hevc/rext: add support for Range extension tools

SPS features/flags:
- transform_skip_rotation_enabled_flag
- transform_skip_context_enabled_flag
- implicit_rdpcm_enabled_flag
- explicit_rdpcm_enabled_flag
- intra_smoothing_disabled_flag
- persistent_rice_adaptation_enabled_flag

PPS features/flags:
- log2_max_transform_skip_block_size
- cross_component_prediction_enabled_flag
- chroma_qp_offset_list_enabled_flag
- diff_cu_chroma_qp_offset_depth
- chroma_qp_offset_list_len_minus1
- cb_qp_offset_list
- cr_qp_offset_list
- log2_sao_offset_scale_luma
- log2_sao_offset_scale_chroma
(cherry picked from commit 005294c5b939a23099871c6130c8a7cc331f73ee)
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

libavcodec/hevc.c

@@ -628,6 +628,11 @@ static int hls_slice_header(HEVCContext *s)
             sh->slice_cr_qp_offset = 0;
         }
 
+        if (s->pps->chroma_qp_offset_list_enabled_flag)
+            sh->cu_chroma_qp_offset_enabled_flag = get_bits1(gb);
+        else
+            sh->cu_chroma_qp_offset_enabled_flag = 0;
+
         if (s->pps->deblocking_filter_control_present_flag) {
             int deblocking_filter_override_flag = 0;
 
@@ -738,6 +743,8 @@ static int hls_slice_header(HEVCContext *s)
         s->HEVClc->qp_y = s->sh.slice_qp;
 
     s->slice_initialized = 1;
+    s->HEVClc->tu.cu_qp_offset_cb = 0;
+    s->HEVClc->tu.cu_qp_offset_cr = 0;
 
     return 0;
 }
@@ -830,6 +837,22 @@ static void hls_sao_param(HEVCContext *s, int rx, int ry)
 #undef SET_SAO
 #undef CTB
 
+static int hls_cross_component_pred(HEVCContext *s, int idx) {
+    HEVCLocalContext *lc    = s->HEVClc;
+    int log2_res_scale_abs_plus1 = ff_hevc_log2_res_scale_abs(s, idx);
+
+    if (log2_res_scale_abs_plus1 !=  0) {
+        int res_scale_sign_flag = ff_hevc_res_scale_sign_flag(s, idx);
+        lc->tu.res_scale_val = (1 << (log2_res_scale_abs_plus1 - 1)) *
+                               (1 - 2 * res_scale_sign_flag);
+    } else {
+        lc->tu.res_scale_val = 0;
+    }
+
+
+    return 0;
+}
+
 static int hls_transform_unit(HEVCContext *s, int x0, int y0,
                               int xBase, int yBase, int cb_xBase, int cb_yBase,
                               int log2_cb_size, int log2_trafo_size,
@@ -882,6 +905,25 @@ static int hls_transform_unit(HEVCContext *s, int x0, int y0,
             ff_hevc_set_qPy(s, x0, y0, cb_xBase, cb_yBase, log2_cb_size);
         }
 
+        if (s->sh.cu_chroma_qp_offset_enabled_flag && cbf_chroma &&
+            !lc->cu.cu_transquant_bypass_flag  &&  !lc->tu.is_cu_chroma_qp_offset_coded) {
+            int cu_chroma_qp_offset_flag = ff_hevc_cu_chroma_qp_offset_flag(s);
+            if (cu_chroma_qp_offset_flag) {
+                int cu_chroma_qp_offset_idx  = 0;
+                if (s->pps->chroma_qp_offset_list_len_minus1 > 0) {
+                    cu_chroma_qp_offset_idx = ff_hevc_cu_chroma_qp_offset_idx(s);
+                    av_log(s->avctx, AV_LOG_ERROR,
+                        "cu_chroma_qp_offset_idx not yet tested.\n");
+                }
+                lc->tu.cu_qp_offset_cb = s->pps->cb_qp_offset_list[cu_chroma_qp_offset_idx];
+                lc->tu.cu_qp_offset_cr = s->pps->cr_qp_offset_list[cu_chroma_qp_offset_idx];
+            } else {
+                lc->tu.cu_qp_offset_cb = 0;
+                lc->tu.cu_qp_offset_cr = 0;
+            }
+            lc->tu.is_cu_chroma_qp_offset_coded = 1;
+        }
+
         if (lc->cu.pred_mode == MODE_INTRA && log2_trafo_size < 4) {
             if (lc->tu.intra_pred_mode >= 6 &&
                 lc->tu.intra_pred_mode <= 14) {
@@ -900,13 +942,20 @@ static int hls_transform_unit(HEVCContext *s, int x0, int y0,
             }
         }
 
