aom/vp8/encoder/onyx_if.c

5260 строки
170 KiB
C
Исходник Обычный вид История

2010-05-18 19:58:33 +04:00
/*
* Copyright (c) 2010 The WebM project authors. All Rights Reserved.
2010-05-18 19:58:33 +04:00
*
* 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.
2010-05-18 19:58:33 +04:00
*/
#include "vp8/common/onyxc_int.h"
2010-05-18 19:58:33 +04:00
#include "onyx_int.h"
#include "vp8/common/systemdependent.h"
2010-05-18 19:58:33 +04:00
#include "quantize.h"
#include "vp8/common/alloccommon.h"
2010-05-18 19:58:33 +04:00
#include "mcomp.h"
#include "firstpass.h"
#include "psnr.h"
#include "vpx_scale/vpxscale.h"
#include "vp8/common/extend.h"
2010-05-18 19:58:33 +04:00
#include "ratectrl.h"
#include "vp8/common/quant_common.h"
#include "segmentation.h"
#include "vp8/common/g_common.h"
2010-05-18 19:58:33 +04:00
#include "vpx_scale/yv12extend.h"
#include "vp8/common/postproc.h"
2010-05-18 19:58:33 +04:00
#include "vpx_mem/vpx_mem.h"
#include "vp8/common/swapyv12buffer.h"
#include "vp8/common/threading.h"
2010-05-18 19:58:33 +04:00
#include "vpx_ports/vpx_timer.h"
#include "temporal_filter.h"
Add runtime CPU detection support for ARM. The primary goal is to allow a binary to be built which supports NEON, but can fall back to non-NEON routines, since some Android devices do not have NEON, even if they are otherwise ARMv7 (e.g., Tegra). The configure-generated flags HAVE_ARMV7, etc., are used to decide which versions of each function to build, and when CONFIG_RUNTIME_CPU_DETECT is enabled, the correct version is chosen at run time. In order for this to work, the CFLAGS must be set to something appropriate (e.g., without -mfpu=neon for ARMv7, and with appropriate -march and -mcpu for even earlier configurations), or the native C code will not be able to run. The ASFLAGS must remain set for the most advanced instruction set required at build time, since the ARM assembler will refuse to emit them otherwise. I have not attempted to make any changes to configure to do this automatically. Doing so will probably require the addition of new configure options. Many of the hooks for RTCD on ARM were already there, but a lot of the code had bit-rotted, and a good deal of the ARM-specific code is not integrated into the RTCD structs at all. I did not try to resolve the latter, merely to add the minimal amount of protection around them to allow RTCD to work. Those functions that were called based on an ifdef at the calling site were expanded to check the RTCD flags at that site, but they should be added to an RTCD struct somewhere in the future. The functions invoked with global function pointers still are, but these should be moved into an RTCD struct for thread safety (I believe every platform currently supported has atomic pointer stores, but this is not guaranteed). The encoder's boolhuff functions did not even have _c and armv7 suffixes, and the correct version was resolved at link time. The token packing functions did have appropriate suffixes, but the version was selected with a define, with no associated RTCD struct. However, for both of these, the only armv7 instruction they actually used was rbit, and this was completely superfluous, so I reworked them to avoid it. The only non-ARMv4 instruction remaining in them is clz, which is ARMv5 (not even ARMv5TE is required). Considering that there are no ARM-specific configs which are not at least ARMv5TE, I did not try to detect these at runtime, and simply enable them for ARMv5 and above. Finally, the NEON register saving code was completely non-reentrant, since it saved the registers to a global, static variable. I moved the storage for this onto the stack. A single binary built with this code was tested on an ARM11 (ARMv6) and a Cortex A8 (ARMv7 w/NEON), for both the encoder and decoder, and produced identical output, while using the correct accelerated functions on each. I did not test on any earlier processors. Change-Id: I45cbd63a614f4554c3b325c45d46c0806f009eaa
2010-10-21 02:39:11 +04:00
#if ARCH_ARM
#include "vpx_ports/arm.h"
#endif
2010-05-18 19:58:33 +04:00
#include <math.h>
#include <stdio.h>
#include <limits.h>
#if CONFIG_RUNTIME_CPU_DETECT
#define IF_RTCD(x) (x)
#define RTCD(x) &cpi->common.rtcd.x
#else
#define IF_RTCD(x) NULL
#define RTCD(x) NULL
#endif
extern void vp8cx_pick_filter_level_fast(YV12_BUFFER_CONFIG *sd, VP8_COMP *cpi);
extern void vp8cx_set_alt_lf_level(VP8_COMP *cpi, int filt_val);
extern void vp8cx_pick_filter_level(YV12_BUFFER_CONFIG *sd, VP8_COMP *cpi);
extern void vp8_init_loop_filter(VP8_COMMON *cm);
extern void vp8_loop_filter_frame(VP8_COMMON *cm, MACROBLOCKD *mbd, int filt_val);
extern void vp8_loop_filter_frame_yonly(VP8_COMMON *cm, MACROBLOCKD *mbd, int filt_val, int sharpness_lvl);
extern void vp8_dmachine_specific_config(VP8_COMP *cpi);
extern void vp8_cmachine_specific_config(VP8_COMP *cpi);
extern void vp8_deblock_frame(YV12_BUFFER_CONFIG *source, YV12_BUFFER_CONFIG *post, int filt_lvl, int low_var_thresh, int flag);
extern void print_parms(VP8_CONFIG *ocf, char *filenam);
extern unsigned int vp8_get_processor_freq();
extern void print_tree_update_probs();
extern void vp8cx_create_encoder_threads(VP8_COMP *cpi);
extern void vp8cx_remove_encoder_threads(VP8_COMP *cpi);
#if HAVE_ARMV7
extern void vp8_yv12_copy_frame_func_neon(YV12_BUFFER_CONFIG *src_ybc, YV12_BUFFER_CONFIG *dst_ybc);
extern void vp8_yv12_copy_src_frame_func_neon(YV12_BUFFER_CONFIG *src_ybc, YV12_BUFFER_CONFIG *dst_ybc);
#endif
int vp8_estimate_entropy_savings(VP8_COMP *cpi);
int vp8_calc_ss_err(YV12_BUFFER_CONFIG *source, YV12_BUFFER_CONFIG *dest, const vp8_variance_rtcd_vtable_t *rtcd);
extern void vp8_temporal_filter_prepare_c(VP8_COMP *cpi, int distance);
2010-05-18 19:58:33 +04:00
static void set_default_lf_deltas(VP8_COMP *cpi);
2010-05-18 19:58:33 +04:00
extern const int vp8_gf_interval_table[101];
#if CONFIG_INTERNAL_STATS
2010-05-18 19:58:33 +04:00
#include "math.h"
extern double vp8_calc_ssim
(
YV12_BUFFER_CONFIG *source,
YV12_BUFFER_CONFIG *dest,
int lumamask,
double *weight,
const vp8_variance_rtcd_vtable_t *rtcd
2010-05-18 19:58:33 +04:00
);
2010-05-18 19:58:33 +04:00
extern double vp8_calc_ssimg
(
YV12_BUFFER_CONFIG *source,
YV12_BUFFER_CONFIG *dest,
double *ssim_y,
double *ssim_u,
double *ssim_v,
const vp8_variance_rtcd_vtable_t *rtcd
2010-05-18 19:58:33 +04:00
);
#endif
#ifdef OUTPUT_YUV_SRC
FILE *yuv_file;
#endif
#if 0
FILE *framepsnr;
FILE *kf_list;
FILE *keyfile;
#endif
#if 0
extern int skip_true_count;
extern int skip_false_count;
#endif
#ifdef ENTROPY_STATS
extern int intra_mode_stats[10][10][10];
#endif
#ifdef SPEEDSTATS
unsigned int frames_at_speed[16] = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0};
unsigned int tot_pm = 0;
unsigned int cnt_pm = 0;
unsigned int tot_ef = 0;
unsigned int cnt_ef = 0;
#endif
#ifdef MODE_STATS
extern unsigned __int64 Sectionbits[50];
extern int y_modes[5] ;
extern int uv_modes[4] ;
extern int b_modes[10] ;
extern int inter_y_modes[10] ;
extern int inter_uv_modes[4] ;
extern unsigned int inter_b_modes[15];
#endif
extern void (*vp8_short_fdct4x4)(short *input, short *output, int pitch);
extern void (*vp8_short_fdct8x4)(short *input, short *output, int pitch);
extern const int vp8_bits_per_mb[2][QINDEX_RANGE];
extern const int qrounding_factors[129];
extern const int qzbin_factors[129];
extern void vp8cx_init_quantizer(VP8_COMP *cpi);
extern const int vp8cx_base_skip_false_prob[128];
// Tables relating active max Q to active min Q
static const int kf_low_motion_minq[QINDEX_RANGE] =
{
0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
0,0,0,0,1,1,1,1,1,1,1,1,2,2,2,2,
3,3,3,3,3,3,4,4,4,5,5,5,5,5,6,6,
6,6,7,7,8,8,8,8,9,9,10,10,10,10,11,11,
11,11,12,12,13,13,13,13,14,14,15,15,15,15,16,16,
16,16,17,17,18,18,18,18,19,20,20,21,21,22,23,23
};
static const int kf_high_motion_minq[QINDEX_RANGE] =
{
0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
1,1,1,1,1,1,1,1,2,2,2,2,3,3,3,3,
3,3,3,3,4,4,4,4,5,5,5,5,5,5,6,6,
6,6,7,7,8,8,8,8,9,9,10,10,10,10,11,11,
11,11,12,12,13,13,13,13,14,14,15,15,15,15,16,16,
16,16,17,17,18,18,18,18,19,19,20,20,20,20,21,21,
21,21,22,22,23,23,24,25,25,26,26,27,28,28,29,30
};
static const int gf_low_motion_minq[QINDEX_RANGE] =
{
0,0,0,0,1,1,1,1,1,1,1,1,2,2,2,2,
3,3,3,3,4,4,4,4,5,5,5,5,6,6,6,6,
7,7,7,7,8,8,8,8,9,9,9,9,10,10,10,10,
11,11,12,12,13,13,14,14,15,15,16,16,17,17,18,18,
19,19,20,20,21,21,22,22,23,23,24,24,25,25,26,26,
27,27,28,28,29,29,30,30,31,31,32,32,33,33,34,34,
35,35,36,36,37,37,38,38,39,39,40,40,41,41,42,42,
43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58
};
static const int gf_mid_motion_minq[QINDEX_RANGE] =
{
0,0,0,0,1,1,1,1,1,1,2,2,3,3,3,4,
4,4,5,5,5,6,6,6,7,7,7,8,8,8,9,9,
9,10,10,10,10,11,11,11,12,12,12,12,13,13,13,14,
14,14,15,15,16,16,17,17,18,18,19,19,20,20,21,21,
22,22,23,23,24,24,25,25,26,26,27,27,28,28,29,29,
30,30,31,31,32,32,33,33,34,34,35,35,36,36,37,37,
38,39,39,40,40,41,41,42,42,43,43,44,45,46,47,48,
49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64
};
static const int gf_high_motion_minq[QINDEX_RANGE] =
{
0,0,0,0,1,1,1,1,1,2,2,2,3,3,3,4,
4,4,5,5,5,6,6,6,7,7,7,8,8,8,9,9,
9,10,10,10,11,11,12,12,13,13,14,14,15,15,16,16,
17,17,18,18,19,19,20,20,21,21,22,22,23,23,24,24,
25,25,26,26,27,27,28,28,29,29,30,30,31,31,32,32,
33,33,34,34,35,35,36,36,37,37,38,38,39,39,40,40,
41,41,42,42,43,44,45,46,47,48,49,50,51,52,53,54,
55,56,57,58,59,60,62,64,66,68,70,72,74,76,78,80
};
static const int inter_minq[QINDEX_RANGE] =
{
0,0,1,1,2,3,3,4,4,5,6,6,7,8,8,9,
9,10,11,11,12,13,13,14,15,15,16,17,17,18,19,20,
20,21,22,22,23,24,24,25,26,27,27,28,29,30,30,31,
32,33,33,34,35,36,36,37,38,39,39,40,41,42,42,43,
44,45,46,46,47,48,49,50,50,51,52,53,54,55,55,56,
57,58,59,60,60,61,62,63,64,65,66,67,67,68,69,70,
71,72,73,74,75,75,76,77,78,79,80,81,82,83,84,85,
86,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100
};
2010-05-18 19:58:33 +04:00
void vp8_initialize()
{
static int init_done = 0;
if (!init_done)
{
vp8_scale_machine_specific_config();
vp8_initialize_common();
//vp8_dmachine_specific_config();
vp8_tokenize_initialize();
init_done = 1;
}
}
#ifdef PACKET_TESTING
extern FILE *vpxlogc;
#endif
static void setup_features(VP8_COMP *cpi)
{
// Set up default state for MB feature flags
cpi->mb.e_mbd.segmentation_enabled = 0;
cpi->mb.e_mbd.update_mb_segmentation_map = 0;
cpi->mb.e_mbd.update_mb_segmentation_data = 0;
vpx_memset(cpi->mb.e_mbd.mb_segment_tree_probs, 255, sizeof(cpi->mb.e_mbd.mb_segment_tree_probs));
vpx_memset(cpi->mb.e_mbd.segment_feature_data, 0, sizeof(cpi->mb.e_mbd.segment_feature_data));
cpi->mb.e_mbd.mode_ref_lf_delta_enabled = 0;
cpi->mb.e_mbd.mode_ref_lf_delta_update = 0;
vpx_memset(cpi->mb.e_mbd.ref_lf_deltas, 0, sizeof(cpi->mb.e_mbd.ref_lf_deltas));
vpx_memset(cpi->mb.e_mbd.mode_lf_deltas, 0, sizeof(cpi->mb.e_mbd.mode_lf_deltas));
vpx_memset(cpi->mb.e_mbd.last_ref_lf_deltas, 0, sizeof(cpi->mb.e_mbd.ref_lf_deltas));
vpx_memset(cpi->mb.e_mbd.last_mode_lf_deltas, 0, sizeof(cpi->mb.e_mbd.mode_lf_deltas));
2010-05-18 19:58:33 +04:00
set_default_lf_deltas(cpi);
2010-05-18 19:58:33 +04:00
}
static void dealloc_compressor_data(VP8_COMP *cpi)
2010-05-18 19:58:33 +04:00
{
vpx_free(cpi->tplist);
cpi->tplist = NULL;
// Delete last frame MV storage buffers
vpx_free(cpi->lfmv);
cpi->lfmv = 0;
vpx_free(cpi->lf_ref_frame_sign_bias);
cpi->lf_ref_frame_sign_bias = 0;
vpx_free(cpi->lf_ref_frame);
cpi->lf_ref_frame = 0;
2010-05-18 19:58:33 +04:00
// Delete sementation map
vpx_free(cpi->segmentation_map);
2010-05-18 19:58:33 +04:00
cpi->segmentation_map = 0;
vpx_free(cpi->active_map);
2010-05-18 19:58:33 +04:00
cpi->active_map = 0;
vp8_de_alloc_frame_buffers(&cpi->common);
vp8_yv12_de_alloc_frame_buffer(&cpi->last_frame_uf);
vp8_yv12_de_alloc_frame_buffer(&cpi->scaled_source);
#if VP8_TEMPORAL_ALT_REF
vp8_yv12_de_alloc_frame_buffer(&cpi->alt_ref_buffer);
2010-05-18 19:58:33 +04:00
#endif
vp8_lookahead_destroy(cpi->lookahead);
2010-05-18 19:58:33 +04:00
vpx_free(cpi->tok);
cpi->tok = 0;
// Structure used to monitor GF usage
vpx_free(cpi->gf_active_flags);
cpi->gf_active_flags = 0;
// Activity mask based per mb zbin adjustments
vpx_free(cpi->mb_activity_map);
cpi->mb_activity_map = 0;
vpx_free(cpi->mb_norm_activity_map);
cpi->mb_norm_activity_map = 0;
vpx_free(cpi->mb.pip);
cpi->mb.pip = 0;
#if !(CONFIG_REALTIME_ONLY)
vpx_free(cpi->twopass.total_stats);
cpi->twopass.total_stats = 0;
vpx_free(cpi->twopass.this_frame_stats);
cpi->twopass.this_frame_stats = 0;
#endif
2010-05-18 19:58:33 +04:00
}
static void enable_segmentation(VP8_PTR ptr)
{
VP8_COMP *cpi = (VP8_COMP *)(ptr);
// Set the appropriate feature bit
cpi->mb.e_mbd.segmentation_enabled = 1;
cpi->mb.e_mbd.update_mb_segmentation_map = 1;
cpi->mb.e_mbd.update_mb_segmentation_data = 1;
}
static void disable_segmentation(VP8_PTR ptr)
{
VP8_COMP *cpi = (VP8_COMP *)(ptr);
// Clear the appropriate feature bit
cpi->mb.e_mbd.segmentation_enabled = 0;
}
// Valid values for a segment are 0 to 3
// Segmentation map is arrange as [Rows][Columns]
static void set_segmentation_map(VP8_PTR ptr, unsigned char *segmentation_map)
{
VP8_COMP *cpi = (VP8_COMP *)(ptr);
// Copy in the new segmentation map
vpx_memcpy(cpi->segmentation_map, segmentation_map, (cpi->common.mb_rows * cpi->common.mb_cols));
// Signal that the map should be updated.
cpi->mb.e_mbd.update_mb_segmentation_map = 1;
cpi->mb.e_mbd.update_mb_segmentation_data = 1;
}
// The values given for each segment can be either deltas (from the default value chosen for the frame) or absolute values.
//
// Valid range for abs values is (0-127 for MB_LVL_ALT_Q) , (0-63 for SEGMENT_ALT_LF)
// Valid range for delta values are (+/-127 for MB_LVL_ALT_Q) , (+/-63 for SEGMENT_ALT_LF)
//
// abs_delta = SEGMENT_DELTADATA (deltas) abs_delta = SEGMENT_ABSDATA (use the absolute values given).
//
//
static void set_segment_data(VP8_PTR ptr, signed char *feature_data, unsigned char abs_delta)
{
VP8_COMP *cpi = (VP8_COMP *)(ptr);
cpi->mb.e_mbd.mb_segement_abs_delta = abs_delta;
vpx_memcpy(cpi->segment_feature_data, feature_data, sizeof(cpi->segment_feature_data));
}
static void segmentation_test_function(VP8_PTR ptr)
{
VP8_COMP *cpi = (VP8_COMP *)(ptr);
unsigned char *seg_map;
signed char feature_data[MB_LVL_MAX][MAX_MB_SEGMENTS];
// Create a temporary map for segmentation data.
