Commit 982e7bbf authored by Michael Niedermayer's avatar Michael Niedermayer

avcodec/vc1: Skip chroma operations if CODEC_FLAG_GRAY is set

Signed-off-by: 's avatarMichael Niedermayer <michaelni@gmx.at>
parent 23e6cf83
...@@ -99,12 +99,14 @@ static void vc1_put_signed_blocks_clamped(VC1Context *v) ...@@ -99,12 +99,14 @@ static void vc1_put_signed_blocks_clamped(VC1Context *v)
s->idsp.put_signed_pixels_clamped(v->block[v->topleft_blk_idx][3], s->idsp.put_signed_pixels_clamped(v->block[v->topleft_blk_idx][3],
s->dest[0] - v_dist * s->linesize - 8, s->dest[0] - v_dist * s->linesize - 8,
stride_y); stride_y);
if (!CONFIG_GRAY || !(s->flags & CODEC_FLAG_GRAY)) {
s->idsp.put_signed_pixels_clamped(v->block[v->topleft_blk_idx][4], s->idsp.put_signed_pixels_clamped(v->block[v->topleft_blk_idx][4],
s->dest[1] - 8 * s->uvlinesize - 8, s->dest[1] - 8 * s->uvlinesize - 8,
s->uvlinesize); s->uvlinesize);
s->idsp.put_signed_pixels_clamped(v->block[v->topleft_blk_idx][5], s->idsp.put_signed_pixels_clamped(v->block[v->topleft_blk_idx][5],
s->dest[2] - 8 * s->uvlinesize - 8, s->dest[2] - 8 * s->uvlinesize - 8,
s->uvlinesize); s->uvlinesize);
}
} }
if (s->mb_x == s->mb_width - 1) { if (s->mb_x == s->mb_width - 1) {
top_mb_pos = (s->mb_y - 1) * s->mb_stride + s->mb_x; top_mb_pos = (s->mb_y - 1) * s->mb_stride + s->mb_x;
...@@ -124,12 +126,14 @@ static void vc1_put_signed_blocks_clamped(VC1Context *v) ...@@ -124,12 +126,14 @@ static void vc1_put_signed_blocks_clamped(VC1Context *v)
s->idsp.put_signed_pixels_clamped(v->block[v->top_blk_idx][3], s->idsp.put_signed_pixels_clamped(v->block[v->top_blk_idx][3],
s->dest[0] - v_dist * s->linesize + 8, s->dest[0] - v_dist * s->linesize + 8,
stride_y); stride_y);
if (!CONFIG_GRAY || !(s->flags & CODEC_FLAG_GRAY)) {
s->idsp.put_signed_pixels_clamped(v->block[v->top_blk_idx][4], s->idsp.put_signed_pixels_clamped(v->block[v->top_blk_idx][4],
s->dest[1] - 8 * s->uvlinesize, s->dest[1] - 8 * s->uvlinesize,
s->uvlinesize); s->uvlinesize);
s->idsp.put_signed_pixels_clamped(v->block[v->top_blk_idx][5], s->idsp.put_signed_pixels_clamped(v->block[v->top_blk_idx][5],
s->dest[2] - 8 * s->uvlinesize, s->dest[2] - 8 * s->uvlinesize,
s->uvlinesize); s->uvlinesize);
}
} }
} }
......
