MayaFlux::IO::VideoEncodeContext Class Reference

RAII owner of one video stream's encoder and pixel-format converter on the write path. More...

#include <VideoEncodeContext.hpp>

Collaboration diagram for MayaFlux::IO::VideoEncodeContext:

Public Member Functions
void	close ()
	Release all owned resources.
bool	drain (FFmpegMuxContext &mux)
	Flush all buffered frames from the encoder to the mux.
bool	encode_frame (const uint8_t *src_data, size_t src_size, uint32_t src_width, uint32_t src_height, FFmpegMuxContext &mux)
	Encode one raw pixel frame into the mux context.
uint32_t	height () const
bool	is_valid () const
	True if codec, scaler, and scratch frame are all ready.
const std::string &	last_error () const
bool	open (FFmpegMuxContext &mux, uint32_t width, uint32_t height, double frame_rate, AVPixelFormat src_pixel_format, AVCodecID codec_id)
	Open the encoder and register a video stream in the mux context.
VideoEncodeContext &	operator= (const VideoEncodeContext &)=delete
VideoEncodeContext &	operator= (VideoEncodeContext &&)=delete
	VideoEncodeContext ()=default
	VideoEncodeContext (const VideoEncodeContext &)=delete
	VideoEncodeContext (VideoEncodeContext &&)=delete
uint32_t	width () const
	~VideoEncodeContext ()

Public Attributes
AVCodecContext *	codec_context {}
	Owned; freed in destructor.
SwsContext *	sws_context {}
	Owned; freed in destructor.

Private Member Functions
bool	drain_packets (FFmpegMuxContext &mux)

Private Attributes
int	m_cached_src_height { -1 }
int	m_cached_src_width { -1 }
uint32_t	m_height {}
std::string	m_last_error
int64_t	m_pts {}
AVFrame *	m_scratch_frame {}
	Owned scratch buffer for encoder input.
AVPixelFormat	m_src_pixel_fmt { AV_PIX_FMT_BGRA }
int	m_src_src_bpp { 4 }
AVStream *	m_stream {}
	Weak ref into FFmpegMuxContext; not owned.
int	m_stream_index { -1 }
uint32_t	m_width {}

Detailed Description

RAII owner of one video stream's encoder and pixel-format converter on the write path.

Encapsulates all video-stream-specific FFmpeg objects:

• AVCodecContext for the selected or inferred encoder
• SwsContext converting from the source pixel format (e.g. AV_PIX_FMT_BGRA, as delivered by the Vulkan swapchain readback) to the encoder's required format (typically AV_PIX_FMT_YUV420P for H.264/H.265)
• AVFrame scratch buffer for sws_scale output
• Monotonically increasing PTS counter

Does NOT own AVFormatContext — that belongs to FFmpegMuxContext.

Open sequence:

1. FFmpegMuxContext::open()
2. VideoEncodeContext::open() — adds stream, configures codec + sws
3. FFmpegMuxContext::write_header()
4. encode_frame() loop
5. drain() — flushes encoder delay
6. FFmpegMuxContext::close() — writes trailer

The source pixel format passed to open() must match the format delivered to every subsequent encode_frame() call. Swapchain BGRA readbacks map to AV_PIX_FMT_BGRA; RGBA readbacks to AV_PIX_FMT_RGBA.

Not copyable or movable; always stack- or member-allocated inside the owner (VideoFileWriter worker thread).

The associated FFmpegMuxContext must outlive this object.

Definition at line 44 of file VideoEncodeContext.hpp.

---

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

• VideoEncodeContext.hpp
• VideoEncodeContext.cpp