X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=libavcodec%2Fdsputil.c;h=453b38ab2aae701e651d02557e92c159a376ff2b;hb=5d2f52b98980fd49c3d1f35a742912f7952262b0;hp=e5e05921dbd925a303761749512c9236a80f78d7;hpb=0fd6aea1f1cc1ceaf2da578ce22d53427ddc4047;p=ffmpeg

diff --git a/libavcodec/dsputil.c b/libavcodec/dsputil.c
index e5e05921dbd..453b38ab2aa 100644
--- a/libavcodec/dsputil.c
+++ b/libavcodec/dsputil.c
@@ -15,21 +15,25 @@
  *
  * You should have received a copy of the GNU Lesser General Public
  * License along with this library; if not, write to the Free Software
- * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  *
  * gmc & q-pel & 32/64 bit based MC by Michael Niedermayer <michaelni@gmx.at>
  */
- 
+
 /**
  * @file dsputil.c
  * DSP utils
  */
- 
+
 #include "avcodec.h"
 #include "dsputil.h"
 #include "mpegvideo.h"
 #include "simple_idct.h"
 #include "faandct.h"
+#include "snow.h"
+
+/* snow.c */
+void ff_spatial_dwt(int *buffer, int width, int height, int stride, int type, int decomposition_count);
 
 uint8_t cropTbl[256 + 2 * MAX_NEG_CROP] = {0, };
 uint32_t squareTbl[512] = {0, };
@@ -59,76 +63,76 @@ const uint8_t ff_zigzag248_direct[64] = {
 };
 
 /* not permutated inverse zigzag_direct + 1 for MMX quantizer */
-uint16_t __align8 inv_zigzag_direct16[64] = {0, };
+DECLARE_ALIGNED_8(uint16_t, inv_zigzag_direct16[64]) = {0, };
 
 const uint8_t ff_alternate_horizontal_scan[64] = {
-    0,  1,   2,  3,  8,  9, 16, 17, 
+    0,  1,   2,  3,  8,  9, 16, 17,
     10, 11,  4,  5,  6,  7, 15, 14,
-    13, 12, 19, 18, 24, 25, 32, 33, 
+    13, 12, 19, 18, 24, 25, 32, 33,
     26, 27, 20, 21, 22, 23, 28, 29,
-    30, 31, 34, 35, 40, 41, 48, 49, 
+    30, 31, 34, 35, 40, 41, 48, 49,
     42, 43, 36, 37, 38, 39, 44, 45,
-    46, 47, 50, 51, 56, 57, 58, 59, 
+    46, 47, 50, 51, 56, 57, 58, 59,
     52, 53, 54, 55, 60, 61, 62, 63,
 };
 
 const uint8_t ff_alternate_vertical_scan[64] = {
-    0,  8,  16, 24,  1,  9,  2, 10, 
+    0,  8,  16, 24,  1,  9,  2, 10,
     17, 25, 32, 40, 48, 56, 57, 49,
-    41, 33, 26, 18,  3, 11,  4, 12, 
+    41, 33, 26, 18,  3, 11,  4, 12,
     19, 27, 34, 42, 50, 58, 35, 43,
-    51, 59, 20, 28,  5, 13,  6, 14, 
+    51, 59, 20, 28,  5, 13,  6, 14,
     21, 29, 36, 44, 52, 60, 37, 45,
-    53, 61, 22, 30,  7, 15, 23, 31, 
+    53, 61, 22, 30,  7, 15, 23, 31,
     38, 46, 54, 62, 39, 47, 55, 63,
 };
 
 /* a*inverse[b]>>32 == a/b for all 0<=a<=65536 && 2<=b<=255 */
 const uint32_t inverse[256]={
-         0, 4294967295U,2147483648U,1431655766, 1073741824,  858993460,  715827883,  613566757, 
- 536870912,  477218589,  429496730,  390451573,  357913942,  330382100,  306783379,  286331154, 
- 268435456,  252645136,  238609295,  226050911,  214748365,  204522253,  195225787,  186737709, 
- 178956971,  171798692,  165191050,  159072863,  153391690,  148102321,  143165577,  138547333, 
- 134217728,  130150525,  126322568,  122713352,  119304648,  116080198,  113025456,  110127367, 
- 107374183,  104755300,  102261127,   99882961,   97612894,   95443718,   93368855,   91382283, 
-  89478486,   87652394,   85899346,   84215046,   82595525,   81037119,   79536432,   78090315, 
-  76695845,   75350304,   74051161,   72796056,   71582789,   70409300,   69273667,   68174085, 
-  67108864,   66076420,   65075263,   64103990,   63161284,   62245903,   61356676,   60492498, 
-  59652324,   58835169,   58040099,   57266231,   56512728,   55778797,   55063684,   54366675, 
-  53687092,   53024288,   52377650,   51746594,   51130564,   50529028,   49941481,   49367441, 
-  48806447,   48258060,   47721859,   47197443,   46684428,   46182445,   45691142,   45210183, 
-  44739243,   44278014,   43826197,   43383509,   42949673,   42524429,   42107523,   41698712, 
-  41297763,   40904451,   40518560,   40139882,   39768216,   39403370,   39045158,   38693400, 
-  38347923,   38008561,   37675152,   37347542,   37025581,   36709123,   36398028,   36092163, 
-  35791395,   35495598,   35204650,   34918434,   34636834,   34359739,   34087043,   33818641, 
-  33554432,   33294321,   33038210,   32786010,   32537632,   32292988,   32051995,   31814573, 
-  31580642,   31350127,   31122952,   30899046,   30678338,   30460761,   30246249,   30034737, 
-  29826162,   29620465,   29417585,   29217465,   29020050,   28825284,   28633116,   28443493, 
-  28256364,   28071682,   27889399,   27709467,   27531842,   27356480,   27183338,   27012373, 
-  26843546,   26676816,   26512144,   26349493,   26188825,   26030105,   25873297,   25718368, 
-  25565282,   25414008,   25264514,   25116768,   24970741,   24826401,   24683721,   24542671, 
-  24403224,   24265352,   24129030,   23994231,   23860930,   23729102,   23598722,   23469767, 
-  23342214,   23216040,   23091223,   22967740,   22845571,   22724695,   22605092,   22486740, 
-  22369622,   22253717,   22139007,   22025474,   21913099,   21801865,   21691755,   21582751, 
-  21474837,   21367997,   21262215,   21157475,   21053762,   20951060,   20849356,   20748635, 
-  20648882,   20550083,   20452226,   20355296,   20259280,   20164166,   20069941,   19976593, 
-  19884108,   19792477,   19701685,   19611723,   19522579,   19434242,   19346700,   19259944, 
-  19173962,   19088744,   19004281,   18920561,   18837576,   18755316,   18673771,   18592933, 
-  18512791,   18433337,   18354562,   18276457,   18199014,   18122225,   18046082,   17970575, 
-  17895698,   17821442,   17747799,   17674763,   17602325,   17530479,   17459217,   17388532, 
+         0, 4294967295U,2147483648U,1431655766, 1073741824,  858993460,  715827883,  613566757,
+ 536870912,  477218589,  429496730,  390451573,  357913942,  330382100,  306783379,  286331154,
+ 268435456,  252645136,  238609295,  226050911,  214748365,  204522253,  195225787,  186737709,
+ 178956971,  171798692,  165191050,  159072863,  153391690,  148102321,  143165577,  138547333,
+ 134217728,  130150525,  126322568,  122713352,  119304648,  116080198,  113025456,  110127367,
+ 107374183,  104755300,  102261127,   99882961,   97612894,   95443718,   93368855,   91382283,
+  89478486,   87652394,   85899346,   84215046,   82595525,   81037119,   79536432,   78090315,
+  76695845,   75350304,   74051161,   72796056,   71582789,   70409300,   69273667,   68174085,
+  67108864,   66076420,   65075263,   64103990,   63161284,   62245903,   61356676,   60492498,
+  59652324,   58835169,   58040099,   57266231,   56512728,   55778797,   55063684,   54366675,
+  53687092,   53024288,   52377650,   51746594,   51130564,   50529028,   49941481,   49367441,
+  48806447,   48258060,   47721859,   47197443,   46684428,   46182445,   45691142,   45210183,
+  44739243,   44278014,   43826197,   43383509,   42949673,   42524429,   42107523,   41698712,
+  41297763,   40904451,   40518560,   40139882,   39768216,   39403370,   39045158,   38693400,
+  38347923,   38008561,   37675152,   37347542,   37025581,   36709123,   36398028,   36092163,
+  35791395,   35495598,   35204650,   34918434,   34636834,   34359739,   34087043,   33818641,
+  33554432,   33294321,   33038210,   32786010,   32537632,   32292988,   32051995,   31814573,
+  31580642,   31350127,   31122952,   30899046,   30678338,   30460761,   30246249,   30034737,
+  29826162,   29620465,   29417585,   29217465,   29020050,   28825284,   28633116,   28443493,
+  28256364,   28071682,   27889399,   27709467,   27531842,   27356480,   27183338,   27012373,
+  26843546,   26676816,   26512144,   26349493,   26188825,   26030105,   25873297,   25718368,
+  25565282,   25414008,   25264514,   25116768,   24970741,   24826401,   24683721,   24542671,
+  24403224,   24265352,   24129030,   23994231,   23860930,   23729102,   23598722,   23469767,
+  23342214,   23216040,   23091223,   22967740,   22845571,   22724695,   22605092,   22486740,
+  22369622,   22253717,   22139007,   22025474,   21913099,   21801865,   21691755,   21582751,
+  21474837,   21367997,   21262215,   21157475,   21053762,   20951060,   20849356,   20748635,
+  20648882,   20550083,   20452226,   20355296,   20259280,   20164166,   20069941,   19976593,
+  19884108,   19792477,   19701685,   19611723,   19522579,   19434242,   19346700,   19259944,
+  19173962,   19088744,   19004281,   18920561,   18837576,   18755316,   18673771,   18592933,
+  18512791,   18433337,   18354562,   18276457,   18199014,   18122225,   18046082,   17970575,
+  17895698,   17821442,   17747799,   17674763,   17602325,   17530479,   17459217,   17388532,
   17318417,   17248865,   17179870,   17111424,   17043522,   16976156,   16909321,   16843010,
 };
 
