26 # define FUNC(n) AV_JOIN(n ## _simple_, BITS)
27 # define PIXEL_SHIFT (BITS >> 4)
29 # define FUNC(n) n ## _complex
30 # define PIXEL_SHIFT h->pixel_shift
43 const int mb_x = sl->mb_x;
44 const int mb_y = sl->mb_y;
45 const int mb_xy = sl->mb_xy;
46 const int mb_type =
h->cur_pic.mb_type[mb_xy];
47 uint8_t *dest_y, *dest_cb, *dest_cr;
48 int linesize, uvlinesize ;
50 const int *block_offset = &
h->block_offset[0];
51 const int transform_bypass = !
SIMPLE && (sl->qscale == 0 &&
h->ps.sps->transform_bypass);
53 const int block_h = 16 >>
h->chroma_y_shift;
56 dest_y =
h->cur_pic.f->data[0] + ((mb_x <<
PIXEL_SHIFT) + mb_y * sl->linesize) * 16;
57 dest_cb =
h->cur_pic.f->data[1] + (mb_x <<
PIXEL_SHIFT) * 8 + mb_y * sl->uvlinesize * block_h;
58 dest_cr =
h->cur_pic.f->data[2] + (mb_x <<
PIXEL_SHIFT) * 8 + mb_y * sl->uvlinesize * block_h;
60 h->vdsp.prefetch(dest_y + (sl->mb_x & 3) * 4 * sl->linesize + (64 <<
PIXEL_SHIFT), sl->linesize, 4);
61 h->vdsp.prefetch(dest_cb + (sl->mb_x & 7) * sl->uvlinesize + (64 <<
PIXEL_SHIFT), dest_cr - dest_cb, 2);
63 h->list_counts[mb_xy] = sl->list_count;
66 linesize = sl->mb_linesize = sl->linesize * 2;
67 uvlinesize = sl->mb_uvlinesize = sl->uvlinesize * 2;
68 block_offset = &
h->block_offset[48];
70 dest_y -= sl->linesize * 15;
71 dest_cb -= sl->uvlinesize * (block_h - 1);
72 dest_cr -= sl->uvlinesize * (block_h - 1);
76 for (list = 0; list < sl->list_count; list++) {
80 int8_t *
ref = &sl->ref_cache[list][
scan8[0]];
83 for (i = 0; i < 16; i += 4) {
84 int ref = sl->ref_cache[list][
scan8[i]];
87 8, (16 + ref) ^ (sl->mb_y & 1), 1);
93 linesize = sl->mb_linesize = sl->linesize;
94 uvlinesize = sl->mb_uvlinesize = sl->uvlinesize;
99 const int bit_depth =
h->ps.sps->bit_depth_luma;
106 for (i = 0; i < 16; i++) {
107 uint16_t *tmp_y = (uint16_t *)(dest_y + i * linesize);
108 for (j = 0; j < 16; j++)
109 tmp_y[j] =
get_bits(&gb, bit_depth);
112 if (!
h->ps.sps->chroma_format_idc) {
113 for (i = 0; i < block_h; i++) {
114 uint16_t *tmp_cb = (uint16_t *)(dest_cb + i * uvlinesize);
115 uint16_t *tmp_cr = (uint16_t *)(dest_cr + i * uvlinesize);
116 for (j = 0; j < 8; j++) {
117 tmp_cb[j] = tmp_cr[j] = 1 << (bit_depth - 1);
121 for (i = 0; i < block_h; i++) {
122 uint16_t *tmp_cb = (uint16_t *)(dest_cb + i * uvlinesize);
123 for (j = 0; j < 8; j++)
124 tmp_cb[j] =
get_bits(&gb, bit_depth);
126 for (i = 0; i < block_h; i++) {
127 uint16_t *tmp_cr = (uint16_t *)(dest_cr + i * uvlinesize);
128 for (j = 0; j < 8; j++)
129 tmp_cr[j] =
get_bits(&gb, bit_depth);
134 for (i = 0; i < 16; i++)
135 memcpy(dest_y + i * linesize, sl->intra_pcm_ptr + i * 16, 16);
137 if (!