CHECK_MEM_ERROR(seg_map, vpx_calloc(cpi->common.mb_rows * cpi->common.mb_cols, 1));
// MB loop to set local segmentation map
/*for ( i = 0; i < cpi->common.mb_rows; i++ )
{
for ( j = 0; j < cpi->common.mb_cols; j++ )
{
//seg_map[(i*cpi->common.mb_cols) + j] = (j % 2) + ((i%2)* 2);
//if ( j < cpi->common.mb_cols/2 )
// Segment 1 around the edge else 0
if ( (i == 0) || (j == 0) || (i == (cpi->common.mb_rows-1)) || (j == (cpi->common.mb_cols-1)) )
seg_map[(i*cpi->common.mb_cols) + j] = 1;
//else if ( (i < 2) || (j < 2) || (i > (cpi->common.mb_rows-3)) || (j > (cpi->common.mb_cols-3)) )
// seg_map[(i*cpi->common.mb_cols) + j] = 2;
//else if ( (i < 5) || (j < 5) || (i > (cpi->common.mb_rows-6)) || (j > (cpi->common.mb_cols-6)) )
// seg_map[(i*cpi->common.mb_cols) + j] = 3;
else
seg_map[(i*cpi->common.mb_cols) + j] = 0;
}
}*/
// Set the segmentation Map
set_segmentation_map(ptr, seg_map);
// Activate segmentation.
enable_segmentation(ptr);
// Set up the quant segment data
feature_data[MB_LVL_ALT_Q][0] = 0;
feature_data[MB_LVL_ALT_Q][1] = 4;
feature_data[MB_LVL_ALT_Q][2] = 0;
feature_data[MB_LVL_ALT_Q][3] = 0;
// Set up the loop segment data
feature_data[MB_LVL_ALT_LF][0] = 0;
feature_data[MB_LVL_ALT_LF][1] = 0;
feature_data[MB_LVL_ALT_LF][2] = 0;
feature_data[MB_LVL_ALT_LF][3] = 0;
// Initialise the feature data structure
// SEGMENT_DELTADATA 0, SEGMENT_ABSDATA 1
set_segment_data(ptr, &feature_data[0][0], SEGMENT_DELTADATA);
// Delete sementation map
vpx_free(seg_map);
seg_map = 0;
}
// A simple function to cyclically refresh the background at a lower Q
static void cyclic_background_refresh(VP8_COMP *cpi, int Q, int lf_adjustment)
{
unsigned char *seg_map;
signed char feature_data[MB_LVL_MAX][MAX_MB_SEGMENTS];
int i;
int block_count = cpi->cyclic_refresh_mode_max_mbs_perframe;
int mbs_in_frame = cpi->common.mb_rows * cpi->common.mb_cols;
// Create a temporary map for segmentation data.
CHECK_MEM_ERROR(seg_map, vpx_calloc(cpi->common.mb_rows * cpi->common.mb_cols, 1));
cpi->cyclic_refresh_q = Q;
for (i = Q; i > 0; i--)
{
if (vp8_bits_per_mb[cpi->common.frame_type][i] >= ((vp8_bits_per_mb[cpi->common.frame_type][Q]*(Q + 128)) / 64))
//if ( vp8_bits_per_mb[cpi->common.frame_type][i] >= ((vp8_bits_per_mb[cpi->common.frame_type][Q]*((2*Q)+96))/64) )
{
break;
}
}
cpi->cyclic_refresh_q = i;
// Only update for inter frames
if (cpi->common.frame_type != KEY_FRAME)
{
// Cycle through the macro_block rows
// MB loop to set local segmentation map
for (i = cpi->cyclic_refresh_mode_index; i < mbs_in_frame; i++)
{
// If the MB is as a candidate for clean up then mark it for possible boost/refresh (segment 1)
// The segment id may get reset to 0 later if the MB gets coded anything other than last frame 0,0
// as only (last frame 0,0) MBs are eligable for refresh : that is to say Mbs likely to be background blocks.
if (cpi->cyclic_refresh_map[i] == 0)
{
seg_map[i] = 1;
}
else
{
seg_map[i] = 0;
// Skip blocks that have been refreshed recently anyway.
if (cpi->cyclic_refresh_map[i] < 0)
//cpi->cyclic_refresh_map[i] = cpi->cyclic_refresh_map[i] / 16;
cpi->cyclic_refresh_map[i]++;
}
if (block_count > 0)
block_count--;
else
break;
}
// If we have gone through the frame reset to the start
cpi->cyclic_refresh_mode_index = i;
if (cpi->cyclic_refresh_mode_index >= mbs_in_frame)
cpi->cyclic_refresh_mode_index = 0;
}
// Set the segmentation Map
set_segmentation_map((VP8_PTR)cpi, seg_map);
// Activate segmentation.
enable_segmentation((VP8_PTR)cpi);
// Set up the quant segment data
feature_data[MB_LVL_ALT_Q][0] = 0;
feature_data[MB_LVL_ALT_Q][1] = (cpi->cyclic_refresh_q - Q);
feature_data[MB_LVL_ALT_Q][2] = 0;
feature_data[MB_LVL_ALT_Q][3] = 0;
// Set up the loop segment data
feature_data[MB_LVL_ALT_LF][0] = 0;
feature_data[MB_LVL_ALT_LF][1] = lf_adjustment;
feature_data[MB_LVL_ALT_LF][2] = 0;
feature_data[MB_LVL_ALT_LF][3] = 0;
// Initialise the feature data structure
// SEGMENT_DELTADATA 0, SEGMENT_ABSDATA 1
set_segment_data((VP8_PTR)cpi, &feature_data[0][0], SEGMENT_DELTADATA);
// Delete sementation map
vpx_free(seg_map);
seg_map = 0;
}
static void set_default_lf_deltas(VP8_COMP *cpi)
2010-05-18 19:58:33 +04:00
{
cpi->mb.e_mbd.mode_ref_lf_delta_enabled = 1;
cpi->mb.e_mbd.mode_ref_lf_delta_update = 1;
vpx_memset(cpi->mb.e_mbd.ref_lf_deltas, 0, sizeof(cpi->mb.e_mbd.ref_lf_deltas));
vpx_memset(cpi->mb.e_mbd.mode_lf_deltas, 0, sizeof(cpi->mb.e_mbd.mode_lf_deltas));
// Test of ref frame deltas
cpi->mb.e_mbd.ref_lf_deltas[INTRA_FRAME] = 2;
cpi->mb.e_mbd.ref_lf_deltas[LAST_FRAME] = 0;
cpi->mb.e_mbd.ref_lf_deltas[GOLDEN_FRAME] = -2;
cpi->mb.e_mbd.ref_lf_deltas[ALTREF_FRAME] = -2;
cpi->mb.e_mbd.mode_lf_deltas[0] = 4; // BPRED
cpi->mb.e_mbd.mode_lf_deltas[1] = -2; // Zero
cpi->mb.e_mbd.mode_lf_deltas[2] = 2; // New mv
cpi->mb.e_mbd.mode_lf_deltas[3] = 4; // Split mv
}
void vp8_set_speed_features(VP8_COMP *cpi)
{
SPEED_FEATURES *sf = &cpi->sf;
int Mode = cpi->compressor_speed;
int Speed = cpi->Speed;
int i;
VP8_COMMON *cm = &cpi->common;
int last_improved_quant = sf->improved_quant;
2010-05-18 19:58:33 +04:00
// Initialise default mode frequency sampling variables
for (i = 0; i < MAX_MODES; i ++)
{
cpi->mode_check_freq[i] = 0;
cpi->mode_test_hit_counts[i] = 0;
cpi->mode_chosen_counts[i] = 0;
}
cpi->mbs_tested_so_far = 0;
// best quality defaults
2010-05-18 19:58:33 +04:00
sf->RD = 1;
sf->search_method = NSTEP;
sf->improved_quant = 1;
sf->improved_dct = 1;
sf->auto_filter = 1;
sf->recode_loop = 1;
sf->quarter_pixel_search = 1;
sf->half_pixel_search = 1;
sf->iterative_sub_pixel = 1;
sf->optimize_coefficients = 1;
sf->use_fastquant_for_pick = 0;
sf->no_skip_block4x4_search = 1;
2010-05-18 19:58:33 +04:00
sf->first_step = 0;
sf->max_step_search_steps = MAX_MVSEARCH_STEPS;
sf->improved_mv_pred = 1;
2010-05-18 19:58:33 +04:00
// default thresholds to 0
for (i = 0; i < MAX_MODES; i++)
sf->thresh_mult[i] = 0;
switch (Mode)
{
#if !(CONFIG_REALTIME_ONLY)
case 0: // best quality mode
sf->thresh_mult[THR_ZEROMV ] = 0;
sf->thresh_mult[THR_ZEROG ] = 0;
sf->thresh_mult[THR_ZEROA ] = 0;
sf->thresh_mult[THR_NEARESTMV] = 0;
sf->thresh_mult[THR_NEARESTG ] = 0;
sf->thresh_mult[THR_NEARESTA ] = 0;
sf->thresh_mult[THR_NEARMV ] = 0;
sf->thresh_mult[THR_NEARG ] = 0;
sf->thresh_mult[THR_NEARA ] = 0;
sf->thresh_mult[THR_DC ] = 0;
sf->thresh_mult[THR_V_PRED ] = 1000;
sf->thresh_mult[THR_H_PRED ] = 1000;
sf->thresh_mult[THR_B_PRED ] = 2000;
sf->thresh_mult[THR_TM ] = 1000;
sf->thresh_mult[THR_NEWMV ] = 1000;
sf->thresh_mult[THR_NEWG ] = 1000;
sf->thresh_mult[THR_NEWA ] = 1000;
sf->thresh_mult[THR_SPLITMV ] = 2500;
sf->thresh_mult[THR_SPLITG ] = 5000;
sf->thresh_mult[THR_SPLITA ] = 5000;
sf->first_step = 0;
sf->max_step_search_steps = MAX_MVSEARCH_STEPS;
break;
case 1:
case 3:
sf->thresh_mult[THR_NEARESTMV] = 0;
sf->thresh_mult[THR_ZEROMV ] = 0;
sf->thresh_mult[THR_DC ] = 0;
sf->thresh_mult[THR_NEARMV ] = 0;
sf->thresh_mult[THR_V_PRED ] = 1000;
sf->thresh_mult[THR_H_PRED ] = 1000;
sf->thresh_mult[THR_B_PRED ] = 2500;
sf->thresh_mult[THR_TM ] = 1000;
sf->thresh_mult[THR_NEARESTG ] = 1000;
sf->thresh_mult[THR_NEARESTA ] = 1000;
sf->thresh_mult[THR_ZEROG ] = 1000;
sf->thresh_mult[THR_ZEROA ] = 1000;
sf->thresh_mult[THR_NEARG ] = 1000;
sf->thresh_mult[THR_NEARA ] = 1000;
#if 1
sf->thresh_mult[THR_ZEROMV ] = 0;
sf->thresh_mult[THR_ZEROG ] = 0;
sf->thresh_mult[THR_ZEROA ] = 0;
sf->thresh_mult[THR_NEARESTMV] = 0;
sf->thresh_mult[THR_NEARESTG ] = 0;
sf->thresh_mult[THR_NEARESTA ] = 0;
sf->thresh_mult[THR_NEARMV ] = 0;
sf->thresh_mult[THR_NEARG ] = 0;
sf->thresh_mult[THR_NEARA ] = 0;
// sf->thresh_mult[THR_DC ] = 0;
// sf->thresh_mult[THR_V_PRED ] = 1000;
// sf->thresh_mult[THR_H_PRED ] = 1000;
// sf->thresh_mult[THR_B_PRED ] = 2000;
// sf->thresh_mult[THR_TM ] = 1000;
sf->thresh_mult[THR_NEWMV ] = 1000;
sf->thresh_mult[THR_NEWG ] = 1000;
sf->thresh_mult[THR_NEWA ] = 1000;
sf->thresh_mult[THR_SPLITMV ] = 1700;
sf->thresh_mult[THR_SPLITG ] = 4500;
sf->thresh_mult[THR_SPLITA ] = 4500;
#else
2010-05-18 19:58:33 +04:00
sf->thresh_mult[THR_NEWMV ] = 1500;
sf->thresh_mult[THR_NEWG ] = 1500;
sf->thresh_mult[THR_NEWA ] = 1500;
sf->thresh_mult[THR_SPLITMV ] = 5000;
sf->thresh_mult[THR_SPLITG ] = 10000;
sf->thresh_mult[THR_SPLITA ] = 10000;
#endif
2010-05-18 19:58:33 +04:00
if (Speed > 0)
2010-05-18 19:58:33 +04:00
{
/* Disable coefficient optimization above speed 0 */
sf->optimize_coefficients = 0;
sf->use_fastquant_for_pick = 1;
sf->no_skip_block4x4_search = 0;
sf->first_step = 1;
cpi->mode_check_freq[THR_SPLITG] = 2;
cpi->mode_check_freq[THR_SPLITA] = 2;
cpi->mode_check_freq[THR_SPLITMV] = 0;
2010-05-18 19:58:33 +04:00
}
if (Speed > 1)
2010-05-18 19:58:33 +04:00
{
cpi->mode_check_freq[THR_SPLITG] = 4;
cpi->mode_check_freq[THR_SPLITA] = 4;
cpi->mode_check_freq[THR_SPLITMV] = 2;
2010-05-18 19:58:33 +04:00
sf->thresh_mult[THR_TM ] = 1500;
sf->thresh_mult[THR_V_PRED ] = 1500;
sf->thresh_mult[THR_H_PRED ] = 1500;
sf->thresh_mult[THR_B_PRED ] = 5000;
if (cpi->ref_frame_flags & VP8_LAST_FLAG)
{
sf->thresh_mult[THR_NEWMV ] = 2000;
sf->thresh_mult[THR_SPLITMV ] = 10000;
}
if (cpi->ref_frame_flags & VP8_GOLD_FLAG)
{
sf->thresh_mult[THR_NEARESTG ] = 1500;
sf->thresh_mult[THR_ZEROG ] = 1500;
sf->thresh_mult[THR_NEARG ] = 1500;
sf->thresh_mult[THR_NEWG ] = 2000;
sf->thresh_mult[THR_SPLITG ] = 20000;
}
if (cpi->ref_frame_flags & VP8_ALT_FLAG)
{
sf->thresh_mult[THR_NEARESTA ] = 1500;
sf->thresh_mult[THR_ZEROA ] = 1500;
sf->thresh_mult[THR_NEARA ] = 1500;
sf->thresh_mult[THR_NEWA ] = 2000;
sf->thresh_mult[THR_SPLITA ] = 20000;
}
}
if (Speed > 2)
2010-05-18 19:58:33 +04:00
{
cpi->mode_check_freq[THR_SPLITG] = 15;
cpi->mode_check_freq[THR_SPLITA] = 15;
cpi->mode_check_freq[THR_SPLITMV] = 7;
sf->thresh_mult[THR_TM ] = 2000;
sf->thresh_mult[THR_V_PRED ] = 2000;
