Merge "API extensions and sample app for spacial scalable encoder"

examples.mk

@@ -49,6 +49,9 @@ vpxenc.DESCRIPTION           = Full featured encoder
 UTILS-$(CONFIG_VP8_ENCODER)    += vp8_scalable_patterns.c
 vp8_scalable_patterns.GUID   = 0D6A210B-F482-4D6F-8570-4A9C01ACC88C
 vp8_scalable_patterns.DESCRIPTION = Temporal Scalability Encoder
+UTILS-$(CONFIG_VP8_ENCODER)    += vp9_spatial_scalable_encoder.c
+vp8_scalable_patterns.GUID   = 4A38598D-627D-4505-9C7B-D4020C84100D
+vp8_scalable_patterns.DESCRIPTION = Spatial Scalable Encoder
 
 # Clean up old ivfenc, ivfdec binaries.
 ifeq ($(CONFIG_MSVS),yes)

vp8/vp8_cx_iface.c

@@ -1269,7 +1269,7 @@ static vpx_codec_enc_cfg_map_t vp8e_usage_cfg_map[] =
         1,                  /* g_delete_first_pass_file */
         "vp8.fpf"           /* first pass filename */
 #endif
-
+        VPX_SS_DEFAULT_LAYERS, /* ss_number_layers */
         1,                  /* ts_number_layers */
         {0},                /* ts_target_bitrate */
         {0},                /* ts_rate_decimator */

vp9/common/vp9_convolve.c

@@ -191,17 +191,21 @@ static void convolve_c(const uint8_t *src, ptrdiff_t src_stride,
                        const int16_t *filter_y, int y_step_q4,
                        int w, int h, int taps) {
   /* Fixed size intermediate buffer places limits on parameters.
-   * Maximum intermediate_height is 135, for y_step_q4 == 32,
+   * Maximum intermediate_height is 324, for y_step_q4 == 80,
    * h == 64, taps == 8.
+   * y_step_q4 of 80 allows for 1/10 scale for 5 layer svc
    */
-  uint8_t temp[64 * 135];
+  uint8_t temp[64 * 324];
   int intermediate_height = (((h - 1) * y_step_q4 + 15) >> 4) + taps;
 
   assert(w <= 64);
   assert(h <= 64);
   assert(taps <= 8);
-  assert(y_step_q4 <= 32);
+  assert(y_step_q4 <= 80);
-  assert(x_step_q4 <= 32);
+  assert(x_step_q4 <= 80);
+
+  if (intermediate_height < h)
+    intermediate_height = h;
 
   convolve_horiz_c(src - src_stride * (taps / 2 - 1), src_stride, temp, 64,
                    filter_x, x_step_q4, filter_y, y_step_q4, w,

vp9/common/vp9_onyx.h

@@ -131,6 +131,8 @@ extern "C"
     // END DATARATE CONTROL OPTIONS
     // ----------------------------------------------------------------
 
+    // Spatial scalability
+    int ss_number_layers;
 
     // these parameters aren't to be used in final build don't use!!!
     int play_alternate;
@@ -211,6 +213,13 @@ extern "C"
   int vp9_set_internal_size(VP9_PTR comp,
                             VPX_SCALING horiz_mode, VPX_SCALING vert_mode);
 
+  int vp9_set_size_literal(VP9_PTR comp, unsigned int width,
+                           unsigned int height);
+
+  int vp9_switch_layer(VP9_PTR comp, int layer);
+
+  void vp9_set_svc(VP9_PTR comp, int use_svc);
+
   int vp9_get_quantizer(VP9_PTR c);
 
 #ifdef __cplusplus

vp9/encoder/vp9_bitstream.c

@@ -1208,7 +1208,8 @@ static int get_refresh_mask(VP9_COMP *cpi) {
     if (!cpi->multi_arf_enabled && cpi->refresh_golden_frame &&
         !cpi->refresh_alt_ref_frame) {
 #else
-    if (cpi->refresh_golden_frame && !cpi->refresh_alt_ref_frame) {
+    if (cpi->refresh_golden_frame && !cpi->refresh_alt_ref_frame &&
+        !cpi->use_svc) {
 #endif
       // Preserve the previously existing golden frame and update the frame in
       // the alt ref slot instead. This is highly specific to the use of
@@ -1320,9 +1321,16 @@ static void write_frame_size_with_refs(VP9_COMP *cpi,
     YV12_BUFFER_CONFIG *cfg = &cm->yv12_fb[cm->ref_frame_map[refs[i]]];
     found = cm->width == cfg->y_crop_width &&
             cm->height == cfg->y_crop_height;
+
+    // TODO(ivan): This prevents a bug while more than 3 buffers are used. Do it
+    // in a better way.
+    if (cpi->use_svc) {
+      found = 0;
+    }
     vp9_wb_write_bit(wb, found);
-    if (found)
+    if (found) {
       break;
+    }
   }
 
   if (!found) {

vp9/encoder/vp9_onyx_if.c

@@ -1190,6 +1190,9 @@ static void init_config(VP9_PTR ptr, VP9_CONFIG *oxcf) {
   cpi->gld_fb_idx = 1;
   cpi->alt_fb_idx = 2;
 
+  cpi->current_layer = 0;
+  cpi->use_svc = 0;
+
   set_tile_limits(cpi);
 
   cpi->fixed_divide[0] = 0;
@@ -1460,6 +1463,9 @@ VP9_PTR vp9_create_compressor(VP9_CONFIG *oxcf) {
   cpi->alt_is_last  = 0;
   cpi->gold_is_alt  = 0;
 
+  // Spatial scalability
+  cpi->number_spatial_layers = oxcf->ss_number_layers;
+
   // Create the encoder segmentation map and set all entries to 0
   CHECK_MEM_ERROR(cm, cpi->segmentation_map,
                   vpx_calloc(cm->mi_rows * cm->mi_cols, 1));
@@ -2264,6 +2270,12 @@ static void update_golden_frame_stats(VP9_COMP *cpi) {
         cpi->oxcf.play_alternate && !cpi->refresh_alt_ref_frame) {
       cpi->source_alt_ref_pending = 1;
       cpi->frames_till_gf_update_due = cpi->baseline_gf_interval;
+
+      // TODO(ivan): for SVC encoder, GF automatic update is disabled by using a
+      // large GF_interval
+      if (cpi->use_svc) {
+        cpi->frames_till_gf_update_due = INT_MAX;
+      }
     }
 
     if (!cpi->source_alt_ref_pending)
@@ -2429,7 +2441,8 @@ static void update_reference_frames(VP9_COMP * const cpi) {
   else if (!cpi->multi_arf_enabled && cpi->refresh_golden_frame &&
       !cpi->refresh_alt_ref_frame) {
 #else
-  else if (cpi->refresh_golden_frame && !cpi->refresh_alt_ref_frame) {
+  else if (cpi->refresh_golden_frame && !cpi->refresh_alt_ref_frame &&
+           !cpi->use_svc) {
 #endif
     /* Preserve the previously existing golden frame and update the frame in
      * the alt ref slot instead. This is highly specific to the current use of
@@ -2502,9 +2515,11 @@ static void loopfilter_frame(VP9_COMP *cpi, VP9_COMMON *cm) {
 static void scale_references(VP9_COMP *cpi) {
   VP9_COMMON *cm = &cpi->common;
   int i;
+  int refs[ALLOWED_REFS_PER_FRAME] = {cpi->lst_fb_idx, cpi->gld_fb_idx,
+                                      cpi->alt_fb_idx};
 
   for (i = 0; i < 3; i++) {
-    YV12_BUFFER_CONFIG *ref = &cm->yv12_fb[cm->ref_frame_map[i]];
+    YV12_BUFFER_CONFIG *ref = &cm->yv12_fb[cm->ref_frame_map[refs[i]]];
 
     if (ref->y_crop_width != cm->width ||
         ref->y_crop_height != cm->height) {
@@ -2517,8 +2532,8 @@ static void scale_references(VP9_COMP *cpi) {
       scale_and_extend_frame(ref, &cm->yv12_fb[new_fb]);
       cpi->scaled_ref_idx[i] = new_fb;
     } else {
-      cpi->scaled_ref_idx[i] = cm->ref_frame_map[i];
+      cpi->scaled_ref_idx[i] = cm->ref_frame_map[refs[i]];
-      cm->fb_idx_ref_cnt[cm->ref_frame_map[i]]++;
+      cm->fb_idx_ref_cnt[cm->ref_frame_map[refs[i]]]++;
     }
   }
 }
@@ -3639,24 +3654,28 @@ static void Pass2Encode(VP9_COMP *cpi, unsigned long *size,
   }
 }
 
-
+static void check_initial_width(VP9_COMP *cpi, YV12_BUFFER_CONFIG *sd) {
-int vp9_receive_raw_frame(VP9_PTR ptr, unsigned int frame_flags,
-                          YV12_BUFFER_CONFIG *sd, int64_t time_stamp,
-                          int64_t end_time) {
-  VP9_COMP              *cpi = (VP9_COMP *) ptr;
   VP9_COMMON            *cm = &cpi->common;
-  struct vpx_usec_timer  timer;
-  int                    res = 0;
-
   if (!cpi->initial_width) {
     // TODO(jkoleszar): Support 1/4 subsampling?
-    cm->subsampling_x = sd->uv_width < sd->y_width;
+    cm->subsampling_x = (sd != NULL) && sd->uv_width < sd->y_width;
-    cm->subsampling_y = sd->uv_height < sd->y_height;
+    cm->subsampling_y = (sd != NULL) && sd->uv_height < sd->y_height;
     alloc_raw_frame_buffers(cpi);
 
     cpi->initial_width = cm->width;
     cpi->initial_height = cm->height;
   }
+}
+
+
+int vp9_receive_raw_frame(VP9_PTR ptr, unsigned int frame_flags,
+                          YV12_BUFFER_CONFIG *sd, int64_t time_stamp,
+                          int64_t end_time) {
+  VP9_COMP              *cpi = (VP9_COMP *) ptr;
+  struct vpx_usec_timer  timer;
+  int                    res = 0;
+
+  check_initial_width(cpi, sd);
   vpx_usec_timer_start(&timer);
   if (vp9_lookahead_push(cpi->lookahead, sd, time_stamp, end_time, frame_flags,
                          cpi->active_map_enabled ? cpi->active_map : NULL))
@@ -4194,7 +4213,76 @@ int vp9_set_internal_size(VP9_PTR comp,
   return 0;
 }
 
+int vp9_set_size_literal(VP9_PTR comp, unsigned int width,
+                         unsigned int height) {
+  VP9_COMP *cpi = (VP9_COMP *)comp;
+  VP9_COMMON *cm = &cpi->common;
 
+  check_initial_width(cpi, NULL);
+
+  if (width) {
+    cm->width = width;
+    if (cm->width * 5 < cpi->initial_width) {
+      cm->width = cpi->initial_width / 5 + 1;
+      printf("Warning: Desired width too small, changed to %d \n", cm->width);
+    }
+    if (cm->width > cpi->initial_width) {
+      cm->width = cpi->initial_width;
+      printf("Warning: Desired width too large, changed to %d \n", cm->width);
+    }
+  }
+
+  if (height) {
+    cm->height = height;
+    if (cm->height * 5 < cpi->initial_height) {
+      cm->height = cpi->initial_height / 5 + 1;
+      printf("Warning: Desired height too small, changed to %d \n", cm->height);
+    }
+    if (cm->height > cpi->initial_height) {
+      cm->height = cpi->initial_height;
+      printf("Warning: Desired height too large, changed to %d \n", cm->height);
+    }
+  }
+
+  assert(cm->width <= cpi->initial_width);
+  assert(cm->height <= cpi->initial_height);
+  update_frame_size(cpi);
+  return 0;
+}
+
+int vp9_switch_layer(VP9_PTR comp, int layer) {
+  VP9_COMP *cpi = (VP9_COMP *)comp;
+
+  if (cpi->use_svc) {
+    cpi->current_layer = layer;
+
+    // Use buffer i for layer i LST
+    cpi->lst_fb_idx = layer;
+
+    // Use buffer i-1 for layer i Alt (Inter-layer prediction)
+    if (layer != 0) cpi->alt_fb_idx = layer - 1;
+
+    // Use the rest for Golden
+    if (layer < 2 * cpi->number_spatial_layers - NUM_REF_FRAMES)
+      cpi->gld_fb_idx = cpi->lst_fb_idx;
+    else
+      cpi->gld_fb_idx = 2 * cpi->number_spatial_layers - 1 - layer;
+
+    printf("Switching to layer %d:\n", layer);
+    printf("Using references: LST/GLD/ALT [%d|%d|%d]\n", cpi->lst_fb_idx,
+           cpi->gld_fb_idx, cpi->alt_fb_idx);
+  } else {
+    printf("Switching layer not supported. Enable SVC first \n");
+  }
+  return 0;
+}
+
+void vp9_set_svc(VP9_PTR comp, int use_svc) {
+  VP9_COMP *cpi = (VP9_COMP *)comp;
+  cpi->use_svc = use_svc;
+  if (cpi->use_svc) printf("Enabled SVC encoder \n");
+  return;
+}
 
 int vp9_calc_ss_err(YV12_BUFFER_CONFIG *source, YV12_BUFFER_CONFIG *dest) {
   int i, j;

vp9/encoder/vp9_onyx_int.h

@@ -351,6 +351,10 @@ typedef struct VP9_COMP {
   int lst_fb_idx;
   int gld_fb_idx;
   int alt_fb_idx;
+
+  int current_layer;
+  int use_svc;
+
 #if CONFIG_MULTIPLE_ARF
   int alt_ref_fb_idx[NUM_REF_FRAMES - 3];
 #endif
@@ -650,6 +654,8 @@ typedef struct VP9_COMP {
   int initial_width;
   int initial_height;
 
+  int number_spatial_layers;
+
 #if CONFIG_MULTIPLE_ARF
   // ARF tracking variables.
   int multi_arf_enabled;
@@ -684,6 +690,17 @@ static int get_ref_frame_idx(VP9_COMP *cpi, MV_REFERENCE_FRAME ref_frame) {
   }
 }
 
+static int get_scale_ref_frame_idx(VP9_COMP *cpi,
+                                   MV_REFERENCE_FRAME ref_frame) {
+  if (ref_frame == LAST_FRAME) {
+    return 0;
+  } else if (ref_frame == GOLDEN_FRAME) {
+    return 1;
+  } else {
+    return 2;
+  }
+}
+
 void vp9_encode_frame(VP9_COMP *cpi);
 
 void vp9_pack_bitstream(VP9_COMP *cpi, unsigned char *dest,

vp9/encoder/vp9_rdopt.c

@@ -2278,8 +2278,9 @@ static void setup_buffer_inter(VP9_COMP *cpi, MACROBLOCK *x,
 static YV12_BUFFER_CONFIG *get_scaled_ref_frame(VP9_COMP *cpi, int ref_frame) {
   YV12_BUFFER_CONFIG *scaled_ref_frame = NULL;
   int fb = get_ref_frame_idx(cpi, ref_frame);
-  if (cpi->scaled_ref_idx[fb] != cpi->common.ref_frame_map[fb])
+  int fb_scale = get_scale_ref_frame_idx(cpi, ref_frame);
-    scaled_ref_frame = &cpi->common.yv12_fb[cpi->scaled_ref_idx[fb]];
+  if (cpi->scaled_ref_idx[fb_scale] != cpi->common.ref_frame_map[fb])
+    scaled_ref_frame = &cpi->common.yv12_fb[cpi->scaled_ref_idx[fb_scale]];
   return scaled_ref_frame;
 }
 

vp9/vp9_cx_iface.c

@@ -160,6 +160,8 @@ static vpx_codec_err_t validate_config(vpx_codec_alg_priv_t      *ctx,
   RANGE_CHECK_HI(cfg, rc_resize_down_thresh, 100);
   RANGE_CHECK(cfg,        g_pass,         VPX_RC_ONE_PASS, VPX_RC_LAST_PASS);
 
+  RANGE_CHECK(cfg, ss_number_layers,      1,
+              VPX_SS_MAX_LAYERS); /*Spatial layers max */
   /* VP8 does not support a lower bound on the keyframe interval in
    * automatic keyframe placement mode.
    */
@@ -319,6 +321,8 @@ static vpx_codec_err_t set_vp9e_config(VP9_CONFIG *oxcf,
 
   oxcf->error_resilient_mode = cfg.g_error_resilient;
   oxcf->frame_parallel_decoding_mode = vp8_cfg.frame_parallel_decoding_mode;
+
+  oxcf->ss_number_layers = cfg.ss_number_layers;
   /*
   printf("Current VP9 Settings: \n");
   printf("target_bandwidth: %d\n", oxcf->target_bandwidth);
@@ -425,6 +429,8 @@ static vpx_codec_err_t set_param(vpx_codec_alg_priv_t *ctx,
       MAP(VP8E_SET_ARNR_TYPE,               xcfg.arnr_type);
       MAP(VP8E_SET_TUNING,                  xcfg.tuning);
       MAP(VP8E_SET_CQ_LEVEL,                xcfg.cq_level);
+      MAP(VP9E_SET_MAX_Q,                   ctx->cfg.rc_max_quantizer);
+      MAP(VP9E_SET_MIN_Q,                   ctx->cfg.rc_min_quantizer);
       MAP(VP8E_SET_MAX_INTRA_BITRATE_PCT,   xcfg.rc_max_intra_bitrate_pct);
       MAP(VP9E_SET_LOSSLESS,                xcfg.lossless);
       MAP(VP9E_SET_FRAME_PARALLEL_DECODING, xcfg.frame_parallel_decoding_mode);
@@ -1006,6 +1012,68 @@ static vpx_codec_err_t vp9e_set_scalemode(vpx_codec_alg_priv_t *ctx,
     return VPX_CODEC_INVALID_PARAM;
 }
 
+static vpx_codec_err_t vp9e_set_width(vpx_codec_alg_priv_t *ctx, int ctr_id,
+                                      va_list args) {
+  unsigned int *data = va_arg(args, unsigned int *);
+  if (data) {
+    int res;
+    res = vp9_set_size_literal(ctx->cpi, *data, 0);
+    if (!res) {
+      return VPX_CODEC_OK;
+    } else {
+      return VPX_CODEC_INVALID_PARAM;
+    }
+  } else {
+    return VPX_CODEC_INVALID_PARAM;
+  }
+}
+
+static vpx_codec_err_t vp9e_set_height(vpx_codec_alg_priv_t *ctx,
+                                       int ctr_id,
+                                       va_list args) {
+  unsigned int *data =  va_arg(args, unsigned int *);
+
+  if (data) {
+    int res;
+    res = vp9_set_size_literal(ctx->cpi, 0, *data);
+
+    if (!res) {
+      return VPX_CODEC_OK;
+    } else {
+      return VPX_CODEC_INVALID_PARAM;
+    }
+  } else {
+    return VPX_CODEC_INVALID_PARAM;
+  }
+}
+
+static vpx_codec_err_t vp9e_set_layer(vpx_codec_alg_priv_t *ctx,
+                                      int ctr_id,
+                                      va_list args) {
+  unsigned int *data =  va_arg(args, unsigned int *);
+
+  if (data) {
+    int res;
+    res = 0;
+
+    res = vp9_switch_layer(ctx->cpi, *data);
+
+    if (!res) {
+      return VPX_CODEC_OK;
+    } else {
+      return VPX_CODEC_INVALID_PARAM;
+    }
+  } else {
+    return VPX_CODEC_INVALID_PARAM;
+  }
+}
+
+static vpx_codec_err_t vp9e_set_svc(vpx_codec_alg_priv_t *ctx, int ctr_id,
+                                    va_list args) {
+  int data = va_arg(args, int);
+  vp9_set_svc(ctx->cpi, data);
+  return VPX_CODEC_OK;
+}
 
 static vpx_codec_ctrl_fn_map_t vp9e_ctf_maps[] = {
   {VP8_SET_REFERENCE,                 vp9e_set_reference},
@@ -1031,10 +1099,16 @@ static vpx_codec_ctrl_fn_map_t vp9e_ctf_maps[] = {
   {VP8E_SET_ARNR_TYPE,                set_param},
   {VP8E_SET_TUNING,                   set_param},
   {VP8E_SET_CQ_LEVEL,                 set_param},
+  {VP9E_SET_MAX_Q,                    set_param},
+  {VP9E_SET_MIN_Q,                    set_param},
   {VP8E_SET_MAX_INTRA_BITRATE_PCT,    set_param},
   {VP9E_SET_LOSSLESS,                 set_param},
   {VP9E_SET_FRAME_PARALLEL_DECODING,  set_param},
   {VP9_GET_REFERENCE,                 get_reference},
+  {VP9E_SET_WIDTH,                    vp9e_set_width},
+  {VP9E_SET_HEIGHT,                   vp9e_set_height},
+  {VP9E_SET_LAYER,                    vp9e_set_layer},
+  {VP9E_SET_SVC,                      vp9e_set_svc},
   { -1, NULL},
 };
 
@@ -1084,6 +1158,8 @@ static vpx_codec_enc_cfg_map_t vp9e_usage_cfg_map[] = {
       0,                  /* kf_min_dist */
       9999,               /* kf_max_dist */
 
+      VPX_SS_DEFAULT_LAYERS, /* ss_number_layers */
+
 #if VPX_ENCODER_ABI_VERSION == (1 + VPX_CODEC_ABI_VERSION)
       1,                  /* g_delete_first_pass_file */
       "vp8.fpf"           /* first pass filename */

vp9_spatial_scalable_encoder.c

@@ -0,0 +1,487 @@
+/*
+ *  Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+ *
+ *  Use of this source code is governed by a BSD-style license
+ *  that can be found in the LICENSE file in the root of the source
+ *  tree. An additional intellectual property rights grant can be found
+ *  in the file PATENTS.  All contributing project authors may
+ *  be found in the AUTHORS file in the root of the source tree.
+ */
+
+/*
+ * This is an example demonstrating how to implement a multi-layer
+ * VP9 encoding scheme based on spatial scalability for video applications
+ * that benefit from a scalable bitstream.
+ */
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdarg.h>
+#include <time.h>
+#include <string.h>
+#include <unistd.h>
+#include <libgen.h>
+#define VPX_CODEC_DISABLE_COMPAT 1
+#include "vpx/vpx_encoder.h"
+#include "vpx/vp8cx.h"
+#define interface (vpx_codec_vp9_cx())
+#define fourcc 0x30395056
+#define IVF_FILE_HDR_SZ (32)
+#define IVF_FRAME_HDR_SZ (12)
+#define NUM_BUFFERS 8
+
+char *input_filename;
+char *output_filename;
+unsigned int number_frames_to_code = 60 * 60;
+unsigned int number_frames_to_skip = 0;
+unsigned int number_spatial_layers = 5;
+unsigned int key_period = 100;
+
+typedef enum ENCODING_MODE {
+  INTER_LAYER_PREDICTION_I,
+  INTER_LAYER_PREDICTION_IP,
+  USE_GOLDEN_FRAME
+} ENCODING_MODE;
+
+static void mem_put_le16(char *mem, unsigned int val) {
+  mem[0] = val;
+  mem[1] = val >> 8;
+}
+
+static void mem_put_le32(char *mem, unsigned int val) {
+  mem[0] = val;
+  mem[1] = val >> 8;
+  mem[2] = val >> 16;
+  mem[3] = val >> 24;
+}
+
+static void usage(char *program_name) {
+  printf(
+      "Usage: %s [-f frames] [-s skip_frames] [-w width] [-h height] \n\t"
+      "[-n rate_num] [-d rate_den] [-b bitrate] [-l layers] "
+      "<input_filename> <output_filename>\n",
+      basename(program_name));
+  exit(EXIT_FAILURE);
+}
+
+static void die(const char *fmt, ...) {
+  va_list ap;
+
+  va_start(ap, fmt);
+  vprintf(fmt, ap);
+  if (fmt[strlen(fmt) - 1] != '\n') printf("\n");
+  exit(EXIT_FAILURE);
+}
+
+static void die_codec(vpx_codec_ctx_t *ctx, const char *s) {
+  const char *detail = vpx_codec_error_detail(ctx);
+
+  printf("%s: %s\n", s, vpx_codec_error(ctx));
+  if (detail) printf("    %s\n", detail);
+  exit(EXIT_FAILURE);
+}
+
+static int read_frame(FILE *f, vpx_image_t *img) {
+  size_t nbytes, to_read;
+  int res = 1;
+
+  to_read = img->w * img->h * 3 / 2;
+  nbytes = fread(img->planes[0], 1, to_read, f);
+  if (nbytes != to_read) {
+    res = 0;
+    if (nbytes > 0)
+      printf("Warning: Read partial frame. Check your width & height!\n");
+  }
+  return res;
+}
+
+static int read_dummy_frame(vpx_image_t *img) {
+  size_t to_read;
+
+  to_read = img->w * img->h * 3 / 2;
+  memset(img->planes[0], 129, to_read);
+  return 1;
+}
+
+static void write_ivf_file_header(FILE *outfile, const vpx_codec_enc_cfg_t *cfg,
+                                  int frame_cnt) {
+  char header[32];
+
+  if (cfg->g_pass != VPX_RC_ONE_PASS && cfg->g_pass != VPX_RC_LAST_PASS) return;
+  header[0] = 'D';
+  header[1] = 'K';
+  header[2] = 'I';
+  header[3] = 'F';
+  mem_put_le16(header + 4, 0);                    /* version */
+  mem_put_le16(header + 6, 32);                   /* headersize */
+  mem_put_le32(header + 8, fourcc);               /* headersize */
+  mem_put_le16(header + 12, cfg->g_w);            /* width */
+  mem_put_le16(header + 14, cfg->g_h);            /* height */
+  mem_put_le32(header + 16, cfg->g_timebase.den); /* rate */
+  mem_put_le32(header + 20, cfg->g_timebase.num); /* scale */
+  mem_put_le32(header + 24, frame_cnt);           /* length */
+  mem_put_le32(header + 28, 0);                   /* unused */
+
+  (void)fwrite(header, 1, 32, outfile);
+}
+
+static void write_ivf_frame_header(FILE *outfile,
+                                   const vpx_codec_cx_pkt_t *pkt) {
+  char header[12];
+  vpx_codec_pts_t pts;
+
+  if (pkt->kind != VPX_CODEC_CX_FRAME_PKT) return;
+
+  pts = pkt->data.frame.pts;
+  mem_put_le32(header, pkt->data.frame.sz);
+  mem_put_le32(header + 4, pts & 0xFFFFFFFF);
+  mem_put_le32(header + 8, pts >> 32);
+
+  (void)fwrite(header, 1, 12, outfile);
+}
+
+static void check_parameters() {
+  if (number_spatial_layers > 5) die("Cannot support more than 5 layers");
+}
+
+static void parse_command_line(int argc, char **argv,
+                               vpx_codec_enc_cfg_t *cfg) {
+  unsigned int width = 1920;
+  unsigned int height = 1080;
+  unsigned int timebase_num = 1;
+  unsigned int timebase_den = 60;
+  unsigned int bitrate = 1000;
+  int c;
+  vpx_codec_err_t res;
+
+  opterr = 0;
+  while ((c = getopt(argc, argv, "f:w:h:n:d:b:s:l:p:")) != -1) switch (c) {
+      case 'f':
+        number_frames_to_code = atoi(optarg);
+        break;
+      case 'w':
+        width = atoi(optarg);
+        break;
+      case 'h':
+        height = atoi(optarg);
+        break;
+      case 'n':
+        timebase_num = atoi(optarg);
+        break;
+      case 'd':
+        timebase_den = atoi(optarg);
+        break;
+      case 'b':
+        bitrate = atoi(optarg);
+        break;
+      case 's':
+        number_frames_to_skip = atoi(optarg);
+        break;
+      case 'l':
+        number_spatial_layers = atoi(optarg);
+        break;
+      case 'p':
+        key_period = atoi(optarg);
+        break;
+      case '?':
+        usage(argv[0]);
+    }
+
+  // Parse required parameters
+  if (argc - optind != 2) {
+    usage(argv[0]);
+  }
+
+  input_filename = argv[optind];
+  output_filename = argv[optind + 1];
+
+  if (width < 16 || width % 2 || height < 16 || height % 2)
+    die("Invalid resolution: %d x %d", width, height);
+
+  /* Populate encoder configuration */
+  res = vpx_codec_enc_config_default(interface, cfg, 0);
+  if (res) {
+    die("Failed to get config: %s\n", vpx_codec_err_to_string(res));
+  }
+  printf(
+      "Codec %s\nframes: %d, skip: %d, layers: %d\n"
+      "width %d, height: %d, \n"
+      "num: %d, den: %d, bitrate: %d, \n"
+      "key period: %d \n",
+      vpx_codec_iface_name(interface), number_frames_to_code,
+      number_frames_to_skip, number_spatial_layers, width, height, timebase_num,
+      timebase_den, bitrate, key_period);
+
+  // Do minimal check at the application level. Encoder parameters will be
+  // checked internally
+  check_parameters();
+
+  cfg->rc_target_bitrate = bitrate;
+  cfg->g_w = width;
+  cfg->g_h = height;
+  cfg->g_timebase.num = timebase_num;
+  cfg->g_timebase.den = timebase_den;
+  cfg->ss_number_layers = number_spatial_layers;
+}
+
+static void set_default_configuration(vpx_codec_enc_cfg_t *cfg) {
+  /* Real time parameters */
+  cfg->rc_dropframe_thresh = 0;
+  cfg->rc_end_usage = VPX_CBR;
+  cfg->rc_resize_allowed = 0;
+  cfg->rc_min_quantizer = 33;
+  cfg->rc_max_quantizer = 33;
+  cfg->rc_undershoot_pct = 100;
+  cfg->rc_overshoot_pct = 15;
+  cfg->rc_buf_initial_sz = 500;
+  cfg->rc_buf_optimal_sz = 600;
+  cfg->rc_buf_sz = 1000;
+
+  /* Enable error resilient mode */
+  cfg->g_error_resilient = 1;
+  cfg->g_lag_in_frames = 0;
+
+  /* Disable automatic keyframe placement */
+  cfg->kf_mode = VPX_KF_DISABLED;
+  cfg->kf_min_dist = cfg->kf_max_dist = 3000;
+}
+
+static void initialize_codec(vpx_codec_ctx_t *codec, vpx_codec_enc_cfg_t *cfg) {
+  int max_intra_size_pct;
+
+  /* Initialize codec */
+  if (vpx_codec_enc_init(codec, interface, cfg, VPX_CODEC_USE_PSNR))
+    die_codec(codec, "Failed to initialize encoder");
+
+  vpx_codec_control(codec, VP9E_SET_SVC, 1);
+  /* Cap CPU & first I-frame size */
+  vpx_codec_control(codec, VP8E_SET_CPUUSED, 1);
+  vpx_codec_control(codec, VP8E_SET_STATIC_THRESHOLD, 1);
+  vpx_codec_control(codec, VP8E_SET_NOISE_SENSITIVITY, 1);
+  vpx_codec_control(codec, VP8E_SET_TOKEN_PARTITIONS, 1);
+
+  max_intra_size_pct =
+      (int)(((double)cfg->rc_buf_optimal_sz * 0.5) *
+            ((double)cfg->g_timebase.den / cfg->g_timebase.num) / 10.0);
+  /* printf ("max_intra_size_pct=%d\n", max_intra_size_pct); */
+
+  vpx_codec_control(codec, VP8E_SET_MAX_INTRA_BITRATE_PCT, max_intra_size_pct);
+}
+
+static int calculate_layer(int frame_cnt, int number_spatial_layers) {
+  if (frame_cnt == 0)
+    return 0;
+  else
+    return (frame_cnt + number_spatial_layers - 1) % number_spatial_layers;
+}
+
+static void switch_to_layer(int layer, unsigned int initial_width,
+                            unsigned int initial_height,
+                            vpx_codec_ctx_t *codec) {
+  // Set layer size
+  int scaling_factor_num[MAX_LAYERS] = {2, 1, 4, 2, 1};
+  int scaling_factor_den[MAX_LAYERS] = {9, 3, 9, 3, 1};
+
+  int quantizer[MAX_LAYERS] = {60, 53, 39, 33, 27};
+
+  unsigned int current_width;
+  unsigned int current_height;
+
+  current_width = initial_width *
+                  scaling_factor_num[layer + 5 - number_spatial_layers] /
+                  scaling_factor_den[layer + 5 - number_spatial_layers];
+  current_height = initial_height *
+                   scaling_factor_num[layer + 5 - number_spatial_layers] /
+                   scaling_factor_den[layer + 5 - number_spatial_layers];
+
+  current_width += current_width % 2;
+  current_height += current_height % 2;
+
+  vpx_codec_control(codec, VP9E_SET_WIDTH, &current_width);
+  vpx_codec_control(codec, VP9E_SET_HEIGHT, &current_height);
+
+  // Set layer context
+  vpx_codec_control(codec, VP9E_SET_LAYER, &layer);
+  vpx_codec_control(codec, VP9E_SET_MAX_Q,
+                    quantizer[layer + 5 - number_spatial_layers]);
+  vpx_codec_control(codec, VP9E_SET_MIN_Q,
+                    quantizer[layer + 5 - number_spatial_layers]);
+}
+
+static int get_flag(int is_I_frame_in_layer, int layer, ENCODING_MODE mode) {
+  // First layer
+  switch (mode) {
+    case INTER_LAYER_PREDICTION_I:
+      if (is_I_frame_in_layer && layer == 0) return VPX_EFLAG_FORCE_KF;
+      if (layer == 0)
+        return VP8_EFLAG_NO_UPD_GF | VP8_EFLAG_NO_UPD_ARF |
+               VP8_EFLAG_NO_REF_GF | VP8_EFLAG_NO_REF_ARF;
+      else if (is_I_frame_in_layer)
+        return VP8_EFLAG_NO_UPD_GF | VP8_EFLAG_NO_UPD_ARF |
+               VP8_EFLAG_NO_REF_GF | VP8_EFLAG_NO_REF_LAST;
+      else
+        return VP8_EFLAG_NO_UPD_GF | VP8_EFLAG_NO_UPD_ARF |
+               VP8_EFLAG_NO_REF_GF | VP8_EFLAG_NO_REF_ARF;
+      break;
+
+    case INTER_LAYER_PREDICTION_IP:
+      if (is_I_frame_in_layer && layer == 0) return VPX_EFLAG_FORCE_KF;
+      if (layer == 0)
+        return VP8_EFLAG_NO_UPD_GF | VP8_EFLAG_NO_UPD_ARF |
+               VP8_EFLAG_NO_REF_GF | VP8_EFLAG_NO_REF_ARF;
+      else if (is_I_frame_in_layer)
+        return VP8_EFLAG_NO_UPD_GF | VP8_EFLAG_NO_UPD_ARF |
+               VP8_EFLAG_NO_REF_GF | VP8_EFLAG_NO_REF_LAST;
+      else
+        return VP8_EFLAG_NO_UPD_GF | VP8_EFLAG_NO_UPD_ARF | VP8_EFLAG_NO_REF_GF;
+      break;
+
+    case USE_GOLDEN_FRAME:
+      if (is_I_frame_in_layer && layer == 0) return VPX_EFLAG_FORCE_KF;
+      if (2 * number_spatial_layers - NUM_BUFFERS <= layer) {
+        if (layer == 0)
+          return VP8_EFLAG_NO_UPD_GF | VP8_EFLAG_NO_UPD_ARF |
+                 VP8_EFLAG_NO_REF_ARF;
+        else if (is_I_frame_in_layer)
+          return VP8_EFLAG_NO_UPD_ARF | VP8_EFLAG_NO_REF_GF |
+                 VP8_EFLAG_NO_REF_LAST;
+        else
+          return VP8_EFLAG_NO_UPD_GF | VP8_EFLAG_NO_UPD_ARF;
+      } else {
+        if (layer == 0)
+          return VP8_EFLAG_NO_UPD_GF | VP8_EFLAG_NO_UPD_ARF |
+                 VP8_EFLAG_NO_REF_GF | VP8_EFLAG_NO_REF_ARF;
+        else if (is_I_frame_in_layer)
+          return VP8_EFLAG_NO_UPD_GF | VP8_EFLAG_NO_UPD_ARF |
+                 VP8_EFLAG_NO_REF_GF | VP8_EFLAG_NO_REF_LAST;
+        else
+          return VP8_EFLAG_NO_UPD_GF | VP8_EFLAG_NO_UPD_ARF |
+                 VP8_EFLAG_NO_REF_GF | VP8_EFLAG_NO_REF_ARF;
+      }
+      break;
+    default:
+      return VPX_EFLAG_FORCE_KF;
+  }
+}
+
+int main(int argc, char **argv) {
+  FILE *infile, *outfile[MAX_LAYERS];
+  vpx_codec_ctx_t codec;
+  vpx_codec_enc_cfg_t cfg;
+  int frame_cnt = 0;
+  vpx_image_t raw;
+  int frame_avail = 1;
+  int got_data = 0;
+  int i;
+  int frames_in_layer[MAX_LAYERS] = {0};
+  clock_t before;
+  clock_t after;
+  int pts = 0;            /* PTS starts at 0 */
+  int frame_duration = 1; /* 1 timebase tick per frame */
+
+  parse_command_line(argc, argv, &cfg);
+
+  // Allocate image buffer
+  if (!vpx_img_alloc(&raw, VPX_IMG_FMT_I420, cfg.g_w, cfg.g_h, 32))
+    die("Failed to allocate image", cfg.g_w, cfg.g_h);
+
+  set_default_configuration(&cfg);
+
+  /* Open input file */
+  if (!(infile = fopen(input_filename, "rb")))
+    die("Failed to open %s for reading", argv[1]);
+
+  /* Open output file  */
+  for (i = 0; i < number_spatial_layers; i++) {
+    char file_name[512];
+    snprintf(file_name, sizeof(file_name), "%s_%d.ivf", output_filename, i);
+    if (!(outfile[i] = fopen(file_name, "wb")))
+      die("Failed to open %s for writing", file_name);
+    write_ivf_file_header(outfile[i], &cfg, 0);
+  }
+
+  initialize_codec(&codec, &cfg);
+
+  // skip initial frames
+  for (i = 0; i < number_frames_to_skip; i++) {
+    read_frame(infile, &raw);
+  }
+
+  before = clock();
+  // Encoding frames
+  while ((frame_avail || got_data) &&
+         frame_cnt <= number_frames_to_code * number_spatial_layers) {
+    int flags = 0;
+    vpx_codec_iter_t iter = NULL;
+    const vpx_codec_cx_pkt_t *pkt;
+
+    int layer = calculate_layer(frame_cnt, number_spatial_layers);
+    int is_I_frame_in_layer =
+        (((frame_cnt - 1) / number_spatial_layers % key_period) == 0);
+    int is_dummy = (frame_cnt == 0);
+
+    if (is_dummy) {  // Dummy frame
+      flags = VPX_EFLAG_FORCE_KF;
+      frame_avail = read_dummy_frame(&raw);
+
+    } else {  // Regular frame
+      // Read a new frame only at the base layer
+      if (layer == 0) frame_avail = read_frame(infile, &raw);
+      switch_to_layer(layer, cfg.g_w, cfg.g_h, &codec);
+      flags = get_flag(is_I_frame_in_layer, layer, INTER_LAYER_PREDICTION_I);
+    }
+
+    // Actual Encoding
+    if (vpx_codec_encode(&codec, frame_avail ? &raw : NULL, pts, 1, flags,
+                         VPX_DL_REALTIME))
+      die_codec(&codec, "Failed to encode frame");
+
+    got_data = 0;
+    // Process data / Get PSNR statistics
+    while ((pkt = vpx_codec_get_cx_data(&codec, &iter))) {
+      got_data = 1;
+      switch (pkt->kind) {
+        case VPX_CODEC_CX_FRAME_PKT:
+          for (i = layer; i < number_spatial_layers; i++) {
+            write_ivf_frame_header(outfile[i], pkt);
+            (void)fwrite(pkt->data.frame.buf, 1, pkt->data.frame.sz,
+                         outfile[i]);
+            frames_in_layer[i]++;
+          }
+          break;
+        case VPX_CODEC_PSNR_PKT:
+          if (frame_cnt != 0)
+            printf(
+                "Processed Frame %d, layer %d, PSNR(Total/Y/U/V): "
+                "%2.3f  %2.3f  %2.3f  %2.3f \n",
+                (frame_cnt - 1) / number_spatial_layers + 1, layer,
+                pkt->data.psnr.psnr[0], pkt->data.psnr.psnr[1],
+                pkt->data.psnr.psnr[2], pkt->data.psnr.psnr[3]);
+          break;
+        default:
+          break;
+      }
+    }
+    frame_cnt++;
+    // TODO(ivan): Modify ts later if(!layer)
+    pts += frame_duration;
+  }
+  // end while
+
+  after = clock();
+  printf("Processed %d frames in different resolutions in %ld ms.\n",
+         frame_cnt - 1, (int)(after - before) / (CLOCKS_PER_SEC / 1000));
+
+  fclose(infile);
+
+  if (vpx_codec_destroy(&codec)) die_codec(&codec, "Failed to destroy codec");
+
+  /* Try to rewrite the output file headers with the actual frame count */
+  for (i = 0; i < number_spatial_layers; i++) {
+    if (!fseek(outfile[i], 0, SEEK_SET)) {
+      write_ivf_file_header(outfile[i], &cfg, frames_in_layer[i]);
+    }
+    fclose(outfile[i]);
+  }
+
+  return EXIT_SUCCESS;
+}

vpx/vp8cx.h

@@ -190,7 +190,15 @@ enum vp8e_enc_control_id {
   VP9E_SET_LOSSLESS,
   VP9E_SET_TILE_COLUMNS,
   VP9E_SET_TILE_ROWS,
-  VP9E_SET_FRAME_PARALLEL_DECODING
+  VP9E_SET_FRAME_PARALLEL_DECODING,
+
+  VP9E_SET_WIDTH              = 99,
+  VP9E_SET_HEIGHT,
+  VP9E_SET_LAYER,
+  VP9E_SET_SVC,
+
+  VP9E_SET_MAX_Q,
+  VP9E_SET_MIN_Q
 };
 
 /*!\brief vpx 1-D scaling mode
@@ -292,6 +300,12 @@ VPX_CTRL_USE_TYPE(VP8E_SET_ROI_MAP,            vpx_roi_map_t *)
 VPX_CTRL_USE_TYPE(VP8E_SET_ACTIVEMAP,          vpx_active_map_t *)
 VPX_CTRL_USE_TYPE(VP8E_SET_SCALEMODE,          vpx_scaling_mode_t *)
 
+VPX_CTRL_USE_TYPE(VP9E_SET_LAYER,              int *)
+VPX_CTRL_USE_TYPE(VP9E_SET_SVC,                int)
+
+VPX_CTRL_USE_TYPE(VP9E_SET_WIDTH,              unsigned int *)
+VPX_CTRL_USE_TYPE(VP9E_SET_HEIGHT,             unsigned int *)
+
 VPX_CTRL_USE_TYPE(VP8E_SET_CPUUSED,            int)
 VPX_CTRL_USE_TYPE(VP8E_SET_ENABLEAUTOALTREF,   unsigned int)
 VPX_CTRL_USE_TYPE(VP8E_SET_NOISE_SENSITIVITY,  unsigned int)
@@ -316,6 +330,9 @@ VPX_CTRL_USE_TYPE(VP8E_SET_MAX_INTRA_BITRATE_PCT, unsigned int)
 VPX_CTRL_USE_TYPE(VP9E_SET_LOSSLESS, unsigned int)
 
 VPX_CTRL_USE_TYPE(VP9E_SET_FRAME_PARALLEL_DECODING, unsigned int)
+
+VPX_CTRL_USE_TYPE(VP9E_SET_MAX_Q,      unsigned int)
+VPX_CTRL_USE_TYPE(VP9E_SET_MIN_Q,      unsigned int)
 /*! @} - end defgroup vp8_encoder */
 #include "vpx_codec_impl_bottom.h"
 #endif

vpx/vpx_encoder.h

@@ -46,6 +46,12 @@ extern "C" {
   /*!\deprecated Use #VPX_TS_MAX_LAYERS instead. */
 #define MAX_LAYERS      VPX_TS_MAX_LAYERS
 
+/*! Spatial Scalability: Maximum number of coding layers */
+#define VPX_SS_MAX_LAYERS       5
+
+/*! Spatial Scalability: Default number of coding layers */
+#define VPX_SS_DEFAULT_LAYERS       3
+
   /*!\brief Current ABI version number
    *
    * \internal
@@ -596,9 +602,15 @@ extern "C" {
     unsigned int           kf_max_dist;
 
     /*
-     * Temporal scalability settings (ts)
+     * Spatial scalability settings (ss)
      */
 
+    /*!\brief Number of coding layers (spatial)
+     *
+     * This value specifies the number of coding layers to be used.
+     */
+    unsigned int           ss_number_layers;
+
     /*!\brief Number of coding layers
      *
      * This value specifies the number of coding layers to be used.