h->ps.sps->chroma_format_idc) {
138 for (i = 0; i < 8; i++) {
139 memset(dest_cb + i * uvlinesize, 1 << (bit_depth - 1), 8);
140 memset(dest_cr + i * uvlinesize, 1 << (bit_depth - 1), 8);
143 const uint8_t *src_cb = sl->intra_pcm_ptr + 256;
144 const uint8_t *src_cr = sl->intra_pcm_ptr + 256 + block_h * 8;
145 for (i = 0; i < block_h; i++) {
146 memcpy(dest_cb + i * uvlinesize, src_cb + i * 8, 8);
147 memcpy(dest_cr + i * uvlinesize, src_cr + i * 8, 8);
154 if (sl->deblocking_filter)
159 h->hpc.pred8x8[sl->chroma_pred_mode](dest_cb, uvlinesize);
160 h->hpc.pred8x8[sl->chroma_pred_mode](dest_cr, uvlinesize);
165 block_offset, linesize, dest_y, 0);
167 if (sl->deblocking_filter)
172 FUNC(hl_motion_422)(
h, sl, dest_y, dest_cb, dest_cr,
173 h->h264qpel.put_h264_qpel_pixels_tab,
174 h->h264chroma.put_h264_chroma_pixels_tab,
175 h->h264qpel.avg_h264_qpel_pixels_tab,
176 h->h264chroma.avg_h264_chroma_pixels_tab,
177 h->h264dsp.weight_h264_pixels_tab,
178 h->h264dsp.biweight_h264_pixels_tab);
180 FUNC(hl_motion_420)(
h, sl, dest_y, dest_cb, dest_cr,
181 h->h264qpel.put_h264_qpel_pixels_tab,
182 h->h264chroma.put_h264_chroma_pixels_tab,
183 h->h264qpel.avg_h264_qpel_pixels_tab,
184 h->h264chroma.avg_h264_chroma_pixels_tab,
185 h->h264dsp.weight_h264_pixels_tab,
186 h->h264dsp.biweight_h264_pixels_tab);
195 uint8_t *dest[2] = { dest_cb, dest_cr };
196 if (transform_bypass) {
197 if (
IS_INTRA(mb_type) &&
h->ps.sps->profile_idc == 244 &&
200 h->hpc.pred8x8_add[sl->chroma_pred_mode](dest[0],
204 h->hpc.pred8x8_add[sl->chroma_pred_mode](dest[1],
209 idct_add =
h->h264dsp.h264_add_pixels4_clear;
210 for (j = 1; j < 3; j++) {
211 for (i = j * 16; i < j * 16 + 4; i++)
212 if (sl->non_zero_count_cache[
scan8[i]] ||
214 idct_add(dest[j - 1] + block_offset[i],
218 for (i = j * 16 + 4; i < j * 16 + 8; i++)
219 if (sl->non_zero_count_cache[
scan8[i + 4]] ||
221 idct_add(dest[j - 1] + block_offset[i + 4],
230 qp[0] = sl->chroma_qp[0] + 3;
231 qp[1] = sl->chroma_qp[1] + 3;
233 qp[0] = sl->chroma_qp[0];
234 qp[1] = sl->chroma_qp[1];
237 h->h264dsp.h264_chroma_dc_dequant_idct(sl->mb + (16 * 16 * 1 <<
PIXEL_SHIFT),
238 h->ps.pps->dequant4_coeff[
IS_INTRA(mb_type) ? 1 : 4][qp[0]][0]);
240 h->h264dsp.h264_chroma_dc_dequant_idct(sl->mb + (16 * 16 * 2 <<
PIXEL_SHIFT),
241 h->ps.pps->dequant4_coeff[
IS_INTRA(mb_type) ? 2 : 5][qp[1]][0]);
242 h->h264dsp.h264_idct_add8(dest, block_offset,
244 sl->non_zero_count_cache);
250 #if !SIMPLE || BITS == 8
258 const int mb_x = sl->mb_x;
259 const int mb_y = sl->mb_y;
260 const int mb_xy = sl->mb_xy;
261 const int mb_type =
h->cur_pic.mb_type[mb_xy];
265 const int *block_offset = &
h->block_offset[0];
266 const int transform_bypass = !