sf->thresh_mult[THR_H_PRED ] = 2000;
sf->thresh_mult[THR_B_PRED ] = 7500;
if (cpi->ref_frame_flags & VP8_LAST_FLAG)
{
sf->thresh_mult[THR_NEWMV ] = 2000;
sf->thresh_mult[THR_SPLITMV ] = 25000;
}
if (cpi->ref_frame_flags & VP8_GOLD_FLAG)
{
sf->thresh_mult[THR_NEARESTG ] = 2000;
sf->thresh_mult[THR_ZEROG ] = 2000;
sf->thresh_mult[THR_NEARG ] = 2000;
sf->thresh_mult[THR_NEWG ] = 2500;
sf->thresh_mult[THR_SPLITG ] = 50000;
}
if (cpi->ref_frame_flags & VP8_ALT_FLAG)
{
sf->thresh_mult[THR_NEARESTA ] = 2000;
sf->thresh_mult[THR_ZEROA ] = 2000;
sf->thresh_mult[THR_NEARA ] = 2000;
sf->thresh_mult[THR_NEWA ] = 2500;
sf->thresh_mult[THR_SPLITA ] = 50000;
}
sf->improved_quant = 0;
sf->improved_dct = 0;
// Only do recode loop on key frames, golden frames and
// alt ref frames
2010-05-18 19:58:33 +04:00
sf->recode_loop = 2;
}
if (Speed > 3)
2010-05-18 19:58:33 +04:00
{
sf->thresh_mult[THR_SPLITA ] = INT_MAX;
sf->thresh_mult[THR_SPLITG ] = INT_MAX;
sf->thresh_mult[THR_SPLITMV ] = INT_MAX;
cpi->mode_check_freq[THR_V_PRED] = 0;
cpi->mode_check_freq[THR_H_PRED] = 0;
cpi->mode_check_freq[THR_B_PRED] = 0;
cpi->mode_check_freq[THR_NEARG] = 0;
cpi->mode_check_freq[THR_NEWG] = 0;
cpi->mode_check_freq[THR_NEARA] = 0;
cpi->mode_check_freq[THR_NEWA] = 0;
sf->auto_filter = 1;
sf->recode_loop = 0; // recode loop off
sf->RD = 0; // Turn rd off
2010-05-18 19:58:33 +04:00
}
if (Speed > 4)
{
sf->auto_filter = 0; // Faster selection of loop filter
cpi->mode_check_freq[THR_V_PRED] = 2;
cpi->mode_check_freq[THR_H_PRED] = 2;
cpi->mode_check_freq[THR_B_PRED] = 2;
if (cpi->ref_frame_flags & VP8_GOLD_FLAG)
{
cpi->mode_check_freq[THR_NEARG] = 2;
cpi->mode_check_freq[THR_NEWG] = 4;
}
if (cpi->ref_frame_flags & VP8_ALT_FLAG)
{
cpi->mode_check_freq[THR_NEARA] = 2;
cpi->mode_check_freq[THR_NEWA] = 4;
}
if (cpi->ref_frame_flags & VP8_GOLD_FLAG)
2010-05-18 19:58:33 +04:00
{
sf->thresh_mult[THR_NEARESTG ] = 2000;
sf->thresh_mult[THR_ZEROG ] = 2000;
sf->thresh_mult[THR_NEARG ] = 2000;
sf->thresh_mult[THR_NEWG ] = 4000;
}
if (cpi->ref_frame_flags & VP8_ALT_FLAG)
2010-05-18 19:58:33 +04:00
{
sf->thresh_mult[THR_NEARESTA ] = 2000;
sf->thresh_mult[THR_ZEROA ] = 2000;
sf->thresh_mult[THR_NEARA ] = 2000;
sf->thresh_mult[THR_NEWA ] = 4000;
}
}
break;
#endif
case 2:
sf->optimize_coefficients = 0;
sf->recode_loop = 0;
sf->auto_filter = 1;
sf->iterative_sub_pixel = 1;
sf->thresh_mult[THR_NEARESTMV] = 0;
sf->thresh_mult[THR_ZEROMV ] = 0;
sf->thresh_mult[THR_DC ] = 0;
sf->thresh_mult[THR_TM ] = 0;
sf->thresh_mult[THR_NEARMV ] = 0;
sf->thresh_mult[THR_V_PRED ] = 1000;
sf->thresh_mult[THR_H_PRED ] = 1000;
sf->thresh_mult[THR_B_PRED ] = 2500;
sf->thresh_mult[THR_NEARESTG ] = 1000;
sf->thresh_mult[THR_ZEROG ] = 1000;
sf->thresh_mult[THR_NEARG ] = 1000;
sf->thresh_mult[THR_NEARESTA ] = 1000;
sf->thresh_mult[THR_ZEROA ] = 1000;
sf->thresh_mult[THR_NEARA ] = 1000;
sf->thresh_mult[THR_NEWMV ] = 2000;
sf->thresh_mult[THR_NEWG ] = 2000;
sf->thresh_mult[THR_NEWA ] = 2000;
sf->thresh_mult[THR_SPLITMV ] = 5000;
sf->thresh_mult[THR_SPLITG ] = 10000;
sf->thresh_mult[THR_SPLITA ] = 10000;
sf->search_method = NSTEP;
if (Speed > 0)
{
cpi->mode_check_freq[THR_SPLITG] = 4;
cpi->mode_check_freq[THR_SPLITA] = 4;
cpi->mode_check_freq[THR_SPLITMV] = 2;
sf->thresh_mult[THR_DC ] = 0;
sf->thresh_mult[THR_TM ] = 1000;
sf->thresh_mult[THR_V_PRED ] = 2000;
sf->thresh_mult[THR_H_PRED ] = 2000;
sf->thresh_mult[THR_B_PRED ] = 5000;
if (cpi->ref_frame_flags & VP8_LAST_FLAG)
{
sf->thresh_mult[THR_NEARESTMV] = 0;
sf->thresh_mult[THR_ZEROMV ] = 0;
sf->thresh_mult[THR_NEARMV ] = 0;
sf->thresh_mult[THR_NEWMV ] = 2000;
sf->thresh_mult[THR_SPLITMV ] = 10000;
}
if (cpi->ref_frame_flags & VP8_GOLD_FLAG)
{
sf->thresh_mult[THR_NEARESTG ] = 1000;
sf->thresh_mult[THR_ZEROG ] = 1000;
sf->thresh_mult[THR_NEARG ] = 1000;
sf->thresh_mult[THR_NEWG ] = 2000;
sf->thresh_mult[THR_SPLITG ] = 20000;
}
if (cpi->ref_frame_flags & VP8_ALT_FLAG)
{
sf->thresh_mult[THR_NEARESTA ] = 1000;
sf->thresh_mult[THR_ZEROA ] = 1000;
sf->thresh_mult[THR_NEARA ] = 1000;
sf->thresh_mult[THR_NEWA ] = 2000;
sf->thresh_mult[THR_SPLITA ] = 20000;
}
sf->improved_quant = 0;
sf->improved_dct = 0;
}
if (Speed > 1)
{
cpi->mode_check_freq[THR_SPLITMV] = 7;
cpi->mode_check_freq[THR_SPLITG] = 15;
cpi->mode_check_freq[THR_SPLITA] = 15;
sf->thresh_mult[THR_TM ] = 2000;
sf->thresh_mult[THR_V_PRED ] = 2000;
sf->thresh_mult[THR_H_PRED ] = 2000;
sf->thresh_mult[THR_B_PRED ] = 5000;
if (cpi->ref_frame_flags & VP8_LAST_FLAG)
{
sf->thresh_mult[THR_NEWMV ] = 2000;
sf->thresh_mult[THR_SPLITMV ] = 25000;
}
if (cpi->ref_frame_flags & VP8_GOLD_FLAG)
{
sf->thresh_mult[THR_NEARESTG ] = 2000;
sf->thresh_mult[THR_ZEROG ] = 2000;
sf->thresh_mult[THR_NEARG ] = 2000;
sf->thresh_mult[THR_NEWG ] = 2500;
sf->thresh_mult[THR_SPLITG ] = 50000;
}
if (cpi->ref_frame_flags & VP8_ALT_FLAG)
{
sf->thresh_mult[THR_NEARESTA ] = 2000;
sf->thresh_mult[THR_ZEROA ] = 2000;
sf->thresh_mult[THR_NEARA ] = 2000;
sf->thresh_mult[THR_NEWA ] = 2500;
sf->thresh_mult[THR_SPLITA ] = 50000;
}
}
if (Speed > 2)
{
sf->auto_filter = 0; // Faster selection of loop filter
cpi->mode_check_freq[THR_V_PRED] = 2;
cpi->mode_check_freq[THR_H_PRED] = 2;
cpi->mode_check_freq[THR_B_PRED] = 2;
if (cpi->ref_frame_flags & VP8_GOLD_FLAG)
{
cpi->mode_check_freq[THR_NEARG] = 2;
cpi->mode_check_freq[THR_NEWG] = 4;
}
if (cpi->ref_frame_flags & VP8_ALT_FLAG)
{
cpi->mode_check_freq[THR_NEARA] = 2;
cpi->mode_check_freq[THR_NEWA] = 4;
}
sf->thresh_mult[THR_SPLITMV ] = INT_MAX;
sf->thresh_mult[THR_SPLITG ] = INT_MAX;
sf->thresh_mult[THR_SPLITA ] = INT_MAX;
}
if (Speed > 3)
{
sf->RD = 0;
sf->auto_filter = 1;
}
if (Speed > 4)
{
sf->auto_filter = 0; // Faster selection of loop filter
sf->search_method = HEX;
//sf->search_method = DIAMOND;
sf->iterative_sub_pixel = 0;
2010-05-18 19:58:33 +04:00
cpi->mode_check_freq[THR_V_PRED] = 4;
cpi->mode_check_freq[THR_H_PRED] = 4;
cpi->mode_check_freq[THR_B_PRED] = 4;
if (cpi->ref_frame_flags & VP8_GOLD_FLAG)
{
cpi->mode_check_freq[THR_NEARG] = 2;
cpi->mode_check_freq[THR_NEWG] = 4;
}
if (cpi->ref_frame_flags & VP8_ALT_FLAG)
{
cpi->mode_check_freq[THR_NEARA] = 2;
cpi->mode_check_freq[THR_NEWA] = 4;
}
sf->thresh_mult[THR_TM ] = 2000;
sf->thresh_mult[THR_B_PRED ] = 5000;
if (cpi->ref_frame_flags & VP8_GOLD_FLAG)
{
sf->thresh_mult[THR_NEARESTG ] = 2000;
sf->thresh_mult[THR_ZEROG ] = 2000;
sf->thresh_mult[THR_NEARG ] = 2000;
sf->thresh_mult[THR_NEWG ] = 4000;
}
if (cpi->ref_frame_flags & VP8_ALT_FLAG)
{
sf->thresh_mult[THR_NEARESTA ] = 2000;
sf->thresh_mult[THR_ZEROA ] = 2000;
sf->thresh_mult[THR_NEARA ] = 2000;
sf->thresh_mult[THR_NEWA ] = 4000;
}
}
if (Speed > 5)
{
// Disable split MB intra prediction mode
sf->thresh_mult[THR_B_PRED] = INT_MAX;
}
if (Speed > 6)
{
unsigned int i, sum = 0;
unsigned int total_mbs = cm->MBs;
int thresh;
int total_skip;
int min = 2000;
if (cpi->oxcf.encode_breakout > 2000)
min = cpi->oxcf.encode_breakout;
min >>= 7;
for (i = 0; i < min; i++)
{
sum += cpi->error_bins[i];
}
total_skip = sum;
sum = 0;
// i starts from 2 to make sure thresh started from 2048
for (; i < 1024; i++)
{
sum += cpi->error_bins[i];
if (10 * sum >= (unsigned int)(cpi->Speed - 6)*(total_mbs - total_skip))
break;
}
i--;
thresh = (i << 7);
if (thresh < 2000)
thresh = 2000;
if (cpi->ref_frame_flags & VP8_LAST_FLAG)
{
sf->thresh_mult[THR_NEWMV] = thresh;
sf->thresh_mult[THR_NEARESTMV ] = thresh >> 1;
sf->thresh_mult[THR_NEARMV ] = thresh >> 1;
}
if (cpi->ref_frame_flags & VP8_GOLD_FLAG)
{
sf->thresh_mult[THR_NEWG] = thresh << 1;
sf->thresh_mult[THR_NEARESTG ] = thresh;
sf->thresh_mult[THR_NEARG ] = thresh;
}
if (cpi->ref_frame_flags & VP8_ALT_FLAG)
{
sf->thresh_mult[THR_NEWA] = thresh << 1;
sf->thresh_mult[THR_NEARESTA ] = thresh;
sf->thresh_mult[THR_NEARA ] = thresh;
}
// Disable other intra prediction modes
sf->thresh_mult[THR_TM] = INT_MAX;
sf->thresh_mult[THR_V_PRED] = INT_MAX;
sf->thresh_mult[THR_H_PRED] = INT_MAX;
sf->improved_mv_pred = 0;
2010-05-18 19:58:33 +04:00
}
if (Speed > 8)
{
sf->quarter_pixel_search = 0;
}
if (Speed > 9)
{
int Tmp = cpi->Speed - 8;
if (Tmp > 4)
Tmp = 4;
if (cpi->ref_frame_flags & VP8_GOLD_FLAG)
{
cpi->mode_check_freq[THR_ZEROG] = 1 << (Tmp - 1);
cpi->mode_check_freq[THR_NEARESTG] = 1 << (Tmp - 1);
cpi->mode_check_freq[THR_NEARG] = 1 << Tmp;
cpi->mode_check_freq[THR_NEWG] = 1 << (Tmp + 1);
}
if (cpi->ref_frame_flags & VP8_ALT_FLAG)
{
cpi->mode_check_freq[THR_ZEROA] = 1 << (Tmp - 1);
cpi->mode_check_freq[THR_NEARESTA] = 1 << (Tmp - 1);
cpi->mode_check_freq[THR_NEARA] = 1 << Tmp;
cpi->mode_check_freq[THR_NEWA] = 1 << (Tmp + 1);
}
cpi->mode_check_freq[THR_NEWMV] = 1 << (Tmp - 1);
}
cm->filter_type = NORMAL_LOOPFILTER;
if (Speed >= 14)
cm->filter_type = SIMPLE_LOOPFILTER;
if (Speed >= 15)
{
sf->half_pixel_search = 0; // This has a big hit on quality. Last resort
}
vpx_memset(cpi->error_bins, 0, sizeof(cpi->error_bins));
}; /* switch */
/* disable frame modes if flags not set */
if (!(cpi->ref_frame_flags & VP8_LAST_FLAG))
{
sf->thresh_mult[THR_NEWMV ] = INT_MAX;
sf->thresh_mult[THR_NEARESTMV] = INT_MAX;
sf->thresh_mult[THR_ZEROMV ] = INT_MAX;
sf->thresh_mult[THR_NEARMV ] = INT_MAX;
sf->thresh_mult[THR_SPLITMV ] = INT_MAX;
}
if (!(cpi->ref_frame_flags & VP8_GOLD_FLAG))
{
sf->thresh_mult[THR_NEARESTG ] = INT_MAX;
sf->thresh_mult[THR_ZEROG ] = INT_MAX;
sf->thresh_mult[THR_NEARG ] = INT_MAX;
sf->thresh_mult[THR_NEWG ] = INT_MAX;
sf->thresh_mult[THR_SPLITG ] = INT_MAX;
}
if (!(cpi->ref_frame_flags & VP8_ALT_FLAG))
{
sf->thresh_mult[THR_NEARESTA ] = INT_MAX;
sf->thresh_mult[THR_ZEROA ] = INT_MAX;
sf->thresh_mult[THR_NEARA ] = INT_MAX;
sf->thresh_mult[THR_NEWA ] = INT_MAX;
sf->thresh_mult[THR_SPLITA ] = INT_MAX;
}
2010-05-18 19:58:33 +04:00
// Slow quant, dct and trellis not worthwhile for first pass
// so make sure they are always turned off.
