X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=libavcodec%2Fh264dsp.h;h=c4be235519a0cb0575400f563a12a91bfc22ba68;hb=4628443ca3534060888dd0015b229337eac13fd2;hp=7eb50be342c774802248e72b69a4ad032947a9e4;hpb=e39e3abad4ce667f06d41097428695a7aad095fb;p=ffmpeg diff --git a/libavcodec/h264dsp.h b/libavcodec/h264dsp.h index 7eb50be342c..c4be235519a 100644 --- a/libavcodec/h264dsp.h +++ b/libavcodec/h264dsp.h @@ -28,56 +28,103 @@ #define AVCODEC_H264DSP_H #include -#include "dsputil.h" -//typedef void (*h264_chroma_mc_func)(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int srcStride, int h, int x, int y); -typedef void (*h264_weight_func)(uint8_t *block, int stride, int log2_denom, int weight, int offset); -typedef void (*h264_biweight_func)(uint8_t *dst, uint8_t *src, int stride, int log2_denom, int weightd, int weights, int offset); +typedef void (*h264_weight_func)(uint8_t *block, int stride, int height, + int log2_denom, int weight, int offset); +typedef void (*h264_biweight_func)(uint8_t *dst, uint8_t *src, + int stride, int height, int log2_denom, + int weightd, int weights, int offset); /** * Context for storing H.264 DSP functions */ -typedef struct H264DSPContext{ +typedef struct H264DSPContext { /* weighted MC */ - h264_weight_func weight_h264_pixels_tab[10]; - h264_biweight_func biweight_h264_pixels_tab[10]; + h264_weight_func weight_h264_pixels_tab[4]; + h264_biweight_func biweight_h264_pixels_tab[4]; /* loop filter */ - void (*h264_v_loop_filter_luma)(uint8_t *pix/*align 16*/, int stride, int alpha, int beta, int8_t *tc0); - void (*h264_h_loop_filter_luma)(uint8_t *pix/*align 4 */, int stride, int alpha, int beta, int8_t *tc0); - void (*h264_h_loop_filter_luma_mbaff)(uint8_t *pix/*align 16*/, int stride, int alpha, int beta, int8_t *tc0); + void (*h264_v_loop_filter_luma)(uint8_t *pix /*align 16*/, int stride, + int alpha, int beta, int8_t *tc0); + void (*h264_h_loop_filter_luma)(uint8_t *pix /*align 4 */, int stride, + int alpha, int beta, int8_t *tc0); + void (*h264_h_loop_filter_luma_mbaff)(uint8_t *pix /*align 16*/, int stride, + int alpha, int beta, int8_t *tc0); /* v/h_loop_filter_luma_intra: align 16 */ - void (*h264_v_loop_filter_luma_intra)(uint8_t *pix, int stride, int alpha, int beta); - void (*h264_h_loop_filter_luma_intra)(uint8_t *pix, int stride, int alpha, int beta); - void (*h264_h_loop_filter_luma_mbaff_intra)(uint8_t *pix/*align 16*/, int stride, int alpha, int beta); - void (*h264_v_loop_filter_chroma)(uint8_t *pix/*align 8*/, int stride, int alpha, int beta, int8_t *tc0); - void (*h264_h_loop_filter_chroma)(uint8_t *pix/*align 4*/, int stride, int alpha, int beta, int8_t *tc0); - void (*h264_h_loop_filter_chroma_mbaff)(uint8_t *pix/*align 8*/, int stride, int alpha, int beta, int8_t *tc0); - void (*h264_v_loop_filter_chroma_intra)(uint8_t *pix/*align 8*/, int stride, int alpha, int beta); - void (*h264_h_loop_filter_chroma_intra)(uint8_t *pix/*align 8*/, int stride, int alpha, int beta); - void (*h264_h_loop_filter_chroma_mbaff_intra)(uint8_t *pix/*align 8*/, int stride, int alpha, int beta); + void (*h264_v_loop_filter_luma_intra)(uint8_t *pix, int stride, + int alpha, int beta); + void (*h264_h_loop_filter_luma_intra)(uint8_t *pix, int stride, + int alpha, int beta); + void (*h264_h_loop_filter_luma_mbaff_intra)(uint8_t *pix /*align 16*/, + int stride, int alpha, int beta); + void (*h264_v_loop_filter_chroma)(uint8_t *pix /*align 8*/, int stride, + int alpha, int beta, int8_t *tc0); + void (*h264_h_loop_filter_chroma)(uint8_t *pix /*align 4*/, int stride, + int alpha, int beta, int8_t *tc0); + void (*h264_h_loop_filter_chroma_mbaff)(uint8_t *pix /*align 8*/, + int stride, int alpha, int beta, + int8_t *tc0); + void (*h264_v_loop_filter_chroma_intra)(uint8_t *pix /*align 8*/, + int stride, int alpha, int beta); + void (*h264_h_loop_filter_chroma_intra)(uint8_t *pix /*align 8*/, + int stride, int alpha, int beta); + void (*h264_h_loop_filter_chroma_mbaff_intra)(uint8_t *pix /*align 8*/, + int stride, int alpha, int beta); // h264_loop_filter_strength: simd