SIMPLE && (sl->qscale == 0 &&
h->ps.sps->transform_bypass);
269 for (p = 0; p < plane_count; p++) {
270 dest[p] =
h->cur_pic.f->data[p] +
271 ((mb_x <<
PIXEL_SHIFT) + mb_y * sl->linesize) * 16;
272 h->vdsp.prefetch(dest[p] + (sl->mb_x & 3) * 4 * sl->linesize + (64 <<
PIXEL_SHIFT),
276 h->list_counts[mb_xy] = sl->list_count;
279 linesize = sl->mb_linesize = sl->mb_uvlinesize = sl->linesize * 2;
280 block_offset = &
h->block_offset[48];
282 for (p = 0; p < 3; p++)
283 dest[p] -= sl->linesize * 15;
286 for (list = 0; list < sl->list_count; list++) {
290 int8_t *
ref = &sl->ref_cache[list][
scan8[0]];
293 for (i = 0; i < 16; i += 4) {
294 int ref = sl->ref_cache[list][
scan8[i]];
297 8, (16 + ref) ^ (sl->mb_y & 1), 1);
303 linesize = sl->mb_linesize = sl->mb_uvlinesize = sl->linesize;
308 const int bit_depth =
h->ps.sps->bit_depth_luma;
312 for (p = 0; p < plane_count; p++)
313 for (i = 0; i < 16; i++) {
314 uint16_t *
tmp = (uint16_t *)(dest[p] + i * linesize);
315 for (j = 0; j < 16; j++)
319 for (p = 0; p < plane_count; p++)
320 for (i = 0; i < 16; i++)
321 memcpy(dest[p] + i * linesize,
322 sl->intra_pcm_ptr + p * 256 + i * 16, 16);
326 if (sl->deblocking_filter)
330 for (p = 0; p < plane_count; p++)
333 block_offset, linesize, dest[p], p);
335 if (sl->deblocking_filter)
339 FUNC(hl_motion_444)(
h, sl, dest[0], dest[1], dest[2],
340 h->h264qpel.put_h264_qpel_pixels_tab,
341 h->h264chroma.put_h264_chroma_pixels_tab,
342 h->h264qpel.avg_h264_qpel_pixels_tab,
343 h->h264chroma.avg_h264_chroma_pixels_tab,
344 h->h264dsp.weight_h264_pixels_tab,
345 h->h264dsp.biweight_h264_pixels_tab);
348 for (p = 0; p < plane_count; p++)
static unsigned int get_bits(GetBitContext *s, int n)
Read 1-25 bits.
#define USES_LIST(a, list)
#define AV_CODEC_FLAG_GRAY
Only decode/encode grayscale.
typedef void(APIENTRY *FF_PFNGLACTIVETEXTUREPROC)(GLenum texture)
static void bit_depth(AudioStatsContext *s, uint64_t mask, uint64_t imask, AVRational *depth)
static av_always_inline void hl_decode_mb_idct_luma(const H264Context *h, H264SliceContext *sl, int mb_type, int simple, int transform_bypass, int pixel_shift, const int *block_offset, int linesize, uint8_t *dest_y, int p)
#define CHROMA_DC_BLOCK_INDEX
static void fill_rectangle(int x, int y, int w, int h)
static av_always_inline int dctcoef_get(int16_t *mb, int high_bit_depth, int index)
static const uint8_t scan8[16 *3+3]
static int init_get_bits(GetBitContext *s, const uint8_t *buffer, int bit_size)
Initialize GetBitContext.
static av_always_inline void xchg_mb_border(const H264Context *h, H264SliceContext *sl, uint8_t *src_y, uint8_t *src_cb, uint8_t *src_cr, int linesize, int uvlinesize, int xchg, int chroma444, int simple, int pixel_shift)
static av_always_inline void hl_decode_mb_predict_luma(const H264Context *h, H264SliceContext *sl, int mb_type, int simple, int transform_bypass, int pixel_shift, const int *block_offset, int linesize, uint8_t *dest_y, int p)
static int ref[MAX_W *MAX_W]
static av_noinline void FUNC() hl_decode_mb(const H264Context *h, H264SliceContext *sl)
static av_noinline void FUNC() hl_decode_mb_444(const H264Context *h, H264SliceContext *sl)
const uint16_t ff_h264_mb_sizes[4]