X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=libavcodec%2Fvc1dsp.c;h=5e79736a50fda2874d622c26f2972aa31c2bb01e;hb=49b19191bc0e23fea139b79b2c42969bc0c501d9;hp=5420e7e260bb77633e5789f3cd7e69a1f07d49e8;hpb=0e58865d6e86bbb664d92311c0f81c65e0213c35;p=ffmpeg

diff --git a/libavcodec/vc1dsp.c b/libavcodec/vc1dsp.c
index 5420e7e260b..5e79736a50f 100644
--- a/libavcodec/vc1dsp.c
+++ b/libavcodec/vc1dsp.c
@@ -130,7 +130,7 @@ static av_always_inline int vc1_filter_line(uint8_t* src, int stride, int pq){
  * @param pq block quantizer
  * @see 8.6
  */
-static void vc1_loop_filter(uint8_t* src, int step, int stride, int len, int pq)
+static inline void vc1_loop_filter(uint8_t* src, int step, int stride, int len, int pq)
 {
     int i;
     int filt3;
@@ -146,8 +146,58 @@ static void vc1_loop_filter(uint8_t* src, int step, int stride, int len, int pq)
     }
 }
 
+static void vc1_v_loop_filter4_c(uint8_t *src, int stride, int pq)
+{
+    vc1_loop_filter(src, 1, stride, 4, pq);
+}
+
+static void vc1_h_loop_filter4_c(uint8_t *src, int stride, int pq)
+{
+    vc1_loop_filter(src, stride, 1, 4, pq);
+}
+
+static void vc1_v_loop_filter8_c(uint8_t *src, int stride, int pq)
+{
+    vc1_loop_filter(src, 1, stride, 8, pq);
+}
+
+static void vc1_h_loop_filter8_c(uint8_t *src, int stride, int pq)
+{
+    vc1_loop_filter(src, stride, 1, 8, pq);
+}
+
+static void vc1_v_loop_filter16_c(uint8_t *src, int stride, int pq)
+{
+    vc1_loop_filter(src, 1, stride, 16, pq);
+}
+
+static void vc1_h_loop_filter16_c(uint8_t *src, int stride, int pq)
+{
+    vc1_loop_filter(src, stride, 1, 16, pq);
+}
+
 /** Do inverse transform on 8x8 block
 */
+static void vc1_inv_trans_8x8_dc_c(uint8_t *dest, int linesize, DCTELEM *block)
+{
+    int i;
+    int dc = block[0];
+    const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
+    dc = (3 * dc +  1) >> 1;
+    dc = (3 * dc + 16) >> 5;
+    for(i = 0; i < 8; i++){
+        dest[0] = cm[dest[0]+dc];
+        dest[1] = cm[dest[1]+dc];
+        dest[2] = cm[dest[2]+dc];
+        dest[3] = cm[dest[3]+dc];
+        dest[4] = cm[dest[4]+dc];
+        dest[5] = cm[dest[5]+dc];
+        dest[6] = cm[dest[6]+dc];
+        dest[7] = cm[dest[7]+dc];
+        dest += linesize;
+    }
+}
+
 static void vc1_inv_trans_8x8_c(DCTELEM block[64])
 {
     int i;
@@ -219,6 +269,26 @@ static void vc1_inv_trans_8x8_c(DCTELEM block[64])
 
 /** Do inverse transform on 8x4 part of block
 */
+static void vc1_inv_trans_8x4_dc_c(uint8_t *dest, int linesize, DCTELEM *block)
+{
+    int i;
+    int dc = block[0];
+    const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
+    dc = ( 3 * dc +  1) >> 1;
+    dc = (17 * dc + 64) >> 7;
+    for(i = 0; i < 4; i++){
+        dest[0] = cm[dest[0]+dc];
+        dest[1] = cm[dest[1]+dc];
+        dest[2] = cm[dest[2]+dc];
+        dest[3] = cm[dest[3]+dc];
+        dest[4] = cm[dest[4]+dc];
+        dest[5] = cm[dest[5]+dc];
+        dest[6] = cm[dest[6]+dc];
+        dest[7] = cm[dest[7]+dc];
+        dest += linesize;
+    }
+}
+
 static void vc1_inv_trans_8x4_c(uint8_t *dest, int linesize, DCTELEM *block)
 {
     int i;
@@ -276,6 +346,22 @@ static void vc1_inv_trans_8x4_c(uint8_t *dest, int linesize, DCTELEM *block)
 