 /* Input permutation for the simple_idct_mmx */
 static const uint8_t simple_mmx_permutation[64]={
-	0x00, 0x08, 0x04, 0x09, 0x01, 0x0C, 0x05, 0x0D, 
-	0x10, 0x18, 0x14, 0x19, 0x11, 0x1C, 0x15, 0x1D, 
-	0x20, 0x28, 0x24, 0x29, 0x21, 0x2C, 0x25, 0x2D, 
-	0x12, 0x1A, 0x16, 0x1B, 0x13, 0x1E, 0x17, 0x1F, 
-	0x02, 0x0A, 0x06, 0x0B, 0x03, 0x0E, 0x07, 0x0F, 
-	0x30, 0x38, 0x34, 0x39, 0x31, 0x3C, 0x35, 0x3D, 
-	0x22, 0x2A, 0x26, 0x2B, 0x23, 0x2E, 0x27, 0x2F, 
-	0x32, 0x3A, 0x36, 0x3B, 0x33, 0x3E, 0x37, 0x3F,
+        0x00, 0x08, 0x04, 0x09, 0x01, 0x0C, 0x05, 0x0D,
+        0x10, 0x18, 0x14, 0x19, 0x11, 0x1C, 0x15, 0x1D,
+        0x20, 0x28, 0x24, 0x29, 0x21, 0x2C, 0x25, 0x2D,
+        0x12, 0x1A, 0x16, 0x1B, 0x13, 0x1E, 0x17, 0x1F,
+        0x02, 0x0A, 0x06, 0x0B, 0x03, 0x0E, 0x07, 0x0F,
+        0x30, 0x38, 0x34, 0x39, 0x31, 0x3C, 0x35, 0x3D,
+        0x22, 0x2A, 0x26, 0x2B, 0x23, 0x2E, 0x27, 0x2F,
+        0x32, 0x3A, 0x36, 0x3B, 0x33, 0x3E, 0x37, 0x3F,
 };
 
 static int pix_sum_c(uint8_t * pix, int line_size)
@@ -137,18 +141,18 @@ static int pix_sum_c(uint8_t * pix, int line_size)
 
     s = 0;
     for (i = 0; i < 16; i++) {
-	for (j = 0; j < 16; j += 8) {
-	    s += pix[0];
-	    s += pix[1];
-	    s += pix[2];
-	    s += pix[3];
-	    s += pix[4];
-	    s += pix[5];
-	    s += pix[6];
-	    s += pix[7];
-	    pix += 8;
-	}
-	pix += line_size - 16;
+        for (j = 0; j < 16; j += 8) {
+            s += pix[0];
+            s += pix[1];
+            s += pix[2];
+            s += pix[3];
+            s += pix[4];
+            s += pix[5];
+            s += pix[6];
+            s += pix[7];
+            pix += 8;
+        }
+        pix += line_size - 16;
     }
     return s;
 }
@@ -160,33 +164,33 @@ static int pix_norm1_c(uint8_t * pix, int line_size)
 
     s = 0;
     for (i = 0; i < 16; i++) {
-	for (j = 0; j < 16; j += 8) {
+        for (j = 0; j < 16; j += 8) {
 #if 0
-	    s += sq[pix[0]];
-	    s += sq[pix[1]];
-	    s += sq[pix[2]];
-	    s += sq[pix[3]];
-	    s += sq[pix[4]];
-	    s += sq[pix[5]];
-	    s += sq[pix[6]];
-	    s += sq[pix[7]];
+            s += sq[pix[0]];
+            s += sq[pix[1]];
+            s += sq[pix[2]];
+            s += sq[pix[3]];
+            s += sq[pix[4]];
+            s += sq[pix[5]];
+            s += sq[pix[6]];
+            s += sq[pix[7]];
 #else
 #if LONG_MAX > 2147483647
-	    register uint64_t x=*(uint64_t*)pix;
-	    s += sq[x&0xff];
-	    s += sq[(x>>8)&0xff];
-	    s += sq[(x>>16)&0xff];
-	    s += sq[(x>>24)&0xff];
+            register uint64_t x=*(uint64_t*)pix;
+            s += sq[x&0xff];
+            s += sq[(x>>8)&0xff];
+            s += sq[(x>>16)&0xff];
+            s += sq[(x>>24)&0xff];
             s += sq[(x>>32)&0xff];
             s += sq[(x>>40)&0xff];
             s += sq[(x>>48)&0xff];
             s += sq[(x>>56)&0xff];
 #else
-	    register uint32_t x=*(uint32_t*)pix;
-	    s += sq[x&0xff];
-	    s += sq[(x>>8)&0xff];
-	    s += sq[(x>>16)&0xff];
-	    s += sq[(x>>24)&0xff];
+            register uint32_t x=*(uint32_t*)pix;
+            s += sq[x&0xff];
+            s += sq[(x>>8)&0xff];
+            s += sq[(x>>16)&0xff];
+            s += sq[(x>>24)&0xff];
             x=*(uint32_t*)(pix+4);
             s += sq[x&0xff];
             s += sq[(x>>8)&0xff];
@@ -194,16 +198,16 @@ static int pix_norm1_c(uint8_t * pix, int line_size)
             s += sq[(x>>24)&0xff];
 #endif
 #endif
-	    pix += 8;
-	}
-	pix += line_size - 16;
+            pix += 8;
+        }
+        pix += line_size - 16;
     }
     return s;
 }
 
 static void bswap_buf(uint32_t *dst, uint32_t *src, int w){
     int i;
-    
+
     for(i=0; i+8<=w; i+=8){
         dst[i+0]= bswap_32(src[i+0]);
         dst[i+1]= bswap_32(src[i+1]);
@@ -288,34 +292,34 @@ static int sse16_c(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
 }
 
 
+#ifdef CONFIG_SNOW_ENCODER //dwt is in snow.c
 static inline int w_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, int w, int h, int type){
     int s, i, j;
     const int dec_count= w==8 ? 3 : 4;
-    int tmp[16*16];
-#if 0
+    int tmp[32*32];
     int level, ori;
-    static const int scale[2][2][4][4]={ 
+    static const int scale[2][2][4][4]={
       {
         {
-            //8x8 dec=3
+            // 9/7 8x8 dec=3
             {268, 239, 239, 213},
             {  0, 224, 224, 152},
             {  0, 135, 135, 110},
         },{
-            //16x16 dec=4
+            // 9/7 16x16 or 32x32 dec=4
             {344, 310, 310, 280},
             {  0, 320, 320, 228},
             {  0, 175, 175, 136},
             {  0, 129, 129, 102},
         }
       },{
-        {//FIXME 5/3
-            //8x8 dec=3
+        {
+            // 5/3 8x8 dec=3
             {275, 245, 245, 218},
             {  0, 230, 230, 156},
             {  0, 138, 138, 113},
         },{
-            //16x16 dec=4
+            // 5/3 16x16 or 32x32 dec=4
             {352, 317, 317, 286},
             {  0, 328, 328, 233},
             {  0, 180, 180, 140},
@@ -323,29 +327,29 @@ static inline int w_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, in
         }
       }
     };
-#endif
 
     for (i = 0; i < h; i++) {
         for (j = 0; j < w; j+=4) {
-            tmp[16*i+j+0] = (pix1[j+0] - pix2[j+0])<<4;
-            tmp[16*i+j+1] = (pix1[j+1] - pix2[j+1])<<4;
-            tmp[16*i+j+2] = (pix1[j+2] - pix2[j+2])<<4;
-            tmp[16*i+j+3] = (pix1[j+3] - pix2[j+3])<<4;
+            tmp[32*i+j+0] = (pix1[j+0] - pix2[j+0])<<4;
+            tmp[32*i+j+1] = (pix1[j+1] - pix2[j+1])<<4;
+            tmp[32*i+j+2] = (pix1[j+2] - pix2[j+2])<<4;
+            tmp[32*i+j+3] = (pix1[j+3] - pix2[j+3])<<4;
         }
         pix1 += line_size;
         pix2 += line_size;
     }
-    ff_spatial_dwt(tmp, w, h, 16, type, dec_count);
+
+    ff_spatial_dwt(tmp, w, h, 32, type, dec_count);
 
     s=0;
-#if 0
+    assert(w==h);
     for(level=0; level<dec_count; level++){
         for(ori= level ? 1 : 0; ori<4; ori++){
-            int sx= (ori&1) ? 1<<level: 0;
-            int stride= 16<<(dec_count-level);
+            int size= w>>(dec_count-level);
+            int sx= (ori&1) ? size : 0;
+            int stride= 32<<(dec_count-level);
             int sy= (ori&2) ? stride>>1 : 0;
-            int size= 1<<level;
-            
+
             for(i=0; i<size; i++){
                 for(j=0; j<size; j++){
                     int v= tmp[sx + sy + i*stride + j] * scale[type][dec_count-3][level][ori];
@@ -354,18 +358,8 @@ static inline int w_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, in
             }
         }
     }
-#endif
-    for (i = 0; i < h; i++) {
-        for (j = 0; j < w; j+=4) {
-            s+= ABS(tmp[16*i+j+0]);
-            s+= ABS(tmp[16*i+j+1]);
-            s+= ABS(tmp[16*i+j+2]);
-            s+= ABS(tmp[16*i+j+3]);
-        }
-    }
-    assert(s>=0); 
-    
-    return s>>2;
+    assert(s>=0);
+    return s>>9;
 }
 
 static int w53_8_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, int h){
@@ -384,6 +378,15 @@ static int w97_16_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, int
     return w_c(v, pix1, pix2, line_size, 16, h, 0);
 }
 
+int w53_32_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, int h){
+    return w_c(v, pix1, pix2, line_size, 32, h, 1);
+}
+
+int w97_32_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, int h){
+    return w_c(v, pix1, pix2, line_size, 32, h, 0);
+}
+#endif
+
 static void get_pixels_c(DCTELEM *restrict block, const uint8_t *pixels, int line_size)
 {
     int i;
@@ -404,7 +407,7 @@ static void get_pixels_c(DCTELEM *restrict block, const uint8_t *pixels, int lin
 }
 
 static void diff_pixels_c(DCTELEM *restrict block, const uint8_t *s1,
-			  const uint8_t *s2, int stride){
+                          const uint8_t *s2, int stride){
     int i;
 
     /* read the pixels */
@@ -425,11 +428,11 @@ static void diff_pixels_c(DCTELEM *restrict block, const uint8_t *s1,
 
 
 static void put_pixels_clamped_c(const DCTELEM *block, uint8_t *restrict pixels,
-				 int line_size)
+                                 int line_size)
 {
     int i;
     uint8_t *cm = cropTbl + MAX_NEG_CROP;
-    
+
     /* read the pixels */
     for(i=0;i<8;i++) {
         pixels[0] = cm[block[0]];
@@ -447,11 +450,11 @@ static void put_pixels_clamped_c(const DCTELEM *block, uint8_t *restrict pixels,
 }
 
 static void put_pixels_clamped4_c(const DCTELEM *block, uint8_t *restrict pixels,
-				 int line_size)
+                                 int line_size)
 {
     int i;
     uint8_t *cm = cropTbl + MAX_NEG_CROP;
-    
+
     /* read the pixels */
     for(i=0;i<4;i++) {
         pixels[0] = cm[block[0]];
@@ -465,11 +468,11 @@ static void put_pixels_clamped4_c(const DCTELEM *block, uint8_t *restrict pixels
 }
 
 static void put_pixels_clamped2_c(const DCTELEM *block, uint8_t *restrict pixels,
-				 int line_size)
+                                 int line_size)
 {
     int i;
     uint8_t *cm = cropTbl + MAX_NEG_CROP;
-    
+
     /* read the pixels */
     for(i=0;i<2;i++) {
         pixels[0] = cm[block[0]];
@@ -480,7 +483,7 @@ static void put_pixels_clamped2_c(const DCTELEM *block, uint8_t *restrict pixels
     }
 }
 
-static void put_signed_pixels_clamped_c(const DCTELEM *block, 
+static void put_signed_pixels_clamped_c(const DCTELEM *block,
                                         uint8_t *restrict pixels,
                                         int line_size)
 {
@@ -506,7 +509,7 @@ static void add_pixels_clamped_c(const DCTELEM *block, uint8_t *restrict pixels,
 {
     int i;
     uint8_t *cm = cropTbl + MAX_NEG_CROP;
-    
+
     /* read the pixels */
     for(i=0;i<8;i++) {
         pixels[0] = cm[pixels[0] + block[0]];
@@ -527,7 +530,7 @@ static void add_pixels_clamped4_c(const DCTELEM *block, uint8_t *restrict pixels
 {
     int i;
     uint8_t *cm = cropTbl + MAX_NEG_CROP;
-    
+
     /* read the pixels */
     for(i=0;i<4;i++) {
         pixels[0] = cm[pixels[0] + block[0]];
@@ -544,7 +547,7 @@ static void add_pixels_clamped2_c(const DCTELEM *block, uint8_t *restrict pixels
 {
     int i;
     uint8_t *cm = cropTbl + MAX_NEG_CROP;
-    
+
     /* read the pixels */
     for(i=0;i<2;i++) {
         pixels[0] = cm[pixels[0] + block[0]];
@@ -553,6 +556,37 @@ static void add_pixels_clamped2_c(const DCTELEM *block, uint8_t *restrict pixels
         block += 8;
     }
 }
+
+static void add_pixels8_c(uint8_t *restrict pixels, DCTELEM *block, int line_size)
+{
+    int i;
+    for(i=0;i<8;i++) {
+        pixels[0] += block[0];
+        pixels[1] += block[1];
+        pixels[2] += block[2];
+        pixels[3] += block[3];
+        pixels[4] += block[4];
+        pixels[5] += block[5];
+        pixels[6] += block[6];
+        pixels[7] += block[7];
+        pixels += line_size;
+        block += 8;
+    }
+}
+
+static void add_pixels4_c(uint8_t *restrict pixels, DCTELEM *block, int line_size)
+{
+    int i;
+    for(i=0;i<4;i++) {
+        pixels[0] += block[0];
+        pixels[1] += block[1];
+        pixels[2] += block[2];
+        pixels[3] += block[3];
+        pixels += line_size;
+        block += 4;
+    }
+}
+
 #if 0
 
 #define PIXOP2(OPNAME, OP) \
@@ -1106,12 +1140,12 @@ static void gmc1_c(uint8_t *dst, uint8_t *src, int stride, int h, int x16, int y
     }
 }
 
-static void gmc_c(uint8_t *dst, uint8_t *src, int stride, int h, int ox, int oy, 
+void ff_gmc_c(uint8_t *dst, uint8_t *src, int stride, int h, int ox, int oy,
                   int dxx, int dxy, int dyx, int dyy, int shift, int r, int width, int height)
 {
     int y, vx, vy;
     const int s= 1<<shift;
-    
+
     width--;
     height--;
 
@@ -1129,7 +1163,7 @@ static void gmc_c(uint8_t *dst, uint8_t *src, int stride, int h, int ox, int oy,
             frac_y= src_y&(s-1);
             src_x>>=shift;
             src_y>>=shift;
-  
+
             if((unsigned)src_x < width){
                 if((unsigned)src_y < height){
                     index= src_x + src_y*stride;
@@ -1139,23 +1173,23 @@ static void gmc_c(uint8_t *dst, uint8_t *src, int stride, int h, int ox, int oy,
                                            + src[index+stride+1]*   frac_x )*   frac_y
                                         + r)>>(shift*2);
                 }else{
-                    index= src_x + clip(src_y, 0, height)*stride;                    
-                    dst[y*stride + x]= ( (  src[index         ]*(s-frac_x) 
+                    index= src_x + clip(src_y, 0, height)*stride;
+                    dst[y*stride + x]= ( (  src[index         ]*(s-frac_x)
                                           + src[index       +1]*   frac_x )*s
                                         + r)>>(shift*2);
                 }
             }else{
                 if((unsigned)src_y < height){
-                    index= clip(src_x, 0, width) + src_y*stride;                    
-                    dst[y*stride + x]= (  (  src[index         ]*(s-frac_y) 
+                    index= clip(src_x, 0, width) + src_y*stride;
+                    dst[y*stride + x]= (  (  src[index         ]*(s-frac_y)
                                            + src[index+stride  ]*   frac_y )*s
                                         + r)>>(shift*2);
                 }else{
-                    index= clip(src_x, 0, width) + clip(src_y, 0, height)*stride;                    
+                    index= clip(src_x, 0, width) + clip(src_y, 0, height)*stride;
                     dst[y*stride + x]=    src[index         ];
                 }
             }
-            
+
             vx+= dxx;
             vy+= dyx;
         }
@@ -1177,7 +1211,7 @@ static inline void put_tpel_pixels_mc10_c(uint8_t *dst, const uint8_t *src, int
     int i,j;
     for (i=0; i < height; i++) {
       for (j=0; j < width; j++) {
-	dst[j] = (683*(2*src[j] + src[j+1] + 1)) >> 11;
+        dst[j] = (683*(2*src[j] + src[j+1] + 1)) >> 11;
       }
       src += stride;
       dst += stride;
@@ -1188,29 +1222,29 @@ static inline void put_tpel_pixels_mc20_c(uint8_t *dst, const uint8_t *src, int
     int i,j;
     for (i=0; i < height; i++) {
       for (j=0; j < width; j++) {
-	dst[j] = (683*(src[j] + 2*src[j+1] + 1)) >> 11;
+        dst[j] = (683*(src[j] + 2*src[j+1] + 1)) >> 11;
       }
       src += stride;
       dst += stride;
     }
 }
-    
+
 static inline void put_tpel_pixels_mc01_c(uint8_t *dst, const uint8_t *src, int stride, int width, int height){
     int i,j;
     for (i=0; i < height; i++) {
       for (j=0; j < width; j++) {
-	dst[j] = (683*(2*src[j] + src[j+stride] + 1)) >> 11;
+        dst[j] = (683*(2*src[j] + src[j+stride] + 1)) >> 11;
       }
       src += stride;
       dst += stride;
     }
 }
-    
+
 static inline void put_tpel_pixels_mc11_c(uint8_t *dst, const uint8_t *src, int stride, int width, int height){
     int i,j;
     for (i=0; i < height; i++) {
       for (j=0; j < width; j++) {
-	dst[j] = (2731*(4*src[j] + 3*src[j+1] + 3*src[j+stride] + 2*src[j+stride+1] + 6)) >> 15;
+        dst[j] = (2731*(4*src[j] + 3*src[j+1] + 3*src[j+stride] + 2*src[j+stride+1] + 6)) >> 15;
       }
       src += stride;
       dst += stride;
@@ -1221,7 +1255,7 @@ static inline void put_tpel_pixels_mc12_c(uint8_t *dst, const uint8_t *src, int
     int i,j;
     for (i=0; i < height; i++) {
       for (j=0; j < width; j++) {
-	dst[j] = (2731*(3*src[j] + 2*src[j+1] + 4*src[j+stride] + 3*src[j+stride+1] + 6)) >> 15;
+        dst[j] = (2731*(3*src[j] + 2*src[j+1] + 4*src[j+stride] + 3*src[j+stride+1] + 6)) >> 15;
       }
       src += stride;
       dst += stride;
@@ -1232,7 +1266,7 @@ static inline void put_tpel_pixels_mc02_c(uint8_t *dst, const uint8_t *src, int
     int i,j;
     for (i=0; i < height; i++) {
       for (j=0; j < width; j++) {
-	dst[j] = (683*(src[j] + 2*src[j+stride] + 1)) >> 11;
+        dst[j] = (683*(src[j] + 2*src[j+stride] + 1)) >> 11;
       }
       src += stride;
       dst += stride;
@@ -1243,7 +1277,7 @@ static inline void put_tpel_pixels_mc21_c(uint8_t *dst, const uint8_t *src, int
     int i,j;
     for (i=0; i < height; i++) {
       for (j=0; j < width; j++) {
-	dst[j] = (2731*(3*src[j] + 4*src[j+1] + 2*src[j+stride] + 3*src[j+stride+1] + 6)) >> 15;
+        dst[j] = (2731*(3*src[j] + 4*src[j+1] + 2*src[j+stride] + 3*src[j+stride+1] + 6)) >> 15;
       }
       src += stride;
       dst += stride;
@@ -1254,7 +1288,7 @@ static inline void put_tpel_pixels_mc22_c(uint8_t *dst, const uint8_t *src, int
     int i,j;
     for (i=0; i < height; i++) {
       for (j=0; j < width; j++) {
-	dst[j] = (2731*(2*src[j] + 3*src[j+1] + 3*src[j+stride] + 4*src[j+stride+1] + 6)) >> 15;
+        dst[j] = (2731*(2*src[j] + 3*src[j+1] + 3*src[j+stride] + 4*src[j+stride+1] + 6)) >> 15;
       }
       src += stride;
       dst += stride;
@@ -1274,7 +1308,7 @@ static inline void avg_tpel_pixels_mc10_c(uint8_t *dst, const uint8_t *src, int
     int i,j;
     for (i=0; i < height; i++) {
       for (j=0; j < width; j++) {
-	dst[j] = (dst[j] + ((683*(2*src[j] + src[j+1] + 1)) >> 11) + 1) >> 1;
+        dst[j] = (dst[j] + ((683*(2*src[j] + src[j+1] + 1)) >> 11) + 1) >> 1;
       }
       src += stride;
       dst += stride;
@@ -1285,29 +1319,29 @@ static inline void avg_tpel_pixels_mc20_c(uint8_t *dst, const uint8_t *src, int
     int i,j;
     for (i=0; i < height; i++) {
       for (j=0; j < width; j++) {
-	dst[j] = (dst[j] + ((683*(src[j] + 2*src[j+1] + 1)) >> 11) + 1) >> 1;
+        dst[j] = (dst[j] + ((683*(src[j] + 2*src[j+1] + 1)) >> 11) + 1) >> 1;
       }
       src += stride;
       dst += stride;
     }
 }
-    
+
 static inline void avg_tpel_pixels_mc01_c(uint8_t *dst, const uint8_t *src, int stride, int width, int height){
     int i,j;
     for (i=0; i < height; i++) {
       for (j=0; j < width; j++) {
-	dst[j] = (dst[j] + ((683*(2*src[j] + src[j+stride] + 1)) >> 11) + 1) >> 1;
+        dst[j] = (dst[j] + ((683*(2*src[j] + src[j+stride] + 1)) >> 11) + 1) >> 1;
       }
       src += stride;
       dst += stride;
     }
 }
-    
+
 static inline void avg_tpel_pixels_mc11_c(uint8_t *dst, const uint8_t *src, int stride, int width, int height){
     int i,j;
     for (i=0; i < height; i++) {
       for (j=0; j < width; j++) {
-	dst[j] = (dst[j] + ((2731*(4*src[j] + 3*src[j+1] + 3*src[j+stride] + 2*src[j+stride+1] + 6)) >> 15) + 1) >> 1;
+        dst[j] = (dst[j] + ((2731*(4*src[j] + 3*src[j+1] + 3*src[j+stride] + 2*src[j+stride+1] + 6)) >> 15) + 1) >> 1;
       }
       src += stride;
       dst += stride;
@@ -1318,7 +1352,7 @@ static inline void avg_tpel_pixels_mc12_c(uint8_t *dst, const uint8_t *src, int
     int i,j;
     for (i=0; i < height; i++) {
       for (j=0; j < width; j++) {
-	dst[j] = (dst[j] + ((2731*(3*src[j] + 2*src[j+1] + 4*src[j+stride] + 3*src[j+stride+1] + 6)) >> 15) + 1) >> 1;
+        dst[j] = (dst[j] + ((2731*(3*src[j] + 2*src[j+1] + 4*src[j+stride] + 3*src[j+stride+1] + 6)) >> 15) + 1) >> 1;
       }
       src += stride;
       dst += stride;
@@ -1329,7 +1363,7 @@ static inline void avg_tpel_pixels_mc02_c(uint8_t *dst, const uint8_t *src, int
     int i,j;
     for (i=0; i < height; i++) {
       for (j=0; j < width; j++) {
-	dst[j] = (dst[j] + ((683*(src[j] + 2*src[j+stride] + 1)) >> 11) + 1) >> 1;
+        dst[j] = (dst[j] + ((683*(src[j] + 2*src[j+stride] + 1)) >> 11) + 1) >> 1;
       }
       src += stride;
       dst += stride;
@@ -1340,7 +1374,7 @@ static inline void avg_tpel_pixels_mc21_c(uint8_t *dst, const uint8_t *src, int
     int i,j;
     for (i=0; i < height; i++) {
       for (j=0; j < width; j++) {
-	dst[j] = (dst[j] + ((2731*(3*src[j] + 4*src[j+1] + 2*src[j+stride] + 3*src[j+stride+1] + 6)) >> 15) + 1) >> 1;
+        dst[j] = (dst[j] + ((2731*(3*src[j] + 4*src[j+1] + 2*src[j+stride] + 3*src[j+stride+1] + 6)) >> 15) + 1) >> 1;
       }
       src += stride;
       dst += stride;
@@ -1351,7 +1385,7 @@ static inline void avg_tpel_pixels_mc22_c(uint8_t *dst, const uint8_t *src, int
     int i,j;
     for (i=0; i < height; i++) {
       for (j=0; j < width; j++) {
-	dst[j] = (dst[j] + ((2731*(2*src[j] + 3*src[j+1] + 3*src[j+stride] + 4*src[j+stride+1] + 6)) >> 15) + 1) >> 1;
+        dst[j] = (dst[j] + ((2731*(2*src[j] + 3*src[j+1] + 3*src[j+stride] + 4*src[j+stride+1] + 6)) >> 15) + 1) >> 1;
       }
       src += stride;
       dst += stride;
@@ -1450,6 +1484,17 @@ H264_CHROMA_MC(avg_       , op_avg)
 #undef op_avg
 #undef op_put
 
+static inline void copy_block2(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h)
+{
+    int i;
+    for(i=0; i<h; i++)
+    {
+        ST16(dst   , LD16(src   ));
+        dst+=dstStride;
+        src+=srcStride;
+    }
+}
+
 static inline void copy_block4(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h)
 {
     int i;
@@ -2015,6 +2060,68 @@ QPEL_MC(0, avg_       , _       , op_avg)
 
 #if 1
 #define H264_LOWPASS(OPNAME, OP, OP2) \
+static void OPNAME ## h264_qpel2_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
+    const int h=2;\
+    uint8_t *cm = cropTbl + MAX_NEG_CROP;\
+    int i;\
+    for(i=0; i<h; i++)\
+    {\
+        OP(dst[0], (src[0]+src[1])*20 - (src[-1]+src[2])*5 + (src[-2]+src[3]));\
+        OP(dst[1], (src[1]+src[2])*20 - (src[0 ]+src[3])*5 + (src[-1]+src[4]));\
+        dst+=dstStride;\
+        src+=srcStride;\
+    }\
+}\
+\
+static void OPNAME ## h264_qpel2_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
+    const int w=2;\
+    uint8_t *cm = cropTbl + MAX_NEG_CROP;\
+    int i;\
+    for(i=0; i<w; i++)\
+    {\
+        const int srcB= src[-2*srcStride];\
+        const int srcA= src[-1*srcStride];\
+        const int src0= src[0 *srcStride];\
+        const int src1= src[1 *srcStride];\
+        const int src2= src[2 *srcStride];\
+        const int src3= src[3 *srcStride];\
+        const int src4= src[4 *srcStride];\
+        OP(dst[0*dstStride], (src0+src1)*20 - (srcA+src2)*5 + (srcB+src3));\
+        OP(dst[1*dstStride], (src1+src2)*20 - (src0+src3)*5 + (srcA+src4));\
+        dst++;\
+        src++;\
+    }\
+}\
+\
+static void OPNAME ## h264_qpel2_hv_lowpass(uint8_t *dst, int16_t *tmp, uint8_t *src, int dstStride, int tmpStride, int srcStride){\
+    const int h=2;\
+    const int w=2;\
+    uint8_t *cm = cropTbl + MAX_NEG_CROP;\
+    int i;\
+    src -= 2*srcStride;\
+    for(i=0; i<h+5; i++)\
+    {\
+        tmp[0]= (src[0]+src[1])*20 - (src[-1]+src[2])*5 + (src[-2]+src[3]);\
+        tmp[1]= (src[1]+src[2])*20 - (src[0 ]+src[3])*5 + (src[-1]+src[4]);\
+        tmp+=tmpStride;\
+        src+=srcStride;\
+    }\
+    tmp -= tmpStride*(h+5-2);\
+    for(i=0; i<w; i++)\
+    {\
+        const int tmpB= tmp[-2*tmpStride];\
+        const int tmpA= tmp[-1*tmpStride];\
+        const int tmp0= tmp[0 *tmpStride];\
+        const int tmp1= tmp[1 *tmpStride];\
+        const int tmp2= tmp[2 *tmpStride];\
+        const int tmp3= tmp[3 *tmpStride];\
+        const int tmp4= tmp[4 *tmpStride];\
+        OP2(dst[0*dstStride], (tmp0+tmp1)*20 - (tmpA+tmp2)*5 + (tmpB+tmp3));\
+        OP2(dst[1*dstStride], (tmp1+tmp2)*20 - (tmp0+tmp3)*5 + (tmpA+tmp4));\
+        dst++;\
+        tmp++;\
+    }\
+}\
 static void OPNAME ## h264_qpel4_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
     const int h=4;\
     uint8_t *cm = cropTbl + MAX_NEG_CROP;\
@@ -2361,6 +2468,7 @@ static void OPNAME ## h264_qpel ## SIZE ## _mc32_c(uint8_t *dst, uint8_t *src, i
 
