---

fl::VideoInFileFFMPEG Class Reference

#include <video.h>

Inheritance diagram for fl::VideoInFileFFMPEG:

List of all members.

Public Methods
	VideoInFileFFMPEG (const std::string &fileName, const PixelFormat &hint)
virtual	~VideoInFileFFMPEG ()
virtual void	seekFrame (int frame)
	Position stream just before the given frame. Numbers follow same convention as Video class.
virtual void	seekTime (double timestamp)
virtual void	readNext (Image &image)
	Reads the next frame and stores it in image. image may end up attached to a buffer used internally by the video device or library, so it may be freed unexpectedly. However, this clss guarantees that the memory will not be freed before the next call to a method of this class.
virtual bool	good () const
virtual void	setTimestampMode (bool frames=false)
	Changes image.timestamp from presentation time to frame number.
virtual void	get (const std::string &name, double &value)
	Retrieve values of stream attributes (such as duration in seconds).
void	open (const std::string &fileName)
void	close ()
void	readNext (Image *image)
	same as readNext() above, except if image is null, then don't do extraction step
void	extractImage (Image &image)
Public Attributes
AVFormatContext *	fc
AVCodecContext *	cc
AVStream *	stream
AVCodec *	codec
AVFrame	picture
AVPacket	packet
	Ensure that if nextImage() attaches image to packet, the memory won't be freed before next read.
int	size
unsigned char *	data
int	gotPicture
	indicates that picture contains a full image
int	state
	state == 0 means good; anything else means we can't read more frames
bool	timestampMode
	Indicates that image.timestamp should be frame number rather than presentation time.
double	expectedSkew
	Compensates for difference between PTS and DTS when seeking. Units = frames.
bool	seekLinear
	Indicates that the file only supports linear seeking, not random seeking. Generally due to lack of timestamps in stream.
const PixelFormat *	hint
std::string	fileName

---

Constructor & Destructor Documentation

VideoInFileFFMPEG::VideoInFileFFMPEG (  const std::string &    fileName, const PixelFormat &    hint )

VideoInFileFFMPEG::~VideoInFileFFMPEG (    )  [virtual]

---

Member Function Documentation

void VideoInFileFFMPEG::close (    )

void VideoInFileFFMPEG::extractImage (  Image &    image )

void VideoInFileFFMPEG::get (  const std::string &    name, double &    value )  [virtual]

Retrieve values of stream attributes (such as duration in seconds). Implements fl::VideoInFile.

bool VideoInFileFFMPEG::good (    )  const [virtual]

Registry of states: 0 = everything good [-7,-1] = FFMPEG libavformat errors, see avformat.h -10 = did not find a video stream -11 = did not find a codec -12 = VideoInFile is closed Implements fl::VideoInFile.

void VideoInFileFFMPEG::open (  const std::string &    fileName )

void VideoInFileFFMPEG::readNext (  Image *    image )

same as readNext() above, except if image is null, then don't do extraction step

void VideoInFileFFMPEG::readNext (  Image &    image )  [virtual]

Reads the next frame and stores it in image. image may end up attached to a buffer used internally by the video device or library, so it may be freed unexpectedly. However, this clss guarantees that the memory will not be freed before the next call to a method of this class. Implements fl::VideoInFile.

void VideoInFileFFMPEG::seekFrame (  int    frame )  [virtual]

Position stream just before the given frame. Numbers follow same convention as Video class. Implements fl::VideoInFile.

void VideoInFileFFMPEG::seekTime (  double    timestamp )  [virtual]

Todo: Seek in c.avi (bear and dog rotating) is screwed up at the FFMPEG level. It has more index than video, and the index seems to be on a different time base than the video itself. Implements fl::VideoInFile.

void VideoInFileFFMPEG::setTimestampMode (  bool    frames = false )  [virtual]

Changes image.timestamp from presentation time to frame number. Implements fl::VideoInFile.

---

Member Data Documentation

AVCodecContext* fl::VideoInFileFFMPEG::cc

AVCodec* fl::VideoInFileFFMPEG::codec

unsigned char* fl::VideoInFileFFMPEG::data

double fl::VideoInFileFFMPEG::expectedSkew

Compensates for difference between PTS and DTS when seeking. Units = frames.

AVFormatContext* fl::VideoInFileFFMPEG::fc

std::string fl::VideoInFileFFMPEG::fileName

int fl::VideoInFileFFMPEG::gotPicture

indicates that picture contains a full image

const PixelFormat* fl::VideoInFileFFMPEG::hint

AVPacket fl::VideoInFileFFMPEG::packet

Ensure that if nextImage() attaches image to packet, the memory won't be freed before next read.

AVFrame fl::VideoInFileFFMPEG::picture

bool fl::VideoInFileFFMPEG::seekLinear

Indicates that the file only supports linear seeking, not random seeking. Generally due to lack of timestamps in stream.

int fl::VideoInFileFFMPEG::size

int fl::VideoInFileFFMPEG::state

state == 0 means good; anything else means we can't read more frames

AVStream* fl::VideoInFileFFMPEG::stream

bool fl::VideoInFileFFMPEG::timestampMode

Indicates that image.timestamp should be frame number rather than presentation time.

---

The documentation for this class was generated from the following files:

• video.h
• VideoFileFormatFFMPEG.cc

---