename = "sample.jpg";
   if (stat(filename, &buffer) != 0) {
     printf(
         "Sample file 'sample.jpg' doesn't exist\n"
         "Make sure it is in the directory and then run the program\n"
         "again.\n");
     exit(1);
   }
   std::ifstream ifs(filename, std::ios::binary | std::ios::ate);
   std::ifstream::pos_type pos = ifs.tellg();
   std::vector<char> result(pos);
   ifs.seekg(0, std::ios::beg);
   ifs.read(&result[0], pos);
   return result;
 }

 static void VerifyProgramArgs(int argc, char **argv) {
   if (argc != 2) {
     printf(
         "Usage: %s file.mp4\n"
         "file.mp4 is the output location of the generated video\n",
         argv[0]);
     exit(1);
   }
 }
 }  // namespace

 int main(int argc, char **argv) {
 #ifndef __STDC_IEC_559__
   fprintf(stderr,
           "Please ensure your compiler uses IEEE 754 \n"
           "floating point representation. If so, you may safely comment out \n"
           "this guard.\n");
   exit(1);
 #endif

   VerifyProgramArgs(argc, argv);
   std::vector<char> jpg_bytes = ReadSampleFileBytes();
   MP4FileHandle handle = MP4Create(argv[1], /* flags */ 0);
   int video_track_id = AddVideoTrack(handle);
   int camm_track_id = AddCammTrack(handle);

   WriteStreams(handle, video_track_id, jpg_bytes, camm_track_id);

   printf("Dumping MP4 file information:\n");
   MP4Dump(handle);
   MP4Close(handle);
   printf("Done!\n");
   return 0;
 }
	/*
	* This code sample muxes an artificially generated video stream with a
	* data stream for camera motion metadata (CAMM) into a single mp4 file.
	* The sample can be built with the "make" command.
	* After building the sample, it can be run as follows:
	*
	* > ./camm_muxing output_file.mp4
	*
	* Note that this must be run in the sample directory as the "sample.jpg" file.
	*
	* The video stream is a single JPEG file for the entire video duration. It
	* is for illustration purposes only. The calling code is expected to
	* already have the encoded bytes of video frames to mux into the mp4 file.
	*/

	#include <endian.h>
	#include <math.h>
	#include <mp4v2/mp4v2.h>
	#include <stdio.h>
	#include <stdlib.h>
	#include <sys/stat.h>
	#include <cstring>
	#include <fstream>
	#include <iostream>
	#include <vector>

	#define FRAME_WIDTH 2880
	#define FRAME_HEIGHT 1800
	#define VIDEO_TIME_SECONDS 10
	#define VIDEO_TIME_SCALE 90000
	#define NUM_CAMM_SAMPLES 200
	#define BYTE_BUFFER_SIZE 1000

	namespace {
	// Number of bytes for the packet type in a CAMM data packet.
	const int kCammPacketBytes = 4;
	// These sizes are fixed when building with IEEE floating point format.
	const int kMetadataTypeSizes[] = {
	// 3 float angle_axis values.
	3 * sizeof(float),
	// 2 int64 values for pixel_exposure_time and rolling_shutter_skew_time.
	2 * sizeof(uint64_t),
	// 3 float gyro values.
	3 * sizeof(float),
	// 3 float acceleration values.
	3 * sizeof(float),
	// 3 float position values.
	3 * sizeof(float),
	// 3 floats for latitude, longitude, and altitude.
	3 * sizeof(float),
	// 3 doubles for time_gps_epoch, latitude, and longitude
	// 1 int32 for gps_fix_type
	// 7 floats for altitude, horizontal_accuracy, vertical_accuracy,
	// vertical_east, vertical_north, vertical_up, and speed_accuracy.
	3 * sizeof(double) + sizeof(uint32_t) + 7 * sizeof(float),
	// 3 floats for magnetic field.
	3 * sizeof(float)};

	static uint64_t double_to_bytes(double d) {
	uint64_t result;
	memcpy(&result, &d, 8);
	return htole64(result);
	}

	static uint32_t float_to_bytes(float f) {
	uint32_t result;
	memcpy(&result, &f, 4);
	return htole32(result);
	}