if ( cpi->pass == 1 )
{
sf->improved_quant = 0;
sf->optimize_coefficients = 0;
sf->improved_dct = 0;
}
2010-05-18 19:58:33 +04:00
if (cpi->sf.search_method == NSTEP)
{
vp8_init3smotion_compensation(&cpi->mb, cm->yv12_fb[cm->lst_fb_idx].y_stride);
2010-05-18 19:58:33 +04:00
}
else if (cpi->sf.search_method == DIAMOND)
{
vp8_init_dsmotion_compensation(&cpi->mb, cm->yv12_fb[cm->lst_fb_idx].y_stride);
2010-05-18 19:58:33 +04:00
}
if (cpi->sf.improved_dct)
{
cpi->mb.vp8_short_fdct8x4 = FDCT_INVOKE(&cpi->rtcd.fdct, short8x4);
cpi->mb.vp8_short_fdct4x4 = FDCT_INVOKE(&cpi->rtcd.fdct, short4x4);
}
else
{
cpi->mb.vp8_short_fdct8x4 = FDCT_INVOKE(&cpi->rtcd.fdct, fast8x4);
cpi->mb.vp8_short_fdct4x4 = FDCT_INVOKE(&cpi->rtcd.fdct, fast4x4);
}
cpi->mb.short_walsh4x4 = FDCT_INVOKE(&cpi->rtcd.fdct, walsh_short4x4);
if (cpi->sf.improved_quant)
{
cpi->mb.quantize_b = QUANTIZE_INVOKE(&cpi->rtcd.quantize,
quantb);
cpi->mb.quantize_b_pair = QUANTIZE_INVOKE(&cpi->rtcd.quantize,
quantb_pair);
2010-05-18 19:58:33 +04:00
}
else
{
cpi->mb.quantize_b = QUANTIZE_INVOKE(&cpi->rtcd.quantize,
fastquantb);
cpi->mb.quantize_b_pair = QUANTIZE_INVOKE(&cpi->rtcd.quantize,
fastquantb_pair);
2010-05-18 19:58:33 +04:00
}
if (cpi->sf.improved_quant != last_improved_quant)
vp8cx_init_quantizer(cpi);
2010-05-18 19:58:33 +04:00
#if CONFIG_RUNTIME_CPU_DETECT
cpi->mb.e_mbd.rtcd = &cpi->common.rtcd;
#endif
if (cpi->sf.iterative_sub_pixel == 1)
{
cpi->find_fractional_mv_step = vp8_find_best_sub_pixel_step_iteratively;
}
else if (cpi->sf.quarter_pixel_search)
{
cpi->find_fractional_mv_step = vp8_find_best_sub_pixel_step;
}
else if (cpi->sf.half_pixel_search)
{
cpi->find_fractional_mv_step = vp8_find_best_half_pixel_step;
}
else
{
cpi->find_fractional_mv_step = vp8_skip_fractional_mv_step;
}
if (cpi->sf.optimize_coefficients == 1 && cpi->pass!=1)
cpi->mb.optimize = 1;
2010-05-18 19:58:33 +04:00
else
cpi->mb.optimize = 0;
if (cpi->common.full_pixel)
cpi->find_fractional_mv_step = vp8_skip_fractional_mv_step;
#ifdef SPEEDSTATS
frames_at_speed[cpi->Speed]++;
#endif
}
static void alloc_raw_frame_buffers(VP8_COMP *cpi)
{
int width = (cpi->oxcf.Width + 15) & ~15;
int height = (cpi->oxcf.Height + 15) & ~15;
2010-05-18 19:58:33 +04:00
cpi->lookahead = vp8_lookahead_init(cpi->oxcf.Width, cpi->oxcf.Height,
cpi->oxcf.lag_in_frames);
if(!cpi->lookahead)
vpx_internal_error(&cpi->common.error, VPX_CODEC_MEM_ERROR,
"Failed to allocate lag buffers");
2010-05-18 19:58:33 +04:00
#if VP8_TEMPORAL_ALT_REF
if (vp8_yv12_alloc_frame_buffer(&cpi->alt_ref_buffer,
width, height, 16))
2010-05-18 19:58:33 +04:00
vpx_internal_error(&cpi->common.error, VPX_CODEC_MEM_ERROR,
"Failed to allocate altref buffer");
#endif
}
static int vp8_alloc_partition_data(VP8_COMP *cpi)
{
vpx_free(cpi->mb.pip);
cpi->mb.pip = vpx_calloc((cpi->common.mb_cols + 1) *
(cpi->common.mb_rows + 1),
sizeof(PARTITION_INFO));
if(!cpi->mb.pip)
return 1;
cpi->mb.pi = cpi->mb.pip + cpi->common.mode_info_stride + 1;
return 0;
}
2010-05-18 19:58:33 +04:00
void vp8_alloc_compressor_data(VP8_COMP *cpi)
{
VP8_COMMON *cm = & cpi->common;
int width = cm->Width;
int height = cm->Height;
if (vp8_alloc_frame_buffers(cm, width, height))
vpx_internal_error(&cpi->common.error, VPX_CODEC_MEM_ERROR,
"Failed to allocate frame buffers");
if (vp8_alloc_partition_data(cpi))
vpx_internal_error(&cpi->common.error, VPX_CODEC_MEM_ERROR,
"Failed to allocate partition data");
2010-05-18 19:58:33 +04:00
if ((width & 0xf) != 0)
width += 16 - (width & 0xf);
if ((height & 0xf) != 0)
height += 16 - (height & 0xf);
if (vp8_yv12_alloc_frame_buffer(&cpi->last_frame_uf,
width, height, VP8BORDERINPIXELS))
vpx_internal_error(&cpi->common.error, VPX_CODEC_MEM_ERROR,
"Failed to allocate last frame buffer");
if (vp8_yv12_alloc_frame_buffer(&cpi->scaled_source, width, height, 16))
vpx_internal_error(&cpi->common.error, VPX_CODEC_MEM_ERROR,
"Failed to allocate scaled source buffer");
vpx_free(cpi->tok);
{
unsigned int tokens = cm->mb_rows * cm->mb_cols * 24 * 16;
CHECK_MEM_ERROR(cpi->tok, vpx_calloc(tokens, sizeof(*cpi->tok)));
}
// Data used for real time vc mode to see if gf needs refreshing
cpi->inter_zz_count = 0;
cpi->gf_bad_count = 0;
cpi->gf_update_recommended = 0;
// Structures used to minitor GF usage
vpx_free(cpi->gf_active_flags);
CHECK_MEM_ERROR(cpi->gf_active_flags,
vpx_calloc(1, cm->mb_rows * cm->mb_cols));
cpi->gf_active_count = cm->mb_rows * cm->mb_cols;
vpx_free(cpi->mb_activity_map);
CHECK_MEM_ERROR(cpi->mb_activity_map,
vpx_calloc(sizeof(unsigned int),
cm->mb_rows * cm->mb_cols));
vpx_free(cpi->mb_norm_activity_map);
CHECK_MEM_ERROR(cpi->mb_norm_activity_map,
vpx_calloc(sizeof(unsigned int),
cm->mb_rows * cm->mb_cols));
#if !(CONFIG_REALTIME_ONLY)
vpx_free(cpi->twopass.total_stats);
cpi->twopass.total_stats = vpx_calloc(1, sizeof(FIRSTPASS_STATS));
vpx_free(cpi->twopass.this_frame_stats);
cpi->twopass.this_frame_stats = vpx_calloc(1, sizeof(FIRSTPASS_STATS));
if(!cpi->twopass.total_stats || !cpi->twopass.this_frame_stats)
vpx_internal_error(&cpi->common.error, VPX_CODEC_MEM_ERROR,
"Failed to allocate firstpass stats");
#endif
#if CONFIG_MULTITHREAD
if (width < 640)
cpi->mt_sync_range = 1;
else if (width <= 1280)
cpi->mt_sync_range = 4;
else if (width <= 2560)
cpi->mt_sync_range = 8;
else
cpi->mt_sync_range = 16;
#endif
vpx_free(cpi->tplist);
CHECK_MEM_ERROR(cpi->tplist, vpx_malloc(sizeof(TOKENLIST) * cpi->common.mb_rows));
2010-05-18 19:58:33 +04:00
}
// Quant MOD
static const int q_trans[] =
{
0, 1, 2, 3, 4, 5, 7, 8,
9, 10, 12, 13, 15, 17, 18, 19,
20, 21, 23, 24, 25, 26, 27, 28,
29, 30, 31, 33, 35, 37, 39, 41,
43, 45, 47, 49, 51, 53, 55, 57,
59, 61, 64, 67, 70, 73, 76, 79,
82, 85, 88, 91, 94, 97, 100, 103,
106, 109, 112, 115, 118, 121, 124, 127,
};
int vp8_reverse_trans(int x)
{
int i;
for (i = 0; i < 64; i++)
if (q_trans[i] >= x)
return i;
return 63;
};
void vp8_new_frame_rate(VP8_COMP *cpi, double framerate)
{
2010-10-15 00:19:06 +04:00
if(framerate < .1)
framerate = 30;
2010-05-18 19:58:33 +04:00
cpi->oxcf.frame_rate = framerate;
cpi->output_frame_rate = cpi->oxcf.frame_rate;
cpi->per_frame_bandwidth = (int)(cpi->oxcf.target_bandwidth / cpi->output_frame_rate);
cpi->av_per_frame_bandwidth = (int)(cpi->oxcf.target_bandwidth / cpi->output_frame_rate);
cpi->min_frame_bandwidth = (int)(cpi->av_per_frame_bandwidth * cpi->oxcf.two_pass_vbrmin_section / 100);
// Set Maximum gf/arf interval
cpi->max_gf_interval = ((int)(cpi->output_frame_rate / 2.0) + 2);
if(cpi->max_gf_interval < 12)
cpi->max_gf_interval = 12;
2010-05-18 19:58:33 +04:00
// Extended interval for genuinely static scenes
cpi->twopass.static_scene_max_gf_interval = cpi->key_frame_frequency >> 1;
2010-05-18 19:58:33 +04:00
// Special conditions when altr ref frame enabled in lagged compress mode
if (cpi->oxcf.play_alternate && cpi->oxcf.lag_in_frames)
2010-05-18 19:58:33 +04:00
{
if (cpi->max_gf_interval > cpi->oxcf.lag_in_frames - 1)
cpi->max_gf_interval = cpi->oxcf.lag_in_frames - 1;
if (cpi->twopass.static_scene_max_gf_interval > cpi->oxcf.lag_in_frames - 1)
cpi->twopass.static_scene_max_gf_interval = cpi->oxcf.lag_in_frames - 1;
2010-05-18 19:58:33 +04:00
}
if ( cpi->max_gf_interval > cpi->twopass.static_scene_max_gf_interval )
cpi->max_gf_interval = cpi->twopass.static_scene_max_gf_interval;
2010-05-18 19:58:33 +04:00
}
static int
rescale(int val, int num, int denom)
{
int64_t llnum = num;
int64_t llden = denom;
int64_t llval = val;
return llval * llnum / llden;
}
static void init_config(VP8_PTR ptr, VP8_CONFIG *oxcf)
2010-05-18 19:58:33 +04:00
{
VP8_COMP *cpi = (VP8_COMP *)(ptr);
VP8_COMMON *cm = &cpi->common;
cpi->oxcf = *oxcf;
2010-05-18 19:58:33 +04:00
cpi->auto_gold = 1;
cpi->auto_adjust_gold_quantizer = 1;
cpi->goldfreq = 7;
cm->version = oxcf->Version;
vp8_setup_version(cm);
// change includes all joint functionality
vp8_change_config(ptr, oxcf);
2010-05-18 19:58:33 +04:00
// Initialize active best and worst q and average q values.
cpi->active_worst_quality = cpi->oxcf.worst_allowed_q;
cpi->active_best_quality = cpi->oxcf.best_allowed_q;
cpi->avg_frame_qindex = cpi->oxcf.worst_allowed_q;
// Initialise the starting buffer levels
cpi->buffer_level = cpi->oxcf.starting_buffer_level;
2010-05-18 19:58:33 +04:00
cpi->bits_off_target = cpi->oxcf.starting_buffer_level;
cpi->rolling_target_bits = cpi->av_per_frame_bandwidth;
cpi->rolling_actual_bits = cpi->av_per_frame_bandwidth;
cpi->long_rolling_target_bits = cpi->av_per_frame_bandwidth;
cpi->long_rolling_actual_bits = cpi->av_per_frame_bandwidth;
cpi->buffered_av_per_frame_bandwidth = cpi->av_per_frame_bandwidth;
2010-05-18 19:58:33 +04:00
cpi->total_actual_bits = 0;
cpi->total_target_vs_actual = 0;
2010-05-18 19:58:33 +04:00
#if VP8_TEMPORAL_ALT_REF
{
int i;
cpi->fixed_divide[0] = 0;
for (i = 1; i < 512; i++)
cpi->fixed_divide[i] = 0x80000 / i;
2010-05-18 19:58:33 +04:00
}
#endif
}
void vp8_change_config(VP8_PTR ptr, VP8_CONFIG *oxcf)
{
VP8_COMP *cpi = (VP8_COMP *)(ptr);
VP8_COMMON *cm = &cpi->common;
if (!cpi)
return;
if (!oxcf)
return;
if (cm->version != oxcf->Version)
{
cm->version = oxcf->Version;
vp8_setup_version(cm);
}
cpi->oxcf = *oxcf;
switch (cpi->oxcf.Mode)
{
case MODE_REALTIME:
cpi->pass = 0;
cpi->compressor_speed = 2;
if (cpi->oxcf.cpu_used < -16)
{
cpi->oxcf.cpu_used = -16;
}
if (cpi->oxcf.cpu_used > 16)
cpi->oxcf.cpu_used = 16;
break;
case MODE_GOODQUALITY:
cpi->pass = 0;
cpi->compressor_speed = 1;
if (cpi->oxcf.cpu_used < -5)
{
cpi->oxcf.cpu_used = -5;
}
if (cpi->oxcf.cpu_used > 5)
cpi->oxcf.cpu_used = 5;
break;
case MODE_BESTQUALITY:
cpi->pass = 0;
cpi->compressor_speed = 0;
break;
case MODE_FIRSTPASS:
cpi->pass = 1;
cpi->compressor_speed = 1;
break;
case MODE_SECONDPASS:
cpi->pass = 2;
cpi->compressor_speed = 1;
if (cpi->oxcf.cpu_used < -5)
{
cpi->oxcf.cpu_used = -5;
}
if (cpi->oxcf.cpu_used > 5)
cpi->oxcf.cpu_used = 5;
break;
case MODE_SECONDPASS_BEST:
cpi->pass = 2;
cpi->compressor_speed = 0;
break;
}
if (cpi->pass == 0 && cpi->oxcf.end_usage != USAGE_STREAM_FROM_SERVER)
2010-05-18 19:58:33 +04:00
cpi->auto_worst_q = 1;
cpi->oxcf.worst_allowed_q = q_trans[oxcf->worst_allowed_q];
cpi->oxcf.best_allowed_q = q_trans[oxcf->best_allowed_q];
cpi->oxcf.cq_level = q_trans[cpi->oxcf.cq_level];
2010-05-18 19:58:33 +04:00
if (oxcf->fixed_q >= 0)
{
if (oxcf->worst_allowed_q < 0)
cpi->oxcf.fixed_q = q_trans[0];
else
cpi->oxcf.fixed_q = q_trans[oxcf->worst_allowed_q];
if (oxcf->alt_q < 0)
cpi->oxcf.alt_q = q_trans[0];
else
cpi->oxcf.alt_q = q_trans[oxcf->alt_q];
if (oxcf->key_q < 0)
cpi->oxcf.key_q = q_trans[0];
else
cpi->oxcf.key_q = q_trans[oxcf->key_q];
if (oxcf->gold_q < 0)
cpi->oxcf.gold_q = q_trans[0];
else
cpi->oxcf.gold_q = q_trans[oxcf->gold_q];
}
cpi->baseline_gf_interval =
cpi->oxcf.alt_freq ? cpi->oxcf.alt_freq : DEFAULT_GF_INTERVAL;
2010-05-18 19:58:33 +04:00
cpi->ref_frame_flags = VP8_ALT_FLAG | VP8_GOLD_FLAG | VP8_LAST_FLAG;
//cpi->use_golden_frame_only = 0;
//cpi->use_last_frame_only = 0;
cm->refresh_golden_frame = 0;
cm->refresh_last_frame = 1;
cm->refresh_entropy_probs = 1;
if (cpi->oxcf.token_partitions >= 0 && cpi->oxcf.token_partitions <= 3)
cm->multi_token_partition =
(TOKEN_PARTITION) cpi->oxcf.token_partitions;
2010-05-18 19:58:33 +04:00
setup_features(cpi);
{
int i;
for (i = 0; i < MAX_MB_SEGMENTS; i++)
cpi->segment_encode_breakout[i] = cpi->oxcf.encode_breakout;
}
// At the moment the first order values may not be > MAXQ
if (cpi->oxcf.fixed_q > MAXQ)
cpi->oxcf.fixed_q = MAXQ;
// local file playback mode == really big buffer
if (cpi->oxcf.end_usage == USAGE_LOCAL_FILE_PLAYBACK)
{
cpi->oxcf.starting_buffer_level = 60000;
cpi->oxcf.optimal_buffer_level = 60000;
cpi->oxcf.maximum_buffer_size = 240000;
2010-05-18 19:58:33 +04:00
}
// Convert target bandwidth from Kbit/s to Bit/s
cpi->oxcf.target_bandwidth *= 1000;
cpi->oxcf.starting_buffer_level =
rescale(cpi->oxcf.starting_buffer_level,
cpi->oxcf.target_bandwidth, 1000);
// Set or reset optimal and maximum buffer levels.
2010-05-18 19:58:33 +04:00
if (cpi->oxcf.optimal_buffer_level == 0)
cpi->oxcf.optimal_buffer_level = cpi->oxcf.target_bandwidth / 8;
else
cpi->oxcf.optimal_buffer_level =
rescale(cpi->oxcf.optimal_buffer_level,
cpi->oxcf.target_bandwidth, 1000);
2010-05-18 19:58:33 +04:00
if (cpi->oxcf.maximum_buffer_size == 0)
cpi->oxcf.maximum_buffer_size = cpi->oxcf.target_bandwidth / 8;
else
cpi->oxcf.maximum_buffer_size =
rescale(cpi->oxcf.maximum_buffer_size,
cpi->oxcf.target_bandwidth, 1000);
2010-05-18 19:58:33 +04:00
// Set up frame rate and related parameters rate control values.
2010-05-18 19:58:33 +04:00
vp8_new_frame_rate(cpi, cpi->oxcf.frame_rate);
// Set absolute upper and lower quality limits
2010-05-18 19:58:33 +04:00
cpi->worst_quality = cpi->oxcf.worst_allowed_q;
cpi->best_quality = cpi->oxcf.best_allowed_q;
// active values should only be modified if out of new range
if (cpi->active_worst_quality > cpi->oxcf.worst_allowed_q)
{
cpi->active_worst_quality = cpi->oxcf.worst_allowed_q;
}
// less likely
else if (cpi->active_worst_quality < cpi->oxcf.best_allowed_q)
{
cpi->active_worst_quality = cpi->oxcf.best_allowed_q;
}
if (cpi->active_best_quality < cpi->oxcf.best_allowed_q)
{
cpi->active_best_quality = cpi->oxcf.best_allowed_q;
}
// less likely
else if (cpi->active_best_quality > cpi->oxcf.worst_allowed_q)
{
cpi->active_best_quality = cpi->oxcf.worst_allowed_q;
}
cpi->buffered_mode = (cpi->oxcf.optimal_buffer_level > 0) ? TRUE : FALSE;
2010-05-18 19:58:33 +04:00
cpi->cq_target_quality = cpi->oxcf.cq_level;
2010-05-18 19:58:33 +04:00
// Only allow dropped frames in buffered mode
cpi->drop_frames_allowed = cpi->oxcf.allow_df && cpi->buffered_mode;
2010-05-18 19:58:33 +04:00
if (!cm->use_bilinear_mc_filter)
cm->mcomp_filter_type = SIXTAP;
else
cm->mcomp_filter_type = BILINEAR;
cpi->target_bandwidth = cpi->oxcf.target_bandwidth;
cm->Width = cpi->oxcf.Width ;
cm->Height = cpi->oxcf.Height ;
cm->horiz_scale = cpi->horiz_scale;
cm->vert_scale = cpi->vert_scale ;
// VP8 sharpness level mapping 0-7 (vs 0-10 in general VPx dialogs)
if (cpi->oxcf.Sharpness > 7)
cpi->oxcf.Sharpness = 7;
cm->sharpness_level = cpi->oxcf.Sharpness;
if (cm->horiz_scale != NORMAL || cm->vert_scale != NORMAL)
{
int UNINITIALIZED_IS_SAFE(hr), UNINITIALIZED_IS_SAFE(hs);
int UNINITIALIZED_IS_SAFE(vr), UNINITIALIZED_IS_SAFE(vs);
Scale2Ratio(cm->horiz_scale, &hr, &hs);
Scale2Ratio(cm->vert_scale, &vr, &vs);
// always go to the next whole number
cm->Width = (hs - 1 + cpi->oxcf.Width * hr) / hs;
cm->Height = (vs - 1 + cpi->oxcf.Height * vr) / vs;
}
if (((cm->Width + 15) & 0xfffffff0) !=
cm->yv12_fb[cm->lst_fb_idx].y_width ||
((cm->Height + 15) & 0xfffffff0) !=
cm->yv12_fb[cm->lst_fb_idx].y_height ||
cm->yv12_fb[cm->lst_fb_idx].y_width == 0)
2010-05-18 19:58:33 +04:00
{
alloc_raw_frame_buffers(cpi);
vp8_alloc_compressor_data(cpi);
}
if (cpi->oxcf.fixed_q >= 0)
{
cpi->last_q[0] = cpi->oxcf.fixed_q;
cpi->last_q[1] = cpi->oxcf.fixed_q;
}
cpi->Speed = cpi->oxcf.cpu_used;
// force to allowlag to 0 if lag_in_frames is 0;
if (cpi->oxcf.lag_in_frames == 0)
{
cpi->oxcf.allow_lag = 0;
}
// Limit on lag buffers as these are not currently dynamically allocated
else if (cpi->oxcf.lag_in_frames > MAX_LAG_BUFFERS)
cpi->oxcf.lag_in_frames = MAX_LAG_BUFFERS;
// YX Temp
cpi->alt_ref_source = NULL;
2010-05-18 19:58:33 +04:00
cpi->is_src_frame_alt_ref = 0;
2010-05-18 19:58:33 +04:00
#if 0
// Experimental RD Code
cpi->frame_distortion = 0;
cpi->last_frame_distortion = 0;
#endif
}
#define M_LOG2_E 0.693147180559945309417
#define log2f(x) (log (x) / (float) M_LOG2_E)
static void cal_mvsadcosts(int *mvsadcost[2])
{
int i = 1;
mvsadcost [0] [0] = 300;
mvsadcost [1] [0] = 300;
do
{
double z = 256 * (2 * (log2f(8 * i) + .6));
2010-05-18 19:58:33 +04:00
mvsadcost [0][i] = (int) z;
mvsadcost [1][i] = (int) z;
mvsadcost [0][-i] = (int) z;
mvsadcost [1][-i] = (int) z;
}
while (++i <= mvfp_max);
2010-05-18 19:58:33 +04:00
}
VP8_PTR vp8_create_compressor(VP8_CONFIG *oxcf)
{
int i;
volatile union
{
VP8_COMP *cpi;
VP8_PTR ptr;
} ctx;
2010-05-18 19:58:33 +04:00
VP8_COMP *cpi;
VP8_COMMON *cm;
cpi = ctx.cpi = vpx_memalign(32, sizeof(VP8_COMP));
// Check that the CPI instance is valid
if (!cpi)
return 0;
cm = &cpi->common;
vpx_memset(cpi, 0, sizeof(VP8_COMP));
if (setjmp(cm->error.jmp))
{
VP8_PTR ptr = ctx.ptr;
ctx.cpi->common.error.setjmp = 0;
vp8_remove_compressor(&ptr);
return 0;
}
cpi->common.error.setjmp = 1;
CHECK_MEM_ERROR(cpi->mb.ss, vpx_calloc(sizeof(search_site), (MAX_MVSEARCH_STEPS * 8) + 1));
vp8_create_common(&cpi->common);
Add runtime CPU detection support for ARM. The primary goal is to allow a binary to be built which supports NEON, but can fall back to non-NEON routines, since some Android devices do not have NEON, even if they are otherwise ARMv7 (e.g., Tegra). The configure-generated flags HAVE_ARMV7, etc., are used to decide which versions of each function to build, and when CONFIG_RUNTIME_CPU_DETECT is enabled, the correct version is chosen at run time. In order for this to work, the CFLAGS must be set to something appropriate (e.g., without -mfpu=neon for ARMv7, and with appropriate -march and -mcpu for even earlier configurations), or the native C code will not be able to run. The ASFLAGS must remain set for the most advanced instruction set required at build time, since the ARM assembler will refuse to emit them otherwise. I have not attempted to make any changes to configure to do this automatically. Doing so will probably require the addition of new configure options. Many of the hooks for RTCD on ARM were already there, but a lot of the code had bit-rotted, and a good deal of the ARM-specific code is not integrated into the RTCD structs at all. I did not try to resolve the latter, merely to add the minimal amount of protection around them to allow RTCD to work. Those functions that were called based on an ifdef at the calling site were expanded to check the RTCD flags at that site, but they should be added to an RTCD struct somewhere in the future. The functions invoked with global function pointers still are, but these should be moved into an RTCD struct for thread safety (I believe every platform currently supported has atomic pointer stores, but this is not guaranteed). The encoder's boolhuff functions did not even have _c and armv7 suffixes, and the correct version was resolved at link time. The token packing functions did have appropriate suffixes, but the version was selected with a define, with no associated RTCD struct. However, for both of these, the only armv7 instruction they actually used was rbit, and this was completely superfluous, so I reworked them to avoid it. The only non-ARMv4 instruction remaining in them is clz, which is ARMv5 (not even ARMv5TE is required). Considering that there are no ARM-specific configs which are not at least ARMv5TE, I did not try to detect these at runtime, and simply enable them for ARMv5 and above. Finally, the NEON register saving code was completely non-reentrant, since it saved the registers to a global, static variable. I moved the storage for this onto the stack. A single binary built with this code was tested on an ARM11 (ARMv6) and a Cortex A8 (ARMv7 w/NEON), for both the encoder and decoder, and produced identical output, while using the correct accelerated functions on each. I did not test on any earlier processors. Change-Id: I45cbd63a614f4554c3b325c45d46c0806f009eaa
2010-10-21 02:39:11 +04:00
vp8_cmachine_specific_config(cpi);
2010-05-18 19:58:33 +04:00
init_config((VP8_PTR)cpi, oxcf);
2010-05-18 19:58:33 +04:00
memcpy(cpi->base_skip_false_prob, vp8cx_base_skip_false_prob, sizeof(vp8cx_base_skip_false_prob));
cpi->common.current_video_frame = 0;
cpi->kf_overspend_bits = 0;
cpi->kf_bitrate_adjustment = 0;
cpi->frames_till_gf_update_due = 0;
cpi->gf_overspend_bits = 0;
cpi->non_gf_bitrate_adjustment = 0;
cpi->prob_last_coded = 128;
cpi->prob_gf_coded = 128;
cpi->prob_intra_coded = 63;
// Prime the recent reference frame useage counters.
