From a7dc49cca7cb17d9695e5d53521e264da4d15788 Mon Sep 17 00:00:00 2001 From: cantabile Date: Mon, 28 Apr 2014 17:22:40 +0300 Subject: [PATCH] Add code --- src/detection.c | 471 ++++++++++++++++++++++++++++++++++++++++++++++++ src/detection.h | 44 +++++ src/wwxd.c | 145 +++++++++++++++ 3 files changed, 660 insertions(+) create mode 100644 src/detection.c create mode 100644 src/detection.h create mode 100644 src/wwxd.c diff --git a/src/detection.c b/src/detection.c new file mode 100644 index 0000000..399a0ed --- /dev/null +++ b/src/detection.c @@ -0,0 +1,471 @@ +#include +#include + +#include + +#include "detection.h" + + +#define MVequal(A,B) ( ((A).x)==((B).x) && ((A).y)==((B).y) ) + +#define iDiamondSize 2 + +#define CHECK_CANDIDATE(X,Y,D) { \ + CheckCandidate((X),(Y), data, (D) ); } + +#define MV_MAX_ERROR (4096 * 256) + +#define MODE_INTER 0 +#define MODE_INTRA 3 + + +typedef void(CheckFunc)(const int x, const int y, + SearchData * const Data, + const unsigned int Direction); + + +static uint32_t +sad16_c(const uint8_t * const cur, + const uint8_t * const ref, + const uint32_t stride) +{ + + uint32_t sad = 0; + uint32_t j; + uint8_t const *ptr_cur = cur; + uint8_t const *ptr_ref = ref; + + for (j = 0; j < 16; j++) { + sad += abs(ptr_cur[0] - ptr_ref[0]); + sad += abs(ptr_cur[1] - ptr_ref[1]); + sad += abs(ptr_cur[2] - ptr_ref[2]); + sad += abs(ptr_cur[3] - ptr_ref[3]); + sad += abs(ptr_cur[4] - ptr_ref[4]); + sad += abs(ptr_cur[5] - ptr_ref[5]); + sad += abs(ptr_cur[6] - ptr_ref[6]); + sad += abs(ptr_cur[7] - ptr_ref[7]); + sad += abs(ptr_cur[8] - ptr_ref[8]); + sad += abs(ptr_cur[9] - ptr_ref[9]); + sad += abs(ptr_cur[10] - ptr_ref[10]); + sad += abs(ptr_cur[11] - ptr_ref[11]); + sad += abs(ptr_cur[12] - ptr_ref[12]); + sad += abs(ptr_cur[13] - ptr_ref[13]); + sad += abs(ptr_cur[14] - ptr_ref[14]); + sad += abs(ptr_cur[15] - ptr_ref[15]); + + ptr_cur += stride; + ptr_ref += stride; + + } + + return sad; + +} + +static uint32_t sad32v_c(const uint8_t * const cur, + const uint8_t * const ref, + const uint32_t stride, + int32_t *sad) +{ + sad[0] = sad16_c(cur, ref, stride); + sad[1] = sad16_c(cur + 16, ref + 16, stride); + sad[2] = sad16_c(cur + 16*stride, ref + 16*stride, stride); + sad[3] = sad16_c(cur + 16*stride + 16, ref + 16*stride + 16, stride); + + return sad[0]+sad[1]+sad[2]+sad[3]; +} + + +static void +CheckCandidate32I(const int x, const int y, SearchData * const data, const unsigned int Direction) +{ + /* maximum speed */ + int32_t sad; + + if ( (x > data->max_dx) || (x < data->min_dx) + || (y > data->max_dy) || (y < data->min_dy) ) return; + + sad = sad32v_c(data->Cur, data->RefP[0] + x + y*((int)data->iEdgedWidth), + data->iEdgedWidth, data->temp); + + if (sad < *(data->iMinSAD)) { + *(data->iMinSAD) = sad; + data->currentMV[0].x = x; data->currentMV[0].y = y; + data->dir = Direction; + } + if (data->temp[0] < data->iMinSAD[1]) { + data->iMinSAD[1] = data->temp[0]; data->currentMV[1].x = x; data->currentMV[1].y = y; } + if (data->temp[1] < data->iMinSAD[2]) { + data->iMinSAD[2] = data->temp[1]; data->currentMV[2].x = x; data->currentMV[2].y = y; } + if (data->temp[2] < data->iMinSAD[3]) { + data->iMinSAD[3] = data->temp[2]; data->currentMV[3].x = x; data->currentMV[3].y = y; } + if (data->temp[3] < data->iMinSAD[4]) { + data->iMinSAD[4] = data->temp[3]; data->currentMV[4].x = x; data->currentMV[4].y = y; } +} + + +/* + * Calculate the min/max range + * relative to the _MACROBLOCK_ position + */ +static inline void +get_range(int32_t * const min_dx, + int32_t * const max_dx, + int32_t * const min_dy, + int32_t * const max_dy, + const uint32_t x, + const uint32_t y, + const uint32_t width, + const uint32_t height, + int fcode) +{ + const uint32_t block_sz = 4; + const int precision = 0; + int k; + const int search_range = 1 << (4+fcode); + int high = search_range - 1; + int low = -search_range; + + k = (int)(width - (x<= bound && lx >= 0) { + num_cand++; + pmv[1] = mbs[lpos].mvs[lz]; + } else pmv[1] = (VECTOR){ 0, 0 }; + + if (tpos >= bound) { + num_cand++; + last_cand = 2; + pmv[2] = mbs[tpos].mvs[tz]; + } else pmv[2] = (VECTOR){ 0, 0 }; + + if (rpos >= bound && rx < mb_width) { + num_cand++; + last_cand = 3; + pmv[3] = mbs[rpos].mvs[rz]; + } else pmv[3] = (VECTOR){ 0, 0 }; + + /* If there're more than one candidate, we return the median vector */ + + if (num_cand > 1) { + /* set median */ + pmv[0].x = + VSMIN(VSMAX(pmv[1].x, pmv[2].x), + VSMIN(VSMAX(pmv[2].x, pmv[3].x), VSMAX(pmv[1].x, pmv[3].x))); + pmv[0].y = + VSMIN(VSMAX(pmv[1].y, pmv[2].y), + VSMIN(VSMAX(pmv[2].y, pmv[3].y), VSMAX(pmv[1].y, pmv[3].y))); + return pmv[0]; + } + + return pmv[last_cand]; /* no point calculating median mv */ +} + + +/* check if given vector is equal to any vector checked before */ +static inline int +vector_repeats(const VECTOR * const pmv, const unsigned int i) +{ + unsigned int j; + for (j = 0; j < i; j++) + if (MVequal(pmv[i], pmv[j])) return 1; /* same vector has been checked already */ + return 0; +} + + +/* make a binary mask that prevents diamonds/squares + from checking a vector which has been checked as a prediction */ +static inline int +make_mask(const VECTOR * const pmv, const unsigned int i, const unsigned int current) +{ + unsigned int mask = 255, j; + for (j = 0; j < i; j++) { + if (pmv[current].x == pmv[j].x) { + if (pmv[current].y == pmv[j].y + iDiamondSize) mask &= ~4; + else if (pmv[current].y == pmv[j].y - iDiamondSize) mask &= ~8; + } else + if (pmv[current].y == pmv[j].y) { + if (pmv[current].x == pmv[j].x + iDiamondSize) mask &= ~1; + else if (pmv[current].x == pmv[j].x - iDiamondSize) mask &= ~2; + } + } + return mask; +} + + +static void +xvid_me_DiamondSearch(int x, int y, SearchData * const data, + int bDirection, CheckFunc * const CheckCandidate) +{ + + /* directions: 1 - left (x-1); 2 - right (x+1), 4 - up (y-1); 8 - down (y+1) */ + + unsigned int * const iDirection = &data->dir; + + for (;;) { + *iDirection = 0; + if (bDirection & 1) CHECK_CANDIDATE(x - iDiamondSize, y, 1); + if (bDirection & 2) CHECK_CANDIDATE(x + iDiamondSize, y, 2); + if (bDirection & 4) CHECK_CANDIDATE(x, y - iDiamondSize, 4); + if (bDirection & 8) CHECK_CANDIDATE(x, y + iDiamondSize, 8); + + if (*iDirection == 0) + break; + + /* now we're doing diagonal checks near our candidate */ + bDirection = *iDirection; + x = data->currentMV->x; y = data->currentMV->y; + if (bDirection & 3) { /* our candidate is left or right */ + CHECK_CANDIDATE(x, y + iDiamondSize, 8); + CHECK_CANDIDATE(x, y - iDiamondSize, 4); + } else { /* what remains here is up or down */ + CHECK_CANDIDATE(x + iDiamondSize, y, 2); + CHECK_CANDIDATE(x - iDiamondSize, y, 1); + } + bDirection |= *iDirection; + x = data->currentMV->x; y = data->currentMV->y; + } +} + + +static void +xvid_me_SubpelRefine(VECTOR centerMV, SearchData * const data, CheckFunc * const CheckCandidate, int dir) +{ + /* Do a half-pel or q-pel refinement */ + + CHECK_CANDIDATE(centerMV.x, centerMV.y - 1, dir); + CHECK_CANDIDATE(centerMV.x + 1, centerMV.y - 1, dir); + CHECK_CANDIDATE(centerMV.x + 1, centerMV.y, dir); + CHECK_CANDIDATE(centerMV.x + 1, centerMV.y + 1, dir); + CHECK_CANDIDATE(centerMV.x, centerMV.y + 1, dir); + CHECK_CANDIDATE(centerMV.x - 1, centerMV.y + 1, dir); + CHECK_CANDIDATE(centerMV.x - 1, centerMV.y, dir); + CHECK_CANDIDATE(centerMV.x - 1, centerMV.y - 1, dir); +} + + +static inline void +MEanalyzeMB ( const uint8_t * const pRef, + const uint8_t * const pCur, + const int x, + const int y, + const MBParam * const pParam, + MACROBLOCK * const pMBs, + SearchData * const Data, + const int fcode) +{ + + int i; + VECTOR pmv[3]; + MACROBLOCK * const pMB = &pMBs[x + y * pParam->mb_width]; + + unsigned int simplicity = 0; + + for (i = 0; i < 5; i++) Data->iMinSAD[i] = MV_MAX_ERROR; + + get_range(&Data->min_dx, &Data->max_dx, &Data->min_dy, &Data->max_dy, x, y, + pParam->width, pParam->height, fcode - 1); + + Data->Cur = pCur + (x + y * pParam->edged_width) * 16; + Data->RefP[0] = pRef + (x + y * pParam->edged_width) * 16; + + pmv[0].x = pMB->mvs[0].x; + pmv[0].y = pMB->mvs[0].y; + + CheckCandidate32I(pmv[0].x, pmv[0].y, Data, 0); + + + if (*Data->iMinSAD > 200) { + + pmv[1].x = pmv[1].y = 0; + + /* median is only used as prediction. it doesn't have to be real */ + if (x == 1 && y == 1) Data->predMV.x = Data->predMV.y = 0; + else + if (x == 1) /* left macroblock does not have any vector now */ + Data->predMV = (pMB - pParam->mb_width)->mvs[0]; /* top instead of median */ + else if (y == 1) /* top macroblock doesn't have it's vector */ + Data->predMV = (pMB - 1)->mvs[0]; /* left instead of median */ + else + Data->predMV = get_pmv2(pMBs, pParam->mb_width, x, y); /* else median */ + + pmv[2].x = Data->predMV.x; + pmv[2].y = Data->predMV.y; + + //if (!vector_repeats(pmv, 1)) // branch