[FFmpeg-cvslog] mpegvideo: implement ff_put_h264_chroma_mc1 & ff_avg_h264_chroma_mc2

[Michael Niedermayer]

ffmpeg | branch: master | Michael Niedermayer <michaelni at gmx.at> | Wed May 29 19:41:55 2013 +0200| [c2625c26c5e58edfa39360b51125f1ddd593e4db] | committer: Michael Niedermayer

mpegvideo: implement ff_put_h264_chroma_mc1 & ff_avg_h264_chroma_mc2

These are needed for lowres 3

This may fix Ticket2538

Signed-off-by: Michael Niedermayer <michaelni at gmx.at>

> http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=c2625c26c5e58edfa39360b51125f1ddd593e4db
---

 libavcodec/h264chroma.c          |    2 ++
 libavcodec/h264chroma.h          |    4 ++--
 libavcodec/h264chroma_template.c |   28 ++++++++++++++++++++++++++++
 libavcodec/mpegvideo.c           |    2 +-
 4 files changed, 33 insertions(+), 3 deletions(-)

diff --git a/libavcodec/h264chroma.c b/libavcodec/h264chroma.c
index fcc4807..5f8ed91 100644
--- a/libavcodec/h264chroma.c
+++ b/libavcodec/h264chroma.c
@@ -32,9 +32,11 @@
     c->put_h264_chroma_pixels_tab[0] = put_h264_chroma_mc8_ ## depth ## _c; \
     c->put_h264_chroma_pixels_tab[1] = put_h264_chroma_mc4_ ## depth ## _c; \
     c->put_h264_chroma_pixels_tab[2] = put_h264_chroma_mc2_ ## depth ## _c; \
+    c->put_h264_chroma_pixels_tab[3] = put_h264_chroma_mc1_ ## depth ## _c; \
     c->avg_h264_chroma_pixels_tab[0] = avg_h264_chroma_mc8_ ## depth ## _c; \
     c->avg_h264_chroma_pixels_tab[1] = avg_h264_chroma_mc4_ ## depth ## _c; \
     c->avg_h264_chroma_pixels_tab[2] = avg_h264_chroma_mc2_ ## depth ## _c; \
+    c->avg_h264_chroma_pixels_tab[3] = avg_h264_chroma_mc1_ ## depth ## _c; \
 
 av_cold void ff_h264chroma_init(H264ChromaContext *c, int bit_depth)
 {
diff --git a/libavcodec/h264chroma.h b/libavcodec/h264chroma.h
index 46fae42..45bbc7d 100644
--- a/libavcodec/h264chroma.h
+++ b/libavcodec/h264chroma.h
@@ -24,8 +24,8 @@
 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 struct H264ChromaContext {
-    h264_chroma_mc_func put_h264_chroma_pixels_tab[3];
-    h264_chroma_mc_func avg_h264_chroma_pixels_tab[3];
+    h264_chroma_mc_func put_h264_chroma_pixels_tab[4];
+    h264_chroma_mc_func avg_h264_chroma_pixels_tab[4];
 } H264ChromaContext;
 
 void ff_h264chroma_init(H264ChromaContext *c, int bit_depth);
diff --git a/libavcodec/h264chroma_template.c b/libavcodec/h264chroma_template.c
index 93559d7..b64172a 100644
--- a/libavcodec/h264chroma_template.c
+++ b/libavcodec/h264chroma_template.c
@@ -24,6 +24,34 @@
 #include "bit_depth_template.c"
 
 #define H264_CHROMA_MC(OPNAME, OP)\
+static void FUNCC(OPNAME ## h264_chroma_mc1)(uint8_t *_dst/*align 8*/, uint8_t *_src/*align 1*/, int stride, int h, int x, int y){\
+    pixel *dst = (pixel*)_dst;\
+    pixel *src = (pixel*)_src;\
+    const int A=(8-x)*(8-y);\
+    const int B=(  x)*(8-y);\
+    const int C=(8-x)*(  y);\
+    const int D=(  x)*(  y);\
+    int i;\
+    stride >>= sizeof(pixel)-1;\
+    \
+    av_assert2(x<8 && y<8 && x>=0 && y>=0);\
+\
+    if(D){\
+        for(i=0; i<h; i++){\
+            OP(dst[0], (A*src[0] + B*src[1] + C*src[stride+0] + D*src[stride+1]));\
+            dst+= stride;\
+            src+= stride;\
+        }\
+    }else{\
+        const int E= B+C;\
+        const int step= C ? stride : 1;\
+        for(i=0; i<h; i++){\
+            OP(dst[0], (A*src[0] + E*src[step+0]));\
+            dst+= stride;\
+            src+= stride;\
+        }\
+    }\
+}\
 static void FUNCC(OPNAME ## h264_chroma_mc2)(uint8_t *_dst/*align 8*/, uint8_t *_src/*align 1*/, int stride, int h, int x, int y){\
     pixel *dst = (pixel*)_dst;\
     pixel *src = (pixel*)_src;\
diff --git a/libavcodec/mpegvideo.c b/libavcodec/mpegvideo.c
index e76fc34..8c9553a 100644
--- a/libavcodec/mpegvideo.c
+++ b/libavcodec/mpegvideo.c
@@ -2179,7 +2179,7 @@ static inline int hpel_motion_lowres(MpegEncContext *s,
                                      int motion_x, int motion_y)
 {
     const int lowres   = s->avctx->lowres;
-    const int op_index = FFMIN(lowres, 2);
+    const int op_index = FFMIN(lowres, 3);
     const int s_mask   = (2 << lowres) - 1;
     int emu = 0;
     int sx, sy;