hevc/rext: basic infrastructure for supporting range extension

- support for 4:2:2 and 4:4:4 up to 12 bits
- add a new profile for range extension
(cherry picked from commit d3c067fa65bbc871758d28aa07f54123430ca346)
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

libavcodec/avcodec.h

@@ -2866,6 +2866,7 @@ typedef struct AVCodecContext {
 #define FF_PROFILE_HEVC_MAIN                        1
 #define FF_PROFILE_HEVC_MAIN_10                     2
 #define FF_PROFILE_HEVC_MAIN_STILL_PICTURE          3
+#define FF_PROFILE_HEVC_REXT                        4
 
     /**
      * level

libavcodec/hevc.h

@@ -83,6 +83,7 @@
 #define SAMPLE(tab, x, y) ((tab)[(y) * s->sps->width + (x)])
 #define SAMPLE_CTB(tab, x, y) ((tab)[(y) * min_cb_width + (x)])
 #define SAMPLE_CBF(tab, x, y) ((tab)[((y) & ((1<<log2_trafo_size)-1)) * MAX_CU_SIZE + ((x) & ((1<<log2_trafo_size)-1))])
+#define SAMPLE_CBF2(tab, x, y) ((tab)[(y) * MAX_CU_SIZE +  (x)])
 
 #define IS_IDR(s) ((s)->nal_unit_type == NAL_IDR_W_RADL || (s)->nal_unit_type == NAL_IDR_N_LP)
 #define IS_BLA(s) ((s)->nal_unit_type == NAL_BLA_W_RADL || (s)->nal_unit_type == NAL_BLA_W_LP || \
@@ -457,6 +458,13 @@ typedef struct HEVCSPS {
     int max_transform_hierarchy_depth_inter;
     int max_transform_hierarchy_depth_intra;
 
+    int transform_skip_rotation_enabled_flag;
+    int transform_skip_context_enabled_flag;
+    int implicit_rdpcm_enabled_flag;
+    int explicit_rdpcm_enabled_flag;
+    int intra_smoothing_disabled_flag;
+    int persistent_rice_adaptation_enabled_flag;
+
     ///< coded frame dimension in various units
     int width;
     int height;
@@ -526,6 +534,15 @@ typedef struct HEVCPPS {
     int log2_parallel_merge_level; ///< log2_parallel_merge_level_minus2 + 2
     int num_extra_slice_header_bits;
     uint8_t slice_header_extension_present_flag;
+    uint8_t log2_max_transform_skip_block_size;
+    uint8_t cross_component_prediction_enabled_flag;
+    uint8_t chroma_qp_offset_list_enabled_flag;
+    uint8_t diff_cu_chroma_qp_offset_depth;
+    uint8_t chroma_qp_offset_list_len_minus1;
+    int8_t  cb_qp_offset_list[5];
+    int8_t  cr_qp_offset_list[5];
+    uint8_t log2_sao_offset_scale_luma;
+    uint8_t log2_sao_offset_scale_chroma;
 
     // Inferred parameters
     unsigned int *column_width;  ///< ColumnWidth
@@ -661,7 +678,8 @@ typedef struct PredictionUnit {
     uint8_t intra_pred_mode[4];
     Mv mvd;
     uint8_t merge_flag;
-    uint8_t intra_pred_mode_c;
+    uint8_t intra_pred_mode_c[4];
+    uint8_t chroma_mode_c[4];
 } PredictionUnit;
 
 typedef struct TransformTree {
@@ -677,7 +695,9 @@ typedef struct TransformUnit {
     int cu_qp_delta;
 
     // Inferred parameters;
-    int cur_intra_pred_mode;
+    int intra_pred_mode;
+    int intra_pred_mode_c;
+    int chroma_mode_c;
     uint8_t is_cu_qp_delta_coded;
 } TransformUnit;
 
@@ -968,7 +988,6 @@ 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_transform_skip_flag_decode(HEVCContext *s, int c_idx);
 
 /**
  * Get the number of candidate references for the current frame.

libavcodec/hevc_cabac.c

@@ -894,7 +894,7 @@ int ff_hevc_cbf_luma_decode(HEVCContext *s, int trafo_depth)
     return GET_CABAC(elem_offset[CBF_LUMA] + !trafo_depth);
 }
 