...@@ -40,6 +40,7 @@ void ff_vc1_loop_filter_iblk(VC1Context *v, int pq) ...@@ -40,6 +40,7 @@ void ff_vc1_loop_filter_iblk(VC1Context *v, int pq)
if (s->mb_x) if (s->mb_x)
v->vc1dsp.vc1_h_loop_filter16(s->dest[0] - 16 * s->linesize, s->linesize, pq); v->vc1dsp.vc1_h_loop_filter16(s->dest[0] - 16 * s->linesize, s->linesize, pq);
v->vc1dsp.vc1_h_loop_filter16(s->dest[0] - 16 * s->linesize + 8, s->linesize, pq); v->vc1dsp.vc1_h_loop_filter16(s->dest[0] - 16 * s->linesize + 8, s->linesize, pq);
if (!CONFIG_GRAY || !(s->flags & CODEC_FLAG_GRAY))
for (j = 0; j < 2; j++) { for (j = 0; j < 2; j++) {
v->vc1dsp.vc1_v_loop_filter8(s->dest[j + 1], s->uvlinesize, pq); v->vc1dsp.vc1_v_loop_filter8(s->dest[j + 1], s->uvlinesize, pq);
if (s->mb_x) if (s->mb_x)
...@@ -51,8 +52,10 @@ void ff_vc1_loop_filter_iblk(VC1Context *v, int pq) ...@@ -51,8 +52,10 @@ void ff_vc1_loop_filter_iblk(VC1Context *v, int pq)
if (s->mb_y == s->end_mb_y - 1) { if (s->mb_y == s->end_mb_y - 1) {
if (s->mb_x) { if (s->mb_x) {
v->vc1dsp.vc1_h_loop_filter16(s->dest[0], s->linesize, pq); v->vc1dsp.vc1_h_loop_filter16(s->dest[0], s->linesize, pq);
if (!CONFIG_GRAY || !(s->flags & CODEC_FLAG_GRAY)) {
v->vc1dsp.vc1_h_loop_filter8(s->dest[1], s->uvlinesize, pq); v->vc1dsp.vc1_h_loop_filter8(s->dest[1], s->uvlinesize, pq);
v->vc1dsp.vc1_h_loop_filter8(s->dest[2], s->uvlinesize, pq); v->vc1dsp.vc1_h_loop_filter8(s->dest[2], s->uvlinesize, pq);
}
} }
v->vc1dsp.vc1_h_loop_filter16(s->dest[0] + 8, s->linesize, pq); v->vc1dsp.vc1_h_loop_filter16(s->dest[0] + 8, s->linesize, pq);
} }
...@@ -73,6 +76,7 @@ void ff_vc1_loop_filter_iblk_delayed(VC1Context *v, int pq) ...@@ -73,6 +76,7 @@ void ff_vc1_loop_filter_iblk_delayed(VC1Context *v, int pq)
if (s->mb_x >= 2) if (s->mb_x >= 2)
v->vc1dsp.vc1_h_loop_filter16(s->dest[0] - 32 * s->linesize - 16, s->linesize, pq); v->vc1dsp.vc1_h_loop_filter16(s->dest[0] - 32 * s->linesize - 16, s->linesize, pq);
v->vc1dsp.vc1_h_loop_filter16(s->dest[0] - 32 * s->linesize - 8, s->linesize, pq); v->vc1dsp.vc1_h_loop_filter16(s->dest[0] - 32 * s->linesize - 8, s->linesize, pq);
if (!CONFIG_GRAY || !(s->flags & CODEC_FLAG_GRAY))
for (j = 0; j < 2; j++) { for (j = 0; j < 2; j++) {
v->vc1dsp.vc1_v_loop_filter8(s->dest[j + 1] - 8 * s->uvlinesize - 8, s->uvlinesize, pq); v->vc1dsp.vc1_v_loop_filter8(s->dest[j + 1] - 8 * s->uvlinesize - 8, s->uvlinesize, pq);
if (s->mb_x >= 2) { if (s->mb_x >= 2) {
...@@ -90,6 +94,7 @@ void ff_vc1_loop_filter_iblk_delayed(VC1Context *v, int pq) ...@@ -90,6 +94,7 @@ void ff_vc1_loop_filter_iblk_delayed(VC1Context *v, int pq)
if (s->mb_x) if (s->mb_x)
v->vc1dsp.vc1_h_loop_filter16(s->dest[0] - 32 * s->linesize, s->linesize, pq); v->vc1dsp.vc1_h_loop_filter16(s->dest[0] - 32 * s->linesize, s->linesize, pq);
