gavl
Data Structures | Functions
Video frames
Video

Container for video images. More...

Data Structures

struct  gavl_video_frame_t

Functions

GAVL_PUBLIC gavl_video_frame_tgavl_video_frame_create (const gavl_video_format_t *format)
 Create video frame.
GAVL_PUBLIC gavl_video_frame_tgavl_video_frame_create_nopad (const gavl_video_format_t *format)
 Create video frame without padding.
GAVL_PUBLIC void gavl_video_frame_destroy (gavl_video_frame_t *frame)
 Destroy a video frame.
GAVL_PUBLIC void gavl_video_frame_null (gavl_video_frame_t *frame)
 Zero all pointers in the video frame.
GAVL_PUBLIC void gavl_video_frame_clear (gavl_video_frame_t *frame, const gavl_video_format_t *format)
 Fill the frame with black color.
GAVL_PUBLIC void gavl_video_frame_fill (gavl_video_frame_t *frame, const gavl_video_format_t *format, const float *color)
 Fill the frame with a user spefified color.
GAVL_PUBLIC void gavl_video_frame_absdiff (gavl_video_frame_t *dst, const gavl_video_frame_t *src1, const gavl_video_frame_t *src2, const gavl_video_format_t *format)
 Fill the frame with the absolute differene of 2 source frames.
GAVL_PUBLIC void gavl_video_frame_psnr (double *psnr, const gavl_video_frame_t *src1, const gavl_video_frame_t *src2, const gavl_video_format_t *format)
 Calculate the PSNR of 2 source frames.
GAVL_PUBLIC int gavl_video_frame_ssim (const gavl_video_frame_t *src1, const gavl_video_frame_t *src2, gavl_video_frame_t *dst, const gavl_video_format_t *format)
 Calculate the SSIM of 2 source frames.
GAVL_PUBLIC void gavl_video_frame_copy (const gavl_video_format_t *format, gavl_video_frame_t *dst, const gavl_video_frame_t *src)
 Copy one video frame to another.
GAVL_PUBLIC void gavl_video_frame_copy_plane (const gavl_video_format_t *format, gavl_video_frame_t *dst, const gavl_video_frame_t *src, int plane)
 Copy a single plane from one video frame to another.
GAVL_PUBLIC void gavl_video_frame_copy_flip_x (const gavl_video_format_t *format, gavl_video_frame_t *dst, const gavl_video_frame_t *src)
 Copy one video frame to another with horizontal flipping.
GAVL_PUBLIC void gavl_video_frame_copy_flip_y (const gavl_video_format_t *format, gavl_video_frame_t *dst, const gavl_video_frame_t *src)
 Copy one video frame to another with vertical flipping.
GAVL_PUBLIC void gavl_video_frame_copy_flip_xy (const gavl_video_format_t *format, gavl_video_frame_t *dst, const gavl_video_frame_t *src)
 Copy one video frame to another with horizontal and vertical flipping.
GAVL_PUBLIC void gavl_video_frame_copy_metadata (gavl_video_frame_t *dst, const gavl_video_frame_t *src)
 Copy metadata of one video frame to another.
GAVL_PUBLIC void gavl_video_frame_get_subframe (gavl_pixelformat_t pixelformat, const gavl_video_frame_t *src, gavl_video_frame_t *dst, gavl_rectangle_i_t *src_rect)
 Get a subframe of another frame.
GAVL_PUBLIC void gavl_video_frame_get_field (gavl_pixelformat_t pixelformat, const gavl_video_frame_t *src, gavl_video_frame_t *dst, int field)
 Get a field from a frame.
GAVL_PUBLIC void gavl_video_frame_dump (gavl_video_frame_t *frame, const gavl_video_format_t *format, const char *namebase)
 Dump a video frame to files.
GAVL_PUBLIC void gavl_video_frame_set_strides (gavl_video_frame_t *frame, const gavl_video_format_t *format)
 Set the strides according to the format.
GAVL_PUBLIC void gavl_video_frame_set_planes (gavl_video_frame_t *frame, const gavl_video_format_t *format, uint8_t *buffer)
 Set the frames according to the format.
GAVL_PUBLIC int gavl_video_frame_extract_channel (const gavl_video_format_t *format, gavl_color_channel_t ch, const gavl_video_frame_t *src, gavl_video_frame_t *dst)
 Extract one channel of a video frame into a grayscale image.
GAVL_PUBLIC int gavl_video_frame_insert_channel (const gavl_video_format_t *format, gavl_color_channel_t ch, const gavl_video_frame_t *src, gavl_video_frame_t *dst)
 Insert one channel from a grayscale image into a video frame.
GAVL_PUBLIC int gavl_video_frames_equal (const gavl_video_format_t *format, const gavl_video_frame_t *f1, const gavl_video_frame_t *f2)
 Check if 2 video frames are bit-identical.