// Hereafter they will be maintained as a sort of moving average
cpi->recent_ref_frame_usage[INTRA_FRAME] = 1;
cpi->recent_ref_frame_usage[LAST_FRAME] = 1;
cpi->recent_ref_frame_usage[GOLDEN_FRAME] = 1;
cpi->recent_ref_frame_usage[ALTREF_FRAME] = 1;
// Set reference frame sign bias for ALTREF frame to 1 (for now)
cpi->common.ref_frame_sign_bias[ALTREF_FRAME] = 1;
cpi->twopass.gf_decay_rate = 0;
2010-05-18 19:58:33 +04:00
cpi->baseline_gf_interval = DEFAULT_GF_INTERVAL;
cpi->gold_is_last = 0 ;
cpi->alt_is_last = 0 ;
cpi->gold_is_alt = 0 ;
// allocate memory for storing last frame's MVs for MV prediction.
CHECK_MEM_ERROR(cpi->lfmv, vpx_calloc((cpi->common.mb_rows+2) * (cpi->common.mb_cols+2), sizeof(int_mv)));
CHECK_MEM_ERROR(cpi->lf_ref_frame_sign_bias, vpx_calloc((cpi->common.mb_rows+2) * (cpi->common.mb_cols+2), sizeof(int)));
CHECK_MEM_ERROR(cpi->lf_ref_frame, vpx_calloc((cpi->common.mb_rows+2) * (cpi->common.mb_cols+2), sizeof(int)));
2010-05-18 19:58:33 +04:00
// Create the encoder segmentation map and set all entries to 0
CHECK_MEM_ERROR(cpi->segmentation_map, vpx_calloc(cpi->common.mb_rows * cpi->common.mb_cols, 1));
CHECK_MEM_ERROR(cpi->active_map, vpx_calloc(cpi->common.mb_rows * cpi->common.mb_cols, 1));
vpx_memset(cpi->active_map , 1, (cpi->common.mb_rows * cpi->common.mb_cols));
cpi->active_map_enabled = 0;
#if 0
// Experimental code for lagged and one pass
// Initialise one_pass GF frames stats
// Update stats used for GF selection
if (cpi->pass == 0)
{
cpi->one_pass_frame_index = 0;
for (i = 0; i < MAX_LAG_BUFFERS; i++)
{
cpi->one_pass_frame_stats[i].frames_so_far = 0;
cpi->one_pass_frame_stats[i].frame_intra_error = 0.0;
cpi->one_pass_frame_stats[i].frame_coded_error = 0.0;
cpi->one_pass_frame_stats[i].frame_pcnt_inter = 0.0;
cpi->one_pass_frame_stats[i].frame_pcnt_motion = 0.0;
cpi->one_pass_frame_stats[i].frame_mvr = 0.0;
cpi->one_pass_frame_stats[i].frame_mvr_abs = 0.0;
cpi->one_pass_frame_stats[i].frame_mvc = 0.0;
cpi->one_pass_frame_stats[i].frame_mvc_abs = 0.0;
}
}
#endif
// Should we use the cyclic refresh method.
// Currently this is tied to error resilliant mode
cpi->cyclic_refresh_mode_enabled = cpi->oxcf.error_resilient_mode;
cpi->cyclic_refresh_mode_max_mbs_perframe = (cpi->common.mb_rows * cpi->common.mb_cols) / 40;
cpi->cyclic_refresh_mode_index = 0;
cpi->cyclic_refresh_q = 32;
if (cpi->cyclic_refresh_mode_enabled)
{
CHECK_MEM_ERROR(cpi->cyclic_refresh_map, vpx_calloc((cpi->common.mb_rows * cpi->common.mb_cols), 1));
}
else
cpi->cyclic_refresh_map = (signed char *) NULL;
// Test function for segmentation
//segmentation_test_function((VP8_PTR) cpi);
#ifdef ENTROPY_STATS
init_context_counters();
#endif
/*Initialize the feed-forward activity masking.*/
cpi->activity_avg = 90<<12;
2010-05-18 19:58:33 +04:00
cpi->frames_since_key = 8; // Give a sensible default for the first frame.
cpi->key_frame_frequency = cpi->oxcf.key_freq;
cpi->this_key_frame_forced = FALSE;
cpi->next_key_frame_forced = FALSE;
2010-05-18 19:58:33 +04:00
cpi->source_alt_ref_pending = FALSE;
cpi->source_alt_ref_active = FALSE;
cpi->common.refresh_alt_ref_frame = 0;
cpi->b_calculate_psnr = CONFIG_INTERNAL_STATS;
#if CONFIG_INTERNAL_STATS
2010-05-18 19:58:33 +04:00
cpi->b_calculate_ssimg = 0;
cpi->count = 0;
cpi->bytes = 0;
if (cpi->b_calculate_psnr)
{
cpi->total_sq_error = 0.0;
cpi->total_sq_error2 = 0.0;
cpi->total_y = 0.0;
cpi->total_u = 0.0;
cpi->total_v = 0.0;
cpi->total = 0.0;
cpi->totalp_y = 0.0;
cpi->totalp_u = 0.0;
cpi->totalp_v = 0.0;
cpi->totalp = 0.0;
cpi->tot_recode_hits = 0;
cpi->summed_quality = 0;
cpi->summed_weights = 0;
}
if (cpi->b_calculate_ssimg)
{
cpi->total_ssimg_y = 0;
cpi->total_ssimg_u = 0;
cpi->total_ssimg_v = 0;
cpi->total_ssimg_all = 0;
}
#endif
2010-05-18 19:58:33 +04:00
#ifndef LLONG_MAX
#define LLONG_MAX 9223372036854775807LL
#endif
cpi->first_time_stamp_ever = LLONG_MAX;
cpi->frames_till_gf_update_due = 0;
cpi->key_frame_count = 1;
cpi->ni_av_qi = cpi->oxcf.worst_allowed_q;
cpi->ni_tot_qi = 0;
cpi->ni_frames = 0;
cpi->total_byte_count = 0;
cpi->drop_frame = 0;
cpi->drop_count = 0;
cpi->max_drop_count = 0;
cpi->max_consec_dropped_frames = 4;
cpi->rate_correction_factor = 1.0;
cpi->key_frame_rate_correction_factor = 1.0;
cpi->gf_rate_correction_factor = 1.0;
cpi->twopass.est_max_qcorrection_factor = 1.0;
2010-05-18 19:58:33 +04:00
cpi->mb.mvcost[0] = &cpi->mb.mvcosts[0][mv_max+1];
cpi->mb.mvcost[1] = &cpi->mb.mvcosts[1][mv_max+1];
cpi->mb.mvsadcost[0] = &cpi->mb.mvsadcosts[0][mvfp_max+1];
cpi->mb.mvsadcost[1] = &cpi->mb.mvsadcosts[1][mvfp_max+1];
2010-05-18 19:58:33 +04:00
cal_mvsadcosts(cpi->mb.mvsadcost);
for (i = 0; i < KEY_FRAME_CONTEXT; i++)
{
cpi->prior_key_frame_distance[i] = (int)cpi->output_frame_rate;
}
#ifdef OUTPUT_YUV_SRC
yuv_file = fopen("bd.yuv", "ab");
#endif
#if 0
framepsnr = fopen("framepsnr.stt", "a");
kf_list = fopen("kf_list.stt", "w");
#endif
cpi->output_pkt_list = oxcf->output_pkt_list;
#if !(CONFIG_REALTIME_ONLY)
if (cpi->pass == 1)
{
vp8_init_first_pass(cpi);
}
else if (cpi->pass == 2)
{
size_t packet_sz = sizeof(FIRSTPASS_STATS);
int packets = oxcf->two_pass_stats_in.sz / packet_sz;
cpi->twopass.stats_in_start = oxcf->two_pass_stats_in.buf;
cpi->twopass.stats_in = cpi->twopass.stats_in_start;
cpi->twopass.stats_in_end = (void*)((char *)cpi->twopass.stats_in
+ (packets - 1) * packet_sz);
2010-05-18 19:58:33 +04:00
vp8_init_second_pass(cpi);
}
#endif
if (cpi->compressor_speed == 2)
{
cpi->cpu_freq = 0; //vp8_get_processor_freq();
cpi->avg_encode_time = 0;
cpi->avg_pick_mode_time = 0;
}
vp8_set_speed_features(cpi);
// Set starting values of RD threshold multipliers (128 = *1)
for (i = 0; i < MAX_MODES; i++)
{
cpi->rd_thresh_mult[i] = 128;
}
#ifdef ENTROPY_STATS
init_mv_ref_counts();
#endif
#if CONFIG_MULTITHREAD
2010-05-18 19:58:33 +04:00
vp8cx_create_encoder_threads(cpi);
#endif
2010-05-18 19:58:33 +04:00
cpi->fn_ptr[BLOCK_16X16].sdf = VARIANCE_INVOKE(&cpi->rtcd.variance, sad16x16);
cpi->fn_ptr[BLOCK_16X16].vf = VARIANCE_INVOKE(&cpi->rtcd.variance, var16x16);
cpi->fn_ptr[BLOCK_16X16].svf = VARIANCE_INVOKE(&cpi->rtcd.variance, subpixvar16x16);
cpi->fn_ptr[BLOCK_16X16].svf_halfpix_h = VARIANCE_INVOKE(&cpi->rtcd.variance, halfpixvar16x16_h);
cpi->fn_ptr[BLOCK_16X16].svf_halfpix_v = VARIANCE_INVOKE(&cpi->rtcd.variance, halfpixvar16x16_v);
cpi->fn_ptr[BLOCK_16X16].svf_halfpix_hv = VARIANCE_INVOKE(&cpi->rtcd.variance, halfpixvar16x16_hv);
cpi->fn_ptr[BLOCK_16X16].sdx3f = VARIANCE_INVOKE(&cpi->rtcd.variance, sad16x16x3);
cpi->fn_ptr[BLOCK_16X16].sdx8f = VARIANCE_INVOKE(&cpi->rtcd.variance, sad16x16x8);
cpi->fn_ptr[BLOCK_16X16].sdx4df = VARIANCE_INVOKE(&cpi->rtcd.variance, sad16x16x4d);
cpi->fn_ptr[BLOCK_16X8].sdf = VARIANCE_INVOKE(&cpi->rtcd.variance, sad16x8);
cpi->fn_ptr[BLOCK_16X8].vf = VARIANCE_INVOKE(&cpi->rtcd.variance, var16x8);
cpi->fn_ptr[BLOCK_16X8].svf = VARIANCE_INVOKE(&cpi->rtcd.variance, subpixvar16x8);
cpi->fn_ptr[BLOCK_16X8].svf_halfpix_h = NULL;
cpi->fn_ptr[BLOCK_16X8].svf_halfpix_v = NULL;
cpi->fn_ptr[BLOCK_16X8].svf_halfpix_hv = NULL;
cpi->fn_ptr[BLOCK_16X8].sdx3f = VARIANCE_INVOKE(&cpi->rtcd.variance, sad16x8x3);
cpi->fn_ptr[BLOCK_16X8].sdx8f = VARIANCE_INVOKE(&cpi->rtcd.variance, sad16x8x8);
cpi->fn_ptr[BLOCK_16X8].sdx4df = VARIANCE_INVOKE(&cpi->rtcd.variance, sad16x8x4d);
cpi->fn_ptr[BLOCK_8X16].sdf = VARIANCE_INVOKE(&cpi->rtcd.variance, sad8x16);
cpi->fn_ptr[BLOCK_8X16].vf = VARIANCE_INVOKE(&cpi->rtcd.variance, var8x16);
cpi->fn_ptr[BLOCK_8X16].svf = VARIANCE_INVOKE(&cpi->rtcd.variance, subpixvar8x16);
cpi->fn_ptr[BLOCK_8X16].svf_halfpix_h = NULL;
cpi->fn_ptr[BLOCK_8X16].svf_halfpix_v = NULL;
cpi->fn_ptr[BLOCK_8X16].svf_halfpix_hv = NULL;
cpi->fn_ptr[BLOCK_8X16].sdx3f = VARIANCE_INVOKE(&cpi->rtcd.variance, sad8x16x3);
cpi->fn_ptr[BLOCK_8X16].sdx8f = VARIANCE_INVOKE(&cpi->rtcd.variance, sad8x16x8);
cpi->fn_ptr[BLOCK_8X16].sdx4df = VARIANCE_INVOKE(&cpi->rtcd.variance, sad8x16x4d);
cpi->fn_ptr[BLOCK_8X8].sdf = VARIANCE_INVOKE(&cpi->rtcd.variance, sad8x8);
cpi->fn_ptr[BLOCK_8X8].vf = VARIANCE_INVOKE(&cpi->rtcd.variance, var8x8);
cpi->fn_ptr[BLOCK_8X8].svf = VARIANCE_INVOKE(&cpi->rtcd.variance, subpixvar8x8);
cpi->fn_ptr[BLOCK_8X8].svf_halfpix_h = NULL;
cpi->fn_ptr[BLOCK_8X8].svf_halfpix_v = NULL;
cpi->fn_ptr[BLOCK_8X8].svf_halfpix_hv = NULL;
cpi->fn_ptr[BLOCK_8X8].sdx3f = VARIANCE_INVOKE(&cpi->rtcd.variance, sad8x8x3);
cpi->fn_ptr[BLOCK_8X8].sdx8f = VARIANCE_INVOKE(&cpi->rtcd.variance, sad8x8x8);
cpi->fn_ptr[BLOCK_8X8].sdx4df = VARIANCE_INVOKE(&cpi->rtcd.variance, sad8x8x4d);
cpi->fn_ptr[BLOCK_4X4].sdf = VARIANCE_INVOKE(&cpi->rtcd.variance, sad4x4);
cpi->fn_ptr[BLOCK_4X4].vf = VARIANCE_INVOKE(&cpi->rtcd.variance, var4x4);
cpi->fn_ptr[BLOCK_4X4].svf = VARIANCE_INVOKE(&cpi->rtcd.variance, subpixvar4x4);
cpi->fn_ptr[BLOCK_4X4].svf_halfpix_h = NULL;
cpi->fn_ptr[BLOCK_4X4].svf_halfpix_v = NULL;
cpi->fn_ptr[BLOCK_4X4].svf_halfpix_hv = NULL;
cpi->fn_ptr[BLOCK_4X4].sdx3f = VARIANCE_INVOKE(&cpi->rtcd.variance, sad4x4x3);
cpi->fn_ptr[BLOCK_4X4].sdx8f = VARIANCE_INVOKE(&cpi->rtcd.variance, sad4x4x8);
cpi->fn_ptr[BLOCK_4X4].sdx4df = VARIANCE_INVOKE(&cpi->rtcd.variance, sad4x4x4d);
2010-05-18 19:58:33 +04:00
cpi->full_search_sad = SEARCH_INVOKE(&cpi->rtcd.search, full_search);
cpi->diamond_search_sad = SEARCH_INVOKE(&cpi->rtcd.search, diamond_search);
cpi->refining_search_sad = SEARCH_INVOKE(&cpi->rtcd.search, refining_search);
2010-05-18 19:58:33 +04:00
// make sure frame 1 is okay
cpi->error_bins[0] = cpi->common.MBs;
//vp8cx_init_quantizer() is first called here. Add check in vp8cx_frame_init_quantizer() so that vp8cx_init_quantizer is only called later
//when needed. This will avoid unnecessary calls of vp8cx_init_quantizer() for every frame.