v->vc1dsp.vc1_h_loop_filter16(s->dest[0] - 32 * s->linesize + 8, s->linesize, pq); v->vc1dsp.vc1_h_loop_filter16(s->dest[0] - 32 * s->linesize + 8, s->linesize, pq);
if (!CONFIG_GRAY || !(s->flags & CODEC_FLAG_GRAY))
for (j = 0; j < 2; j++) { for (j = 0; j < 2; j++) {
v->vc1dsp.vc1_v_loop_filter8(s->dest[j + 1] - 8 * s->uvlinesize, s->uvlinesize, pq); v->vc1dsp.vc1_v_loop_filter8(s->dest[j + 1] - 8 * s->uvlinesize, s->uvlinesize, pq);
if (s->mb_x >= 2) { if (s->mb_x >= 2) {
...@@ -105,7 +110,7 @@ void ff_vc1_loop_filter_iblk_delayed(VC1Context *v, int pq) ...@@ -105,7 +110,7 @@ void ff_vc1_loop_filter_iblk_delayed(VC1Context *v, int pq)
if (s->mb_x >= 2) if (s->mb_x >= 2)
v->vc1dsp.vc1_h_loop_filter16(s->dest[0] - 16 * s->linesize - 16, s->linesize, pq); v->vc1dsp.vc1_h_loop_filter16(s->dest[0] - 16 * s->linesize - 16, s->linesize, pq);
v->vc1dsp.vc1_h_loop_filter16(s->dest[0] - 16 * s->linesize - 8, s->linesize, pq); v->vc1dsp.vc1_h_loop_filter16(s->dest[0] - 16 * s->linesize - 8, s->linesize, pq);
if (s->mb_x >= 2) { if (s->mb_x >= 2 && !(s->flags & CODEC_FLAG_GRAY)) {
for (j = 0; j < 2; j++) { for (j = 0; j < 2; j++) {
v->vc1dsp.vc1_h_loop_filter8(s->dest[j + 1] - 8 * s->uvlinesize - 8, s->uvlinesize, pq); v->vc1dsp.vc1_h_loop_filter8(s->dest[j + 1] - 8 * s->uvlinesize - 8, s->uvlinesize, pq);
} }
...@@ -116,7 +121,7 @@ void ff_vc1_loop_filter_iblk_delayed(VC1Context *v, int pq) ...@@ -116,7 +121,7 @@ void ff_vc1_loop_filter_iblk_delayed(VC1Context *v, int pq)
if (s->mb_x) if (s->mb_x)
v->vc1dsp.vc1_h_loop_filter16(s->dest[0] - 16 * s->linesize, s->linesize, pq); v->vc1dsp.vc1_h_loop_filter16(s->dest[0] - 16 * s->linesize, s->linesize, pq);
v->vc1dsp.vc1_h_loop_filter16(s->dest[0] - 16 * s->linesize + 8, s->linesize, pq); v->vc1dsp.vc1_h_loop_filter16(s->dest[0] - 16 * s->linesize + 8, s->linesize, pq);
if (s->mb_x) { if (s->mb_x && !(s->flags & CODEC_FLAG_GRAY)) {
for (j = 0; j < 2; j++) { for (j = 0; j < 2; j++) {
v->vc1dsp.vc1_h_loop_filter8(s->dest[j + 1] - 8 * s->uvlinesize, s->uvlinesize, pq); v->vc1dsp.vc1_h_loop_filter8(s->dest[j + 1] - 8 * s->uvlinesize, s->uvlinesize, pq);
} }
...@@ -150,7 +155,7 @@ void ff_vc1_smooth_overlap_filter_iblk(VC1Context *v) ...@@ -150,7 +155,7 @@ void ff_vc1_smooth_overlap_filter_iblk(VC1Context *v)
v->block[v->cur_blk_idx][0]); v->block[v->cur_blk_idx][0]);
v->vc1dsp.vc1_h_s_overlap(v->block[v->left_blk_idx][3], v->vc1dsp.vc1_h_s_overlap(v->block[v->left_blk_idx][3],
v->block[v->cur_blk_idx][2]); v->block[v->cur_blk_idx][2]);
if (!(s->flags & CODEC_FLAG_GRAY)) { if (!CONFIG_GRAY || !(s->flags & CODEC_FLAG_GRAY)) {
v->vc1dsp.vc1_h_s_overlap(v->block[v->left_blk_idx][4], v->vc1dsp.vc1_h_s_overlap(v->block[v->left_blk_idx][4],
v->block[v->cur_blk_idx][4]); v->block[v->cur_blk_idx][4]);