 H264_LOWPASS(put_       , op_put, op2_put)
 H264_LOWPASS(avg_       , op_avg, op2_avg)
+H264_MC(put_, 2)
 H264_MC(put_, 4)
 H264_MC(put_, 8)
 H264_MC(put_, 16)
@@ -2374,6 +2482,76 @@ H264_MC(avg_, 16)
 #undef op2_put
 #endif
 
+#define op_scale1(x)  block[x] = clip_uint8( (block[x]*weight + offset) >> log2_denom )
+#define op_scale2(x)  dst[x] = 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 void wmv2_mspel8_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h){
     uint8_t *cm = cropTbl + MAX_NEG_CROP;
     int i;
@@ -2388,7 +2566,7 @@ static void wmv2_mspel8_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int
         dst[6]= cm[(9*(src[6] + src[7]) - (src[ 5] + src[8]) + 8)>>4];
         dst[7]= cm[(9*(src[7] + src[8]) - (src[ 6] + src[9]) + 8)>>4];
         dst+=dstStride;
-        src+=srcStride;        
+        src+=srcStride;
     }
 }
 
@@ -2472,7 +2650,7 @@ static void put_mspel8_mc22_c(uint8_t *dst, uint8_t *src, int stride){
 static void h263_v_loop_filter_c(uint8_t *src, int stride, int qscale){
     int x;
     const int strength= ff_h263_loop_filter_strength[qscale];
-    
+
     for(x=0; x<8; x++){
         int d1, d2, ad1;
         int p0= src[x-2*stride];
@@ -2486,19 +2664,19 @@ static void h263_v_loop_filter_c(uint8_t *src, int stride, int qscale){
         else if(d<   strength) d1= d;
         else if(d< 2*strength) d1= 2*strength - d;
         else                   d1= 0;
-        
+
         p1 += d1;
         p2 -= d1;
         if(p1&256) p1= ~(p1>>31);
         if(p2&256) p2= ~(p2>>31);
-        
+
         src[x-1*stride] = p1;
         src[x+0*stride] = p2;
 
         ad1= ABS(d1)>>1;
-        
+
         d2= clip((p0-p3)/4, -ad1, ad1);
-        
+
         src[x-2*stride] = p0 - d2;
         src[x+  stride] = p3 + d2;
     }
@@ -2507,7 +2685,7 @@ static void h263_v_loop_filter_c(uint8_t *src, int stride, int qscale){
 static void h263_h_loop_filter_c(uint8_t *src, int stride, int qscale){
     int y;
     const int strength= ff_h263_loop_filter_strength[qscale];
-    
+
     for(y=0; y<8; y++){
         int d1, d2, ad1;
         int p0= src[y*stride-2];
@@ -2521,19 +2699,19 @@ static void h263_h_loop_filter_c(uint8_t *src, int stride, int qscale){
         else if(d<   strength) d1= d;
         else if(d< 2*strength) d1= 2*strength - d;
         else                   d1= 0;
-        
+
         p1 += d1;
         p2 -= d1;
         if(p1&256) p1= ~(p1>>31);
         if(p2&256) p2= ~(p2>>31);
-        
+
         src[y*stride-1] = p1;
         src[y*stride+0] = p2;
 
         ad1= ABS(d1)>>1;
-        
+
         d2= clip((p0-p3)/4, -ad1, ad1);
-        
+
         src[y*stride-2] = p0 - d2;
         src[y*stride+1] = p3 + d2;
     }
@@ -2554,7 +2732,7 @@ static void h261_loop_filter_c(uint8_t *src, int stride){
             temp[yz] = src[xy - stride] + 2*src[xy] + src[xy + stride];
         }
     }
-        
+
     for(y=0; y<8; y++){
         src[  y*stride] = (temp[  y*8] + 2)>>2;
         src[7+y*stride] = (temp[7+y*8] + 2)>>2;
@@ -2566,6 +2744,120 @@ static void h261_loop_filter_c(uint8_t *src, int stride){
     }
 }
 
+static inline void h264_loop_filter_luma_c(uint8_t *pix, int xstride, int ystride, int alpha, int beta, int8_t *tc0)
+{
+    int i, d;
+    for( i = 0; i < 4; i++ ) {
+        if( tc0[i] < 0 ) {
+            pix += 4*ystride;
+            continue;
+        }
+        for( d = 0; d < 4; 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( ABS( p0 - q0 ) < alpha &&
+                ABS( p1 - p0 ) < beta &&
+                ABS( q1 - q0 ) < beta ) {
+
+                int tc = tc0[i];
+                int i_delta;
+
+                if( ABS( p2 - p0 ) < beta ) {
+                    pix[-2*xstride] = p1 + clip( (( p2 + ( ( p0 + q0 + 1 ) >> 1 ) ) >> 1) - p1, -tc0[i], tc0[i] );
+                    tc++;
+                }
+                if( ABS( q2 - q0 ) < beta ) {
+                    pix[   xstride] = q1 + clip( (( q2 + ( ( p0 + q0 + 1 ) >> 1 ) ) >> 1) - q1, -tc0[i], tc0[i] );
+                    tc++;
+                }
+
+                i_delta = clip( (((q0 - p0 ) << 2) + (p1 - q1) + 4) >> 3, -tc, tc );
+                pix[-xstride] = clip_uint8( p0 + i_delta );    /* p0' */
+                pix[0]        = 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, 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, alpha, beta, tc0);
+}
+
+static inline void h264_loop_filter_chroma_c(uint8_t *pix, int xstride, int ystride, 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 += 2*ystride;
+            continue;
+        }
+        for( d = 0; d < 2; 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( ABS( p0 - q0 ) < alpha &&
+                ABS( p1 - p0 ) < beta &&
+                ABS( q1 - q0 ) < beta ) {
+
+                int delta = clip( (((q0 - p0 ) << 2) + (p1 - q1) + 4) >> 3, -tc, tc );
+
+                pix[-xstride] = clip_uint8( p0 + delta );    /* p0' */
+                pix[0]        = 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, 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, alpha, beta, tc0);
+}
+
+static inline void h264_loop_filter_chroma_intra_c(uint8_t *pix, int xstride, int ystride, int alpha, int beta)
+{
+    int d;
+    for( d = 0; d < 8; 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( ABS( p0 - q0 ) < alpha &&
+            ABS( p1 - p0 ) < beta &&
+            ABS( 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, 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, alpha, beta);
+}
+
 static inline int pix_abs16_c(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
 {
     int s, i;
@@ -2766,7 +3058,8 @@ static int pix_abs8_xy2_c(void *v, uint8_t *pix1, uint8_t *pix2, int line_size,
     return s;
 }
 
-static int nsse16_c(MpegEncContext *c, uint8_t *s1, uint8_t *s2, int stride, int h){
+static int nsse16_c(void *v, uint8_t *s1, uint8_t *s2, int stride, int h){
+    MpegEncContext *c = v;
     int score1=0;
     int score2=0;
     int x,y;
@@ -2791,11 +3084,12 @@ static int nsse16_c(MpegEncContext *c, uint8_t *s1, uint8_t *s2, int stride, int
     else  return score1 + ABS(score2)*8;
 }
 
-static int nsse8_c(MpegEncContext *c, uint8_t *s1, uint8_t *s2, int stride, int h){
+static int nsse8_c(void *v, uint8_t *s1, uint8_t *s2, int stride, int h){
+    MpegEncContext *c = v;
     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 ]);
@@ -2811,7 +3105,7 @@ static int nsse8_c(MpegEncContext *c, uint8_t *s1, uint8_t *s2, int stride, int
         s1+= stride;
         s2+= stride;
     }
-    
+
     if(c) return score1 + ABS(score2)*c->avctx->nsse_weight;
     else  return score1 + ABS(score2)*8;
 }
@@ -2836,7 +3130,7 @@ static void add_8x8basis_c(int16_t rem[64], int16_t basis[64], int scale){
 
     for(i=0; i<8*8; i++){
         rem[i] += (basis[i]*scale + (1<<(BASIS_SHIFT - RECON_SHIFT-1)))>>(BASIS_SHIFT - RECON_SHIFT);
-    }    
+    }
 }
 
 /**
@@ -2844,14 +3138,14 @@ static void add_8x8basis_c(int16_t rem[64], int16_t basis[64], int scale){
  * @param block the block which will be permuted according to the given permutation vector
  * @param permutation the permutation vector
  * @param last the last non zero coefficient in scantable order, used to speed the permutation up
- * @param scantable the used scantable, this is only used to speed the permutation up, the block is not 
+ * @param scantable the used scantable, this is only used to speed the permutation up, the block is not
  *                  (inverse) permutated to scantable order!
  */
 void ff_block_permute(DCTELEM *block, uint8_t *permutation, const uint8_t *scantable, int last)
 {
     int i;
     DCTELEM temp[64];
-    
+
     if(last<=0) return;
     //if(permutation[1]==1) return; //FIXME its ok but not clean and might fail for some perms
 
@@ -2860,7 +3154,7 @@ void ff_block_permute(DCTELEM *block, uint8_t *permutation, const uint8_t *scant
         temp[j]= block[j];
         block[j]=0;
     }
-    
+
     for(i=0; i<=last; i++){
         const int j= scantable[i];
         const int perm_j= permutation[j];
@@ -2874,9 +3168,9 @@ static int zero_cmp(void *s, uint8_t *a, uint8_t *b, int stride, int h){
 
 void ff_set_cmp(DSPContext* c, me_cmp_func *cmp, int type){
     int i;
-    
+
     memset(cmp, 0, sizeof(void*)*5);
-        
+
     for(i=0; i<5; i++){
         switch(type&0xFF){
         case FF_CMP_SAD:
@@ -2891,6 +3185,9 @@ void ff_set_cmp(DSPContext* c, me_cmp_func *cmp, int type){
         case FF_CMP_DCT:
             cmp[i]= c->dct_sad[i];
             break;
+        case FF_CMP_DCT264:
+            cmp[i]= c->dct264_sad[i];
+            break;
         case FF_CMP_DCTMAX:
             cmp[i]= c->dct_max[i];
             break;
@@ -2915,12 +3212,14 @@ void ff_set_cmp(DSPContext* c, me_cmp_func *cmp, int type){
         case FF_CMP_NSSE:
             cmp[i]= c->nsse[i];
             break;
+#ifdef CONFIG_SNOW_ENCODER
         case FF_CMP_W53:
             cmp[i]= c->w53[i];
             break;
         case FF_CMP_W97:
             cmp[i]= c->w97[i];
             break;
+#endif
         default:
             av_log(NULL, AV_LOG_ERROR,"internal error in cmp function selection\n");
         }
@@ -2979,7 +3278,7 @@ static void sub_hfyu_median_prediction_c(uint8_t *dst, uint8_t *src1, uint8_t *s
         lt= src1[i];
         l= src2[i];
         dst[i]= l - pred;
-    }    
+    }
 
     *left= l;
     *left_top= lt;
@@ -3004,7 +3303,7 @@ static int hadamard8_diff8x8_c(/*MpegEncContext*/ void *s, uint8_t *dst, uint8_t
     int i;
     int temp[64];
     int sum=0;
-    
+
     assert(h==8);
 
     for(i=0; i<8; i++){
@@ -3013,12 +3312,12 @@ static int hadamard8_diff8x8_c(/*MpegEncContext*/ void *s, uint8_t *dst, uint8_t
         BUTTERFLY2(temp[8*i+2], temp[8*i+3], src[stride*i+2]-dst[stride*i+2],src[stride*i+3]-dst[stride*i+3]);
         BUTTERFLY2(temp[8*i+4], temp[8*i+5], src[stride*i+4]-dst[stride*i+4],src[stride*i+5]-dst[stride*i+5]);
         BUTTERFLY2(temp[8*i+6], temp[8*i+7], src[stride*i+6]-dst[stride*i+6],src[stride*i+7]-dst[stride*i+7]);
-        
+
         BUTTERFLY1(temp[8*i+0], temp[8*i+2]);
         BUTTERFLY1(temp[8*i+1], temp[8*i+3]);
         BUTTERFLY1(temp[8*i+4], temp[8*i+6]);
         BUTTERFLY1(temp[8*i+5], temp[8*i+7]);
-        
+
         BUTTERFLY1(temp[8*i+0], temp[8*i+4]);
         BUTTERFLY1(temp[8*i+1], temp[8*i+5]);
         BUTTERFLY1(temp[8*i+2], temp[8*i+6]);
@@ -3030,13 +3329,13 @@ static int hadamard8_diff8x8_c(/*MpegEncContext*/ void *s, uint8_t *dst, uint8_t
         BUTTERFLY1(temp[8*2+i], temp[8*3+i]);
         BUTTERFLY1(temp[8*4+i], temp[8*5+i]);
         BUTTERFLY1(temp[8*6+i], temp[8*7+i]);
-        
+
         BUTTERFLY1(temp[8*0+i], temp[8*2+i]);
         BUTTERFLY1(temp[8*1+i], temp[8*3+i]);
         BUTTERFLY1(temp[8*4+i], temp[8*6+i]);
         BUTTERFLY1(temp[8*5+i], temp[8*7+i]);
 
-        sum += 
+        sum +=
              BUTTERFLYA(temp[8*0+i], temp[8*4+i])
             +BUTTERFLYA(temp[8*1+i], temp[8*5+i])
             +BUTTERFLYA(temp[8*2+i], temp[8*6+i])
@@ -3056,21 +3355,21 @@ static int hadamard8_intra8x8_c(/*MpegEncContext*/ void *s, uint8_t *src, uint8_
     int i;
     int temp[64];
     int sum=0;
-    
+
     assert(h==8);
-    
+
     for(i=0; i<8; i++){
         //FIXME try pointer walks
         BUTTERFLY2(temp[8*i+0], temp[8*i+1], src[stride*i+0],src[stride*i+1]);
         BUTTERFLY2(temp[8*i+2], temp[8*i+3], src[stride*i+2],src[stride*i+3]);
         BUTTERFLY2(temp[8*i+4], temp[8*i+5], src[stride*i+4],src[stride*i+5]);
         BUTTERFLY2(temp[8*i+6], temp[8*i+7], src[stride*i+6],src[stride*i+7]);
-        
+
         BUTTERFLY1(temp[8*i+0], temp[8*i+2]);
         BUTTERFLY1(temp[8*i+1], temp[8*i+3]);
         BUTTERFLY1(temp[8*i+4], temp[8*i+6]);
         BUTTERFLY1(temp[8*i+5], temp[8*i+7]);
-        
+
         BUTTERFLY1(temp[8*i+0], temp[8*i+4]);
         BUTTERFLY1(temp[8*i+1], temp[8*i+5]);
         BUTTERFLY1(temp[8*i+2], temp[8*i+6]);
@@ -3082,30 +3381,30 @@ static int hadamard8_intra8x8_c(/*MpegEncContext*/ void *s, uint8_t *src, uint8_
         BUTTERFLY1(temp[8*2+i], temp[8*3+i]);
         BUTTERFLY1(temp[8*4+i], temp[8*5+i]);
         BUTTERFLY1(temp[8*6+i], temp[8*7+i]);
-        
+
         BUTTERFLY1(temp[8*0+i], temp[8*2+i]);
         BUTTERFLY1(temp[8*1+i], temp[8*3+i]);
         BUTTERFLY1(temp[8*4+i], temp[8*6+i]);
         BUTTERFLY1(temp[8*5+i], temp[8*7+i]);
-    
-        sum += 
+
+        sum +=
              BUTTERFLYA(temp[8*0+i], temp[8*4+i])
             +BUTTERFLYA(temp[8*1+i], temp[8*5+i])
             +BUTTERFLYA(temp[8*2+i], temp[8*6+i])
             +BUTTERFLYA(temp[8*3+i], temp[8*7+i]);
     }
-    
+
     sum -= ABS(temp[8*0] + temp[8*4]); // -mean
-    
+
     return sum;
 }
 
 static int dct_sad8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int stride, int h){
     MpegEncContext * const s= (MpegEncContext *)c;
-    uint64_t __align8 aligned_temp[sizeof(DCTELEM)*64/8];
+    DECLARE_ALIGNED_8(uint64_t, aligned_temp[sizeof(DCTELEM)*64/8]);
     DCTELEM * const temp= (DCTELEM*)aligned_temp;
     int sum=0, i;
-    
+
     assert(h==8);
 
     s->dsp.diff_pixels(temp, src1, src2, stride);
@@ -3113,16 +3412,69 @@ static int dct_sad8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2
 
     for(i=0; i<64; i++)
         sum+= ABS(temp[i]);
-        
+
     return sum;
 }
 
+#ifdef CONFIG_GPL
+#define DCT8_1D {\
+    const int s07 = SRC(0) + SRC(7);\
+    const int s16 = SRC(1) + SRC(6);\
+    const int s25 = SRC(2) + SRC(5);\
+    const int s34 = SRC(3) + SRC(4);\
+    const int a0 = s07 + s34;\
+    const int a1 = s16 + s25;\
+    const int a2 = s07 - s34;\
+    const int a3 = s16 - s25;\
+    const int d07 = SRC(0) - SRC(7);\
+    const int d16 = SRC(1) - SRC(6);\
+    const int d25 = SRC(2) - SRC(5);\
+    const int d34 = SRC(3) - SRC(4);\
+    const int a4 = d16 + d25 + (d07 + (d07>>1));\
+    const int a5 = d07 - d34 - (d25 + (d25>>1));\
+    const int a6 = d07 + d34 - (d16 + (d16>>1));\
+    const int a7 = d16 - d25 + (d34 + (d34>>1));\
+    DST(0,  a0 + a1     ) ;\
+    DST(1,  a4 + (a7>>2)) ;\
+    DST(2,  a2 + (a3>>1)) ;\
+    DST(3,  a5 + (a6>>2)) ;\
+    DST(4,  a0 - a1     ) ;\
+    DST(5,  a6 - (a5>>2)) ;\
+    DST(6, (a2>>1) - a3 ) ;\
+    DST(7, (a4>>2) - a7 ) ;\
+}
+
+static int dct264_sad8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int stride, int h){
+    MpegEncContext * const s= (MpegEncContext *)c;
+    int16_t dct[8][8];
+    int i;
+    int sum=0;
+
+    s->dsp.diff_pixels(dct, src1, src2, stride);
+
+#define SRC(x) dct[i][x]
+#define DST(x,v) dct[i][x]= v
+    for( i = 0; i < 8; i++ )
+        DCT8_1D
+#undef SRC
+#undef DST
+
+#define SRC(x) dct[x][i]
+#define DST(x,v) sum += ABS(v)
+    for( i = 0; i < 8; i++ )
+        DCT8_1D
+#undef SRC
+#undef DST
+    return sum;
+}
+#endif
+
 static int dct_max8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int stride, int h){
     MpegEncContext * const s= (MpegEncContext *)c;
-    uint64_t __align8 aligned_temp[sizeof(DCTELEM)*64/8];
+    DECLARE_ALIGNED_8(uint64_t, aligned_temp[sizeof(DCTELEM)*64/8]);
     DCTELEM * const temp= (DCTELEM*)aligned_temp;
     int sum=0, i;
-    
+
     assert(h==8);
 
     s->dsp.diff_pixels(temp, src1, src2, stride);
@@ -3130,7 +3482,7 @@ static int dct_max8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2
 
     for(i=0; i<64; i++)
         sum= FFMAX(sum, ABS(temp[i]));
-        
+
     return sum;
 }
 
