FFmpeg
Macros | Functions | Variables
vf_deshake.c File Reference

fast deshake / depan video filter More...

#include "avfilter.h"
#include "formats.h"
#include "internal.h"
#include "video.h"
#include "libavutil/common.h"
#include "libavutil/mem.h"
#include "libavutil/opt.h"
#include "libavutil/pixdesc.h"
#include "libavutil/qsort.h"
#include "deshake.h"

Go to the source code of this file.

Macros

#define OFFSET(x)   offsetof(DeshakeContext, x)
 
#define FLAGS   AV_OPT_FLAG_VIDEO_PARAM|AV_OPT_FLAG_FILTERING_PARAM
 
#define CMP(i, j)
 

Functions

 AVFILTER_DEFINE_CLASS (deshake)
 
static int cmp (const void *a, const void *b)
 
static double clean_mean (double *values, int count)
 Cleaned mean (cuts off 20% of values to remove outliers and then averages) More...
 
static void find_block_motion (DeshakeContext *deshake, uint8_t *src1, uint8_t *src2, int cx, int cy, int stride, IntMotionVector *mv)
 Find the most likely shift in motion between two frames for a given macroblock. More...
 
static int block_contrast (uint8_t *src, int x, int y, int stride, int blocksize)
 Find the contrast of a given block. More...
 
static double block_angle (int x, int y, int cx, int cy, IntMotionVector *shift)
 Find the rotation for a given block. More...
 
static void find_motion (DeshakeContext *deshake, uint8_t *src1, uint8_t *src2, int width, int height, int stride, Transform *t)
 Find the estimated global motion for a scene given the most likely shift for each block in the frame. More...
 
static int deshake_transform_c (AVFilterContext *ctx, int width, int height, int cw, int ch, const float *matrix_y, const float *matrix_uv, enum InterpolateMethod interpolate, enum FillMethod fill, AVFrame *in, AVFrame *out)
 
static av_cold int init (AVFilterContext *ctx)
 
static int query_formats (AVFilterContext *ctx)
 
static int config_props (AVFilterLink *link)
 
static av_cold void uninit (AVFilterContext *ctx)
 
static int filter_frame (AVFilterLink *link, AVFrame *in)
 

Variables

static const AVOption deshake_options []
 
static const AVFilterPad deshake_inputs []
 
static const AVFilterPad deshake_outputs []
 
AVFilter ff_vf_deshake
 

Detailed Description

fast deshake / depan video filter

SAD block-matching motion compensation to fix small changes in horizontal and/or vertical shift. This filter helps remove camera shake from hand-holding a camera, bumping a tripod, moving on a vehicle, etc.

Algorithm:

TODO:

Dark Shikari links to http://wiki.videolan.org/SoC_x264_2010#GPU_Motion_Estimation_2 for an algorithm similar to what could be used here to get the gmv It requires only a couple diamond searches + fast downscaling

Special thanks to Jason Kotenko for his help with the algorithm and my inability to see simple errors in C code.

Definition in file vf_deshake.c.

Macro Definition Documentation

#define OFFSET (   x)    offsetof(DeshakeContext, x)

Definition at line 64 of file vf_deshake.c.

Definition at line 65 of file vf_deshake.c.

#define CMP (   i,
 
)
Value:
deshake->sad(src1 + cy * stride + cx, stride,\
src2 + (j) * stride + (i), stride)
#define i(width, name, range_min, range_max)
Definition: cbs_h2645.c:259
#define src1
Definition: h264pred.c:139
GLint GLenum GLboolean GLsizei stride
Definition: opengl_enc.c:104

Referenced by find_block_motion().

Function Documentation

AVFILTER_DEFINE_CLASS ( deshake  )
static int cmp ( const void a,
const void b 
)
static

Definition at line 91 of file vf_deshake.c.

Referenced by clean_mean().

static double clean_mean ( double *  values,
int  count 
)
static

Cleaned mean (cuts off 20% of values to remove outliers and then averages)

Definition at line 99 of file vf_deshake.c.

Referenced by find_motion().

static void find_block_motion ( DeshakeContext deshake,
uint8_t src1,
uint8_t src2,
int  cx,
int  cy,
int  stride,
IntMotionVector mv 
)
static

Find the most likely shift in motion between two frames for a given macroblock.