+        lc->tu.cross_pf = 0;
 
         if (cbf_luma)
             ff_hevc_hls_residual_coding(s, x0, y0, log2_trafo_size, scan_idx, 0);
         if (log2_trafo_size > 2 || s->sps->chroma_format_idc == 3) {
             int trafo_size_h = 1 << (log2_trafo_size_c + s->sps->hshift[1]);
             int trafo_size_v = 1 << (log2_trafo_size_c + s->sps->vshift[1]);
+            lc->tu.cross_pf  = (s->pps->cross_component_prediction_enabled_flag && cbf_luma &&
+                                (lc->cu.pred_mode == MODE_INTER ||
+                                 (lc->tu.chroma_mode_c ==  4)));
 
+            if (lc->tu.cross_pf) {
+                hls_cross_component_pred(s, 0);
+            }
             for (i = 0; i < (s->sps->chroma_format_idc == 2 ? 2 : 1); i++) {
                 if (lc->cu.pred_mode == MODE_INTRA) {
                     ff_hevc_set_neighbour_available(s, x0, y0 + (i << log2_trafo_size_c), trafo_size_h, trafo_size_v);
@@ -915,8 +964,27 @@ static int hls_transform_unit(HEVCContext *s, int x0, int y0,
                 if (SAMPLE_CBF(lc->tt.cbf_cb[trafo_depth], x0, y0 + (i << log2_trafo_size_c)))
                     ff_hevc_hls_residual_coding(s, x0, y0 + (i << log2_trafo_size_c),
                                                 log2_trafo_size_c, scan_idx_c, 1);
+                else
+                    if (lc->tu.cross_pf) {
+                        ptrdiff_t stride = s->frame->linesize[1];
+                        int hshift = s->sps->hshift[1];
+                        int vshift = s->sps->vshift[1];
+                        int16_t *coeffs_y = lc->tu.coeffs[0];
+                        int16_t *coeffs =   lc->tu.coeffs[1];
+                        int size = 1 << log2_trafo_size_c;
+
+                        uint8_t *dst = &s->frame->data[1][(y0 >> vshift) * stride +
+                                                              ((x0 >> hshift) << s->sps->pixel_shift)];
+                        for (i = 0; i < (size * size); i++) {
+                            coeffs[i] = ((lc->tu.res_scale_val * coeffs_y[i]) >> 3);
+                        }
+                        s->hevcdsp.transform_add[log2_trafo_size-2](dst, coeffs, stride);
+                    }
             }
 
+            if (lc->tu.cross_pf) {
+                hls_cross_component_pred(s, 1);
+            }
             for (i = 0; i < (s->sps->chroma_format_idc == 2 ? 2 : 1); i++) {
                 if (lc->cu.pred_mode == MODE_INTRA) {
                     ff_hevc_set_neighbour_available(s, x0, y0 + (i << log2_trafo_size_c), trafo_size_h, trafo_size_v);
@@ -925,6 +993,22 @@ static int hls_transform_unit(HEVCContext *s, int x0, int y0,
                 if (SAMPLE_CBF(lc->tt.cbf_cr[trafo_depth], x0, y0 + (i << log2_trafo_size_c)))
                     ff_hevc_hls_residual_coding(s, x0, y0 + (i << log2_trafo_size_c),
                                                 log2_trafo_size_c, scan_idx_c, 2);
+                else
+                    if (lc->tu.cross_pf) {
+                        ptrdiff_t stride = s->frame->linesize[2];
+                        int hshift = s->sps->hshift[2];
+                        int vshift = s->sps->vshift[2];
+                        int16_t *coeffs_y = lc->tu.coeffs[0];
+                        int16_t *coeffs =   lc->tu.coeffs[1];
+                        int size = 1 << log2_trafo_size_c;
+
+                        uint8_t *dst = &s->frame->data[2][(y0 >> vshift) * stride +
+                                                          ((x0 >> hshift) << s->sps->pixel_shift)];
+                        for (i = 0; i < (size * size); i++) {
+                            coeffs[i] = ((lc->tu.res_scale_val * coeffs_y[i]) >> 3);
+                        }
+                        s->hevcdsp.transform_add[log2_trafo_size-2](dst, coeffs, stride);
+                    }
             }
         } else if (blk_idx == 3) {
             int trafo_size_h = 1 << (log2_trafo_size + 1);
@@ -2070,6 +2154,11 @@ static int hls_coding_quadtree(HEVCContext *s, int x0, int y0,
         lc->tu.cu_qp_delta          = 0;
     }
 
+    if (s->sh.cu_chroma_qp_offset_enabled_flag &&
+        log2_cb_size >= s->sps->log2_ctb_size - s->pps->diff_cu_chroma_qp_offset_depth) {
+        lc->tu.is_cu_chroma_qp_offset_coded = 0;
+    }
+
     if (split_cu_flag) {
         const int cb_size_split = cb_size >> 1;
         const int x1 = x0 + cb_size_split;

libavcodec/hevc.h

@@ -602,6 +602,8 @@ typedef struct SliceHeader {
     int slice_cb_qp_offset;
     int slice_cr_qp_offset;
 
+    uint8_t cu_chroma_qp_offset_enabled_flag;
+
     int beta_offset;    ///< beta_offset_div2 * 2
     int tc_offset;      ///< tc_offset_div2 * 2
 