-int ff_hevc_transform_skip_flag_decode(HEVCContext *s, int c_idx)
+static int ff_hevc_transform_skip_flag_decode(HEVCContext *s, int c_idx)
 {
     return GET_CABAC(elem_offset[TRANSFORM_SKIP_FLAG] + !!c_idx);
 }
@@ -1093,16 +1093,18 @@ void ff_hevc_hls_residual_coding(HEVCContext *s, int x0, int y0,
     // Derive QP for dequant
     if (!lc->cu.cu_transquant_bypass_flag) {
         static const int qp_c[] = { 29, 30, 31, 32, 33, 33, 34, 34, 35, 35, 36, 36, 37, 37 };
-        static const uint8_t rem6[51 + 2 * 6 + 1] = {
+        static const uint8_t rem6[51 + 4 * 6 + 1] = {
             0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2,
             3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5,
             0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3,
+            4, 5, 0, 1, 2, 3, 4, 5, 0, 1
         };
 
-        static const uint8_t div6[51 + 2 * 6 + 1] = {
+        static const uint8_t div6[51 + 4 * 6 + 1] = {
             0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 3,  3,  3,
             3, 3, 3, 4, 4, 4, 4, 4, 4, 5, 5, 5, 5, 5, 5, 6, 6, 6, 6,  6,  6,
             7, 7, 7, 7, 7, 7, 8, 8, 8, 8, 8, 8, 9, 9, 9, 9, 9, 9, 10, 10, 10, 10,
+            10, 10, 11, 11, 11, 11, 11, 11, 12, 12
         };
         int qp_y = lc->qp_y;
 

libavcodec/hevc_filter.c

@@ -61,12 +61,16 @@ static int chroma_tc(HEVCContext *s, int qp_y, int c_idx, int tc_offset)
         offset = s->pps->cr_qp_offset;
 
     qp_i = av_clip(qp_y + offset, 0, 57);
-    if (qp_i < 30)
-        qp = qp_i;
-    else if (qp_i > 43)
-        qp = qp_i - 6;
-    else
-        qp = qp_c[qp_i - 30];
+    if (s->sps->chroma_format_idc == 1) {
+        if (qp_i < 30)
+            qp = qp_i;
+        else if (qp_i > 43)
+            qp = qp_i - 6;
+        else
+            qp = qp_c[qp_i - 30];
+    } else {
+        qp = av_clip(qp_i, 0, 51);
+    }
 
     idxt = av_clip(qp + DEFAULT_INTRA_TC_OFFSET + tc_offset, 0, 53);
     return tctable[idxt];
@@ -327,6 +331,9 @@ static void deblocking_filter_CTB(HEVCContext *s, int x0, int y0)
     if (x0) {
         left_tc_offset   = s->deblock[ctb - 1].tc_offset;
         left_beta_offset = s->deblock[ctb - 1].beta_offset;
+    } else {
+        left_tc_offset   = 0;
+        left_beta_offset = 0;
     }
 
     x_end = x0 + ctb_size;
@@ -371,22 +378,25 @@ static void deblocking_filter_CTB(HEVCContext *s, int x0, int y0)
 
     // vertical filtering chroma
     for (chroma = 1; chroma <= 2; chroma++) {
-        for (y = y0; y < y_end; y += 16) {
-            for (x = x0 ? x0 : 16; x < x_end; x += 16) {
-                const int bs0 = s->vertical_bs[(x >> 3) + (y       >> 2) * s->bs_width];
-                const int bs1 = s->vertical_bs[(x >> 3) + ((y + 8) >> 2) * s->bs_width];
+        int h = 1 << s->sps->hshift[chroma];
+        int v = 1 << s->sps->vshift[chroma];
+        for (y = y0; y < y_end; y += (8 * v)) {
+            for (x = x0 ? x0 : 8 * h; x < x_end; x += (8 * h)) {
+                const int bs0 = s->vertical_bs[(x >> 3) + (y             >> 2) * s->bs_width];
+                const int bs1 = s->vertical_bs[(x >> 3) + ((y + (4 * v)) >> 2) * s->bs_width];
+
                 if ((bs0 == 2) || (bs1 == 2)) {
-                    const int qp0 = (get_qPy(s, x - 1, y)     + get_qPy(s, x, y)     + 1) >> 1;
-                    const int qp1 = (get_qPy(s, x - 1, y + 8) + get_qPy(s, x, y + 8) + 1) >> 1;
+                    const int qp0 = (get_qPy(s, x - 1, y)           + get_qPy(s, x, y)           + 1) >> 1;
+                    const int qp1 = (get_qPy(s, x - 1, y + (4 * v)) + get_qPy(s, x, y + (4 * v)) + 1) >> 1;
 