never taken + // CheckCandidate32I(pmv[1].x, pmv[1].y, Data, 1); + if (!vector_repeats(pmv, 2)) + CheckCandidate32I(pmv[2].x, pmv[2].y, Data, 2); + + if (*Data->iMinSAD > 500) { /* diamond only if needed */ + unsigned int mask = make_mask(pmv, 3, Data->dir); + xvid_me_DiamondSearch(Data->currentMV->x, Data->currentMV->y, Data, mask, CheckCandidate32I); + } else simplicity++; + + if (*Data->iMinSAD > 500) /* refinement from 2-pixel to 1-pixel */ + xvid_me_SubpelRefine(Data->currentMV[0], Data, CheckCandidate32I, 0); + else simplicity++; + } else simplicity++; + + for (i = 0; i < 4; i++) { + MACROBLOCK * MB = &pMBs[x + (i&1) + (y+(i>>1)) * pParam->mb_width]; + MB->mvs[0] = MB->mvs[1] = MB->mvs[2] = MB->mvs[3] = Data->currentMV[i]; + MB->mode = MODE_INTER; + /* if we skipped some search steps, we have to assume that SAD would be lower with them */ + MB->sad16 = Data->iMinSAD[i+1] - (simplicity<<7); + if (MB->sad16 < 0) MB->sad16 = 0; + } +} + + +/* average deviation from mean */ + +static uint32_t +dev16_c(const uint8_t * const cur, + const uint32_t stride) +{ + + uint32_t mean = 0; + uint32_t dev = 0; + uint32_t i, j; + uint8_t const *ptr_cur = cur; + + for (j = 0; j < 16; j++) { + + for (i = 0; i < 16; i++) + mean += *(ptr_cur + i); + + ptr_cur += stride; + + } + + mean /= (16 * 16); + ptr_cur = cur; + + for (j = 0; j < 16; j++) { + + for (i = 0; i < 16; i++) + dev += abs(*(ptr_cur + i) - (int32_t) mean); + + ptr_cur += stride; + + } + + return dev; +} + + +int +MEanalysis( const uint8_t *pRef, + const uint8_t *pCurrent, + const MBParam * const pParam, + MACROBLOCK * const pMBs, + const int intraCount, + const int fcode) +{ + uint32_t x, y, intra = 0; + int sSAD = 0; + int IntraThresh = 2000, + IntraThresh2 = 90; + + int blocks = 10; + int complexity = 0; + + SearchData Data; + Data.iEdgedWidth = pParam->edged_width; + + if (intraCount > 0 && intraCount < 30) { + if (intraCount < 10) { + IntraThresh += 15 * (10 - intraCount) * (10 - intraCount); + } + IntraThresh2 += 4 * (30 - intraCount); + } + + for (y = 1; y < pParam->mb_height-1; y += 2) { + for (x = 1; x < pParam->mb_width-1; x += 2) { + int i; + blocks += 10; + + pMBs[x + y * pParam->mb_width].mvs[0] = (VECTOR){ 0, 0 }; + + MEanalyzeMB(pRef, pCurrent, x, y, pParam, pMBs, &Data, fcode); + + for (i = 0; i < 4; i++) { + int dev; + MACROBLOCK *pMB = &pMBs[x+(i&1) + (y+(i>>1)) * pParam->mb_width]; + dev = dev16_c(pCurrent + (x + (i&1) + (y + (i>>1)) * pParam->edged_width) * 16, + pParam->edged_width); + + complexity += VSMAX(dev, 300); + if (dev + IntraThresh < pMB->sad16) { + pMB->mode = MODE_INTRA; + if (++intra > ((pParam->mb_height-2)*(pParam->mb_width-2))/2) + return 1; + } + + if (pMB->mvs[0].x == 0 && pMB->mvs[0].y == 0) + if (dev > 1000 && pMB->sad16 < 1000) + sSAD += 512; + + sSAD += (dev < 3000) ? pMB->sad16 : pMB->sad16/2; /* blocks with big contrast differences usually have large SAD - while they look very