only. the C version is inlined in h264.c - void (*h264_loop_filter_strength)(int16_t bS[2][4][4], uint8_t nnz[40], int8_t ref[2][40], int16_t mv[2][40][2], - int bidir, int edges, int step, int mask_mv0, int mask_mv1, int field); + void (*h264_loop_filter_strength)(int16_t bS[2][4][4], uint8_t nnz[40], + int8_t ref[2][40], int16_t mv[2][40][2], + int bidir, int edges, int step, + int mask_mv0, int mask_mv1, int field); /* IDCT */ - void (*h264_idct_add)(uint8_t *dst/*align 4*/, DCTELEM *block/*align 16*/, int stride); - void (*h264_idct8_add)(uint8_t *dst/*align 8*/, DCTELEM *block/*align 16*/, int stride); - void (*h264_idct_dc_add)(uint8_t *dst/*align 4*/, DCTELEM *block/*align 16*/, int stride); - void (*h264_idct8_dc_add)(uint8_t *dst/*align 8*/, DCTELEM *block/*align 16*/, int stride); + void (*h264_idct_add)(uint8_t *dst /*align 4*/, + int16_t *block /*align 16*/, int stride); + void (*h264_idct8_add)(uint8_t *dst /*align 8*/, + int16_t *block /*align 16*/, int stride); + void (*h264_idct_dc_add)(uint8_t *dst /*align 4*/, + int16_t *block /*align 16*/, int stride); + void (*h264_idct8_dc_add)(uint8_t *dst /*align 8*/, + int16_t *block /*align 16*/, int stride); - void (*h264_dct)(DCTELEM block[4][4]); - void (*h264_idct_add16)(uint8_t *dst/*align 16*/, const int *blockoffset, DCTELEM *block/*align 16*/, int stride, const uint8_t nnzc[6*8]); - void (*h264_idct8_add4)(uint8_t *dst/*align 16*/, const int *blockoffset, DCTELEM *block/*align 16*/, int stride, const uint8_t nnzc[6*8]); - void (*h264_idct_add8)(uint8_t **dst/*align 16*/, const int *blockoffset, DCTELEM *block/*align 16*/, int stride, const uint8_t nnzc[6*8]); - void (*h264_idct_add16intra)(uint8_t *dst/*align 16*/, const int *blockoffset, DCTELEM *block/*align 16*/, int stride, const uint8_t nnzc[6*8]); - void (*h264_luma_dc_dequant_idct)(DCTELEM *output, DCTELEM *input/*align 16*/, int qmul); - void (*h264_chroma_dc_dequant_idct)(DCTELEM *block, int qmul); -}H264DSPContext; + void (*h264_idct_add16)(uint8_t *dst /*align 16*/, const int *blockoffset, + int16_t *block /*align 16*/, int stride, + const uint8_t nnzc[15 * 8]); + void (*h264_idct8_add4)(uint8_t *dst /*align 16*/, const int *blockoffset, + int16_t *block /*align 16*/, int stride, + const uint8_t nnzc[15 * 8]); + void (*h264_idct_add8)(uint8_t **dst /*align 16*/, const int *blockoffset, + int16_t *block /*align 16*/, int stride, + const uint8_t nnzc[15 * 8]); + void (*h264_idct_add16intra)(uint8_t *dst /*align 16*/, const int *blockoffset, + int16_t *block /*align 16*/, + int stride, const uint8_t nnzc[15 * 8]); + void (*h264_luma_dc_dequant_idct)(int16_t *output, + int16_t *input /*align 16*/, int qmul); + void (*h264_chroma_dc_dequant_idct)(int16_t *block, int qmul); -void ff_h264dsp_init(H264DSPContext *c); -void ff_h264dsp_init_arm(H264DSPContext *c); -void ff_h264dsp_init_ppc(H264DSPContext *c); -void ff_h264dsp_init_x86(H264DSPContext *c); + /* bypass-transform */ + void (*h264_add_pixels8_clear)(uint8_t *dst, int16_t *block, int stride); + void (*h264_add_pixels4_clear)(uint8_t *dst, int16_t *block, int stride); + + /** + * Search buf from the start for up to size bytes. Return the index + * of a zero byte, or >= size if not found. Ideally, use lookahead + * to filter out any zero bytes that are known to not be followed by + * one or more further zero bytes and a one byte. Better still, filter + * out any bytes that form the trailing_zero_8bits syntax element too. + */ + int (*startcode_find_candidate)(const uint8_t *buf, int size); +} H264DSPContext; + +void ff_h264dsp_init(H264DSPContext *c, const int bit_depth, + const int chroma_format_idc); +void ff_h264dsp_init_aarch64(H264DSPContext *c, const int bit_depth, + const int chroma_format_idc); +void ff_h264dsp_init_arm(H264DSPContext *c, const int bit_depth, + const int chroma_format_idc); +void ff_h264dsp_init_ppc(H264DSPContext *c, const int bit_depth, + const int chroma_format_idc); +void ff_h264dsp_init_x86(H264DSPContext *c, const int bit_depth, + const int chroma_format_idc); #endif /* AVCODEC_H264DSP_H */