Test each block against several shifts given by the rx and ry attributes. Searches using a simple matrix of those shifts and chooses the most likely shift by the smallest difference in blocks.

Definition at line 120 of file vf_deshake.c.

Referenced by find_motion().

static int block_contrast ( uint8_t src,
int  x,
int  y,
int  stride,
int  blocksize 
)
static

Find the contrast of a given block.

When searching for global motion we really only care about the high contrast blocks, so using this method we can actually skip blocks we don't care much about.

Definition at line 190 of file vf_deshake.c.

Referenced by find_motion().

static double block_angle ( int  x,
int  y,
int  cx,
int  cy,
IntMotionVector shift 
)
static

Find the rotation for a given block.

Definition at line 214 of file vf_deshake.c.

Referenced by find_motion().

static void find_motion ( DeshakeContext deshake,
uint8_t src1,
uint8_t src2,
int  width,
int  height,
int  stride,
Transform t 
)
static

Find the estimated global motion for a scene given the most likely shift for each block in the frame.

The global motion is estimated to be the same as the motion from most blocks in the frame, so if most blocks move one pixel to the right and two pixels down, this would yield a motion vector (1, -2).

Definition at line 235 of file vf_deshake.c.

Referenced by filter_frame().

static int deshake_transform_c ( AVFilterContext ctx,
int  width,
int  height,
int  cw,
int  ch,
const float *  matrix_y,
const float *  matrix_uv,
enum InterpolateMethod  interpolate,
enum FillMethod  fill,
AVFrame in,
AVFrame out 
)
static

Definition at line 315 of file vf_deshake.c.

Referenced by init().

static av_cold int init ( AVFilterContext ctx)
static

Definition at line 341 of file vf_deshake.c.

static int query_formats ( AVFilterContext ctx)
static

Definition at line 374 of file vf_deshake.c.

static int config_props ( AVFilterLink link)
static

Definition at line 387 of file vf_deshake.c.

static av_cold void uninit ( AVFilterContext ctx)
static

Definition at line 400 of file vf_deshake.c.

static int filter_frame ( AVFilterLink link,
AVFrame in 
)
static

Definition at line 410 of file vf_deshake.c.

