Preparatory patch for high bit depth h264 decoding support.

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>

1 files changed, 1 insertions, 258 deletions

diff --git a/libavcodec/h264dsp.c b/libavcodec/h264dsp.c
index b555ed3..22b2086 100644
--- a/libavcodec/h264dsp.c
+++ b/libavcodec/h264dsp.c
@@ -29,264 +29,7 @@
 #include "avcodec.h"
 #include "h264dsp.h"
 
-#define op_scale1(x)  block[x] = av_clip_uint8( (block[x]*weight + offset) >> log2_denom )
-#define op_scale2(x)  dst[x] = av_clip_uint8( (src[x]*weights + dst[x]*weightd + offset) >> (log2_denom+1))
-#define H264_WEIGHT(W,H) \
-static void weight_h264_pixels ## W ## x ## H ## _c(uint8_t *block, int stride, int log2_denom, int weight, int offset){ \
-    int y; \
-    offset <<= log2_denom; \
-    if(log2_denom) offset += 1<<(log2_denom-1); \
-    for(y=0; y<H; y++, block += stride){ \
-        op_scale1(0); \
-        op_scale1(1); \
-        if(W==2) continue; \
-        op_scale1(2); \
-        op_scale1(3); \
-        if(W==4) continue; \
-        op_scale1(4); \
-        op_scale1(5); \
-        op_scale1(6); \
-        op_scale1(7); \
-        if(W==8) continue; \
-        op_scale1(8); \
-        op_scale1(9); \
-        op_scale1(10); \
-        op_scale1(11); \
-        op_scale1(12); \
-        op_scale1(13); \
-        op_scale1(14); \
-        op_scale1(15); \
-    } \
-} \
-static void biweight_h264_pixels ## W ## x ## H ## _c(uint8_t *dst, uint8_t *src, int stride, int log2_denom, int weightd, int weights, int offset){ \
-    int y; \
-    offset = ((offset + 1) | 1) << log2_denom; \
-    for(y=0; y<H; y++, dst += stride, src += stride){ \
-        op_scale2(0); \
-        op_scale2(1); \
-        if(W==2) continue; \
-        op_scale2(2); \
-        op_scale2(3); \
-        if(W==4) continue; \
-        op_scale2(4); \
-        op_scale2(5); \
-        op_scale2(6); \
-        op_scale2(7); \
-        if(W==8) continue; \
-        op_scale2(8); \
-        op_scale2(9); \
-        op_scale2(10); \
-        op_scale2(11); \
-        op_scale2(12); \
-        op_scale2(13); \
-        op_scale2(14); \
-        op_scale2(15); \
-    } \
-}
-
-H264_WEIGHT(16,16)
-H264_WEIGHT(16,8)
-H264_WEIGHT(8,16)
-H264_WEIGHT(8,8)
-H264_WEIGHT(8,4)
-H264_WEIGHT(4,8)
-H264_WEIGHT(4,4)
-H264_WEIGHT(4,2)
-H264_WEIGHT(2,4)
-H264_WEIGHT(2,2)
-
-#undef op_scale1
-#undef op_scale2
-#undef H264_WEIGHT
-
-static av_always_inline av_flatten void h264_loop_filter_luma_c(uint8_t *pix, int xstride, int ystride, int inner_iters, int alpha, int beta, int8_t *tc0)
-{
-    int i, d;
-    for( i = 0; i < 4; i++ ) {
-        if( tc0[i] < 0 ) {
-            pix += inner_iters*ystride;
-            continue;
-        }
-        for( d = 0; d < inner_iters; d++ ) {
-            const int p0 = pix[-1*xstride];
-            const int p1 = pix[-2*xstride];
-            const int p2 = pix[-3*xstride];
-            const int q0 = pix[0];
-            const int q1 = pix[1*xstride];
-            const int q2 = pix[2*xstride];
-
-            if( FFABS( p0 - q0 ) < alpha &&
-                FFABS( p1 - p0 ) < beta &&
-                FFABS( q1 - q0 ) < beta ) {
-
-                int tc = tc0[i];
-                int i_delta;
-
-                if( FFABS( p2 - p0 ) < beta ) {