@@ -692,13 +694,20 @@ typedef struct TransformTree {
 } TransformTree;
 
 typedef struct TransformUnit {
+    DECLARE_ALIGNED(32, int16_t, coeffs[2][MAX_TB_SIZE * MAX_TB_SIZE]);
     int cu_qp_delta;
 
+    int res_scale_val;
+
     // Inferred parameters;
     int intra_pred_mode;
     int intra_pred_mode_c;
     int chroma_mode_c;
     uint8_t is_cu_qp_delta_coded;
+    uint8_t is_cu_chroma_qp_offset_coded;
+    int8_t  cu_qp_offset_cb;
+    int8_t  cu_qp_offset_cr;
+    uint8_t cross_pf;
 } TransformUnit;
 
 typedef struct DBParams {
@@ -750,6 +759,8 @@ typedef struct HEVCLocalContext {
     DECLARE_ALIGNED(16, int16_t, mc_buffer[(MAX_PB_SIZE + 7) * MAX_PB_SIZE]);
     uint8_t cabac_state[HEVC_CONTEXTS];
 
+    uint8_t stat_coeff[4];
+
     uint8_t first_qp_group;
 
     GetBitContext gb;
@@ -988,6 +999,8 @@ int ff_hevc_no_residual_syntax_flag_decode(HEVCContext *s);
 int ff_hevc_split_transform_flag_decode(HEVCContext *s, int log2_trafo_size);
 int ff_hevc_cbf_cb_cr_decode(HEVCContext *s, int trafo_depth);
 int ff_hevc_cbf_luma_decode(HEVCContext *s, int trafo_depth);
+int ff_hevc_log2_res_scale_abs(HEVCContext *s, int idx);
+int ff_hevc_res_scale_sign_flag(HEVCContext *s, int idx);
 
 /**
  * Get the number of candidate references for the current frame.
@@ -1019,6 +1032,8 @@ void ff_hevc_deblocking_boundary_strengths(HEVCContext *s, int x0, int y0,
                                            int log2_trafo_size);
 int ff_hevc_cu_qp_delta_sign_flag(HEVCContext *s);
 int ff_hevc_cu_qp_delta_abs(HEVCContext *s);
+int ff_hevc_cu_chroma_qp_offset_flag(HEVCContext *s);
+int ff_hevc_cu_chroma_qp_offset_idx(HEVCContext *s);
 void ff_hevc_hls_filter(HEVCContext *s, int x, int y, int ctb_size);
 void ff_hevc_hls_filters(HEVCContext *s, int x_ctb, int y_ctb, int ctb_size);
 void ff_hevc_hls_residual_coding(HEVCContext *s, int x0, int y0,

libavcodec/hevcdsp.c

@@ -200,6 +200,7 @@ void ff_hevc_dsp_init(HEVCDSPContext *hevcdsp, int bit_depth)
     hevcdsp->transform_add[2]       = FUNC(transform_add16x16, depth);      \
     hevcdsp->transform_add[3]       = FUNC(transform_add32x32, depth);      \
     hevcdsp->transform_skip         = FUNC(transform_skip, depth);          \
+    hevcdsp->transform_rdpcm        = FUNC(transform_rdpcm, depth);         \
     hevcdsp->idct_4x4_luma          = FUNC(transform_4x4_luma, depth);      \
     hevcdsp->idct[0]                = FUNC(idct_4x4, depth);                \
     hevcdsp->idct[1]                = FUNC(idct_8x8, depth);                \

libavcodec/hevcdsp.h

@@ -48,6 +48,8 @@ typedef struct HEVCDSPContext {
 
     void (*transform_skip)(int16_t *coeffs, int16_t log2_size);
 
+    void (*transform_rdpcm)(int16_t *coeffs, int16_t log2_size, int mode);
+
     void (*idct_4x4_luma)(int16_t *coeffs);
 
     void (*idct[4])(int16_t *coeffs, int col_limit);

libavcodec/hevcdsp_template.c

@@ -110,6 +110,29 @@ static void FUNC(transform_add32x32)(uint8_t *_dst, int16_t *coeffs,
     }
 }
 
+
+static void FUNC(transform_rdpcm)(int16_t *_coeffs, int16_t log2_size, int mode)
+{
+    int16_t *coeffs = (int16_t *) _coeffs;
+    int x, y;
+    int size = 1 << log2_size;
+
+    if (mode) {
+        coeffs += size;
+        for (y = 0; y < size - 1; y++) {
+            for (x = 0; x < size; x++)
+                coeffs[x] += coeffs[x - size];
+            coeffs += size;
+        }
+    } else {
+        for (y = 0; y < size; y++) {
+            for (x = 1; x < size; x++)
+                coeffs[x] += coeffs[x - 1];
+            coeffs += size;
+        }
+    }
+}
+
 static void FUNC(transform_skip)(int16_t *_coeffs, int16_t log2_size)
 {
     int shift  = 15 - BIT_DEPTH - log2_size;