	// Writes sample camm data in a byte stream format for a single packet.
	// Returns the number of bytes written.
	// In real code, you are not required to write all packet types like in this
	// example. The pts of each packet also does not need to be equally spaced.
	static int WriteSampleCammBytes(void *bytes, int packet_number) {
	uint32_t angle_x, angle_y, angle_z, gyro_x, gyro_y, gyro_z, acceleration_x,
	acceleration_y, acceleration_z, position_x, position_y, position_z,
	latitude32, longitude32, gps_fix_type, horizontal_accuracy_meters,
	vertical_accuracy_meters, vertical_east_velocity_mps,
	vertical_north_velocity_mps, speed_accuracy_mps, altitude,
	magnetic_field_x, magnetic_field_y, magnetic_field_z,
	pixel_exposure_nanoseconds, rolling_shutter_skew_time,
	vertical_up_velocity_mps;
	uint64_t latitude64, longitude64, time_gps_epoch;
	memset(bytes, 0, BYTE_BUFFER_SIZE);
	uint16_t camm_data = reinterpret_cast<uint16_t >(bytes);
	uint16_t packet_type = packet_number % 8;
	// The packet type comes after the first 2 reserved bytes.
	camm_data[1] = htole16(packet_type);
	camm_data += 2;
	switch (packet_type) {
	case 0:
	angle_x = float_to_bytes(M_PI / 2);
	angle_y = float_to_bytes(-M_PI / 2);
	angle_z =
	float_to_bytes(fmod(packet_number * M_PI / 20, 2 * M_PI) - M_PI);
	memcpy(camm_data, &angle_x, 4);
	memcpy(camm_data + 2, &angle_y, 4);
	memcpy(camm_data + 4, &angle_z, 4);
	break;
	case 1:
	pixel_exposure_nanoseconds = double_to_bytes(500);
	rolling_shutter_skew_time = double_to_bytes(300);
	memcpy(camm_data, &pixel_exposure_nanoseconds, 8);
	memcpy(camm_data + 4, &rolling_shutter_skew_time, 8);
	break;
	case 2:
	gyro_x = float_to_bytes(M_PI / 20);
	gyro_y = float_to_bytes(2 * M_PI / 20);
	gyro_z = float_to_bytes(3 * M_PI / 20);
	memcpy(camm_data, &gyro_x, 4);
	memcpy(camm_data + 2, &gyro_y, 4);
	memcpy(camm_data + 4, &gyro_z, 4);
	break;
	case 3:
	acceleration_x = float_to_bytes(0.1);
	acceleration_y = float_to_bytes(0.2);
	acceleration_z = float_to_bytes(0.3);
	memcpy(camm_data, &acceleration_x, 4);
	memcpy(camm_data + 2, &acceleration_y, 4);
	memcpy(camm_data + 4, &acceleration_z, 4);
	break;
	case 4:
	position_x = 0;
	position_y = 0;
	position_z = 0;
	memcpy(camm_data, &position_x, 4);
	memcpy(camm_data + 2, &position_y, 4);
	memcpy(camm_data + 4, &position_z, 4);
	break;
	case 5:
	latitude32 = float_to_bytes(37.454356 + .001 * packet_number);
	longitude32 = float_to_bytes(-122.167477 + 0.001 * packet_number);
	altitude = 0;
	memcpy(camm_data, &latitude32, 4);
	memcpy(camm_data + 2, &longitude32, 4);
	memcpy(camm_data + 4, &altitude, 4);
	break;
	case 6:
	time_gps_epoch = double_to_bytes(1500507374.825 + packet_number);
	memcpy(camm_data, &time_gps_epoch, 8);
	camm_data += 4;
	gps_fix_type = 0;
	memcpy(camm_data, &gps_fix_type, 4);
	camm_data += 2;
	latitude64 = double_to_bytes(37.454356 + .001 * packet_number);
	memcpy(camm_data, &latitude64, 8);
	camm_data += 4;
	longitude64 = double_to_bytes(-122.167477 + .001 * packet_number);
	memcpy(camm_data, &longitude64, 8);
	camm_data += 4;
	altitude = 0;
	memcpy(camm_data, &altitude, 4);
	camm_data += 2;
	horizontal_accuracy_meters = float_to_bytes(7.5);
	memcpy(camm_data, &horizontal_accuracy_meters, 4);
	camm_data += 2;
	vertical_accuracy_meters = float_to_bytes(10.5);
	memcpy(camm_data, &vertical_accuracy_meters, 4);
	camm_data += 2;
	vertical_east_velocity_mps = float_to_bytes(1.1);
	memcpy(camm_data, &vertical_east_velocity_mps, 4);
	camm_data += 2;
	vertical_north_velocity_mps = float_to_bytes(1.1);
	memcpy(camm_data, &vertical_north_velocity_mps, 4);
	camm_data += 2;
	vertical_up_velocity_mps = 0;
	memcpy(camm_data, &vertical_up_velocity_mps, 4);
	camm_data += 2;
	speed_accuracy_mps = float_to_bytes(2.5);
	memcpy(camm_data, &speed_accuracy_mps, 4);
	break;
	case 7:
	magnetic_field_x = float_to_bytes(0.01);
	magnetic_field_y = float_to_bytes(0.01);
	magnetic_field_z = float_to_bytes(0.01);
	memcpy(camm_data, &magnetic_field_x, 4);
	memcpy(camm_data + 2, &magnetic_field_y, 4);
	memcpy(camm_data + 4, &magnetic_field_z, 4);
	break;
	default:
	break;
	}
	return kCammPacketBytes + kMetadataTypeSizes[packet_type];
	}