Detailed Description

Container for video images.

This is the standardized method of storing one frame with video data. For planar formats, the first scanline starts at planes[0], subsequent scanlines start in intervalls of strides[0] bytes. For planar formats, planes[0] will contain the luminance channel, planes[1] contains Cb (aka U), planes[2] contains Cr (aka V).

Video frames are created with gavl_video_frame_create and destroyed with gavl_video_frame_destroy. The memory can either be allocated by gavl (with memory aligned scanlines) or by the caller.

Gavl video frames are always oriented top->bottom left->right. If you must flip frames, use the functions gavl_video_frame_copy_flip_x, gavl_video_frame_copy_flip_y or gavl_video_frame_copy_flip_xy .


Function Documentation

GAVL_PUBLIC gavl_video_frame_t* gavl_video_frame_create ( const gavl_video_format_t format)

Create video frame.

Parameters:
formatFormat of the data to be stored in this frame or NULL

Creates a video frame for a given format and allocates buffers for the scanlines. To create a video frame from your custom memory, pass NULL for the format and you'll get an empty frame in which you can set the pointers manually. If scanlines are allocated, they are padded to that scanlines start at certain byte boundaries (currently 8).

GAVL_PUBLIC gavl_video_frame_t* gavl_video_frame_create_nopad ( const gavl_video_format_t format)

Create video frame without padding.

Parameters:
formatFormat of the data to be stored in this frame or NULL

Same as gavl_video_frame_create but omits padding, so scanlines will always be adjacent in memory.

GAVL_PUBLIC void gavl_video_frame_destroy ( gavl_video_frame_t frame)

Destroy a video frame.

Parameters:
frameA video frame

Destroys a video frame and frees all associated memory. If you used your custom memory to allocate the frame, call gavl_video_frame_null before.

GAVL_PUBLIC void gavl_video_frame_null ( gavl_video_frame_t frame)

Zero all pointers in the video frame.

Parameters:
frameA video frame

Zero all pointers, so gavl_video_frame_destroy won't free them. Call this for video frames, which were created with a NULL format before destroying them.

GAVL_PUBLIC void gavl_video_frame_clear ( gavl_video_frame_t frame,
const gavl_video_format_t format 
)

Fill the frame with black color.

Parameters:
frameA video frame
formatFormat of the data in the frame
GAVL_PUBLIC void gavl_video_frame_fill ( gavl_video_frame_t frame,
const gavl_video_format_t format,
const float *  color 
)

Fill the frame with a user spefified color.

Parameters:
frameA video frame
formatFormat of the data in the frame
colorColor components in RGBA format scaled 0.0 .. 1.0
GAVL_PUBLIC void gavl_video_frame_absdiff ( gavl_video_frame_t dst,
const gavl_video_frame_t src1,
const gavl_video_frame_t src2,
const gavl_video_format_t format 
)

Fill the frame with the absolute differene of 2 source frames.