-                    if(tc0[i])
-                    pix[-2*xstride] = p1 + av_clip( (( p2 + ( ( p0 + q0 + 1 ) >> 1 ) ) >> 1) - p1, -tc0[i], tc0[i] );
-                    tc++;
-                }
-                if( FFABS( q2 - q0 ) < beta ) {
-                    if(tc0[i])
-                    pix[   xstride] = q1 + av_clip( (( q2 + ( ( p0 + q0 + 1 ) >> 1 ) ) >> 1) - q1, -tc0[i], tc0[i] );
-                    tc++;
-                }
-
-                i_delta = av_clip( (((q0 - p0 ) << 2) + (p1 - q1) + 4) >> 3, -tc, tc );
-                pix[-xstride] = av_clip_uint8( p0 + i_delta );    /* p0' */
-                pix[0]        = av_clip_uint8( q0 - i_delta );    /* q0' */
-            }
-            pix += ystride;
-        }
-    }
-}
-static void h264_v_loop_filter_luma_c(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0)
-{
-    h264_loop_filter_luma_c(pix, stride, 1, 4, alpha, beta, tc0);
-}
-static void h264_h_loop_filter_luma_c(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0)
-{
-    h264_loop_filter_luma_c(pix, 1, stride, 4, alpha, beta, tc0);
-}
-static void h264_h_loop_filter_luma_mbaff_c(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0)
-{
-    h264_loop_filter_luma_c(pix, 1, stride, 2, alpha, beta, tc0);
-}
-
-static av_always_inline av_flatten void h264_loop_filter_luma_intra_c(uint8_t *pix, int xstride, int ystride, int inner_iters, int alpha, int beta)
-{
-    int d;
-    for( d = 0; d < 4 * inner_iters; d++ ) {
-        const int p2 = pix[-3*xstride];
-        const int p1 = pix[-2*xstride];
-        const int p0 = pix[-1*xstride];
-
-        const int q0 = pix[ 0*xstride];
-        const int q1 = pix[ 1*xstride];
-        const int q2 = pix[ 2*xstride];
-
-        if( FFABS( p0 - q0 ) < alpha &&
-            FFABS( p1 - p0 ) < beta &&
-            FFABS( q1 - q0 ) < beta ) {
-
-            if(FFABS( p0 - q0 ) < (( alpha >> 2 ) + 2 )){
-                if( FFABS( p2 - p0 ) < beta)
-                {
-                    const int p3 = pix[-4*xstride];
-                    /* p0', p1', p2' */
-                    pix[-1*xstride] = ( p2 + 2*p1 + 2*p0 + 2*q0 + q1 + 4 ) >> 3;
-                    pix[-2*xstride] = ( p2 + p1 + p0 + q0 + 2 ) >> 2;
-                    pix[-3*xstride] = ( 2*p3 + 3*p2 + p1 + p0 + q0 + 4 ) >> 3;
-                } else {
-                    /* p0' */
-                    pix[-1*xstride] = ( 2*p1 + p0 + q1 + 2 ) >> 2;
-                }
-                if( FFABS( q2 - q0 ) < beta)
-                {
-                    const int q3 = pix[3*xstride];
-                    /* q0', q1', q2' */
-                    pix[0*xstride] = ( p1 + 2*p0 + 2*q0 + 2*q1 + q2 + 4 ) >> 3;
-                    pix[1*xstride] = ( p0 + q0 + q1 + q2 + 2 ) >> 2;
-                    pix[2*xstride] = ( 2*q3 + 3*q2 + q1 + q0 + p0 + 4 ) >> 3;
-                } else {
-                    /* q0' */
-                    pix[0*xstride] = ( 2*q1 + q0 + p1 + 2 ) >> 2;
-                }
-            }else{
-                /* p0', q0' */
-                pix[-1*xstride] = ( 2*p1 + p0 + q1 + 2 ) >> 2;
-                pix[ 0*xstride] = ( 2*q1 + q0 + p1 + 2 ) >> 2;
-            }
-        }
-        pix += ystride;
-    }
-}