	static void WriteCammPacket(MP4FileHandle handle, int camm_track_id, int pts,
	int packet_num, char *buffer) {
	int packet_size = WriteSampleCammBytes(buffer, packet_num);
	if (!MP4WriteSample(handle, camm_track_id,
	reinterpret_cast<uint8_t *>(buffer), packet_size,
	/* duration */ 0,
	/* rendering_offset */ pts,
	/* is_sync_sample */ true)) {
	printf("Failed to write camm track sample\n");
	exit(1);
	}
	}

	static void WriteVideoFrame(MP4FileHandle handle, int video_track_id,
	const std::vector<char> &jpg_bytes, int pts,
	int duration) {
	// This code sample doesn't illustrate how to encode video. An encoder
	// for h264 or another format should be used to do that, and then call
	// MP4WriteSample with the bytes returned from the encoder.
	if (!MP4WriteSample(handle, video_track_id,
	reinterpret_cast<const uint8_t *>(jpg_bytes.data()),
	/* num_bytes */ jpg_bytes.size(),
	/* duration */ VIDEO_TIME_SCALE,
	/* rendering_offset */ pts,
	/* is_sync_sample */ true)) {
	printf("Failed to write mp4 sample to video track\n");
	exit(1);
	}
	}

	static void WriteStreams(MP4FileHandle handle, int video_track_id,
	const std::vector<char> &jpg_bytes,
	int camm_track_id) {
	char camm_buffer[BYTE_BUFFER_SIZE];
	int video_pts = 0;
	int camm_pts = 0;
	int camm_packet_num = 0;
	int max_pts = VIDEO_TIME_SECONDS * VIDEO_TIME_SCALE;
	while (video_pts < max_pts && camm_pts < max_pts) {
	if (video_pts <= camm_pts) {
	WriteVideoFrame(handle, video_track_id, jpg_bytes, video_pts,
	VIDEO_TIME_SCALE);
	video_pts += VIDEO_TIME_SCALE;
	} else {
	WriteCammPacket(handle, camm_track_id, camm_pts, camm_packet_num,
	camm_buffer);
	camm_pts += max_pts / NUM_CAMM_SAMPLES;
	camm_packet_num++;
	}
	}
	}

	static int AddVideoTrack(MP4FileHandle handle) {
	printf("Adding video track\n");
	int video_track_id = MP4AddVideoTrack(
	handle, VIDEO_TIME_SCALE, VIDEO_TIME_SCALE * VIDEO_TIME_SECONDS,
	FRAME_WIDTH, FRAME_HEIGHT, MP4_JPEG_VIDEO_TYPE);
	if (video_track_id == MP4_INVALID_TRACK_ID) {
	printf("Can't add video track\n");
	exit(1);
	}
	return video_track_id;
	}

	static int AddCammTrack(MP4FileHandle handle) {
	printf("Adding camm track\n");
	int data_track_id = MP4AddTrack(handle, "camm", VIDEO_TIME_SCALE);
	if (data_track_id == MP4_INVALID_TRACK_ID) {
	printf("Can't add camm data track\n");
	exit(1);
	}
	if (!MP4SetTrackName(handle, data_track_id, "camm")) {
	printf("Couldn't set the track name of the CAMM track.\n");
	exit(1);
	}
	return data_track_id;
	}

	// Reads the contents of the sample jpeg file into memory.
	static std::vector<char> ReadSampleFileBytes() {
	struct stat buffer;
	const char *filename = "sample.jpg";
	if (stat(filename, &buffer) != 0) {
	printf(
	"Sample file 'sample.jpg' doesn't exist\n"
	"Make sure it is in the directory and then run the program\n"
	"again.\n");
	exit(1);
	}
	std::ifstream ifs(filename, std::ios::binary \| std::ios::ate);
	std::ifstream::pos_type pos = ifs.tellg();
	std::vector<char> result(pos);
	ifs.seekg(0, std::ios::beg);
	ifs.read(&result[0], pos);
	return result;
	}

	static void VerifyProgramArgs(int argc, char **argv) {
	if (argc != 2) {
	printf(
	"Usage: %s file.mp4\n"
	"file.mp4 is the output location of the generated video\n",
	argv[0]);
	exit(1);
	}
	}
	} // namespace

	int main(int argc, char **argv) {
	#ifndef __STDC_IEC_559__
	fprintf(stderr,
	"Please ensure your compiler uses IEEE 754 \n"
	"floating point representation. If so, you may safely comment out \n"
	"this guard.\n");
	exit(1);
	#endif

	VerifyProgramArgs(argc, argv);
	std::vector<char> jpg_bytes = ReadSampleFileBytes();
	MP4FileHandle handle = MP4Create(argv[1], /* flags */ 0);
	int video_track_id = AddVideoTrack(handle);
	int camm_track_id = AddCammTrack(handle);

	WriteStreams(handle, video_track_id, jpg_bytes, camm_track_id);

	printf("Dumping MP4 file information:\n");
	MP4Dump(handle);
	MP4Close(handle);
	printf("Done!\n");
	return 0;
	}