Parameters:
formatFormat of the data in the frame
dstA video frame
src1First source frame
src2Second source frame

Since 1.1.1

GAVL_PUBLIC void gavl_video_frame_psnr ( double *  psnr,
const gavl_video_frame_t src1,
const gavl_video_frame_t src2,
const gavl_video_format_t format 
)

Calculate the PSNR of 2 source frames.

Parameters:
psnrReturns PSNR for all components (maximum 4)
src1First source frame
src2Second source frame
formatFormat of the data in the frame

Since 1.1.1

GAVL_PUBLIC int gavl_video_frame_ssim ( const gavl_video_frame_t src1,
const gavl_video_frame_t src2,
gavl_video_frame_t dst,
const gavl_video_format_t format 
)

Calculate the SSIM of 2 source frames.

Parameters:
src1First source frame
src2Second source frame
dstWill contain the SSIM index for each pixel
formatFormat of the data in the frame
Returns:
1 if the SSIM could be computed, 0 else

This calculates the SSIM indices of each pixel for 2 source frames and stores them into dst. The source frames must have the pixelformat GAVL_GRAY_FLOAT implying that only the luminance component is considered. The destination also has the pixelformat GAVL_GRAY_FLOAT. If other pixelformats are passed to this function it will return 0 and nothing is done.

The SSIM algorithm is taken from the paper "Image Quality Assessment: From Error Visibility to Structural Similarity" by Z. Want et. al. published in IEEE Transactions on image processing, VOL. 13, NO. 4, APRIL 2004. Homepage of the author: http://www.ece.uwaterloo.ca/~z70wang/research/ssim/

Since 1.1.2

GAVL_PUBLIC void gavl_video_frame_copy ( const gavl_video_format_t format,
gavl_video_frame_t dst,
const gavl_video_frame_t src 
)

Copy one video frame to another.

Parameters:
formatThe format of the frames
dstDestination
srcSource

The source and destination formats must be identical, as this routine does no format conversion. The scanlines however can be padded differently in the source and destination. This function only copies the image data. For copying the metadata (timestamp etc.) use# gavl_video_frame_copy_metadata.

GAVL_PUBLIC void gavl_video_frame_copy_plane ( const gavl_video_format_t format,
gavl_video_frame_t dst,
const gavl_video_frame_t src,
int  plane 
)

Copy a single plane from one video frame to another.

Parameters:
formatThe format of the frames
dstDestination
srcSource
planeThe plane to copy

Like gavl_video_frame_copy but copies only a single plane

GAVL_PUBLIC void gavl_video_frame_copy_flip_x ( const gavl_video_format_t format,
gavl_video_frame_t dst,
const gavl_video_frame_t src 
)

Copy one video frame to another with horizontal flipping.

Parameters:
formatThe format of the frames
dstDestination
srcSource

Like gavl_video_frame_copy but flips the image horizontally

GAVL_PUBLIC void gavl_video_frame_copy_flip_y ( const gavl_video_format_t format,
gavl_video_frame_t dst,
const gavl_video_frame_t src 
)

Copy one video frame to another with vertical flipping.

Parameters:
formatThe format of the frames
dstDestination
srcSource

Like gavl_video_frame_copy but flips the image vertically

GAVL_PUBLIC void gavl_video_frame_copy_flip_xy ( const gavl_video_format_t format,
gavl_video_frame_t dst,
const gavl_video_frame_t src 
)

Copy one video frame to another with horizontal and vertical flipping.

Parameters:
formatThe format of the frames
dstDestination
srcSource

Like gavl_video_frame_copy but flips the image both horizontally and vertically

GAVL_PUBLIC void gavl_video_frame_copy_metadata ( gavl_video_frame_t dst,
const gavl_video_frame_t src 
)

Copy metadata of one video frame to another.

Parameters:
dstDestination
srcSource

This function only copies the metadata (timestamp, duration, timecode). For copying the image data use gavl_video_frame_copy.