-static void h264_v_loop_filter_luma_intra_c(uint8_t *pix, int stride, int alpha, int beta)
-{
-    h264_loop_filter_luma_intra_c(pix, stride, 1, 4, alpha, beta);
-}
-static void h264_h_loop_filter_luma_intra_c(uint8_t *pix, int stride, int alpha, int beta)
-{
-    h264_loop_filter_luma_intra_c(pix, 1, stride, 4, alpha, beta);
-}
-static void h264_h_loop_filter_luma_mbaff_intra_c(uint8_t *pix, int stride, int alpha, int beta)
-{
-    h264_loop_filter_luma_intra_c(pix, 1, stride, 2, alpha, beta);
-}
-
-static av_always_inline av_flatten void h264_loop_filter_chroma_c(uint8_t *pix, int xstride, int ystride, int inner_iters, int alpha, int beta, int8_t *tc0)
-{
-    int i, d;
-    for( i = 0; i < 4; i++ ) {
-        const int tc = tc0[i];
-        if( tc <= 0 ) {
-            pix += inner_iters*ystride;
-            continue;
-        }
-        for( d = 0; d < inner_iters; d++ ) {
-            const int p0 = pix[-1*xstride];
-            const int p1 = pix[-2*xstride];
-            const int q0 = pix[0];
-            const int q1 = pix[1*xstride];
-
-            if( FFABS( p0 - q0 ) < alpha &&
-                FFABS( p1 - p0 ) < beta &&
-                FFABS( q1 - q0 ) < beta ) {
-
-                int delta = av_clip( (((q0 - p0 ) << 2) + (p1 - q1) + 4) >> 3, -tc, tc );
-
-                pix[-xstride] = av_clip_uint8( p0 + delta );    /* p0' */
-                pix[0]        = av_clip_uint8( q0 - delta );    /* q0' */
-            }
-            pix += ystride;
-        }
-    }
-}
-static void h264_v_loop_filter_chroma_c(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0)
-{
-    h264_loop_filter_chroma_c(pix, stride, 1, 2, alpha, beta, tc0);
-}
-static void h264_h_loop_filter_chroma_c(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0)
-{
-    h264_loop_filter_chroma_c(pix, 1, stride, 2, alpha, beta, tc0);
-}
-static void h264_h_loop_filter_chroma_mbaff_c(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0)
-{
-    h264_loop_filter_chroma_c(pix, 1, stride, 1, alpha, beta, tc0);
-}
-
-static av_always_inline av_flatten void h264_loop_filter_chroma_intra_c(uint8_t *pix, int xstride, int ystride, int inner_iters, int alpha, int beta)
-{
-    int d;
-    for( d = 0; d < 4 * inner_iters; d++ ) {
-        const int p0 = pix[-1*xstride];
-        const int p1 = pix[-2*xstride];
-        const int q0 = pix[0];
-        const int q1 = pix[1*xstride];
-
-        if( FFABS( p0 - q0 ) < alpha &&
-            FFABS( p1 - p0 ) < beta &&
-            FFABS( q1 - q0 ) < beta ) {
-
-            pix[-xstride] = ( 2*p1 + p0 + q1 + 2 ) >> 2;   /* p0' */
-            pix[0]        = ( 2*q1 + q0 + p1 + 2 ) >> 2;   /* q0' */
-        }
-        pix += ystride;
-    }
-}
-static void h264_v_loop_filter_chroma_intra_c(uint8_t *pix, int stride, int alpha, int beta)
-{
-    h264_loop_filter_chroma_intra_c(pix, stride, 1, 2, alpha, beta);
-}
-static void h264_h_loop_filter_chroma_intra_c(uint8_t *pix, int stride, int alpha, int beta)
-{
-    h264_loop_filter_chroma_intra_c(pix, 1, stride, 2, alpha, beta);
-}
-static void h264_h_loop_filter_chroma_mbaff_intra_c(uint8_t *pix, int stride, int alpha, int beta)
-{
-    h264_loop_filter_chroma_intra_c(pix, 1, stride, 1, alpha, beta);
-}
+#include "h264dsp_template.c"
 
 void ff_h264dsp_init(H264DSPContext *c)
 {