@@ -3138,40 +3490,40 @@ void simple_idct(DCTELEM *block); //FIXME
 
 static int quant_psnr8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int stride, int h){
     MpegEncContext * const s= (MpegEncContext *)c;
-    uint64_t __align8 aligned_temp[sizeof(DCTELEM)*64*2/8];
+    DECLARE_ALIGNED_8 (uint64_t, aligned_temp[sizeof(DCTELEM)*64*2/8]);
     DCTELEM * const temp= (DCTELEM*)aligned_temp;
     DCTELEM * const bak = ((DCTELEM*)aligned_temp)+64;
     int sum=0, i;
 
     assert(h==8);
     s->mb_intra=0;
-    
+
     s->dsp.diff_pixels(temp, src1, src2, stride);
-    
+
     memcpy(bak, temp, 64*sizeof(DCTELEM));
-    
+
     s->block_last_index[0/*FIXME*/]= s->fast_dct_quantize(s, temp, 0/*FIXME*/, s->qscale, &i);
     s->dct_unquantize_inter(s, temp, 0, s->qscale);
-    simple_idct(temp); //FIXME 
-    
+    simple_idct(temp); //FIXME
+
     for(i=0; i<64; i++)
         sum+= (temp[i]-bak[i])*(temp[i]-bak[i]);
-        
+
     return sum;
 }
 
 static int rd8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int stride, int h){
     MpegEncContext * const s= (MpegEncContext *)c;
     const uint8_t *scantable= s->intra_scantable.permutated;
-    uint64_t __align8 aligned_temp[sizeof(DCTELEM)*64/8];
-    uint64_t __align8 aligned_bak[stride];
+    DECLARE_ALIGNED_8 (uint64_t, aligned_temp[sizeof(DCTELEM)*64/8]);
+    DECLARE_ALIGNED_8 (uint64_t, aligned_bak[stride]);
     DCTELEM * const temp= (DCTELEM*)aligned_temp;
     uint8_t * const bak= (uint8_t*)aligned_bak;
     int i, last, run, bits, level, distoration, start_i;
     const int esc_length= s->ac_esc_length;
     uint8_t * length;
     uint8_t * last_length;
-    
+
     assert(h==8);
 
     for(i=0; i<8; i++){
@@ -3184,9 +3536,9 @@ static int rd8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int
     s->block_last_index[0/*FIXME*/]= last= s->fast_dct_quantize(s, temp, 0/*FIXME*/, s->qscale, &i);
 
     bits=0;
-    
+
     if (s->mb_intra) {
-        start_i = 1; 
+        start_i = 1;
         length     = s->intra_ac_vlc_length;
         last_length= s->intra_ac_vlc_last_length;
         bits+= s->luma_dc_vlc_length[temp[0] + 256]; //FIXME chroma
@@ -3195,13 +3547,13 @@ static int rd8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int
         length     = s->inter_ac_vlc_length;
         last_length= s->inter_ac_vlc_last_length;
     }
-    
+
     if(last>=start_i){
         run=0;
         for(i=start_i; i<last; i++){
             int j= scantable[i];
             level= temp[j];
-        
+
             if(level){
                 level+=64;
                 if((level&(~127)) == 0){
@@ -3213,16 +3565,16 @@ static int rd8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int
                 run++;
         }
         i= scantable[last];
-       
+
         level= temp[i] + 64;
 
         assert(level - 64);
-        
+
         if((level&(~127)) == 0){
             bits+= last_length[UNI_AC_ENC_INDEX(run, level)];
         }else
             bits+= esc_length;
-    
+
     }
 
     if(last>=0){
@@ -3231,9 +3583,9 @@ static int rd8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int
         else
             s->dct_unquantize_inter(s, temp, 0, s->qscale);
     }
-    
+
     s->dsp.idct_add(bak, stride, temp);
-    
+
     distoration= s->dsp.sse[1](NULL, bak, src1, stride, 8);
 
     return distoration + ((bits*s->qscale*s->qscale*109 + 64)>>7);
@@ -3242,7 +3594,7 @@ static int rd8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int
 static int bit8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int stride, int h){
     MpegEncContext * const s= (MpegEncContext *)c;
     const uint8_t *scantable= s->intra_scantable.permutated;
-    uint64_t __align8 aligned_temp[sizeof(DCTELEM)*64/8];
+    DECLARE_ALIGNED_8 (uint64_t, aligned_temp[sizeof(DCTELEM)*64/8]);
     DCTELEM * const temp= (DCTELEM*)aligned_temp;
     int i, last, run, bits, level, start_i;
     const int esc_length= s->ac_esc_length;
@@ -3250,15 +3602,15 @@ static int bit8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, in
     uint8_t * last_length;
 
     assert(h==8);
-    
+
     s->dsp.diff_pixels(temp, src1, src2, stride);
 
     s->block_last_index[0/*FIXME*/]= last= s->fast_dct_quantize(s, temp, 0/*FIXME*/, s->qscale, &i);
 
     bits=0;
-    
+
     if (s->mb_intra) {
-        start_i = 1; 
+        start_i = 1;
         length     = s->intra_ac_vlc_length;
         last_length= s->intra_ac_vlc_last_length;
         bits+= s->luma_dc_vlc_length[temp[0] + 256]; //FIXME chroma
@@ -3267,13 +3619,13 @@ static int bit8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, in
         length     = s->inter_ac_vlc_length;
         last_length= s->inter_ac_vlc_last_length;
     }
-    
+
     if(last>=start_i){
         run=0;
         for(i=start_i; i<last; i++){
             int j= scantable[i];
             level= temp[j];
-        
+
             if(level){
                 level+=64;
                 if((level&(~127)) == 0){
@@ -3285,11 +3637,11 @@ static int bit8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, in
                 run++;
         }
         i= scantable[last];
-                
+
         level= temp[i] + 64;
-        
+
         assert(level - 64);
-        
+
         if((level&(~127)) == 0){
             bits+= last_length[UNI_AC_ENC_INDEX(run, level)];
         }else
@@ -3302,22 +3654,22 @@ static int bit8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, in
 static int vsad_intra16_c(/*MpegEncContext*/ void *c, uint8_t *s, uint8_t *dummy, int stride, int h){
     int score=0;
     int x,y;
-    
+
     for(y=1; y<h; y++){
         for(x=0; x<16; x+=4){
-            score+= ABS(s[x  ] - s[x  +stride]) + ABS(s[x+1] - s[x+1+stride]) 
+            score+= ABS(s[x  ] - s[x  +stride]) + ABS(s[x+1] - s[x+1+stride])
                    +ABS(s[x+2] - s[x+2+stride]) + ABS(s[x+3] - s[x+3+stride]);
         }
         s+= stride;
     }
-    
+
     return score;
 }
 
 static int vsad16_c(/*MpegEncContext*/ void *c, uint8_t *s1, uint8_t *s2, int stride, int h){
     int score=0;
     int x,y;
-    
+
     for(y=1; y<h; y++){
         for(x=0; x<16; x++){
             score+= ABS(s1[x  ] - s2[x ] - s1[x  +stride] + s2[x +stride]);
@@ -3325,7 +3677,7 @@ static int vsad16_c(/*MpegEncContext*/ void *c, uint8_t *s1, uint8_t *s2, int st
         s1+= stride;
         s2+= stride;
     }
-    
+
     return score;
 }
 
@@ -3333,22 +3685,22 @@ static int vsad16_c(/*MpegEncContext*/ void *c, uint8_t *s1, uint8_t *s2, int st
 static int vsse_intra16_c(/*MpegEncContext*/ void *c, uint8_t *s, uint8_t *dummy, int stride, int h){
     int score=0;
     int x,y;
-    
+
     for(y=1; y<h; y++){
         for(x=0; x<16; x+=4){
-            score+= SQ(s[x  ] - s[x  +stride]) + SQ(s[x+1] - s[x+1+stride]) 
+            score+= SQ(s[x  ] - s[x  +stride]) + SQ(s[x+1] - s[x+1+stride])
                    +SQ(s[x+2] - s[x+2+stride]) + SQ(s[x+3] - s[x+3+stride]);
         }
         s+= stride;
     }
-    
+
     return score;
 }
 
 static int vsse16_c(/*MpegEncContext*/ void *c, uint8_t *s1, uint8_t *s2, int stride, int h){
     int score=0;
     int x,y;
-    
+
     for(y=1; y<h; y++){
         for(x=0; x<16; x++){
             score+= SQ(s1[x  ] - s2[x ] - s1[x  +stride] + s2[x +stride]);
@@ -3356,13 +3708,16 @@ static int vsse16_c(/*MpegEncContext*/ void *c, uint8_t *s1, uint8_t *s2, int st
         s1+= stride;
         s2+= stride;
     }
-    
+
     return score;
 }
 
 WARPER8_16_SQ(hadamard8_diff8x8_c, hadamard8_diff16_c)
 WARPER8_16_SQ(hadamard8_intra8x8_c, hadamard8_intra16_c)
 WARPER8_16_SQ(dct_sad8x8_c, dct_sad16_c)
+#ifdef CONFIG_GPL
+WARPER8_16_SQ(dct264_sad8x8_c, dct264_sad16_c)
+#endif
 WARPER8_16_SQ(dct_max8x8_c, dct_max16_c)
 WARPER8_16_SQ(quant_psnr8x8_c, quant_psnr16_c)
 WARPER8_16_SQ(rd8x8_c, rd16_c)
@@ -3416,6 +3771,8 @@ static void ff_jref_idct1_add(uint8_t *dest, int line_size, DCTELEM *block)
     dest[0] = cm[dest[0] + ((block[0] + 4)>>3)];
 }
 