vp8cx_init_quantizer(cpi);
{
vp8_loop_filter_init(cm);
2010-05-18 19:58:33 +04:00
cm->last_frame_type = KEY_FRAME;
cm->last_filter_type = cm->filter_type;
cm->last_sharpness_level = cm->sharpness_level;
}
cpi->common.error.setjmp = 0;
return (VP8_PTR) cpi;
}
void vp8_remove_compressor(VP8_PTR *ptr)
{
VP8_COMP *cpi = (VP8_COMP *)(*ptr);
if (!cpi)
return;
if (cpi && (cpi->common.current_video_frame > 0))
{
#if !(CONFIG_REALTIME_ONLY)
if (cpi->pass == 2)
{
vp8_end_second_pass(cpi);
}
#endif
#ifdef ENTROPY_STATS
print_context_counters();
print_tree_update_probs();
print_mode_context();
#endif
#if CONFIG_INTERNAL_STATS
2010-05-18 19:58:33 +04:00
if (cpi->pass != 1)
{
FILE *f = fopen("opsnr.stt", "a");
double time_encoded = (cpi->last_end_time_stamp_seen
- cpi->first_time_stamp_ever) / 10000000.000;
2010-05-18 19:58:33 +04:00
double total_encode_time = (cpi->time_receive_data + cpi->time_compress_data) / 1000.000;
double dr = (double)cpi->bytes * (double) 8 / (double)1000 / time_encoded;
if (cpi->b_calculate_psnr)
{
YV12_BUFFER_CONFIG *lst_yv12 = &cpi->common.yv12_fb[cpi->common.lst_fb_idx];
double samples = 3.0 / 2 * cpi->count * lst_yv12->y_width * lst_yv12->y_height;
2010-05-18 19:58:33 +04:00
double total_psnr = vp8_mse2psnr(samples, 255.0, cpi->total_sq_error);
double total_psnr2 = vp8_mse2psnr(samples, 255.0, cpi->total_sq_error2);
double total_ssim = 100 * pow(cpi->summed_quality / cpi->summed_weights, 8.0);
fprintf(f, "Bitrate\tAVGPsnr\tGLBPsnr\tAVPsnrP\tGLPsnrP\tVPXSSIM\t Time(us)\n");
fprintf(f, "%7.3f\t%7.3f\t%7.3f\t%7.3f\t%7.3f\t%7.3f\t%8.0f\n",
2010-05-18 19:58:33 +04:00
dr, cpi->total / cpi->count, total_psnr, cpi->totalp / cpi->count, total_psnr2, total_ssim,
total_encode_time);
}
if (cpi->b_calculate_ssimg)
{
fprintf(f, "BitRate\tSSIM_Y\tSSIM_U\tSSIM_V\tSSIM_A\t Time(us)\n");
fprintf(f, "%7.3f\t%6.4f\t%6.4f\t%6.4f\t%6.4f\t%8.0f\n", dr,
cpi->total_ssimg_y / cpi->count, cpi->total_ssimg_u / cpi->count,
cpi->total_ssimg_v / cpi->count, cpi->total_ssimg_all / cpi->count, total_encode_time);
}
fclose(f);
#if 0
f = fopen("qskip.stt", "a");
fprintf(f, "minq:%d -maxq:%d skipture:skipfalse = %d:%d\n", cpi->oxcf.best_allowed_q, cpi->oxcf.worst_allowed_q, skiptruecount, skipfalsecount);
fclose(f);
#endif
}
#endif
#ifdef SPEEDSTATS
if (cpi->compressor_speed == 2)
{
int i;
FILE *f = fopen("cxspeed.stt", "a");
cnt_pm /= cpi->common.MBs;
for (i = 0; i < 16; i++)
fprintf(f, "%5d", frames_at_speed[i]);
fprintf(f, "\n");
//fprintf(f, "%10d PM %10d %10d %10d EF %10d %10d %10d\n", cpi->Speed, cpi->avg_pick_mode_time, (tot_pm/cnt_pm), cnt_pm, cpi->avg_encode_time, 0, 0);
fclose(f);
}
#endif
#ifdef MODE_STATS
{
extern int count_mb_seg[4];
FILE *f = fopen("modes.stt", "a");
double dr = (double)cpi->oxcf.frame_rate * (double)bytes * (double)8 / (double)count / (double)1000 ;
fprintf(f, "intra_mode in Intra Frames:\n");
fprintf(f, "Y: %8d, %8d, %8d, %8d, %8d\n", y_modes[0], y_modes[1], y_modes[2], y_modes[3], y_modes[4]);
fprintf(f, "UV:%8d, %8d, %8d, %8d\n", uv_modes[0], uv_modes[1], uv_modes[2], uv_modes[3]);
fprintf(f, "B: ");
{
int i;
for (i = 0; i < 10; i++)
fprintf(f, "%8d, ", b_modes[i]);
fprintf(f, "\n");
}
fprintf(f, "Modes in Inter Frames:\n");
fprintf(f, "Y: %8d, %8d, %8d, %8d, %8d, %8d, %8d, %8d, %8d, %8d\n",
inter_y_modes[0], inter_y_modes[1], inter_y_modes[2], inter_y_modes[3], inter_y_modes[4],
inter_y_modes[5], inter_y_modes[6], inter_y_modes[7], inter_y_modes[8], inter_y_modes[9]);
fprintf(f, "UV:%8d, %8d, %8d, %8d\n", inter_uv_modes[0], inter_uv_modes[1], inter_uv_modes[2], inter_uv_modes[3]);
fprintf(f, "B: ");
{
int i;
for (i = 0; i < 15; i++)
fprintf(f, "%8d, ", inter_b_modes[i]);
fprintf(f, "\n");
}
fprintf(f, "P:%8d, %8d, %8d, %8d\n", count_mb_seg[0], count_mb_seg[1], count_mb_seg[2], count_mb_seg[3]);
fprintf(f, "PB:%8d, %8d, %8d, %8d\n", inter_b_modes[LEFT4X4], inter_b_modes[ABOVE4X4], inter_b_modes[ZERO4X4], inter_b_modes[NEW4X4]);
fclose(f);
}
#endif
#ifdef ENTROPY_STATS
{
int i, j, k;
FILE *fmode = fopen("modecontext.c", "w");
fprintf(fmode, "\n#include \"entropymode.h\"\n\n");
fprintf(fmode, "const unsigned int vp8_kf_default_bmode_counts ");
fprintf(fmode, "[VP8_BINTRAMODES] [VP8_BINTRAMODES] [VP8_BINTRAMODES] =\n{\n");
for (i = 0; i < 10; i++)
{
fprintf(fmode, " { //Above Mode : %d\n", i);
for (j = 0; j < 10; j++)
{
fprintf(fmode, " {");
for (k = 0; k < 10; k++)
{
if (!intra_mode_stats[i][j][k])
fprintf(fmode, " %5d, ", 1);
else
fprintf(fmode, " %5d, ", intra_mode_stats[i][j][k]);
}
fprintf(fmode, "}, // left_mode %d\n", j);
}
fprintf(fmode, " },\n");
}
fprintf(fmode, "};\n");
fclose(fmode);
2010-05-18 19:58:33 +04:00
}
#endif
#if defined(SECTIONBITS_OUTPUT)
if (0)
{
int i;
FILE *f = fopen("tokenbits.stt", "a");
for (i = 0; i < 28; i++)
fprintf(f, "%8d", (int)(Sectionbits[i] / 256));
fprintf(f, "\n");
fclose(f);
}
#endif
#if 0
{
printf("\n_pick_loop_filter_level:%d\n", cpi->time_pick_lpf / 1000);
printf("\n_frames recive_data encod_mb_row compress_frame Total\n");
printf("%6d %10ld %10ld %10ld %10ld\n", cpi->common.current_video_frame, cpi->time_receive_data / 1000, cpi->time_encode_mb_row / 1000, cpi->time_compress_data / 1000, (cpi->time_receive_data + cpi->time_compress_data) / 1000);
}
#endif
}
#if CONFIG_MULTITHREAD
2010-05-18 19:58:33 +04:00
vp8cx_remove_encoder_threads(cpi);
#endif
2010-05-18 19:58:33 +04:00
dealloc_compressor_data(cpi);
2010-05-18 19:58:33 +04:00
vpx_free(cpi->mb.ss);
vpx_free(cpi->tok);
vpx_free(cpi->cyclic_refresh_map);
vp8_remove_common(&cpi->common);
vpx_free(cpi);
*ptr = 0;
#ifdef OUTPUT_YUV_SRC
fclose(yuv_file);
#endif
#if 0
if (keyfile)
fclose(keyfile);
if (framepsnr)
fclose(framepsnr);
if (kf_list)
fclose(kf_list);
#endif
}
static uint64_t calc_plane_error(unsigned char *orig, int orig_stride,
unsigned char *recon, int recon_stride,
unsigned int cols, unsigned int rows,
vp8_variance_rtcd_vtable_t *rtcd)
{
unsigned int row, col;
uint64_t total_sse = 0;
int diff;
for (row = 0; row + 16 <= rows; row += 16)
{
for (col = 0; col + 16 <= cols; col += 16)
{
unsigned int sse;
VARIANCE_INVOKE(rtcd, mse16x16)(orig + col, orig_stride,
recon + col, recon_stride,
&sse);
total_sse += sse;
}
/* Handle odd-sized width */
if (col < cols)
{
unsigned int border_row, border_col;
unsigned char *border_orig = orig;
unsigned char *border_recon = recon;
for (border_row = 0; border_row < 16; border_row++)
{
for (border_col = col; border_col < cols; border_col++)
{
diff = border_orig[border_col] - border_recon[border_col];
total_sse += diff * diff;
}
border_orig += orig_stride;
border_recon += recon_stride;
}
}
orig += orig_stride * 16;
recon += recon_stride * 16;
}
/* Handle odd-sized height */
for (; row < rows; row++)
{
for (col = 0; col < cols; col++)
{
diff = orig[col] - recon[col];
total_sse += diff * diff;
}
orig += orig_stride;
recon += recon_stride;
}
return total_sse;
}
static void generate_psnr_packet(VP8_COMP *cpi)
{
YV12_BUFFER_CONFIG *orig = cpi->Source;
YV12_BUFFER_CONFIG *recon = cpi->common.frame_to_show;
struct vpx_codec_cx_pkt pkt;
uint64_t sse;
int i;
unsigned int width = cpi->common.Width;
unsigned int height = cpi->common.Height;
pkt.kind = VPX_CODEC_PSNR_PKT;
sse = calc_plane_error(orig->y_buffer, orig->y_stride,
recon->y_buffer, recon->y_stride,
width, height,
IF_RTCD(&cpi->rtcd.variance));
pkt.data.psnr.sse[0] = sse;
pkt.data.psnr.sse[1] = sse;
pkt.data.psnr.samples[0] = width * height;
pkt.data.psnr.samples[1] = width * height;
width = (width + 1) / 2;
height = (height + 1) / 2;
sse = calc_plane_error(orig->u_buffer, orig->uv_stride,
recon->u_buffer, recon->uv_stride,
width, height,
IF_RTCD(&cpi->rtcd.variance));
pkt.data.psnr.sse[0] += sse;
pkt.data.psnr.sse[2] = sse;
pkt.data.psnr.samples[0] += width * height;
pkt.data.psnr.samples[2] = width * height;
sse = calc_plane_error(orig->v_buffer, orig->uv_stride,
recon->v_buffer, recon->uv_stride,
width, height,
IF_RTCD(&cpi->rtcd.variance));
pkt.data.psnr.sse[0] += sse;
pkt.data.psnr.sse[3] = sse;
pkt.data.psnr.samples[0] += width * height;
pkt.data.psnr.samples[3] = width * height;
for (i = 0; i < 4; i++)
pkt.data.psnr.psnr[i] = vp8_mse2psnr(pkt.data.psnr.samples[i], 255.0,
pkt.data.psnr.sse[i]);
vpx_codec_pkt_list_add(cpi->output_pkt_list, &pkt);
}
int vp8_use_as_reference(VP8_PTR ptr, int ref_frame_flags)
{
VP8_COMP *cpi = (VP8_COMP *)(ptr);
if (ref_frame_flags > 7)
return -1 ;
cpi->ref_frame_flags = ref_frame_flags;
return 0;
}
int vp8_update_reference(VP8_PTR ptr, int ref_frame_flags)
{
VP8_COMP *cpi = (VP8_COMP *)(ptr);
if (ref_frame_flags > 7)
return -1 ;
cpi->common.refresh_golden_frame = 0;
cpi->common.refresh_alt_ref_frame = 0;
cpi->common.refresh_last_frame = 0;
if (ref_frame_flags & VP8_LAST_FLAG)
cpi->common.refresh_last_frame = 1;
if (ref_frame_flags & VP8_GOLD_FLAG)
cpi->common.refresh_golden_frame = 1;
if (ref_frame_flags & VP8_ALT_FLAG)
cpi->common.refresh_alt_ref_frame = 1;
return 0;
}
int vp8_get_reference(VP8_PTR ptr, VP8_REFFRAME ref_frame_flag, YV12_BUFFER_CONFIG *sd)
{
VP8_COMP *cpi = (VP8_COMP *)(ptr);
VP8_COMMON *cm = &cpi->common;
int ref_fb_idx;
2010-05-18 19:58:33 +04:00
if (ref_frame_flag == VP8_LAST_FLAG)
ref_fb_idx = cm->lst_fb_idx;
2010-05-18 19:58:33 +04:00
else if (ref_frame_flag == VP8_GOLD_FLAG)
ref_fb_idx = cm->gld_fb_idx;
2010-05-18 19:58:33 +04:00
else if (ref_frame_flag == VP8_ALT_FLAG)
ref_fb_idx = cm->alt_fb_idx;
2010-05-18 19:58:33 +04:00
else
return -1;
vp8_yv12_copy_frame_ptr(&cm->yv12_fb[ref_fb_idx], sd);
2010-05-18 19:58:33 +04:00
return 0;
}
int vp8_set_reference(VP8_PTR ptr, VP8_REFFRAME ref_frame_flag, YV12_BUFFER_CONFIG *sd)
{
VP8_COMP *cpi = (VP8_COMP *)(ptr);
VP8_COMMON *cm = &cpi->common;
int ref_fb_idx;
2010-05-18 19:58:33 +04:00
if (ref_frame_flag == VP8_LAST_FLAG)
ref_fb_idx = cm->lst_fb_idx;
2010-05-18 19:58:33 +04:00
else if (ref_frame_flag == VP8_GOLD_FLAG)
ref_fb_idx = cm->gld_fb_idx;
2010-05-18 19:58:33 +04:00
else if (ref_frame_flag == VP8_ALT_FLAG)
ref_fb_idx = cm->alt_fb_idx;
2010-05-18 19:58:33 +04:00
else
return -1;
vp8_yv12_copy_frame_ptr(sd, &cm->yv12_fb[ref_fb_idx]);
2010-05-18 19:58:33 +04:00
return 0;
}
int vp8_update_entropy(VP8_PTR comp, int update)
{
VP8_COMP *cpi = (VP8_COMP *) comp;
VP8_COMMON *cm = &cpi->common;
cm->refresh_entropy_probs = update;
return 0;
}
#if OUTPUT_YUV_SRC
2010-05-18 19:58:33 +04:00
void vp8_write_yuv_frame(const char *name, YV12_BUFFER_CONFIG *s)
{
FILE *yuv_file = fopen(name, "ab");
unsigned char *src = s->y_buffer;
int h = s->y_height;
do
{
fwrite(src, s->y_width, 1, yuv_file);
src += s->y_stride;
}
while (--h);
src = s->u_buffer;
h = s->uv_height;
do
{
fwrite(src, s->uv_width, 1, yuv_file);
src += s->uv_stride;
}
while (--h);
src = s->v_buffer;
h = s->uv_height;
do
{
fwrite(src, s->uv_width, 1, yuv_file);
src += s->uv_stride;
}
while (--h);
fclose(yuv_file);
}
#endif
2010-05-18 19:58:33 +04:00
static void scale_and_extend_source(YV12_BUFFER_CONFIG *sd, VP8_COMP *cpi)
{
VP8_COMMON *cm = &cpi->common;
// are we resizing the image
if (cm->horiz_scale != 0 || cm->vert_scale != 0)
{
#if CONFIG_SPATIAL_RESAMPLING
int UNINITIALIZED_IS_SAFE(hr), UNINITIALIZED_IS_SAFE(hs);
int UNINITIALIZED_IS_SAFE(vr), UNINITIALIZED_IS_SAFE(vs);
int tmp_height;
if (cm->vert_scale == 3)
tmp_height = 9;
else
tmp_height = 11;
Scale2Ratio(cm->horiz_scale, &hr, &hs);
Scale2Ratio(cm->vert_scale, &vr, &vs);
vp8_scale_frame(sd, &cpi->scaled_source, cm->temp_scale_frame.y_buffer,
tmp_height, hs, hr, vs, vr, 0);
vp8_yv12_extend_frame_borders(&cpi->scaled_source);
2010-05-18 19:58:33 +04:00
cpi->Source = &cpi->scaled_source;
#endif
}
}
2010-05-18 19:58:33 +04:00
static void resize_key_frame(VP8_COMP *cpi)
{
#if CONFIG_SPATIAL_RESAMPLING
VP8_COMMON *cm = &cpi->common;
// Do we need to apply resampling for one pass cbr.