                     c_tc[0] = (bs0 == 2) ? chroma_tc(s, qp0, chroma, tc_offset) : 0;
                     c_tc[1] = (bs1 == 2) ? chroma_tc(s, qp1, chroma, tc_offset) : 0;
-                    src     = &s->frame->data[chroma][y / 2 * s->frame->linesize[chroma] + ((x / 2) << s->sps->pixel_shift)];
+                    src       = &s->frame->data[chroma][(y >> s->sps->vshift[chroma]) * s->frame->linesize[chroma] + ((x >> s->sps->hshift[chroma]) << s->sps->pixel_shift)];
                     if (pcmf) {
                         no_p[0] = get_pcm(s, x - 1, y);
-                        no_p[1] = get_pcm(s, x - 1, y + 8);
+                        no_p[1] = get_pcm(s, x - 1, y + (4 * v));
                         no_q[0] = get_pcm(s, x, y);
-                        no_q[1] = get_pcm(s, x, y + 8);
+                        no_q[1] = get_pcm(s, x, y + (4 * v));
                         s->hevcdsp.hevc_v_loop_filter_chroma_c(src,
                                                                s->frame->linesize[chroma],
                                                                c_tc, no_p, no_q);
@@ -436,35 +446,37 @@ static void deblocking_filter_CTB(HEVCContext *s, int x0, int y0)
 
     // horizontal filtering chroma
     for (chroma = 1; chroma <= 2; chroma++) {
-        for (y = y0 ? y0 : 16; y < y_end; y += 16) {
-            for (x = x0 - 8; x < x_end; x += 16) {
+        int h = 1 << s->sps->hshift[chroma];
+        int v = 1 << s->sps->vshift[chroma];
+        for (y = y0 ? y0 : 8 * v; y < y_end; y +=  (8 * v)) {
+            for (x = x0 - 8; x < x_end; x += (8 * h)) {
                 int bs0, bs1;
                 // to make sure no memory access over boundary when x = -8
                 // TODO: simplify with row based deblocking
                 if (x < 0) {
                     bs0 = 0;
-                    bs1 = s->horizontal_bs[(x + 8 + y * s->bs_width) >> 2];
-                } else if (x >= x_end - 8) {
-                    bs0 = s->horizontal_bs[(x +     y * s->bs_width) >> 2];
+                    bs1 = s->horizontal_bs[(x + (4 * h) + y * s->bs_width) >> 2];
+                } else if (x >= x_end - 4 * h) {
+                    bs0 = s->horizontal_bs[(x +           y * s->bs_width) >> 2];
                     bs1 = 0;
                 } else {
-                    bs0 = s->horizontal_bs[(x + y     * s->bs_width) >> 2];
-                    bs1 = s->horizontal_bs[(x + 8 + y * s->bs_width) >> 2];
+                    bs0 = s->horizontal_bs[(x           + y * s->bs_width) >> 2];
+                    bs1 = s->horizontal_bs[(x + (4 * h) + y * s->bs_width) >> 2];
                 }
 
                 if ((bs0 == 2) || (bs1 == 2)) {
-                    const int qp0 = bs0 == 2 ? (get_qPy(s, x,     y - 1) + get_qPy(s, x,     y) + 1) >> 1 : 0;
-                    const int qp1 = bs1 == 2 ? (get_qPy(s, x + 8, y - 1) + get_qPy(s, x + 8, y) + 1) >> 1 : 0;
+                    const int qp0 = bs0 == 2 ? (get_qPy(s, x,           y - 1) + get_qPy(s, x,           y) + 1) >> 1 : 0;
+                    const int qp1 = bs1 == 2 ? (get_qPy(s, x + (4 * h), y - 1) + get_qPy(s, x + (4 * h), y) + 1) >> 1 : 0;
 
