From 2692ceab8bcc0e7787ce707e9040b4d3ca79723e Mon Sep 17 00:00:00 2001
From: Michael Niedermayer <michaelni@gmx.at>
Date: Sat, 7 Jan 2006 18:34:19 +0000
Subject: [PATCH] use h264 MC functions for rectangular blocks too

Originally committed as revision 4821 to svn://svn.ffmpeg.org/ffmpeg/trunk
---
 libavcodec/snow.c             | 42 ++++++++++++++++++++++++++++++-----
 tests/ffmpeg.regression.ref   | 12 +++++-----
 tests/rotozoom.regression.ref | 12 +++++-----
 3 files changed, 48 insertions(+), 18 deletions(-)

diff --git a/libavcodec/snow.c b/libavcodec/snow.c
index aa569c28ce..0aa56dba61 100644
--- a/libavcodec/snow.c
+++ b/libavcodec/snow.c
@@ -2469,10 +2469,29 @@ mca( 8, 8,8)
 static void pred_block(SnowContext *s, uint8_t *dst, uint8_t *src, uint8_t *tmp, int stride, int sx, int sy, int b_w, int b_h, BlockNode *block, int plane_index, int w, int h){
     if(block->type & BLOCK_INTRA){
         int x, y;
-        const int color= block->color[plane_index];
-        for(y=0; y < b_h; y++){
-            for(x=0; x < b_w; x++){
-                dst[x + y*stride]= color;
+        const int color = block->color[plane_index];
+        const int color4= color*0x01010101;
+        if(b_w==16){
+            for(y=0; y < b_h; y++){
+                *(uint32_t*)&dst[0 + y*stride]= color4;
+                *(uint32_t*)&dst[4 + y*stride]= color4;
+                *(uint32_t*)&dst[8 + y*stride]= color4;
+                *(uint32_t*)&dst[12+ y*stride]= color4;
+            }
+        }else if(b_w==8){
+            for(y=0; y < b_h; y++){
+                *(uint32_t*)&dst[0 + y*stride]= color4;
+                *(uint32_t*)&dst[4 + y*stride]= color4;
+            }
+        }else if(b_w==4){
+            for(y=0; y < b_h; y++){
+                *(uint32_t*)&dst[0 + y*stride]= color4;
+            }
+        }else{
+            for(y=0; y < b_h; y++){
+                for(x=0; x < b_w; x++){
+                    dst[x + y*stride]= color;
+                }
             }
         }
     }else{
@@ -2489,10 +2508,21 @@ static void pred_block(SnowContext *s, uint8_t *dst, uint8_t *src, uint8_t *tmp,
             ff_emulated_edge_mc(tmp + MB_SIZE, src, stride, b_w+5, b_h+5, sx, sy, w, h);
             src= tmp + MB_SIZE;
         }
-        if((dx&3) || (dy&3) || b_w!=b_h || (b_w!=4 && b_w!=8 && b_w!=16))
+        assert(b_w == b_h || 2*b_w == b_h || b_w == 2*b_h);
+        assert(!(b_w&(b_w-1)));
+        assert(b_w>1 && b_h>1);
+        if((dx&3) || (dy&3) || b_w==2 || b_h==2)
             mc_block(dst, src, tmp, stride, b_w, b_h, dx, dy);
-        else
+        else if(b_w==b_h)
             s->dsp.put_h264_qpel_pixels_tab[2-(b_w>>3)][dy+(dx>>2)](dst,src + 2 + 2*stride,stride);
+        else if(b_w==2*b_h){
+            s->dsp.put_h264_qpel_pixels_tab[2-(b_h>>3)][dy+(dx>>2)](dst    ,src + 2       + 2*stride,stride);
+            s->dsp.put_h264_qpel_pixels_tab[2-(b_h>>3)][dy+(dx>>2)](dst+b_h,src + 2 + b_h + 2*stride,stride);
+        }else{
+            assert(2*b_w==b_h);
+            s->dsp.put_h264_qpel_pixels_tab[2-(b_w>>3)][dy+(dx>>2)](dst           ,src + 2 + 2*stride           ,stride);
+            s->dsp.put_h264_qpel_pixels_tab[2-(b_w>>3)][dy+(dx>>2)](dst+b_w*stride,src + 2 + 2*stride+b_w*stride,stride);
+        }
     }
 }
 
diff --git a/tests/ffmpeg.regression.ref b/tests/ffmpeg.regression.ref
index db8b1f37dd..c56846fa65 100644
--- a/tests/ffmpeg.regression.ref
+++ b/tests/ffmpeg.regression.ref
@@ -119,12 +119,12 @@ a7ef4746f27be309138c188e327d3ebe *./data/a-ffv1.avi
 2653642 ./data/a-ffv1.avi
 799d3db687f6cdd7a837ec156efc171f *./data/out.yuv
 stddev:  0.00 PSNR:99.99 bytes:7602176
-1b5f58ef86909635fa4a582eeb275118 *./data/a-snow.avi
-1198432 ./data/a-snow.avi
-0b96ab5415cf2d1f486a75810d033e2b *./data/out.yuv
-stddev:  2.90 PSNR:38.87 bytes:7602176
-de8eac1454746a37eae0b1ff88c0f849 *./data/a-snow53.avi
-3535414 ./data/a-snow53.avi
+14fa515bf25a47fc2ad3d18b726e8e31 *./data/a-snow.avi
+1197138 ./data/a-snow.avi
+e7c746171b092266b0cf55bb5de2a40a *./data/out.yuv
+stddev:  2.89 PSNR:38.87 bytes:7602176
+94bedf8929178a8202ae3b5dbcdb84dd *./data/a-snow53.avi
+3533696 ./data/a-snow53.avi
 799d3db687f6cdd7a837ec156efc171f *./data/out.yuv
 stddev:  0.00 PSNR:99.99 bytes:7602176
 e1da20e3f52f4d2aa18e9486986161fc *./data/a-dv.dv
diff --git a/tests/rotozoom.regression.ref b/tests/rotozoom.regression.ref
index 7f168f1516..4eda12d065 100644
--- a/tests/rotozoom.regression.ref
+++ b/tests/rotozoom.regression.ref
@@ -119,12 +119,12 @@ d0831a8339491fd680b650f05262e5d9 *./data/a-ffv1.avi
 3524768 ./data/a-ffv1.avi
 dde5895817ad9d219f79a52d0bdfb001 *./data/out.yuv
 stddev:  0.00 PSNR:99.99 bytes:7602176
-f5dd9e5779f7fdc0fd9f67f44fd3979b *./data/a-snow.avi
-286706 ./data/a-snow.avi
-1b59f5ede351d67fc61c75a34b13ccb5 *./data/out.yuv
-stddev:  2.32 PSNR:40.79 bytes:7602176
-d0a8bdb25d7ed641eab0debcac396eb7 *./data/a-snow53.avi
-2725842 ./data/a-snow53.avi
+b926518ac399c7af0f218a7115315b4f *./data/a-snow.avi
+286800 ./data/a-snow.avi
+6c59db71d950610f854d05e2cef18609 *./data/out.yuv
+stddev:  2.32 PSNR:40.80 bytes:7602176
+4d2bcc832e318fad3c25614e31daa6fe *./data/a-snow53.avi
+2725630 ./data/a-snow53.avi
 dde5895817ad9d219f79a52d0bdfb001 *./data/out.yuv
 stddev:  0.00 PSNR:99.99 bytes:7602176
 a553532dcd54c1c421b52c3b6fece6ef *./data/a-dv.dv