// In one pass this is more limited than in two pass cbr
// The test and any change is only made one per key frame sequence
if (cpi->oxcf.allow_spatial_resampling && (cpi->oxcf.end_usage == USAGE_STREAM_FROM_SERVER))
{
int UNINITIALIZED_IS_SAFE(hr), UNINITIALIZED_IS_SAFE(hs);
int UNINITIALIZED_IS_SAFE(vr), UNINITIALIZED_IS_SAFE(vs);
int new_width, new_height;
// If we are below the resample DOWN watermark then scale down a notch.
if (cpi->buffer_level < (cpi->oxcf.resample_down_water_mark * cpi->oxcf.optimal_buffer_level / 100))
{
cm->horiz_scale = (cm->horiz_scale < ONETWO) ? cm->horiz_scale + 1 : ONETWO;
cm->vert_scale = (cm->vert_scale < ONETWO) ? cm->vert_scale + 1 : ONETWO;
}
// Should we now start scaling back up
else if (cpi->buffer_level > (cpi->oxcf.resample_up_water_mark * cpi->oxcf.optimal_buffer_level / 100))
{
cm->horiz_scale = (cm->horiz_scale > NORMAL) ? cm->horiz_scale - 1 : NORMAL;
cm->vert_scale = (cm->vert_scale > NORMAL) ? cm->vert_scale - 1 : NORMAL;
}
// Get the new hieght and width
Scale2Ratio(cm->horiz_scale, &hr, &hs);
Scale2Ratio(cm->vert_scale, &vr, &vs);
new_width = ((hs - 1) + (cpi->oxcf.Width * hr)) / hs;
new_height = ((vs - 1) + (cpi->oxcf.Height * vr)) / vs;
// If the image size has changed we need to reallocate the buffers
// and resample the source image
if ((cm->Width != new_width) || (cm->Height != new_height))
{
cm->Width = new_width;
cm->Height = new_height;
vp8_alloc_compressor_data(cpi);
scale_and_extend_source(cpi->un_scaled_source, cpi);
}
}
#endif
}
static void update_alt_ref_frame_stats(VP8_COMP *cpi)
2010-05-18 19:58:33 +04:00
{
VP8_COMMON *cm = &cpi->common;
// Select an interval before next GF or altref
if (!cpi->auto_gold)
cpi->frames_till_gf_update_due = cpi->goldfreq;
if ((cpi->pass != 2) && cpi->frames_till_gf_update_due)
{
cpi->current_gf_interval = cpi->frames_till_gf_update_due;
// Set the bits per frame that we should try and recover in subsequent inter frames
// to account for the extra GF spend... note that his does not apply for GF updates
// that occur coincident with a key frame as the extra cost of key frames is dealt
// with elsewhere.
cpi->gf_overspend_bits += cpi->projected_frame_size;
cpi->non_gf_bitrate_adjustment = cpi->gf_overspend_bits / cpi->frames_till_gf_update_due;
}
// Update data structure that monitors level of reference to last GF
vpx_memset(cpi->gf_active_flags, 1, (cm->mb_rows * cm->mb_cols));
cpi->gf_active_count = cm->mb_rows * cm->mb_cols;
2010-05-18 19:58:33 +04:00
// this frame refreshes means next frames don't unless specified by user
2010-05-18 19:58:33 +04:00
cpi->common.frames_since_golden = 0;
// Clear the alternate reference update pending flag.
cpi->source_alt_ref_pending = FALSE;
// Set the alternate refernce frame active flag
cpi->source_alt_ref_active = TRUE;
}
static void update_golden_frame_stats(VP8_COMP *cpi)
2010-05-18 19:58:33 +04:00
{
VP8_COMMON *cm = &cpi->common;
// Update the Golden frame usage counts.
2010-05-18 19:58:33 +04:00
if (cm->refresh_golden_frame)
{
// Select an interval before next GF
if (!cpi->auto_gold)
cpi->frames_till_gf_update_due = cpi->goldfreq;
if ((cpi->pass != 2) && (cpi->frames_till_gf_update_due > 0))
{
cpi->current_gf_interval = cpi->frames_till_gf_update_due;
// Set the bits per frame that we should try and recover in subsequent inter frames
// to account for the extra GF spend... note that his does not apply for GF updates
// that occur coincident with a key frame as the extra cost of key frames is dealt
// with elsewhere.
if ((cm->frame_type != KEY_FRAME) && !cpi->source_alt_ref_active)
{
// Calcluate GF bits to be recovered
// Projected size - av frame bits available for inter frames for clip as a whole
cpi->gf_overspend_bits += (cpi->projected_frame_size - cpi->inter_frame_target);
}
cpi->non_gf_bitrate_adjustment = cpi->gf_overspend_bits / cpi->frames_till_gf_update_due;
}
// Update data structure that monitors level of reference to last GF
vpx_memset(cpi->gf_active_flags, 1, (cm->mb_rows * cm->mb_cols));
cpi->gf_active_count = cm->mb_rows * cm->mb_cols;
2010-05-18 19:58:33 +04:00
// this frame refreshes means next frames don't unless specified by user
cm->refresh_golden_frame = 0;
cpi->common.frames_since_golden = 0;
//if ( cm->frame_type == KEY_FRAME )
//{
cpi->recent_ref_frame_usage[INTRA_FRAME] = 1;
cpi->recent_ref_frame_usage[LAST_FRAME] = 1;
cpi->recent_ref_frame_usage[GOLDEN_FRAME] = 1;
cpi->recent_ref_frame_usage[ALTREF_FRAME] = 1;
//}
//else
//{
// // Carry a potrtion of count over to begining of next gf sequence
// cpi->recent_ref_frame_usage[INTRA_FRAME] >>= 5;
// cpi->recent_ref_frame_usage[LAST_FRAME] >>= 5;
// cpi->recent_ref_frame_usage[GOLDEN_FRAME] >>= 5;
// cpi->recent_ref_frame_usage[ALTREF_FRAME] >>= 5;
//}
// ******** Fixed Q test code only ************
// If we are going to use the ALT reference for the next group of frames set a flag to say so.
if (cpi->oxcf.fixed_q >= 0 &&
cpi->oxcf.play_alternate && !cpi->common.refresh_alt_ref_frame)
{
cpi->source_alt_ref_pending = TRUE;
cpi->frames_till_gf_update_due = cpi->baseline_gf_interval;
}
if (!cpi->source_alt_ref_pending)
cpi->source_alt_ref_active = FALSE;
// Decrement count down till next gf
if (cpi->frames_till_gf_update_due > 0)
cpi->frames_till_gf_update_due--;
}
else if (!cpi->common.refresh_alt_ref_frame)
{
// Decrement count down till next gf
if (cpi->frames_till_gf_update_due > 0)
cpi->frames_till_gf_update_due--;
if (cpi->common.frames_till_alt_ref_frame)
cpi->common.frames_till_alt_ref_frame --;
cpi->common.frames_since_golden ++;
if (cpi->common.frames_since_golden > 1)
{
cpi->recent_ref_frame_usage[INTRA_FRAME] += cpi->count_mb_ref_frame_usage[INTRA_FRAME];
cpi->recent_ref_frame_usage[LAST_FRAME] += cpi->count_mb_ref_frame_usage[LAST_FRAME];
cpi->recent_ref_frame_usage[GOLDEN_FRAME] += cpi->count_mb_ref_frame_usage[GOLDEN_FRAME];
cpi->recent_ref_frame_usage[ALTREF_FRAME] += cpi->count_mb_ref_frame_usage[ALTREF_FRAME];
}
}
}
// This function updates the reference frame probability estimates that
// will be used during mode selection
static void update_rd_ref_frame_probs(VP8_COMP *cpi)
{
VP8_COMMON *cm = &cpi->common;
#if 0
const int *const rfct = cpi->recent_ref_frame_usage;
const int rf_intra = rfct[INTRA_FRAME];
const int rf_inter = rfct[LAST_FRAME] + rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME];
if (cm->frame_type == KEY_FRAME)
{
cpi->prob_intra_coded = 255;
cpi->prob_last_coded = 128;
cpi->prob_gf_coded = 128;
}
else if (!(rf_intra + rf_inter))
{
// This is a trap in case this function is called with cpi->recent_ref_frame_usage[] blank.
cpi->prob_intra_coded = 63;
cpi->prob_last_coded = 128;
cpi->prob_gf_coded = 128;
}
else
{
cpi->prob_intra_coded = (rf_intra * 255) / (rf_intra + rf_inter);
if (cpi->prob_intra_coded < 1)
cpi->prob_intra_coded = 1;
if ((cm->frames_since_golden > 0) || cpi->source_alt_ref_active)
{
cpi->prob_last_coded = rf_inter ? (rfct[LAST_FRAME] * 255) / rf_inter : 128;
if (cpi->prob_last_coded < 1)
cpi->prob_last_coded = 1;
cpi->prob_gf_coded = (rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME])
? (rfct[GOLDEN_FRAME] * 255) / (rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME]) : 128;
if (cpi->prob_gf_coded < 1)
cpi->prob_gf_coded = 1;
}
}
#else
const int *const rfct = cpi->count_mb_ref_frame_usage;
const int rf_intra = rfct[INTRA_FRAME];
const int rf_inter = rfct[LAST_FRAME] + rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME];
if (cm->frame_type == KEY_FRAME)
{
cpi->prob_intra_coded = 255;
cpi->prob_last_coded = 128;
cpi->prob_gf_coded = 128;
}
else if (!(rf_intra + rf_inter))
{
// This is a trap in case this function is called with cpi->recent_ref_frame_usage[] blank.
cpi->prob_intra_coded = 63;
cpi->prob_last_coded = 128;
cpi->prob_gf_coded = 128;
}
else
{
cpi->prob_intra_coded = (rf_intra * 255) / (rf_intra + rf_inter);
if (cpi->prob_intra_coded < 1)
cpi->prob_intra_coded = 1;
cpi->prob_last_coded = rf_inter ? (rfct[LAST_FRAME] * 255) / rf_inter : 128;
if (cpi->prob_last_coded < 1)
cpi->prob_last_coded = 1;
cpi->prob_gf_coded = (rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME])
? (rfct[GOLDEN_FRAME] * 255) / (rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME]) : 128;
if (cpi->prob_gf_coded < 1)
cpi->prob_gf_coded = 1;
}
// update reference frame costs since we can do better than what we got last frame.
if (cpi->common.refresh_alt_ref_frame)
{
cpi->prob_intra_coded += 40;
cpi->prob_last_coded = 200;
cpi->prob_gf_coded = 1;
}
else if (cpi->common.frames_since_golden == 0)
{
cpi->prob_last_coded = 214;
cpi->prob_gf_coded = 1;
}
else if (cpi->common.frames_since_golden == 1)
{
cpi->prob_last_coded = 192;
cpi->prob_gf_coded = 220;
}
else if (cpi->source_alt_ref_active)
{
//int dist = cpi->common.frames_till_alt_ref_frame + cpi->common.frames_since_golden;
cpi->prob_gf_coded -= 20;
if (cpi->prob_gf_coded < 10)
cpi->prob_gf_coded = 10;
}
#endif
}
// 1 = key, 0 = inter
static int decide_key_frame(VP8_COMP *cpi)
{
VP8_COMMON *cm = &cpi->common;
int code_key_frame = FALSE;
cpi->kf_boost = 0;
if (cpi->Speed > 11)
return FALSE;
// Clear down mmx registers
vp8_clear_system_state(); //__asm emms;
if ((cpi->compressor_speed == 2) && (cpi->Speed >= 5) && (cpi->sf.RD == 0))
{
double change = 1.0 * abs((int)(cpi->intra_error - cpi->last_intra_error)) / (1 + cpi->last_intra_error);
double change2 = 1.0 * abs((int)(cpi->prediction_error - cpi->last_prediction_error)) / (1 + cpi->last_prediction_error);
double minerror = cm->MBs * 256;
#if 0
if (10 * cpi->intra_error / (1 + cpi->prediction_error) < 15
&& cpi->prediction_error > minerror
&& (change > .25 || change2 > .25))
{
FILE *f = fopen("intra_inter.stt", "a");
if (cpi->prediction_error <= 0)
cpi->prediction_error = 1;
fprintf(f, "%d %d %d %d %14.4f\n",
cm->current_video_frame,
(int) cpi->prediction_error,
(int) cpi->intra_error,
(int)((10 * cpi->intra_error) / cpi->prediction_error),
change);
fclose(f);
}
#endif
cpi->last_intra_error = cpi->intra_error;
cpi->last_prediction_error = cpi->prediction_error;
if (10 * cpi->intra_error / (1 + cpi->prediction_error) < 15
&& cpi->prediction_error > minerror
&& (change > .25 || change2 > .25))
{
/*(change > 1.4 || change < .75)&& cpi->this_frame_percent_intra > cpi->last_frame_percent_intra + 3*/
return TRUE;
}
return FALSE;
}
// If the following are true we might as well code a key frame
if (((cpi->this_frame_percent_intra == 100) &&
(cpi->this_frame_percent_intra > (cpi->last_frame_percent_intra + 2))) ||
((cpi->this_frame_percent_intra > 95) &&
(cpi->this_frame_percent_intra >= (cpi->last_frame_percent_intra + 5))))
{
code_key_frame = TRUE;
}
// in addition if the following are true and this is not a golden frame then code a key frame
// Note that on golden frames there often seems to be a pop in intra useage anyway hence this
// restriction is designed to prevent spurious key frames. The Intra pop needs to be investigated.
else if (((cpi->this_frame_percent_intra > 60) &&
(cpi->this_frame_percent_intra > (cpi->last_frame_percent_intra * 2))) ||
((cpi->this_frame_percent_intra > 75) &&
(cpi->this_frame_percent_intra > (cpi->last_frame_percent_intra * 3 / 2))) ||
((cpi->this_frame_percent_intra > 90) &&
(cpi->this_frame_percent_intra > (cpi->last_frame_percent_intra + 10))))
{
if (!cm->refresh_golden_frame)
code_key_frame = TRUE;
}
return code_key_frame;
}
#if !(CONFIG_REALTIME_ONLY)
static void Pass1Encode(VP8_COMP *cpi, unsigned long *size, unsigned char *dest, unsigned int *frame_flags)
{
(void) size;
(void) dest;
(void) frame_flags;
vp8_set_quantizer(cpi, 26);
2010-05-18 19:58:33 +04:00
scale_and_extend_source(cpi->un_scaled_source, cpi);
vp8_first_pass(cpi);
}
#endif
#if 0
void write_cx_frame_to_file(YV12_BUFFER_CONFIG *frame, int this_frame)
{
// write the frame
FILE *yframe;
int i;
char filename[255];
sprintf(filename, "cx\\y%04d.raw", this_frame);
yframe = fopen(filename, "wb");
for (i = 0; i < frame->y_height; i++)
fwrite(frame->y_buffer + i * frame->y_stride, frame->y_width, 1, yframe);
fclose(yframe);
sprintf(filename, "cx\\u%04d.raw", this_frame);
yframe = fopen(filename, "wb");
for (i = 0; i < frame->uv_height; i++)
fwrite(frame->u_buffer + i * frame->uv_stride, frame->uv_width, 1, yframe);
fclose(yframe);
sprintf(filename, "cx\\v%04d.raw", this_frame);
yframe = fopen(filename, "wb");
for (i = 0; i < frame->uv_height; i++)
fwrite(frame->v_buffer + i * frame->uv_stride, frame->uv_width, 1, yframe);
fclose(yframe);
}
#endif
// return of 0 means drop frame
// Function to test for conditions that indeicate we should loop
// back and recode a frame.
static BOOL recode_loop_test( VP8_COMP *cpi,
int high_limit, int low_limit,
int q, int maxq, int minq )
{
BOOL force_recode = FALSE;
VP8_COMMON *cm = &cpi->common;
// Is frame recode allowed at all
// Yes if either recode mode 1 is selected or mode two is selcted
// and the frame is a key frame. golden frame or alt_ref_frame
if ( (cpi->sf.recode_loop == 1) ||
( (cpi->sf.recode_loop == 2) &&
( (cm->frame_type == KEY_FRAME) ||
cm->refresh_golden_frame ||
cm->refresh_alt_ref_frame ) ) )
{
// General over and under shoot tests
if ( ((cpi->projected_frame_size > high_limit) && (q < maxq)) ||
((cpi->projected_frame_size < low_limit) && (q > minq)) )
{
force_recode = TRUE;
}
// Special Constrained quality tests
else if (cpi->oxcf.end_usage == USAGE_CONSTRAINED_QUALITY)
{
// Undershoot and below auto cq level
if ( (q > cpi->cq_target_quality) &&
(cpi->projected_frame_size <
((cpi->this_frame_target * 7) >> 3)))
{
force_recode = TRUE;
}
// Severe undershoot and between auto and user cq level
else if ( (q > cpi->oxcf.cq_level) &&
(cpi->projected_frame_size < cpi->min_frame_bandwidth) &&
(cpi->active_best_quality > cpi->oxcf.cq_level))
{
force_recode = TRUE;
cpi->active_best_quality = cpi->oxcf.cq_level;
}
}
}
return force_recode;
}
void update_reference_frames(VP8_COMMON *cm)
{
YV12_BUFFER_CONFIG *yv12_fb = cm->yv12_fb;
// At this point the new frame has been encoded.
