diff --git a/libavcodec/avcodec.h b/libavcodec/avcodec.h index 0a263cbb08b6ea43961dc3006b32b0e8982e647c..70df54250064fee83ca6436fb24ada8253bdde20 100644 --- a/libavcodec/avcodec.h +++ b/libavcodec/avcodec.h @@ -915,8 +915,8 @@ typedef struct AVCodecContext { void (*release_buffer)(struct AVCodecContext *c, AVFrame *pic); /** - * is 1 if the decoded stream contains b frames, 0 otherwise. - * - encoding: unused + * if 1 the stream has a 1 frame delay during decoding. + * - encoding: set by lavc * - decoding: set by lavc */ int has_b_frames; @@ -1251,6 +1251,7 @@ typedef struct AVCodecContext { #define FF_CMP_ZERO 7 #define FF_CMP_VSAD 8 #define FF_CMP_VSSE 9 +#define FF_CMP_NSSE 10 #define FF_CMP_CHROMA 256 /** diff --git a/libavcodec/dsputil.c b/libavcodec/dsputil.c index 3e3e550b115c2b77ae47eb9f7bd2fbc0a5616179..5c84a139be3f8a13a1104b82c3a4e97f2b390f61 100644 --- a/libavcodec/dsputil.c +++ b/libavcodec/dsputil.c @@ -2587,6 +2587,54 @@ static int pix_abs8_xy2_c(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, return s; } +static int nsse16_c(/*MpegEncContext*/ void *c, uint8_t *s1, uint8_t *s2, int stride, int h){ + int score1=0; + int score2=0; + int x,y; + + for(y=0; y<h; y++){ + for(x=0; x<16; x++){ + score1+= (s1[x ] - s2[x ])*(s1[x ] - s2[x ]); + } + if(y+1<h){ + for(x=0; x<15; x++){ + score2+= ABS( s1[x ] - s1[x +stride] + - s1[x+1] + s1[x+1+stride]) + -ABS( s2[x ] - s2[x +stride] + - s2[x+1] + s2[x+1+stride]); + } + } + s1+= stride; + s2+= stride; + } + + return score1 + ABS(score2)*8; +} + +static int nsse8_c(/*MpegEncContext*/ void *c, uint8_t *s1, uint8_t *s2, int stride, int h){ + int score1=0; + int score2=0; + int x,y; + + for(y=0; y<h; y++){ + for(x=0; x<8; x++){ + score1+= (s1[x ] - s2[x ])*(s1[x ] - s2[x ]); + } + if(y+1<h){ + for(x=0; x<7; x++){ + score2+= ABS( s1[x ] - s1[x +stride] + - s1[x+1] + s1[x+1+stride]) + -ABS( s2[x ] - s2[x +stride] + - s2[x+1] + s2[x+1+stride]); + } + } + s1+= stride; + s2+= stride; + } + + return score1 + ABS(score2)*8; +} + static int try_8x8basis_c(int16_t rem[64], int16_t weight[64], int16_t basis[64], int scale){ int i; unsigned int sum=0; @@ -2680,6 +2728,9 @@ void ff_set_cmp(DSPContext* c, me_cmp_func *cmp, int type){ case FF_CMP_ZERO: cmp[i]= zero_cmp; break; + case FF_CMP_NSSE: + cmp[i]= c->nsse[i]; + break; default: av_log(NULL, AV_LOG_ERROR,"internal error in cmp function selection\n"); } @@ -3313,6 +3364,8 @@ void dsputil_init(DSPContext* c, AVCodecContext *avctx) c->vsad[4]= vsad_intra16_c; c->vsse[0]= vsse16_c; c->vsse[4]= vsse_intra16_c; + c->nsse[0]= nsse16_c; + c->nsse[1]= nsse8_c; c->add_bytes= add_bytes_c; c->diff_bytes= diff_bytes_c; diff --git a/libavcodec/dsputil.h b/libavcodec/dsputil.h index cdc5a3b859abe925bed1c9417d5d5327ed2c327a..be53279797f8bdc2afd7d366dbb0055ad5468cd3 100644 --- a/libavcodec/dsputil.h +++ b/libavcodec/dsputil.h @@ -162,6 +162,7 @@ typedef struct DSPContext { me_cmp_func rd[5]; me_cmp_func vsad[5]; me_cmp_func vsse[5]; + me_cmp_func nsse[5]; me_cmp_func me_pre_cmp[5]; me_cmp_func me_cmp[5]; diff --git a/libavcodec/motion_est.c b/libavcodec/motion_est.c index b92b6dfad8dd0bd07b6dd03ce015bb64c7957878..bee1e07f008475ea873d096b47df4b55e30b8947 100644 --- a/libavcodec/motion_est.c +++ b/libavcodec/motion_est.c @@ -223,6 +223,7 @@ static inline int get_penalty_factor(MpegEncContext *s, int type){ switch(type&0xFF){ default: case FF_CMP_SAD: + case FF_CMP_NSSE: return s->lambda>>FF_LAMBDA_SHIFT; case FF_CMP_DCT: return (3*s->lambda)>>(FF_LAMBDA_SHIFT+1); diff --git a/libavcodec/mpegvideo.c b/libavcodec/mpegvideo.c index bd2de7cd0feae75301da847385e213b52c829006..f402a81269eab94b0e681b212fd38e2053a460f3 100644 --- a/libavcodec/mpegvideo.c +++ b/libavcodec/mpegvideo.c @@ -3916,9 +3916,15 @@ static int sse_mb(MpegEncContext *s){ if(s->mb_y*16 + 16 > s->height) h= s->height- s->mb_y*16; if(w==16 && h==16) + if(s->avctx->mb_cmp == FF_CMP_NSSE){ + return s->dsp.nsse[0](NULL, s->new_picture.data[0] + s->mb_x*16 + s->mb_y*s->linesize*16, s->dest[0], s->linesize, 16) + +s->dsp.nsse[1](NULL, s->new_picture.data[1] + s->mb_x*8 + s->mb_y*s->uvlinesize*8,s->dest[1], s->uvlinesize, 8) + +s->dsp.nsse[1](NULL, s->new_picture.data[2] + s->mb_x*8 + s->mb_y*s->uvlinesize*8,s->dest[2], s->uvlinesize, 8); + }else{ return s->dsp.sse[0](NULL, s->new_picture.data[0] + s->mb_x*16 + s->mb_y*s->linesize*16, s->dest[0], s->linesize, 16) +s->dsp.sse[1](NULL, s->new_picture.data[1] + s->mb_x*8 + s->mb_y*s->uvlinesize*8,s->dest[1], s->uvlinesize, 8) +s->dsp.sse[1](NULL, s->new_picture.data[2] + s->mb_x*8 + s->mb_y*s->uvlinesize*8,s->dest[2], s->uvlinesize, 8); + } else return sse(s, s->new_picture.data[0] + s->mb_x*16 + s->mb_y*s->linesize*16, s->dest[0], w, h, s->linesize) +sse(s, s->new_picture.data[1] + s->mb_x*8 + s->mb_y*s->uvlinesize*8,s->dest[1], w>>1, h>>1, s->uvlinesize)