good in b-frames */ + } + } + } + complexity >>= 7; + + sSAD /= complexity + 4*blocks; + + return sSAD > IntraThresh2; +} diff --git a/src/detection.h b/src/detection.h new file mode 100644 index 0000000..93c1ecd --- /dev/null +++ b/src/detection.h @@ -0,0 +1,44 @@ +#include + + +typedef struct { + int x, y; +} VECTOR; + + +typedef struct { + VECTOR mvs[4]; + int32_t sad16; + int mode; +} MACROBLOCK; + + +typedef struct { + uint32_t width; + uint32_t height; + uint32_t edged_width; + uint32_t edged_height; + uint32_t mb_width; + uint32_t mb_height; + int edge_size; +} MBParam; + + +typedef struct { + int min_dx, max_dx, min_dy, max_dy; + int32_t iMinSAD[5]; + VECTOR currentMV[5]; + int temp[4]; + unsigned int dir; + VECTOR predMV; + const uint8_t *RefP[1], *Cur; + uint32_t iEdgedWidth; +} SearchData; + + +int MEanalysis( const uint8_t *pRef, + const uint8_t *pCurrent, + const MBParam * const pParam, + MACROBLOCK * const pMBs, + const int intraCount, + const int fcode); diff --git a/src/wwxd.c b/src/wwxd.c new file mode 100644 index 0000000..2577d02 --- /dev/null +++ b/src/wwxd.c @@ -0,0 +1,145 @@ +#include +#include +#include +#include + +#include +#include + +#include "detection.h" + + +typedef struct { + VSNodeRef *node; + const VSVideoInfo *vi; + + MBParam Param; + int intraCount; + int fcode; + + uint8_t *padded_prevp; + uint8_t *padded_curp; + + MACROBLOCK *pMBs; +} WwxdData; + + +static void VS_CC wwxdInit(VSMap *in, VSMap *out, void **instanceData, VSNode *node, VSCore *core, const VSAPI *vsapi) { + WwxdData *d = (WwxdData *) * instanceData; + vsapi->setVideoInfo(d->vi, 1, node); +} + + +static const VSFrameRef *VS_CC wwxdGetFrame(int n, int activationReason, void **instanceData, void **frameData, VSFrameContext *frameCtx, VSCore *core, const VSAPI *vsapi) { + WwxdData *d = (WwxdData *) * instanceData; + + if (activationReason == arInitial) { + vsapi->requestFrameFilter(VSMAX(0, n - 1), d->node, frameCtx); + vsapi->requestFrameFilter(n, d->node, frameCtx); + } else if (activationReason == arAllFramesReady) { + int scenechange, i; + + const VSFrameRef *prev = vsapi->getFrameFilter(VSMAX(0, n - 1), d->node, frameCtx); + const VSFrameRef *cur = vsapi->getFrameFilter(n, d->node, frameCtx); + + int stride = vsapi->getStride(cur, 0); + + uint8_t *padded_prevp = d->padded_prevp + d->Param.edge_size * d->Param.edged_width + d->Param.edge_size; + uint8_t *padded_curp = d->padded_curp + d->Param.edge_size * d->Param.edged_width + d->Param.edge_size; + + vs_bitblt(padded_prevp, d->Param.edged_width, vsapi->getReadPtr(prev, 0), stride, d->Param.width, d->Param.height); + vs_bitblt(padded_curp, d->Param.edged_width, vsapi->getReadPtr(cur, 0), stride, d->Param.width, d->Param.height); + + if (d->vi->width % 16) { + int pad_width = 16 - (d->vi->width % 16); + for (i = 0; i < d->vi->height; i++) { + memset(padded_prevp + i*d->Param.edged_width + d->vi->width, *(padded_prevp + i*d->Param.edged_width + d->vi->width - 1), pad_width); + memset(padded_curp + i*d->Param.edged_width + d->vi->width, *(padded_curp + i*d->Param.edged_width + d->vi->width - 1), pad_width); + } + } + + if (d->vi->height % 16) { + int pad_height = 16 - (d->vi->height % 16); + for (i = 0; i < pad_height; i++) { + memcpy(padded_prevp + (d->vi->height + i)*d->Param.edged_width, padded_prevp + (d->vi->height - 1)*d->Param.edged_width, d->Param.edged_width); + memcpy(padded_curp + (d->vi->height + i)*d->Param.edged_width, padded_curp + (d->vi->height - 1)*d->Param.edged_width, d->Param.edged_width); + } + } + + if (n == 0) { + scenechange = 1; + } else { + scenechange = MEanalysis(padded_prevp, padded_curp, &d->Param, d->pMBs, d->intraCount, d->fcode); + } + + if (scenechange) { + d->intraCount = 1; + } else { + d->intraCount++; + } + + VSFrameRef *dst = vsapi->copyFrame(cur, core); + vsapi->freeFrame(prev); + vsapi->freeFrame(cur); + + VSMap *props = vsapi->getFramePropsRW(dst); + vsapi->propSetInt(props, "Scenechange", scenechange, paReplace); + + return dst; + } + + return 0; +} + + +static void VS_CC wwxdFree(void *instanceData, VSCore *core, const VSAPI *vsapi) { + WwxdData *d = (WwxdData *)instanceData; + + vsapi->freeNode(d->node); + free(d->padded_prevp); + free(d->padded_curp); + free(d->pMBs); + free(d); +} + + +static void VS_CC wwxdCreate(const VSMap *in, VSMap *out, void *userData, VSCore *core, const VSAPI *vsapi) { + WwxdData d; + WwxdData *data; + + d.node = vsapi->propGetNode(in, "clip", 0, 0); + d.vi = vsapi->getVideoInfo(d.node); + + d.Param.edge_size = 64; + d.Param.width = d.vi->width; + d.Param.height = d.vi->height; + d.Param.mb_width = (d.vi->width + 15) / 16; + d.Param.mb_height = (d.vi->height + 15) / 16; + d.Param.edged_width = 16 * d.Param.mb_width + 2 * d.Param.edge_size; + d.Param.edged_height = 16 * d.Param.mb_height + 2 * d.Param.edge_size; + + d.intraCount = 1; + + d.fcode = 4; + + d.padded_prevp = (uint8_t *)malloc(d.Param.edged_width * (d.Param.edged_height + 1) + 64); + d.padded_curp = (uint8_t *)malloc(d.Param.edged_width * (d.Param.edged_height + 1) + 64); + memset(d.padded_prevp, 0, d.Param.edged_width * (d.Param.edged_height + 1) + 64); + memset(d.padded_curp, 0, d.Param.edged_width * (d.Param.edged_height + 1) + 64); + + d.pMBs = (MACROBLOCK *)malloc(d.Param.mb_width * d.Param.mb_height * sizeof(MACROBLOCK)); + memset(d.pMBs, 0, d.Param.mb_width * d.Param.mb_height * sizeof(MACROBLOCK)); + + data = malloc(sizeof(d)); + *data = d; + + vsapi->createFilter(in, out, "Wwxd", wwxdInit, wwxdGetFrame, wwxdFree, fmSerial, 0, data, core); +} + + +VS_EXTERNAL_API(void) VapourSynthPluginInit(VSConfigPlugin configFunc, VSRegisterFunction registerFunc, VSPlugin *plugin) { + configFunc("com.nodame.wwxd", "wwxd", "Scene change detection approximately like Xvid's", VAPOURSYNTH_API_VERSION, 1, plugin); + registerFunc("WWXD", + "clip:clip;", + wwxdCreate, 0, plugin); +}