Variable Documentation

const AVOption deshake_options[]
static
Initial value:
= {
{ "x", "set x for the rectangular search area", OFFSET(cx), AV_OPT_TYPE_INT, {.i64=-1}, -1, INT_MAX, .flags = FLAGS },
{ "y", "set y for the rectangular search area", OFFSET(cy), AV_OPT_TYPE_INT, {.i64=-1}, -1, INT_MAX, .flags = FLAGS },
{ "w", "set width for the rectangular search area", OFFSET(cw), AV_OPT_TYPE_INT, {.i64=-1}, -1, INT_MAX, .flags = FLAGS },
{ "h", "set height for the rectangular search area", OFFSET(ch), AV_OPT_TYPE_INT, {.i64=-1}, -1, INT_MAX, .flags = FLAGS },
{ "rx", "set x for the rectangular search area", OFFSET(rx), AV_OPT_TYPE_INT, {.i64=16}, 0, MAX_R, .flags = FLAGS },
{ "ry", "set y for the rectangular search area", OFFSET(ry), AV_OPT_TYPE_INT, {.i64=16}, 0, MAX_R, .flags = FLAGS },
{ "edge", "set edge mode", OFFSET(edge), AV_OPT_TYPE_INT, {.i64=FILL_MIRROR}, FILL_BLANK, FILL_COUNT-1, FLAGS, "edge"},
{ "blank", "fill zeroes at blank locations", 0, AV_OPT_TYPE_CONST, {.i64=FILL_BLANK}, INT_MIN, INT_MAX, FLAGS, "edge" },
{ "original", "original image at blank locations", 0, AV_OPT_TYPE_CONST, {.i64=FILL_ORIGINAL}, INT_MIN, INT_MAX, FLAGS, "edge" },
{ "clamp", "extruded edge value at blank locations", 0, AV_OPT_TYPE_CONST, {.i64=FILL_CLAMP}, INT_MIN, INT_MAX, FLAGS, "edge" },
{ "mirror", "mirrored edge at blank locations", 0, AV_OPT_TYPE_CONST, {.i64=FILL_MIRROR}, INT_MIN, INT_MAX, FLAGS, "edge" },
{ "blocksize", "set motion search blocksize", OFFSET(blocksize), AV_OPT_TYPE_INT, {.i64=8}, 4, 128, .flags = FLAGS },
{ "contrast", "set contrast threshold for blocks", OFFSET(contrast), AV_OPT_TYPE_INT, {.i64=125}, 1, 255, .flags = FLAGS },
{ "search", "set search strategy", OFFSET(search), AV_OPT_TYPE_INT, {.i64=EXHAUSTIVE}, EXHAUSTIVE, SEARCH_COUNT-1, FLAGS, "smode" },
{ "exhaustive", "exhaustive search", 0, AV_OPT_TYPE_CONST, {.i64=EXHAUSTIVE}, INT_MIN, INT_MAX, FLAGS, "smode" },
{ "less", "less exhaustive search", 0, AV_OPT_TYPE_CONST, {.i64=SMART_EXHAUSTIVE}, INT_MIN, INT_MAX, FLAGS, "smode" },
{ "filename", "set motion search detailed log file name", OFFSET(filename), AV_OPT_TYPE_STRING, {.str=NULL}, .flags = FLAGS },
{ "opencl", "ignored", OFFSET(opencl), AV_OPT_TYPE_BOOL, {.i64=0}, 0, 1, .flags = FLAGS },
{ NULL }
}
#define NULL
Definition: coverity.c:32
#define OFFSET(x)
Definition: vf_deshake.c:64
#define FLAGS
Definition: vf_deshake.c:65
static float search(FOCContext *foc, int pass, int maxpass, int xmin, int xmax, int ymin, int ymax, int *best_x, int *best_y, float best_score)
Definition: vf_find_rect.c:156
Search all possible positions.
Definition: deshake.h:32
Search most possible positions (faster)
Definition: deshake.h:33
uint8_t pi<< 24) CONV_FUNC(AV_SAMPLE_FMT_S64, int64_t, AV_SAMPLE_FMT_U8,(uint64_t)((*(const uint8_t *) pi-0x80U))<< 56) CONV_FUNC(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_U8,(*(const uint8_t *) pi-0x80)*(1.0f/(1<< 7))) CONV_FUNC(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_U8,(*(const uint8_t *) pi-0x80)*(1.0/(1<< 7))) CONV_FUNC(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_S16,(*(const int16_t *) pi >>8)+0x80) CONV_FUNC(AV_SAMPLE_FMT_S64, int64_t, AV_SAMPLE_FMT_S16,(uint64_t)(*(const int16_t *) pi)<< 48) CONV_FUNC(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_S16,*(const int16_t *) pi *(1.0f/(1<< 15))) CONV_FUNC(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_S16,*(const int16_t *) pi *(1.0/(1<< 15))) CONV_FUNC(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_S32,(*(const int32_t *) pi >>24)+0x80) CONV_FUNC(AV_SAMPLE_FMT_S64, int64_t, AV_SAMPLE_FMT_S32,(uint64_t)(*(const int32_t *) pi)<< 32) CONV_FUNC(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_S32,*(const int32_t *) pi *(1.0f/(1U<< 31))) CONV_FUNC(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_S32,*(const int32_t *) pi *(1.0/(1U<< 31))) CONV_FUNC(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_S64,(*(const int64_t *) pi >>56)+0x80) CONV_FUNC(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_S64,*(const int64_t *) pi *(1.0f/(UINT64_C(1)<< 63))) CONV_FUNC(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_S64,*(const int64_t *) pi *(1.0/(UINT64_C(1)<< 63))) CONV_FUNC(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_FLT, av_clip_uint8(lrintf(*(const float *) pi *(1<< 7))+0x80)) CONV_FUNC(AV_SAMPLE_FMT_S16, int16_t, AV_SAMPLE_FMT_FLT, av_clip_int16(lrintf(*(const float *) pi *(1<< 15)))) CONV_FUNC(AV_SAMPLE_FMT_S32, int32_t, AV_SAMPLE_FMT_FLT, av_clipl_int32(llrintf(*(const float *) pi *(1U<< 31)))) CONV_FUNC(AV_SAMPLE_FMT_S64, int64_t, AV_SAMPLE_FMT_FLT, llrintf(*(const float *) pi *(UINT64_C(1)<< 63))) CONV_FUNC(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_DBL, av_clip_uint8(lrint(*(const double *) pi *(1<< 7))+0x80)) CONV_FUNC(AV_SAMPLE_FMT_S16, int16_t, AV_SAMPLE_FMT_DBL, av_clip_int16(lrint(*(const double *) pi *(1<< 15)))) CONV_FUNC(AV_SAMPLE_FMT_S32, int32_t, AV_SAMPLE_FMT_DBL, av_clipl_int32(llrint(*(const