+static void just_return() { return; }
+
 /* init static data */
 void dsputil_static_init(void)
 {
@@ -3426,11 +3783,11 @@ void dsputil_static_init(void)
         cropTbl[i] = 0;
         cropTbl[i + MAX_NEG_CROP + 256] = 255;
     }
-    
+
     for(i=0;i<512;i++) {
         squareTbl[i] = (i - 256) * (i - 256);
     }
-    
+
     for(i=0; i<64; i++) inv_zigzag_direct16[ff_zigzag_direct[i]]= i+1;
 }
 
@@ -3442,15 +3799,15 @@ void dsputil_init(DSPContext* c, AVCodecContext *avctx)
 #ifdef CONFIG_ENCODERS
     if(avctx->dct_algo==FF_DCT_FASTINT) {
         c->fdct = fdct_ifast;
-	c->fdct248 = fdct_ifast248;
-    } 
+        c->fdct248 = fdct_ifast248;
+    }
     else if(avctx->dct_algo==FF_DCT_FAAN) {
         c->fdct = ff_faandct;
-	c->fdct248 = ff_faandct248; 
-    } 
+        c->fdct248 = ff_faandct248;
+    }
     else {
         c->fdct = ff_jpeg_fdct_islow; //slow/accurate/default
-	c->fdct248 = ff_fdct248_islow;
+        c->fdct248 = ff_fdct248_islow;
     }
 #endif //CONFIG_ENCODERS
 
@@ -3480,6 +3837,11 @@ void dsputil_init(DSPContext* c, AVCodecContext *avctx)
             c->idct_add= ff_jref_idct_add;
             c->idct    = j_rev_dct;
             c->idct_permutation_type= FF_LIBMPEG2_IDCT_PERM;
+        }else if(avctx->idct_algo==FF_IDCT_VP3){
+            c->idct_put= ff_vp3_idct_put_c;
+            c->idct_add= ff_vp3_idct_add_c;
+            c->idct    = ff_vp3_idct_c;
+            c->idct_permutation_type= FF_NO_IDCT_PERM;
         }else{ //accurate/default
             c->idct_put= simple_idct_put;
             c->idct_add= simple_idct_add;
@@ -3489,18 +3851,19 @@ void dsputil_init(DSPContext* c, AVCodecContext *avctx)
     }
 
     c->h264_idct_add= ff_h264_idct_add_c;
-
-    /* VP3 DSP support */
-    c->vp3_dsp_init = vp3_dsp_init_c;
-    c->vp3_idct = vp3_idct_c;
+    c->h264_idct8_add= ff_h264_idct8_add_c;
+    c->h264_idct_dc_add= ff_h264_idct_dc_add_c;
+    c->h264_idct8_dc_add= ff_h264_idct8_dc_add_c;
 
     c->get_pixels = get_pixels_c;
     c->diff_pixels = diff_pixels_c;
     c->put_pixels_clamped = put_pixels_clamped_c;
     c->put_signed_pixels_clamped = put_signed_pixels_clamped_c;
     c->add_pixels_clamped = add_pixels_clamped_c;
+    c->add_pixels8 = add_pixels8_c;
+    c->add_pixels4 = add_pixels4_c;
     c->gmc1 = gmc1_c;
-    c->gmc = gmc_c;
+    c->gmc = ff_gmc_c;
     c->clear_blocks = clear_blocks_c;
     c->pix_sum = pix_sum_c;
     c->pix_norm1 = pix_norm1_c;
@@ -3592,6 +3955,7 @@ void dsputil_init(DSPContext* c, AVCodecContext *avctx)
     dspfunc(put_h264_qpel, 0, 16);
     dspfunc(put_h264_qpel, 1, 8);
     dspfunc(put_h264_qpel, 2, 4);
+    dspfunc(put_h264_qpel, 3, 2);
     dspfunc(avg_h264_qpel, 0, 16);
     dspfunc(avg_h264_qpel, 1, 8);
     dspfunc(avg_h264_qpel, 2, 4);
@@ -3604,6 +3968,27 @@ void dsputil_init(DSPContext* c, AVCodecContext *avctx)
     c->avg_h264_chroma_pixels_tab[1]= avg_h264_chroma_mc4_c;
     c->avg_h264_chroma_pixels_tab[2]= avg_h264_chroma_mc2_c;
 
+    c->weight_h264_pixels_tab[0]= weight_h264_pixels16x16_c;
+    c->weight_h264_pixels_tab[1]= weight_h264_pixels16x8_c;
+    c->weight_h264_pixels_tab[2]= weight_h264_pixels8x16_c;
+    c->weight_h264_pixels_tab[3]= weight_h264_pixels8x8_c;
+    c->weight_h264_pixels_tab[4]= weight_h264_pixels8x4_c;
+    c->weight_h264_pixels_tab[5]= weight_h264_pixels4x8_c;
+    c->weight_h264_pixels_tab[6]= weight_h264_pixels4x4_c;
+    c->weight_h264_pixels_tab[7]= weight_h264_pixels4x2_c;
+    c->weight_h264_pixels_tab[8]= weight_h264_pixels2x4_c;
+    c->weight_h264_pixels_tab[9]= weight_h264_pixels2x2_c;
+    c->biweight_h264_pixels_tab[0]= biweight_h264_pixels16x16_c;
+    c->biweight_h264_pixels_tab[1]= biweight_h264_pixels16x8_c;
+    c->biweight_h264_pixels_tab[2]= biweight_h264_pixels8x16_c;
+    c->biweight_h264_pixels_tab[3]= biweight_h264_pixels8x8_c;
+    c->biweight_h264_pixels_tab[4]= biweight_h264_pixels8x4_c;
+    c->biweight_h264_pixels_tab[5]= biweight_h264_pixels4x8_c;
+    c->biweight_h264_pixels_tab[6]= biweight_h264_pixels4x4_c;
+    c->biweight_h264_pixels_tab[7]= biweight_h264_pixels4x2_c;
+    c->biweight_h264_pixels_tab[8]= biweight_h264_pixels2x4_c;
+    c->biweight_h264_pixels_tab[9]= biweight_h264_pixels2x2_c;
+
     c->put_mspel_pixels_tab[0]= put_mspel8_mc00_c;
     c->put_mspel_pixels_tab[1]= put_mspel8_mc10_c;
     c->put_mspel_pixels_tab[2]= put_mspel8_mc20_c;
@@ -3612,15 +3997,18 @@ void dsputil_init(DSPContext* c, AVCodecContext *avctx)
     c->put_mspel_pixels_tab[5]= put_mspel8_mc12_c;
     c->put_mspel_pixels_tab[6]= put_mspel8_mc22_c;
     c->put_mspel_pixels_tab[7]= put_mspel8_mc32_c;
-        
+
 #define SET_CMP_FUNC(name) \
     c->name[0]= name ## 16_c;\
     c->name[1]= name ## 8x8_c;
-    
+
     SET_CMP_FUNC(hadamard8_diff)
     c->hadamard8_diff[4]= hadamard8_intra16_c;
     SET_CMP_FUNC(dct_sad)
     SET_CMP_FUNC(dct_max)
+#ifdef CONFIG_GPL
+    SET_CMP_FUNC(dct264_sad)
+#endif
     c->sad[0]= pix_abs16_c;
     c->sad[1]= pix_abs8_c;
     c->sse[0]= sse16_c;
@@ -3635,24 +4023,46 @@ void dsputil_init(DSPContext* c, AVCodecContext *avctx)
     c->vsse[4]= vsse_intra16_c;
     c->nsse[0]= nsse16_c;
     c->nsse[1]= nsse8_c;
+#ifdef CONFIG_SNOW_ENCODER
     c->w53[0]= w53_16_c;
     c->w53[1]= w53_8_c;
     c->w97[0]= w97_16_c;
     c->w97[1]= w97_8_c;
+#endif
 
     c->add_bytes= add_bytes_c;
     c->diff_bytes= diff_bytes_c;
     c->sub_hfyu_median_prediction= sub_hfyu_median_prediction_c;
     c->bswap_buf= bswap_buf;
-    
+
+    c->h264_v_loop_filter_luma= h264_v_loop_filter_luma_c;
+    c->h264_h_loop_filter_luma= h264_h_loop_filter_luma_c;
+    c->h264_v_loop_filter_chroma= h264_v_loop_filter_chroma_c;
+    c->h264_h_loop_filter_chroma= h264_h_loop_filter_chroma_c;
+    c->h264_v_loop_filter_chroma_intra= h264_v_loop_filter_chroma_intra_c;
+    c->h264_h_loop_filter_chroma_intra= h264_h_loop_filter_chroma_intra_c;
+
     c->h263_h_loop_filter= h263_h_loop_filter_c;
     c->h263_v_loop_filter= h263_v_loop_filter_c;
-    
+
     c->h261_loop_filter= h261_loop_filter_c;
-    
+
     c->try_8x8basis= try_8x8basis_c;
     c->add_8x8basis= add_8x8basis_c;
 
+#ifdef CONFIG_SNOW_ENCODER
+    c->vertical_compose97i = ff_snow_vertical_compose97i;
+    c->horizontal_compose97i = ff_snow_horizontal_compose97i;
+    c->inner_add_yblock = ff_snow_inner_add_yblock;
+#endif
+
+    c->shrink[0]= ff_img_copy_plane;
+    c->shrink[1]= ff_shrink22;
+    c->shrink[2]= ff_shrink44;
+    c->shrink[3]= ff_shrink88;
+
+    c->prefetch= just_return;
+
 #ifdef HAVE_MMX
     dsputil_init_mmx(c, avctx);
 #endif
@@ -3695,6 +4105,10 @@ void dsputil_init(DSPContext* c, AVCodecContext *avctx)
         for(i=0; i<64; i++)
             c->idct_permutation[i]= ((i&7)<<3) | (i>>3);
         break;
+    case FF_PARTTRANS_IDCT_PERM:
+        for(i=0; i<64; i++)
+            c->idct_permutation[i]= (i&0x24) | ((i&3)<<3) | ((i>>3)&3);
+        break;
     default:
         av_log(avctx, AV_LOG_ERROR, "Internal error, IDCT permutation not set\n");
     }