v->vc1dsp.vc1_h_s_overlap(v->block[v->left_blk_idx][5], v->vc1dsp.vc1_h_s_overlap(v->block[v->left_blk_idx][5],
...@@ -169,7 +174,7 @@ void ff_vc1_smooth_overlap_filter_iblk(VC1Context *v) ...@@ -169,7 +174,7 @@ void ff_vc1_smooth_overlap_filter_iblk(VC1Context *v)
v->block[v->cur_blk_idx][0]); v->block[v->cur_blk_idx][0]);
v->vc1dsp.vc1_v_s_overlap(v->block[v->top_blk_idx][3], v->vc1dsp.vc1_v_s_overlap(v->block[v->top_blk_idx][3],
v->block[v->cur_blk_idx][1]); v->block[v->cur_blk_idx][1]);
if (!(s->flags & CODEC_FLAG_GRAY)) { if (!CONFIG_GRAY || !(s->flags & CODEC_FLAG_GRAY)) {
v->vc1dsp.vc1_v_s_overlap(v->block[v->top_blk_idx][4], v->vc1dsp.vc1_v_s_overlap(v->block[v->top_blk_idx][4],
v->block[v->cur_blk_idx][4]); v->block[v->cur_blk_idx][4]);
v->vc1dsp.vc1_v_s_overlap(v->block[v->top_blk_idx][5], v->vc1dsp.vc1_v_s_overlap(v->block[v->top_blk_idx][5],
...@@ -189,7 +194,7 @@ void ff_vc1_smooth_overlap_filter_iblk(VC1Context *v) ...@@ -189,7 +194,7 @@ void ff_vc1_smooth_overlap_filter_iblk(VC1Context *v)
v->block[v->left_blk_idx][0]); v->block[v->left_blk_idx][0]);
v->vc1dsp.vc1_v_s_overlap(v->block[v->topleft_blk_idx][3], v->vc1dsp.vc1_v_s_overlap(v->block[v->topleft_blk_idx][3],
v->block[v->left_blk_idx][1]); v->block[v->left_blk_idx][1]);
if (!(s->flags & CODEC_FLAG_GRAY)) { if (!CONFIG_GRAY || !(s->flags & CODEC_FLAG_GRAY)) {
v->vc1dsp.vc1_v_s_overlap(v->block[v->topleft_blk_idx][4], v->vc1dsp.vc1_v_s_overlap(v->block[v->topleft_blk_idx][4],
v->block[v->left_blk_idx][4]); v->block[v->left_blk_idx][4]);
v->vc1dsp.vc1_v_s_overlap(v->block[v->topleft_blk_idx][5], v->vc1dsp.vc1_v_s_overlap(v->block[v->topleft_blk_idx][5],
...@@ -331,21 +336,22 @@ void ff_vc1_apply_p_loop_filter(VC1Context *v) ...@@ -331,21 +336,22 @@ void ff_vc1_apply_p_loop_filter(VC1Context *v)
{ {
MpegEncContext *s = &v->s; MpegEncContext *s = &v->s;
int i; int i;
int block_count = CONFIG_GRAY && (s->flags & CODEC_FLAG_GRAY) ? 4 : 6;
for (i = 0; i < 6; i++) { for (i = 0; i < block_count; i++) {
vc1_apply_p_v_loop_filter(v, i); vc1_apply_p_v_loop_filter(v, i);
} }
/* V always precedes H, therefore we run H one MB before V; /* V always precedes H, therefore we run H one MB before V;
* at the end of a row, we catch up to complete the row */ * at the end of a row, we catch up to complete the row */
if (s->mb_x) { if (s->mb_x) {
for (i = 0; i < 6; i++) { for (i = 0; i < block_count; i++) {
vc1_apply_p_h_loop_filter(v, i); vc1_apply_p_h_loop_filter(v, i);
} }
if (s->mb_x == s->mb_width - 1) { if (s->mb_x == s->mb_width - 1) {
s->mb_x++; s->mb_x++;
ff_update_block_index(s); ff_update_block_index(s);
for (i = 0; i < 6; i++) { for (i = 0; i < block_count; i++) {
vc1_apply_p_h_loop_filter(v, i); vc1_apply_p_h_loop_filter(v, i);
} }
} }
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment