Add psnr/ssim tuning option

Add a new encoder control, VP8E_SET_TUNING, to allow the application
to inform the encoder that the material will benefit from certain
tuning. Expose this control as the --tune option to vpxenc. The args
helper is expanded to support enumerated arguments by name or value.

Two tunings are provided by this patch, PSNR (default) and SSIM.
Activity masking is made dependent on setting --tune=ssim, as the
current implementation hurts speed (10%) and PSNR (2.7% avg,
10% peak) too much for it to be a default yet.

Change-Id: I110d969381c4805347ff5a0ffaf1a14ca1965257

diff --git a/args.c b/args.c
index 7829290..7b2cc3a 100644 (file)
--- a/args.c
+++ b/args.c
@@ -135,6 +135,17 @@ void arg_show_usage(FILE *fp, const struct arg_def *const *defs)
                      def->long_name, long_val);
 
         fprintf(fp, "  %-37s\t%s\n", option_text, def->desc);
+
+        if(def->enums)
+        {
+            const struct arg_enum_list *listptr;
+
+            fprintf(fp, "  %-37s\t  ", "");
+
+            for(listptr = def->enums; listptr->name; listptr++)
+                fprintf(fp, "%s%s", listptr->name,
+                        listptr[1].name ? ", " : "\n");
+        }
     }
 }
 
@@ -218,3 +229,37 @@ struct vpx_rational arg_parse_rational(const struct arg *arg)
 
     return rat;
 }
+
+
+int arg_parse_enum(const struct arg *arg)
+{
+    const struct arg_enum_list *listptr;
+    long int                    rawval;
+    char                       *endptr;
+
+    /* First see if the value can be parsed as a raw value */
+    rawval = strtol(arg->val, &endptr, 10);
+    if (arg->val[0] != '\0' && endptr[0] == '\0')
+    {
+        /* Got a raw value, make sure it's valid */
+        for(listptr = arg->def->enums; listptr->name; listptr++)
+            if(listptr->val == rawval)
+                return rawval;
+    }
+
+    /* Next see if it can be parsed as a string */
+    for(listptr = arg->def->enums; listptr->name; listptr++)
+        if(!strcmp(arg->val, listptr->name))
+            return listptr->val;
+
+    die("Option %s: Invalid value '%s'\n", arg->name, arg->val);
+    return 0;
+}
+
+
+int arg_parse_enum_or_int(const struct arg *arg)
+{
+    if(arg->def->enums)
+        return arg_parse_enum(arg);
+    return arg_parse_int(arg);
+}

diff --git a/args.h b/args.h
index 4fafcf8..7963fa6 100644 (file)
--- a/args.h
+++ b/args.h
@@ -22,14 +22,23 @@ struct arg
     const struct arg_def  *def;
 };
 
+struct arg_enum_list
+{
+    const char *name;
+    int         val;
+};
+#define ARG_ENUM_LIST_END {0}
+
 typedef struct arg_def
 {
     const char *short_name;
     const char *long_name;
     int         has_val;
     const char *desc;
+    const struct arg_enum_list *enums;
 } arg_def_t;
-#define ARG_DEF(s,l,v,d) {s,l,v,d}
+#define ARG_DEF(s,l,v,d) {s,l,v,d, NULL}
+#define ARG_DEF_ENUM(s,l,v,d,e) {s,l,v,d,e}
 #define ARG_DEF_LIST_END {0}
 
 struct arg arg_init(char **argv);
@@ -41,4 +50,5 @@ char **argv_dup(int argc, const char **argv);
 unsigned int arg_parse_uint(const struct arg *arg);
 int arg_parse_int(const struct arg *arg);
 struct vpx_rational arg_parse_rational(const struct arg *arg);
+int arg_parse_enum_or_int(const struct arg *arg);
 #endif

diff --git a/vp8/common/onyx.h b/vp8/common/onyx.h
index 3c199d1..3724b11 100644 (file)
--- a/vp8/common/onyx.h
+++ b/vp8/common/onyx.h
@@ -18,6 +18,7 @@ extern "C"
 #endif
 
 #include "vpx/internal/vpx_codec_internal.h"
+#include "vpx/vp8cx.h"
 #include "vpx_scale/yv12config.h"
 #include "type_aliases.h"
 #include "ppflags.h"
@@ -189,6 +190,8 @@ extern "C"
 
         struct vpx_fixed_buf         two_pass_stats_in;
         struct vpx_codec_pkt_list  *output_pkt_list;
+
+        vp8e_tuning tuning;
     } VP8_CONFIG;
 
 

diff --git a/vp8/encoder/encodeframe.c b/vp8/encoder/encodeframe.c
index 2002735..1528389 100644 (file)
--- a/vp8/encoder/encodeframe.c
+++ b/vp8/encoder/encodeframe.c
@@ -497,7 +497,8 @@ void encode_mb_row(VP8_COMP *cpi,
         x->rddiv = cpi->RDDIV;
         x->rdmult = cpi->RDMULT;
 
-        activity_sum += vp8_activity_masking(cpi, x);
+        if(cpi->oxcf.tuning == VP8_TUNE_SSIM)
+            activity_sum += vp8_activity_masking(cpi, x);
 
         // Is segmentation enabled
         // MB level adjutment to quantizer

diff --git a/vp8/encoder/ethreading.c b/vp8/encoder/ethreading.c
index 0611d5b..2cd24da 100644 (file)
--- a/vp8/encoder/ethreading.c
+++ b/vp8/encoder/ethreading.c
@@ -115,7 +115,8 @@ THREAD_FUNCTION thread_encoding_proc(void *p_data)
                         x->rddiv = cpi->RDDIV;
                         x->rdmult = cpi->RDMULT;
 
-                        activity_sum += vp8_activity_masking(cpi, x);
+                        if(cpi->oxcf.tuning == VP8_TUNE_SSIM)
+                            activity_sum += vp8_activity_masking(cpi, x);
 
                         // Is segmentation enabled
                         // MB level adjutment to quantizer

diff --git a/vp8/vp8_cx_iface.c b/vp8/vp8_cx_iface.c
index ed830cc..f959207 100644 (file)
--- a/vp8/vp8_cx_iface.c
+++ b/vp8/vp8_cx_iface.c
@@ -37,6 +37,7 @@ struct vp8_extracfg
     unsigned int                arnr_max_frames;    /* alt_ref Noise Reduction Max Frame Count */
     unsigned int                arnr_strength;    /* alt_ref Noise Reduction Strength */
     unsigned int                arnr_type;        /* alt_ref filter type */
+    vp8e_tuning                 tuning;
 
 };
 
@@ -67,6 +68,7 @@ static const struct extraconfig_map extracfg_map[] =
             0,                          /* arnr_max_frames */
             3,                          /* arnr_strength */
             3,                          /* arnr_type*/
+            0,                          /* tuning*/
         }
     }
 };
@@ -336,6 +338,7 @@ static vpx_codec_err_t set_vp8e_config(VP8_CONFIG *oxcf,
     oxcf->arnr_strength =  vp8_cfg.arnr_strength;
     oxcf->arnr_type =      vp8_cfg.arnr_type;
 
+    oxcf->tuning = vp8_cfg.tuning;
 
     /*
         printf("Current VP8 Settings: \n");
@@ -449,6 +452,7 @@ static vpx_codec_err_t set_param(vpx_codec_alg_priv_t *ctx,
         MAP(VP8E_SET_ARNR_MAXFRAMES,        xcfg.arnr_max_frames);
         MAP(VP8E_SET_ARNR_STRENGTH ,        xcfg.arnr_strength);
         MAP(VP8E_SET_ARNR_TYPE     ,        xcfg.arnr_type);
+        MAP(VP8E_SET_TUNING,                xcfg.tuning);
 
     }
 
@@ -1029,6 +1033,7 @@ static vpx_codec_ctrl_fn_map_t vp8e_ctf_maps[] =
     {VP8E_SET_ARNR_MAXFRAMES,           set_param},
     {VP8E_SET_ARNR_STRENGTH ,           set_param},
     {VP8E_SET_ARNR_TYPE     ,           set_param},
+    {VP8E_SET_TUNING,                   set_param},
     { -1, NULL},
 };
 

diff --git a/vpx/vp8cx.h b/vpx/vp8cx.h
index efd7945..3ed1cf1 100644 (file)
--- a/vpx/vp8cx.h
+++ b/vpx/vp8cx.h
@@ -140,7 +140,8 @@ enum vp8e_enc_control_id
     VP8E_SET_ARNR_MAXFRAMES,         /**< control function to set the max number of frames blurred creating arf*/
     VP8E_SET_ARNR_STRENGTH ,         /**< control function to set the filter strength for the arf */
     VP8E_SET_ARNR_TYPE     ,         /**< control function to set the type of filter to use for the arf*/
-} ;
+    VP8E_SET_TUNING,                 /**< control function to set visual tuning */
+};
 
 /*!\brief vpx 1-D scaling mode
  *
@@ -224,6 +225,18 @@ typedef enum
 } vp8e_token_partitions;
 
 
+/*!\brief VP8 model tuning parameters
+ *
+ * Changes the encoder to tune for certain types of input material.
+ *
+ */
+typedef enum
+{
+    VP8_TUNE_PSNR,
+    VP8_TUNE_SSIM
+} vp8e_tuning;
+
+
 /*!\brief VP8 encoder control function parameter type
  *
  * Defines the data types that VP8E control functions take. Note that
@@ -253,7 +266,7 @@ VPX_CTRL_USE_TYPE(VP8E_SET_TOKEN_PARTITIONS,   vp8e_token_partitions)
 VPX_CTRL_USE_TYPE(VP8E_SET_ARNR_MAXFRAMES,     unsigned int)
 VPX_CTRL_USE_TYPE(VP8E_SET_ARNR_STRENGTH ,     unsigned int)
 VPX_CTRL_USE_TYPE(VP8E_SET_ARNR_TYPE     ,     unsigned int)
-
+VPX_CTRL_USE_TYPE(VP8E_SET_TUNING,             vp8e_tuning)
 
 VPX_CTRL_USE_TYPE(VP8E_GET_LAST_QUANTIZER,     int *)
 VPX_CTRL_USE_TYPE(VP8E_GET_LAST_QUANTIZER_64,  int *)

diff --git a/vpxenc.c b/vpxenc.c
index 4baeefc..5e4fe3f 100755 (executable)
--- a/vpxenc.c
+++ b/vpxenc.c
@@ -993,18 +993,27 @@ static const arg_def_t arnr_strength = ARG_DEF(NULL, "arnr-strength", 1,
                                        "AltRef Strength");
 static const arg_def_t arnr_type = ARG_DEF(NULL, "arnr-type", 1,
                                    "AltRef Type");
+static const struct arg_enum_list tuning_enum[] = {
+    {"psnr", VP8_TUNE_PSNR},
+    {"ssim", VP8_TUNE_SSIM},
+    {NULL, 0}
+};
+static const arg_def_t tune_ssim = ARG_DEF_ENUM(NULL, "tune", 1,
+                                   "Material to favor", tuning_enum);
 
 static const arg_def_t *vp8_args[] =
 {
     &cpu_used, &auto_altref, &noise_sens, &sharpness, &static_thresh,
-    &token_parts, &arnr_maxframes, &arnr_strength, &arnr_type, NULL
+    &token_parts, &arnr_maxframes, &arnr_strength, &arnr_type,
+    &tune_ssim, NULL
 };
 static const int vp8_arg_ctrl_map[] =
 {
     VP8E_SET_CPUUSED, VP8E_SET_ENABLEAUTOALTREF,
     VP8E_SET_NOISE_SENSITIVITY, VP8E_SET_SHARPNESS, VP8E_SET_STATIC_THRESHOLD,
     VP8E_SET_TOKEN_PARTITIONS,
-    VP8E_SET_ARNR_MAXFRAMES, VP8E_SET_ARNR_STRENGTH , VP8E_SET_ARNR_TYPE, 0
+    VP8E_SET_ARNR_MAXFRAMES, VP8E_SET_ARNR_STRENGTH , VP8E_SET_ARNR_TYPE,
+    VP8E_SET_TUNING, 0
 };
 #endif
 
@@ -1317,7 +1326,7 @@ int main(int argc, const char **argv_)
                 if (arg_ctrl_cnt < ARG_CTRL_CNT_MAX)
                 {
                     arg_ctrls[arg_ctrl_cnt][0] = ctrl_args_map[i];
-                    arg_ctrls[arg_ctrl_cnt][1] = arg_parse_int(&arg);
+                    arg_ctrls[arg_ctrl_cnt][1] = arg_parse_enum_or_int(&arg);
                     arg_ctrl_cnt++;
                 }
             }