                     tc_offset = x >= x0 ? cur_tc_offset : left_tc_offset;
                     c_tc[0]   = bs0 == 2 ? chroma_tc(s, qp0, chroma, tc_offset)     : 0;
                     c_tc[1]   = bs1 == 2 ? chroma_tc(s, qp1, chroma, cur_tc_offset) : 0;
-                    src       = &s->frame->data[chroma][y / 2 * s->frame->linesize[chroma] + ((x / 2) << s->sps->pixel_shift)];
+                    src       = &s->frame->data[chroma][(y >> s->sps->vshift[1]) * s->frame->linesize[chroma] + ((x >> s->sps->hshift[1]) << s->sps->pixel_shift)];
                     if (pcmf) {
-                        no_p[0] = get_pcm(s, x, y - 1);
-                        no_p[1] = get_pcm(s, x + 8, y - 1);
-                        no_q[0] = get_pcm(s, x, y);
-                        no_q[1] = get_pcm(s, x + 8, y);
+                        no_p[0] = get_pcm(s, x,           y - 1);
+                        no_p[1] = get_pcm(s, x + (4 * h), y - 1);
+                        no_q[0] = get_pcm(s, x,           y);
+                        no_q[1] = get_pcm(s, x + (4 * h), y);
                         s->hevcdsp.hevc_h_loop_filter_chroma_c(src,
                                                                s->frame->linesize[chroma],
                                                                c_tc, no_p, no_q);

libavcodec/hevc_ps.c

@@ -207,6 +207,8 @@ static void decode_profile_tier_level(HEVCContext *s, PTLCommon *ptl)
         av_log(s->avctx, AV_LOG_DEBUG, "Main 10 profile bitstream\n");
     else if (ptl->profile_idc == FF_PROFILE_HEVC_MAIN_STILL_PICTURE)
         av_log(s->avctx, AV_LOG_DEBUG, "Main Still Picture profile bitstream\n");
+    else if (ptl->profile_idc == FF_PROFILE_HEVC_REXT)
+        av_log(s->avctx, AV_LOG_DEBUG, "Range Extension profile bitstream\n");
     else
         av_log(s->avctx, AV_LOG_WARNING, "Unknown HEVC profile: %d\n", ptl->profile_idc);
 
@@ -668,8 +670,8 @@ int ff_hevc_decode_nal_sps(HEVCContext *s)
     }
 
     sps->chroma_format_idc = get_ue_golomb_long(gb);
-    if (sps->chroma_format_idc != 1) {
-        avpriv_report_missing_feature(s->avctx, "chroma_format_idc != 1\n");
+    if (!(sps->chroma_format_idc == 1 || sps->chroma_format_idc == 2 || sps->chroma_format_idc == 3)) {
+        avpriv_report_missing_feature(s->avctx, "chroma_format_idc != {1, 2, 3}\n");
         ret = AVERROR_PATCHWELCOME;
         goto err;
     }
@@ -677,6 +679,9 @@ int ff_hevc_decode_nal_sps(HEVCContext *s)
     if (sps->chroma_format_idc == 3)
         sps->separate_colour_plane_flag = get_bits1(gb);
 
+    if (sps->separate_colour_plane_flag)
+        sps->chroma_format_idc = 0;
+
     sps->width  = get_ue_golomb_long(gb);
     sps->height = get_ue_golomb_long(gb);
     if ((ret = av_image_check_size(sps->width,
@@ -718,20 +723,30 @@ int ff_hevc_decode_nal_sps(HEVCContext *s)
         goto err;
     }
 
-    if (sps->chroma_format_idc == 1) {
-        switch (sps->bit_depth) {
-        case 8:  sps->pix_fmt = AV_PIX_FMT_YUV420P;   break;
-        case 9:  sps->pix_fmt = AV_PIX_FMT_YUV420P9;  break;
-        case 10: sps->pix_fmt = AV_PIX_FMT_YUV420P10; break;
-        default:
-            av_log(s->avctx, AV_LOG_ERROR, "Unsupported bit depth: %d\n",
-                   sps->bit_depth);
-            ret = AVERROR_PATCHWELCOME;
-            goto err;
-        }
-    } else {
+    switch (sps->bit_depth) {
+    case 8:
+        if (sps->chroma_format_idc == 1) sps->pix_fmt = AV_PIX_FMT_YUV420P;
+        if (sps->chroma_format_idc == 2) sps->pix_fmt = AV_PIX_FMT_YUV422P;
+        if (sps->chroma_format_idc == 3) sps->pix_fmt = AV_PIX_FMT_YUV444P;
+       break;
+    case 9:
+        if (sps->chroma_format_idc == 1) sps->pix_fmt = AV_PIX_FMT_YUV420P9;
+        if (sps->chroma_format_idc == 2) sps->pix_fmt = AV_PIX_FMT_YUV422P9;
+        if (sps->chroma_format_idc == 3) sps->pix_fmt = AV_PIX_FMT_YUV444P9;
+        break;
+    case 10:
+        if (sps->chroma_format_idc == 1) sps->pix_fmt = AV_PIX_FMT_YUV420P10;
+        if (sps->chroma_format_idc == 2) sps->pix_fmt = AV_PIX_FMT_YUV422P10;
+        if (sps->chroma_format_idc == 3) sps->pix_fmt = AV_PIX_FMT_YUV444P10;
+        break;
+    case 12:
+        if (sps->chroma_format_idc == 1) sps->pix_fmt = AV_PIX_FMT_YUV420P12;
+        if (sps->chroma_format_idc == 2) sps->pix_fmt = AV_PIX_FMT_YUV422P12;
+        if (sps->chroma_format_idc == 3) sps->pix_fmt = AV_PIX_FMT_YUV444P12;
+        break;
+    default:
         av_log(s->avctx, AV_LOG_ERROR,
-               "non-4:2:0 support is currently unspecified.\n");
+               "4:2:0, 4:2:2, 4:4:4 supports are currently specified for 8, 10 and 12 bits.\n");
         return AVERROR_PATCHWELCOME;
     }
 
@@ -862,8 +877,42 @@ int ff_hevc_decode_nal_sps(HEVCContext *s)
     vui_present = get_bits1(gb);
     if (vui_present)
         decode_vui(s, sps);
-    skip_bits1(gb); // sps_extension_flag
 
+    if (get_bits1(gb)) { // sps_extension_flag
+        int sps_extension_flag[1];
+        for (i = 0; i < 1; i++)
+            sps_extension_flag[i] = get_bits1(gb);
+        skip_bits(gb, 7); //sps_extension_7bits = get_bits(gb, 7);
+        if (sps_extension_flag[0]) {
+            int extended_precision_processing_flag;
+            int high_precision_offsets_enabled_flag;
+            int cabac_bypass_alignment_enabled_flag;
+
+            sps->transform_skip_rotation_enabled_flag = get_bits1(gb);
+            sps->transform_skip_context_enabled_flag  = get_bits1(gb);
+            sps->implicit_rdpcm_enabled_flag = get_bits1(gb);
+
+            sps->explicit_rdpcm_enabled_flag = get_bits1(gb);
+
+            extended_precision_processing_flag = get_bits1(gb);
+            if (extended_precision_processing_flag)
+                av_log(s->avctx, AV_LOG_WARNING,
+                   "extended_precision_processing_flag not yet implemented\n");
+
+            sps->intra_smoothing_disabled_flag       = get_bits1(gb);
+            high_precision_offsets_enabled_flag  = get_bits1(gb);
+            if (high_precision_offsets_enabled_flag)
+                av_log(s->avctx, AV_LOG_WARNING,
+                   "high_precision_offsets_enabled_flag not yet implemented\n");
+
+            sps->persistent_rice_adaptation_enabled_flag = get_bits1(gb);
+
+            cabac_bypass_alignment_enabled_flag  = get_bits1(gb);
+            if (cabac_bypass_alignment_enabled_flag)
+                av_log(s->avctx, AV_LOG_WARNING,
+                   "cabac_bypass_alignment_enabled_flag not yet implemented\n");
+        }
+    }
     if (s->apply_defdispwin) {
         sps->output_window.left_offset   += sps->vui.def_disp_win.left_offset;
         sps->output_window.right_offset  += sps->vui.def_disp_win.right_offset;
@@ -997,6 +1046,46 @@ static void hevc_pps_free(void *opaque, uint8_t *data)
     av_freep(&pps);
 }
 