Since 1.1.0.

GAVL_PUBLIC void gavl_video_frame_get_subframe ( gavl_pixelformat_t  pixelformat,
const gavl_video_frame_t src,
gavl_video_frame_t dst,
gavl_rectangle_i_t src_rect 
)

Get a subframe of another frame.

Parameters:
pixelformatPixelformat of the frames
dstDestination
srcSource
src_rectRectangular area in the source, which will be in the destination frame

This fills the pointers of dst from src such that the dst will represent the speficied rectangular area. Note that no data are copied here. This means that dst must be created with NULL as the format argument and gavl_video_frame_null must be called before destroying dst.

When dealing with chroma subsampled pixelformats, you must call gavl_rectangle_i_align on src_rect before.

GAVL_PUBLIC void gavl_video_frame_get_field ( gavl_pixelformat_t  pixelformat,
const gavl_video_frame_t src,
gavl_video_frame_t dst,
int  field 
)

Get a field from a frame.

Parameters:
pixelformatPixelformat of the frames
dstDestination
srcSource
fieldField index (0 = top field, 1 = bottom field)

This fills the pointers and strides of the destination frame such that it will represent the speficied field of the source frame. Note that no data are copied here. This means that dst must be created with NULL as the format argument and gavl_video_frame_null must be called before destroying dst.

GAVL_PUBLIC void gavl_video_frame_dump ( gavl_video_frame_t frame,
const gavl_video_format_t format,
const char *  namebase 
)

Dump a video frame to files.

Parameters:
frameVideo frame to dump
formatFormat of the video data in the frame
namebaseBase for the output filenames

This is purely for debugging purposes: It dumps each plane of the frame into files <namebase>.p1, <namebase>.p2 etc

GAVL_PUBLIC void gavl_video_frame_set_strides ( gavl_video_frame_t frame,
const gavl_video_format_t format 
)

Set the strides according to the format.

Parameters:
frameVideo frame
formatFormat of the video data in the frame

This sets the strides array according to the format under the assumption, that no padding is done.

GAVL_PUBLIC void gavl_video_frame_set_planes ( gavl_video_frame_t frame,
const gavl_video_format_t format,
uint8_t *  buffer 
)

Set the frames according to the format.

Parameters:
frameVideo frame
formatFormat of the video data in the frame
bufferStart of the first scanline of the first plane

This sets the planes array from a raw buffer. If frame->strides[0] is zero, gavl_video_frame_set_strides is called before.

GAVL_PUBLIC int gavl_video_frame_extract_channel ( const gavl_video_format_t format,
gavl_color_channel_t  ch,
const gavl_video_frame_t src,
gavl_video_frame_t dst 
)

Extract one channel of a video frame into a grayscale image.

Parameters:
formatFormat of the source frame
chChannel to extract
srcSource frame
dstDestination where the extracted channel will be copied

This extracts one color channel into a grayscale image for separate processing. Use gavl_get_color_channel_format to obtain the format for the destination frame.

GAVL_PUBLIC int gavl_video_frame_insert_channel ( const gavl_video_format_t format,
gavl_color_channel_t  ch,
const gavl_video_frame_t src,
gavl_video_frame_t dst 
)

Insert one channel from a grayscale image into a video frame.

Parameters:
formatFormat of the source frame
chChannel to merge
srcSource frame (grayscale image containing one chanel)
dstDestination

This inserts one color channel from a grayscale image into a video frame (overwriting the previous contents of that channel). Use gavl_get_color_channel_format to obtain the format for the source grayscale frame.

GAVL_PUBLIC int gavl_video_frames_equal ( const gavl_video_format_t format,
const gavl_video_frame_t f1,
const gavl_video_frame_t f2 
)

Check if 2 video frames are bit-identical.

Parameters:
formatFormat
f1First frame
f2Second frame
Returns:
1 if the frames are equal, 0 else

Since 1.2.0