double *) pi *(1U<< 31)))) CONV_FUNC(AV_SAMPLE_FMT_S64, int64_t, AV_SAMPLE_FMT_DBL, llrint(*(const double *) pi *(UINT64_C(1)<< 63)))#define FMT_PAIR_FUNC(out, in) static conv_func_type *const fmt_pair_to_conv_functions[AV_SAMPLE_FMT_NB *AV_SAMPLE_FMT_NB]={FMT_PAIR_FUNC(AV_SAMPLE_FMT_U8, AV_SAMPLE_FMT_U8), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_U8), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S32, AV_SAMPLE_FMT_U8), FMT_PAIR_FUNC(AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_U8), FMT_PAIR_FUNC(AV_SAMPLE_FMT_DBL, AV_SAMPLE_FMT_U8), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S64, AV_SAMPLE_FMT_U8), FMT_PAIR_FUNC(AV_SAMPLE_FMT_U8, AV_SAMPLE_FMT_S16), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_S16), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S32, AV_SAMPLE_FMT_S16), FMT_PAIR_FUNC(AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_S16), FMT_PAIR_FUNC(AV_SAMPLE_FMT_DBL, AV_SAMPLE_FMT_S16), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S64, AV_SAMPLE_FMT_S16), FMT_PAIR_FUNC(AV_SAMPLE_FMT_U8, AV_SAMPLE_FMT_S32), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_S32), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S32, AV_SAMPLE_FMT_S32), FMT_PAIR_FUNC(AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_S32), FMT_PAIR_FUNC(AV_SAMPLE_FMT_DBL, AV_SAMPLE_FMT_S32), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S64, AV_SAMPLE_FMT_S32), FMT_PAIR_FUNC(AV_SAMPLE_FMT_U8, AV_SAMPLE_FMT_FLT), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_FLT), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S32, AV_SAMPLE_FMT_FLT), FMT_PAIR_FUNC(AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_FLT), FMT_PAIR_FUNC(AV_SAMPLE_FMT_DBL, AV_SAMPLE_FMT_FLT), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S64, AV_SAMPLE_FMT_FLT), FMT_PAIR_FUNC(AV_SAMPLE_FMT_U8, AV_SAMPLE_FMT_DBL), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_DBL), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S32, AV_SAMPLE_FMT_DBL), FMT_PAIR_FUNC(AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_DBL), FMT_PAIR_FUNC(AV_SAMPLE_FMT_DBL, AV_SAMPLE_FMT_DBL), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S64, AV_SAMPLE_FMT_DBL), FMT_PAIR_FUNC(AV_SAMPLE_FMT_U8, AV_SAMPLE_FMT_S64), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_S64), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S32, AV_SAMPLE_FMT_S64), FMT_PAIR_FUNC(AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_S64), FMT_PAIR_FUNC(AV_SAMPLE_FMT_DBL, AV_SAMPLE_FMT_S64), FMT_PAIR_FUNC(AV_SAMPLE_FMT_S64, AV_SAMPLE_FMT_S64),};static void cpy1(uint8_t **dst, const uint8_t **src, int len){memcpy(*dst,*src, len);}static void cpy2(uint8_t **dst, const uint8_t **src, int len){memcpy(*dst,*src, 2 *len);}static void cpy4(uint8_t **dst, const uint8_t **src, int len){memcpy(*dst,*src, 4 *len);}static void cpy8(uint8_t **dst, const uint8_t **src, int len){memcpy(*dst,*src, 8 *len);}AudioConvert *swri_audio_convert_alloc(enum AVSampleFormat out_fmt, enum AVSampleFormat in_fmt, int channels, const int *ch_map, int flags){AudioConvert *ctx;conv_func_type *f=fmt_pair_to_conv_functions[av_get_packed_sample_fmt(out_fmt)+AV_SAMPLE_FMT_NB *av_get_packed_sample_fmt(in_fmt)];if(!f) return NULL;ctx=av_mallocz(sizeof(*ctx));if(!ctx) return NULL;if(channels==1){in_fmt=av_get_planar_sample_fmt(in_fmt);out_fmt=av_get_planar_sample_fmt(out_fmt);}ctx->channels=channels;ctx->conv_f=f;ctx->ch_map=ch_map;if(in_fmt==AV_SAMPLE_FMT_U8||in_fmt==AV_SAMPLE_FMT_U8P) memset(ctx->silence, 0x80, sizeof(ctx->silence));if(out_fmt==in_fmt &&!ch_map){switch(av_get_bytes_per_sample(in_fmt)){case 1:ctx->simd_f=cpy1;break;case 2:ctx->simd_f=cpy2;break;case 4:ctx->simd_f=cpy4;break;case 8:ctx->simd_f=cpy8;break;}}if(HAVE_X86ASM &&1) swri_audio_convert_init_x86(ctx, out_fmt, in_fmt, channels);if(ARCH_ARM) swri_audio_convert_init_arm(ctx, out_fmt, in_fmt, channels);if(ARCH_AARCH64) swri_audio_convert_init_aarch64(ctx, out_fmt, in_fmt, channels);return ctx;}void swri_audio_convert_free(AudioConvert **ctx){av_freep(ctx);}int swri_audio_convert(AudioConvert *ctx, AudioData *out, AudioData *in, int len){int ch;int off=0;const int os=(out->planar?1:out->ch_count)*out->bps;unsigned misaligned=0;av_assert0(ctx->channels==out->ch_count);if(ctx->in_simd_align_mask){int planes=in->planar?in->ch_count:1;unsigned m=0;for(ch=0;ch< planes;ch++) m|=(intptr_t) in->ch[ch];misaligned|=m &ctx->in_simd_align_mask;}if(ctx->out_simd_align_mask){int planes=out->planar?out->ch_count:1;unsigned m=0;for(ch=0;ch< planes;ch++) m|=(intptr_t) out->ch[ch];misaligned|=m &ctx->out_simd_align_mask;}if(ctx->simd_f &&!ctx->ch_map &&!misaligned){off=len &~15;av_assert1(off >=0);av_assert1(off<=len);av_assert2(ctx->channels==SWR_CH_MAX||!in->ch[ctx->channels]);if(off >0){if(out->planar==in->planar){int planes=out->planar?out->ch_count:1;for(ch=0;ch< planes;ch++){ctx->simd_f(out-> ch ch
Definition: audioconvert.c:56
#define MAX_R
Definition: deshake.h:53