 /** Do inverse transform on 4x8 parts of block
 */
+static void vc1_inv_trans_4x8_dc_c(uint8_t *dest, int linesize, DCTELEM *block)
+{
+    int i;
+    int dc = block[0];
+    const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
+    dc = (17 * dc +  4) >> 3;
+    dc = (12 * dc + 64) >> 7;
+    for(i = 0; i < 8; i++){
+        dest[0] = cm[dest[0]+dc];
+        dest[1] = cm[dest[1]+dc];
+        dest[2] = cm[dest[2]+dc];
+        dest[3] = cm[dest[3]+dc];
+        dest += linesize;
+    }
+}
+
 static void vc1_inv_trans_4x8_c(uint8_t *dest, int linesize, DCTELEM *block)
 {
     int i;
@@ -333,6 +419,22 @@ static void vc1_inv_trans_4x8_c(uint8_t *dest, int linesize, DCTELEM *block)
 
 /** Do inverse transform on 4x4 part of block
 */
+static void vc1_inv_trans_4x4_dc_c(uint8_t *dest, int linesize, DCTELEM *block)
+{
+    int i;
+    int dc = block[0];
+    const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
+    dc = (17 * dc +  4) >> 3;
+    dc = (17 * dc + 64) >> 7;
+    for(i = 0; i < 4; i++){
+        dest[0] = cm[dest[0]+dc];
+        dest[1] = cm[dest[1]+dc];
+        dest[2] = cm[dest[2]+dc];
+        dest[3] = cm[dest[3]+dc];
+        dest += linesize;
+    }
+}
+
 static void vc1_inv_trans_4x4_c(uint8_t *dest, int linesize, DCTELEM *block)
 {
     int i;
@@ -515,9 +617,18 @@ void ff_vc1dsp_init(DSPContext* dsp, AVCodecContext *avctx) {
     dsp->vc1_inv_trans_4x8 = vc1_inv_trans_4x8_c;
     dsp->vc1_inv_trans_8x4 = vc1_inv_trans_8x4_c;
     dsp->vc1_inv_trans_4x4 = vc1_inv_trans_4x4_c;
+    dsp->vc1_inv_trans_8x8_dc = vc1_inv_trans_8x8_dc_c;
+    dsp->vc1_inv_trans_4x8_dc = vc1_inv_trans_4x8_dc_c;
+    dsp->vc1_inv_trans_8x4_dc = vc1_inv_trans_8x4_dc_c;
+    dsp->vc1_inv_trans_4x4_dc = vc1_inv_trans_4x4_dc_c;
     dsp->vc1_h_overlap = vc1_h_overlap_c;
     dsp->vc1_v_overlap = vc1_v_overlap_c;
-    dsp->vc1_loop_filter = vc1_loop_filter;
+    dsp->vc1_v_loop_filter4 = vc1_v_loop_filter4_c;
+    dsp->vc1_h_loop_filter4 = vc1_h_loop_filter4_c;
+    dsp->vc1_v_loop_filter8 = vc1_v_loop_filter8_c;
+    dsp->vc1_h_loop_filter8 = vc1_h_loop_filter8_c;
+    dsp->vc1_v_loop_filter16 = vc1_v_loop_filter16_c;
+    dsp->vc1_h_loop_filter16 = vc1_h_loop_filter16_c;
 
     dsp->put_vc1_mspel_pixels_tab[ 0] = ff_put_vc1_mspel_mc00_c;
     dsp->put_vc1_mspel_pixels_tab[ 1] = put_vc1_mspel_mc10_c;