+static int pps_range_extensions(HEVCContext *s, HEVCPPS *pps, HEVCSPS *sps) {
+    GetBitContext *gb = &s->HEVClc->gb;
+    int i;
+
+    if (pps->transform_skip_enabled_flag) {
+        pps->log2_max_transform_skip_block_size = get_ue_golomb_long(gb) + 2;
+    }
+    pps->cross_component_prediction_enabled_flag = get_bits1(gb);
+    if (pps->cross_component_prediction_enabled_flag) {
+        av_log(s->avctx, AV_LOG_WARNING,
+               "cross_component_prediction_enabled_flag is not yet implemented.\n");
+    }
+    pps->chroma_qp_offset_list_enabled_flag = get_bits1(gb);
+    if (pps->chroma_qp_offset_list_enabled_flag) {
+        pps->diff_cu_chroma_qp_offset_depth = get_ue_golomb_long(gb);
+        pps->chroma_qp_offset_list_len_minus1 = get_ue_golomb_long(gb);
+        if (pps->chroma_qp_offset_list_len_minus1 && pps->chroma_qp_offset_list_len_minus1 >= 5) {
+            av_log(s->avctx, AV_LOG_ERROR,
+                   "chroma_qp_offset_list_len_minus1 shall be in the range [0, 5].\n");
+            return AVERROR_INVALIDDATA;
+        }
+        for (i = 0; i <= pps->chroma_qp_offset_list_len_minus1; i++) {
+            pps->cb_qp_offset_list[i] = get_se_golomb_long(gb);
+            if (pps->cb_qp_offset_list[i]) {
+                av_log(s->avctx, AV_LOG_WARNING,
+                       "cb_qp_offset_list not tested yet.\n");
+            }
+            pps->cr_qp_offset_list[i] = get_se_golomb_long(gb);
+            if (pps->cr_qp_offset_list[i]) {
+                av_log(s->avctx, AV_LOG_WARNING,
+                       "cb_qp_offset_list not tested yet.\n");
+            }
+        }
+    }
+    pps->log2_sao_offset_scale_luma = get_ue_golomb_long(gb);
+    pps->log2_sao_offset_scale_chroma = get_ue_golomb_long(gb);
+
+    return(0);
+}
+
 int ff_hevc_decode_nal_pps(HEVCContext *s)
 {
     GetBitContext *gb = &s->HEVClc->gb;
@@ -1030,6 +1119,7 @@ int ff_hevc_decode_nal_pps(HEVCContext *s)
     pps->disable_dbf                           = 0;
     pps->beta_offset                           = 0;
     pps->tc_offset                             = 0;
+    pps->log2_max_transform_skip_block_size    = 2;
 
     // Coded parameters
     pps_id = get_ue_golomb_long(gb);
@@ -1190,7 +1280,16 @@ int ff_hevc_decode_nal_pps(HEVCContext *s)
     }
 
     pps->slice_header_extension_present_flag = get_bits1(gb);
-    skip_bits1(gb);     // pps_extension_flag
+
+    if (get_bits1(gb)) { // pps_extension_present_flag
+        int pps_range_extensions_flag = get_bits1(gb);
+        /* int pps_extension_7bits = */ get_bits(gb, 7);
+        if (sps->ptl.general_ptl.profile_idc == FF_PROFILE_HEVC_REXT && pps_range_extensions_flag) {
+            av_log(s->avctx, AV_LOG_ERROR,
+                   "PPS extension flag is partially implemented.\n");
+            pps_range_extensions(s, pps, sps);
+        }
+    }
 
     // Inferred parameters
     pps->col_bd   = av_malloc_array(pps->num_tile_columns + 1, sizeof(*pps->col_bd));

libavcodec/hevcdsp.c

@@ -119,6 +119,10 @@ DECLARE_ALIGNED(16, const int8_t, ff_hevc_qpel_filters[3][16]) = {
 #include "hevcdsp_template.c"
 #undef BIT_DEPTH
 