Definition at line 67 of file vf_deshake.c.

const AVFilterPad deshake_inputs[]
static
Initial value:
= {
{
.name = "default",
.filter_frame = filter_frame,
.config_props = config_props,
},
{ NULL }
}
#define NULL
Definition: coverity.c:32
static int filter_frame(AVFilterLink *link, AVFrame *in)
Definition: vf_deshake.c:410
static int config_props(AVFilterLink *link)
Definition: vf_deshake.c:387

Definition at line 535 of file vf_deshake.c.

const AVFilterPad deshake_outputs[]
static
Initial value:
= {
{
.name = "default",
},
{ NULL }
}
#define NULL
Definition: coverity.c:32

Definition at line 545 of file vf_deshake.c.

AVFilter ff_vf_deshake
Initial value:
= {
.name = "deshake",
.description = NULL_IF_CONFIG_SMALL("Stabilize shaky video."),
.priv_size = sizeof(DeshakeContext),
.init = init,
.priv_class = &deshake_class,
}
#define NULL_IF_CONFIG_SMALL(x)
Return NULL if CONFIG_SMALL is true, otherwise the argument without modification. ...
Definition: internal.h:186
static int query_formats(AVFilterContext *ctx)
Definition: vf_deshake.c:374
static const AVFilterPad deshake_inputs[]
Definition: vf_deshake.c:535
static const AVFilterPad outputs[]
Definition: af_acontrast.c:203
these buffered frames must be flushed immediately if a new input produces new the filter must not call request_frame to get more It must just process the frame or queue it The task of requesting more frames is left to the filter s request_frame method or the application If a filter has several inputs
static av_cold int init(AVFilterContext *ctx)
Definition: vf_deshake.c:341
static const AVFilterPad deshake_outputs[]
Definition: vf_deshake.c:545
static av_cold void uninit(AVFilterContext *ctx)
Definition: vf_deshake.c:400

Definition at line 553 of file vf_deshake.c.