+#define BIT_DEPTH 12
+#include "hevcdsp_template.c"
+#undef BIT_DEPTH
+
 void ff_hevc_dsp_init(HEVCDSPContext *hevcdsp, int bit_depth)
 {
 #undef FUNC
@@ -235,6 +239,9 @@ int i = 0;
     case 10:
         HEVC_DSP(10);
         break;
+    case 12:
+        HEVC_DSP(12);
+        break;
     default:
         HEVC_DSP(8);
         break;

libavcodec/hevcdsp.h

@@ -41,8 +41,8 @@ typedef struct SAOParams {
 } SAOParams;
 
 typedef struct HEVCDSPContext {
-    void (*put_pcm)(uint8_t *dst, ptrdiff_t stride, int size,
-                    GetBitContext *gb, int pcm_bit_depth);
+    void (*put_pcm)(uint8_t *_dst, ptrdiff_t _stride, int width, int height,
+                    struct GetBitContext *gb, int pcm_bit_depth);
 
     void (*transform_add[4])(uint8_t *_dst, int16_t *coeffs, ptrdiff_t _stride);
 

libavcodec/hevcdsp_template.c

@@ -27,7 +27,7 @@
 #include "hevcdsp.h"
 
 
-static void FUNC(put_pcm)(uint8_t *_dst, ptrdiff_t stride, int size,
+static void FUNC(put_pcm)(uint8_t *_dst, ptrdiff_t stride, int width, int height,
                           GetBitContext *gb, int pcm_bit_depth)
 {
     int x, y;
@@ -35,8 +35,8 @@ static void FUNC(put_pcm)(uint8_t *_dst, ptrdiff_t stride, int size,
 
     stride /= sizeof(pixel);
 
-    for (y = 0; y < size; y++) {
-        for (x = 0; x < size; x++)
+    for (y = 0; y < height; y++) {
+        for (x = 0; x < width; x++)
             dst[x] = get_bits(gb, pcm_bit_depth) << (BIT_DEPTH - pcm_bit_depth);
         dst += stride;
     }

libavcodec/hevcpred.c

@@ -36,6 +36,10 @@
 #include "hevcpred_template.c"
 #undef BIT_DEPTH
 
+#define BIT_DEPTH 12
+#include "hevcpred_template.c"
+#undef BIT_DEPTH
+
 void ff_hevc_pred_init(HEVCPredContext *hpc, int bit_depth)
 {
 #undef FUNC
@@ -63,6 +67,9 @@ void ff_hevc_pred_init(HEVCPredContext *hpc, int bit_depth)
     case 10:
         HEVC_PRED(10);
         break;
+    case 12:
+        HEVC_PRED(12);
+        break;
     default:
         HEVC_PRED(8);
         break;

libavcodec/hevcpred_template.c

@@ -91,8 +91,8 @@ do {                                  \
 
     int min_pu_width = s->sps->min_pu_width;
 
-    enum IntraPredMode mode = c_idx ? lc->pu.intra_pred_mode_c :
-                              lc->tu.cur_intra_pred_mode;
+    enum IntraPredMode mode = c_idx ? lc->tu.intra_pred_mode_c :
+                              lc->tu.intra_pred_mode;
     pixel4 a;
     pixel  left_array[2 * MAX_TB_SIZE + 1];
     pixel  filtered_left_array[2 * MAX_TB_SIZE + 1];
@@ -207,7 +207,6 @@ do {                                  \
                         j++;
                     EXTEND_LEFT_CIP(top, j, j + 1);
                     left[-1] = top[-1];
-                    j        = 0;
                 }
             } else {
                 j = 0;
@@ -221,7 +220,6 @@ do {                                  \
                         top[-1] = top[0];
                     }
                 left[-1] = top[-1];
-                j        = 0;
             }
             left[-1] = top[-1];
             if (cand_bottom_left || cand_left) {
@@ -238,7 +236,6 @@ do {                                  \
                 if (!IS_INTRA(-1, - 1))
                     left[-1] = left[0];
             } else if (x0 == 0) {
-                a = PIXEL_SPLAT_X4(left[size_max_y - 1]);
                 EXTEND(left, 0, size_max_y);
             } else {
                 a = PIXEL_SPLAT_X4(left[size_max_y - 1]);
@@ -290,7 +287,7 @@ do {                                  \
     top[-1] = left[-1];
 
     // Filtering process
-    if (c_idx == 0) {
+    if (!s->sps->intra_smoothing_disabled_flag && (c_idx == 0  || s->sps->chroma_format_idc == 3)) {
         if (mode != INTRA_DC && size != 4){
             int intra_hor_ver_dist_thresh[] = { 7, 1, 0 };
             int min_dist_vert_hor = FFMIN(FFABS((int)(mode - 26U)),