// If any buffer copy / swapping is signaled it should be done here.
if (cm->frame_type == KEY_FRAME)
{
yv12_fb[cm->new_fb_idx].flags |= VP8_GOLD_FLAG | VP8_ALT_FLAG ;
yv12_fb[cm->gld_fb_idx].flags &= ~VP8_GOLD_FLAG;
yv12_fb[cm->alt_fb_idx].flags &= ~VP8_ALT_FLAG;
cm->alt_fb_idx = cm->gld_fb_idx = cm->new_fb_idx;
}
else /* For non key frames */
{
if (cm->refresh_alt_ref_frame)
{
assert(!cm->copy_buffer_to_arf);
cm->yv12_fb[cm->new_fb_idx].flags |= VP8_ALT_FLAG;
cm->yv12_fb[cm->alt_fb_idx].flags &= ~VP8_ALT_FLAG;
cm->alt_fb_idx = cm->new_fb_idx;
}
else if (cm->copy_buffer_to_arf)
{
assert(!(cm->copy_buffer_to_arf & ~0x3));
if (cm->copy_buffer_to_arf == 1)
{
if(cm->alt_fb_idx != cm->lst_fb_idx)
{
yv12_fb[cm->lst_fb_idx].flags |= VP8_ALT_FLAG;
yv12_fb[cm->alt_fb_idx].flags &= ~VP8_ALT_FLAG;
cm->alt_fb_idx = cm->lst_fb_idx;
}
}
else /* if (cm->copy_buffer_to_arf == 2) */
{
if(cm->alt_fb_idx != cm->gld_fb_idx)
{
yv12_fb[cm->gld_fb_idx].flags |= VP8_ALT_FLAG;
yv12_fb[cm->alt_fb_idx].flags &= ~VP8_ALT_FLAG;
cm->alt_fb_idx = cm->gld_fb_idx;
}
}
}
if (cm->refresh_golden_frame)
{
assert(!cm->copy_buffer_to_gf);
cm->yv12_fb[cm->new_fb_idx].flags |= VP8_GOLD_FLAG;
cm->yv12_fb[cm->gld_fb_idx].flags &= ~VP8_GOLD_FLAG;
cm->gld_fb_idx = cm->new_fb_idx;
}
else if (cm->copy_buffer_to_gf)
{
assert(!(cm->copy_buffer_to_arf & ~0x3));
if (cm->copy_buffer_to_gf == 1)
{
if(cm->gld_fb_idx != cm->lst_fb_idx)
{
yv12_fb[cm->lst_fb_idx].flags |= VP8_GOLD_FLAG;
yv12_fb[cm->gld_fb_idx].flags &= ~VP8_GOLD_FLAG;
cm->gld_fb_idx = cm->lst_fb_idx;
}
}
else /* if (cm->copy_buffer_to_gf == 2) */
{
if(cm->alt_fb_idx != cm->gld_fb_idx)
{
yv12_fb[cm->alt_fb_idx].flags |= VP8_GOLD_FLAG;
yv12_fb[cm->gld_fb_idx].flags &= ~VP8_GOLD_FLAG;
cm->gld_fb_idx = cm->alt_fb_idx;
}
}
}
}
if (cm->refresh_last_frame)
{
cm->yv12_fb[cm->new_fb_idx].flags |= VP8_LAST_FLAG;
cm->yv12_fb[cm->lst_fb_idx].flags &= ~VP8_LAST_FLAG;
cm->lst_fb_idx = cm->new_fb_idx;
}
}
void loopfilter_frame(VP8_COMP *cpi, VP8_COMMON *cm)
{
if (cm->no_lpf)
{
cm->filter_level = 0;
}
else
{
struct vpx_usec_timer timer;
vp8_clear_system_state();
vpx_usec_timer_start(&timer);
if (cpi->sf.auto_filter == 0)
vp8cx_pick_filter_level_fast(cpi->Source, cpi);
else
vp8cx_pick_filter_level(cpi->Source, cpi);
vpx_usec_timer_mark(&timer);
cpi->time_pick_lpf += vpx_usec_timer_elapsed(&timer);
}
#if CONFIG_MULTITHREAD
if (cpi->b_multi_threaded)
sem_post(&cpi->h_event_end_lpf); /* signal that we have set filter_level */
#endif
if (cm->filter_level > 0)
{
vp8cx_set_alt_lf_level(cpi, cm->filter_level);
vp8_loop_filter_frame(cm, &cpi->mb.e_mbd, cm->filter_level);
cm->last_filter_type = cm->filter_type;
cm->last_sharpness_level = cm->sharpness_level;
}
vp8_yv12_extend_frame_borders_ptr(cm->frame_to_show);
}
static void update_buffer_level(VP8_COMP *cpi)
{
long long tmp;
/* Update the buffered average bitrate.
*
* The buffered average bitrate tracks the bitrate over the buffer
* window. Here we simulate taking a frame of average size out
* of the buffer, and putting in the new frame just encoded.
* It is calculated accordingly:
*
* A = Average Bits Per Frame In The Buffer
* P = New Frame Size
* N = Number of bits in the buffer
*
* We recalculate the average as so:
* (N-A)*A + A*P A * (N - A + P)
* A' = ------------- = ---------------
* N N
*
* This is modeled after a the standard algorithm for a moving
* average with fixed weighting (eg A' = ((N-1)*A + 1*P) / N). This makes
* the step response nonlinear but consistent with expected behavior --
* when A is large, the model adapts more quickly, since there are
* fewer frames in the buffer and conversely when A is small there
* will be more frames in the buffer so the average will adapt
* slowly.
*
* TODO(jkoleszar): This may give poor step response in some situations,
* for example motion following a long static section. It might be
* worth experimenting more with weighting by av_per_frame_bandwidth
* rather than buffered_av_per_frame_bandwidth or using a more accurate
* algorithm to get faster response. Current testing showed worse results
* with that setting though.
*
*/
/* Guard against buffered_av_per_frame_bandwidth falling to 0. Should
* never happen, but without this check, it would be irrecoverable.
*/
if(cpi->buffered_av_per_frame_bandwidth == 0)
cpi->buffered_av_per_frame_bandwidth = 1;
tmp = cpi->oxcf.maximum_buffer_size
- cpi->buffered_av_per_frame_bandwidth
+ cpi->projected_frame_size;
tmp *= cpi->buffered_av_per_frame_bandwidth;
cpi->buffered_av_per_frame_bandwidth = tmp
/ cpi->oxcf.maximum_buffer_size;
if(cpi->oxcf.end_usage == USAGE_STREAM_FROM_SERVER)
{
/* In CBR mode, buffer level is synthesized from the buffered
* average per-frame bandwidth to get the response characteristics
* of that model, rather than using the unbounded (wrt buffer size)
* bits_off_target. ie, the long term average bitrate doesn't
* matter in CBR mode. If the clip is consistently undershooting
* because it is very static, for example, you don't want to blow
* your short term bitrate budget trying to the the long term spend
* up to the target when you hit a motion section.
*
* Instead, the ratio of buffered_av_per_frame_bandwidth to the
* target av_per_frame_bandwidth is taken, scaled by
* maximum_buffer_size and centered around optimal_buffer_level,
* which presents the expected behavior of buffer_level for the other
* parts of the rate control code which handle the targeting.
*
* Note that this only happens after the starting_buffer_level
* has passed, to give the model a chance to stabilize.
*/
if(cpi->total_actual_bits > cpi->oxcf.starting_buffer_level)
{
tmp = (long long)cpi->buffered_av_per_frame_bandwidth
* cpi->oxcf.maximum_buffer_size
/ cpi->av_per_frame_bandwidth;
cpi->buffer_level = cpi->oxcf.maximum_buffer_size
- tmp
+ cpi->oxcf.optimal_buffer_level;
}
else
cpi->buffer_level = cpi->oxcf.optimal_buffer_level;
/* Accumulate recent overshoot error.
*
* If this frame is larger than the target, then accumulate
* that error to apply as a damping factor later. Only care about
* recent overshoot, so this value decays by (N-P)/N
*/
if(cpi->total_actual_bits > cpi->oxcf.starting_buffer_level)
{
long long decayed_overshoot;
decayed_overshoot = cpi->accumulated_overshoot;
decayed_overshoot *= (cpi->oxcf.maximum_buffer_size
- cpi->projected_frame_size);
decayed_overshoot /= cpi->oxcf.maximum_buffer_size;
cpi->accumulated_overshoot = decayed_overshoot;
cpi->accumulated_overshoot +=
(cpi->projected_frame_size > cpi->av_per_frame_bandwidth)
? cpi->projected_frame_size - cpi->av_per_frame_bandwidth
: 0;
}
}
else
cpi->buffer_level = cpi->bits_off_target;
}
static void encode_frame_to_data_rate
(
VP8_COMP *cpi,
unsigned long *size,
unsigned char *dest,
unsigned int *frame_flags
)
2010-05-18 19:58:33 +04:00
{
int Q;
int frame_over_shoot_limit;
int frame_under_shoot_limit;
int Loop = FALSE;
int loop_count;
int this_q;
int last_zbin_oq;
int q_low;
int q_high;
int zbin_oq_high;
int zbin_oq_low = 0;
int top_index;
int bottom_index;
VP8_COMMON *cm = &cpi->common;
int active_worst_qchanged = FALSE;
int overshoot_seen = FALSE;
int undershoot_seen = FALSE;
int drop_mark = cpi->oxcf.drop_frames_water_mark * cpi->oxcf.optimal_buffer_level / 100;
int drop_mark75 = drop_mark * 2 / 3;
int drop_mark50 = drop_mark / 4;
int drop_mark25 = drop_mark / 8;
2010-05-18 19:58:33 +04:00
// Clear down mmx registers to allow floating point in what follows
vp8_clear_system_state();
// Test code for segmentation of gf/arf (0,0)
//segmentation_test_function((VP8_PTR) cpi);
#if CONFIG_REALTIME_ONLY
if(cpi->oxcf.auto_key && cm->frame_type != KEY_FRAME)
{
if(cpi->force_next_frame_intra)
{
cm->frame_type = KEY_FRAME; /* delayed intra frame */
}
}
cpi->force_next_frame_intra = 0;
#endif
2010-05-18 19:58:33 +04:00
// For an alt ref frame in 2 pass we skip the call to the second pass function that sets the target bandwidth
#if !(CONFIG_REALTIME_ONLY)
if (cpi->pass == 2)
{
if (cpi->common.refresh_alt_ref_frame)
{
cpi->per_frame_bandwidth = cpi->twopass.gf_bits; // Per frame bit target for the alt ref frame
cpi->target_bandwidth = cpi->twopass.gf_bits * cpi->output_frame_rate; // per second target bitrate
2010-05-18 19:58:33 +04:00
}
}
else
#endif
cpi->per_frame_bandwidth = (int)(cpi->target_bandwidth / cpi->output_frame_rate);
// Default turn off buffer to buffer copying
cm->copy_buffer_to_gf = 0;
cm->copy_buffer_to_arf = 0;
// Clear zbin over-quant value and mode boost values.
cpi->zbin_over_quant = 0;
cpi->zbin_mode_boost = 0;
// Enable or disable mode based tweaking of the zbin
// For 2 Pass Only used where GF/ARF prediction quality
// is above a threshold
cpi->zbin_mode_boost = 0;
2010-05-18 19:58:33 +04:00
cpi->zbin_mode_boost_enabled = TRUE;
if (cpi->pass == 2)
{
if ( cpi->gfu_boost <= 400 )
{
cpi->zbin_mode_boost_enabled = FALSE;
}
}
2010-05-18 19:58:33 +04:00
// Current default encoder behaviour for the altref sign bias
if (cpi->source_alt_ref_active)
cpi->common.ref_frame_sign_bias[ALTREF_FRAME] = 1;
else
cpi->common.ref_frame_sign_bias[ALTREF_FRAME] = 0;
// Check to see if a key frame is signalled
// For two pass with auto key frame enabled cm->frame_type may already be set, but not for one pass.
if ((cm->current_video_frame == 0) ||
(cm->frame_flags & FRAMEFLAGS_KEY) ||
(cpi->oxcf.auto_key && (cpi->frames_since_key % cpi->key_frame_frequency == 0)))
{
// Key frame from VFW/auto-keyframe/first frame
cm->frame_type = KEY_FRAME;
}
// Set default state for segment and mode based loop filter update flags
cpi->mb.e_mbd.update_mb_segmentation_map = 0;
cpi->mb.e_mbd.update_mb_segmentation_data = 0;
cpi->mb.e_mbd.mode_ref_lf_delta_update = 0;
// Set various flags etc to special state if it is a key frame
if (cm->frame_type == KEY_FRAME)
{
int i;
// Reset the loop filter deltas and segmentation map
setup_features(cpi);
2010-05-18 19:58:33 +04:00
// If segmentation is enabled force a map update for key frames
if (cpi->mb.e_mbd.segmentation_enabled)
{
cpi->mb.e_mbd.update_mb_segmentation_map = 1;
cpi->mb.e_mbd.update_mb_segmentation_data = 1;
}
// The alternate reference frame cannot be active for a key frame
cpi->source_alt_ref_active = FALSE;
// Reset the RD threshold multipliers to default of * 1 (128)
for (i = 0; i < MAX_MODES; i++)
{
cpi->rd_thresh_mult[i] = 128;
}
}
// Test code for segmentation
//if ( (cm->frame_type == KEY_FRAME) || ((cm->current_video_frame % 2) == 0))
//if ( (cm->current_video_frame % 2) == 0 )
// enable_segmentation((VP8_PTR)cpi);
//else
// disable_segmentation((VP8_PTR)cpi);
#if 0
// Experimental code for lagged compress and one pass
// Initialise one_pass GF frames stats
// Update stats used for GF selection
//if ( cpi->pass == 0 )
{
cpi->one_pass_frame_index = cm->current_video_frame % MAX_LAG_BUFFERS;
cpi->one_pass_frame_stats[cpi->one_pass_frame_index ].frames_so_far = 0;
cpi->one_pass_frame_stats[cpi->one_pass_frame_index ].frame_intra_error = 0.0;
cpi->one_pass_frame_stats[cpi->one_pass_frame_index ].frame_coded_error = 0.0;
cpi->one_pass_frame_stats[cpi->one_pass_frame_index ].frame_pcnt_inter = 0.0;
cpi->one_pass_frame_stats[cpi->one_pass_frame_index ].frame_pcnt_motion = 0.0;
cpi->one_pass_frame_stats[cpi->one_pass_frame_index ].frame_mvr = 0.0;
cpi->one_pass_frame_stats[cpi->one_pass_frame_index ].frame_mvr_abs = 0.0;
cpi->one_pass_frame_stats[cpi->one_pass_frame_index ].frame_mvc = 0.0;
cpi->one_pass_frame_stats[cpi->one_pass_frame_index ].frame_mvc_abs = 0.0;
}
#endif
update_rd_ref_frame_probs(cpi);
if (cpi->drop_frames_allowed)
{
// The reset to decimation 0 is only done here for one pass.
// Once it is set two pass leaves decimation on till the next kf.
if ((cpi->buffer_level > drop_mark) && (cpi->decimation_factor > 0))
cpi->decimation_factor --;
if (cpi->buffer_level > drop_mark75 && cpi->decimation_factor > 0)
cpi->decimation_factor = 1;
else if (cpi->buffer_level < drop_mark25 && (cpi->decimation_factor == 2 || cpi->decimation_factor == 3))
{
cpi->decimation_factor = 3;
}
else if (cpi->buffer_level < drop_mark50 && (cpi->decimation_factor == 1 || cpi->decimation_factor == 2))
{
cpi->decimation_factor = 2;
}
else if (cpi->buffer_level < drop_mark75 && (cpi->decimation_factor == 0 || cpi->decimation_factor == 1))
{
cpi->decimation_factor = 1;
}
//vpx_log("Encoder: Decimation Factor: %d \n",cpi->decimation_factor);
}
// The following decimates the frame rate according to a regular pattern (i.e. to 1/2 or 2/3 frame rate)
// This can be used to help prevent buffer under-run in CBR mode. Alternatively it might be desirable in
// some situations to drop frame rate but throw more bits at each frame.
//
// Note that dropping a key frame can be problematic if spatial resampling is also active
if (cpi->decimation_factor > 0)
{
switch (cpi->decimation_factor)
{
case 1:
cpi->per_frame_bandwidth = cpi->per_frame_bandwidth * 3 / 2;
break;
case 2:
cpi->per_frame_bandwidth = cpi->per_frame_bandwidth * 5 / 4;
break;
case 3:
cpi->per_frame_bandwidth = cpi->per_frame_bandwidth * 5 / 4;
break;
}
// Note that we should not throw out a key frame (especially when spatial resampling is enabled).
if ((cm->frame_type == KEY_FRAME)) // && cpi->oxcf.allow_spatial_resampling )
{
cpi->decimation_count = cpi->decimation_factor;
}
else if (cpi->decimation_count > 0)
{
cpi->decimation_count --;
cpi->bits_off_target += cpi->av_per_frame_bandwidth;
cm->current_video_frame++;
cpi->frames_since_key++;
#if CONFIG_INTERNAL_STATS
2010-05-18 19:58:33 +04:00
cpi->count ++;
#endif
cpi->buffer_level = cpi->bits_off_target;
return;
}
else
cpi->decimation_count = cpi->decimation_factor;
}
// Decide how big to make the frame
if (!vp8_pick_frame_size(cpi))
2010-05-18 19:58:33 +04:00
{
cm->current_video_frame++;
cpi->frames_since_key++;
return;
}
// Reduce active_worst_allowed_q for CBR if our buffer is getting too full.
// This has a knock on effect on active best quality as well.
// For CBR if the buffer reaches its maximum level then we can no longer
// save up bits for later frames so we might as well use them up
// on the current frame.
if (cpi->pass == 2
&& (cpi->oxcf.end_usage == USAGE_STREAM_FROM_SERVER) &&
2010-05-18 19:58:33 +04:00
(cpi->buffer_level >= cpi->oxcf.optimal_buffer_level) && cpi->buffered_mode)
{
int Adjustment = cpi->active_worst_quality / 4; // Max adjustment is 1/4
if (Adjustment)
{
int buff_lvl_step;
if (cpi->buffer_level < cpi->oxcf.maximum_buffer_size)
{
buff_lvl_step = (cpi->oxcf.maximum_buffer_size - cpi->oxcf.optimal_buffer_level) / Adjustment;
if (buff_lvl_step)
Adjustment = (cpi->buffer_level - cpi->oxcf.optimal_buffer_level) / buff_lvl_step;
else
Adjustment = 0;
2010-05-18 19:58:33 +04:00
}
cpi->active_worst_quality -= Adjustment;
if(cpi->active_worst_quality < cpi->active_best_quality)
cpi->active_worst_quality = cpi->active_best_quality;
2010-05-18 19:58:33 +04:00
}
}
// Set an active best quality and if necessary active worst quality
// There is some odd behaviour for one pass here that needs attention.
if ( (cpi->pass == 2) || (cpi->ni_frames > 150))
2010-05-18 19:58:33 +04:00
{
vp8_clear_system_state();
2010-05-18 19:58:33 +04:00
Q = cpi->active_worst_quality;
if ( cm->frame_type == KEY_FRAME )
2010-05-18 19:58:33 +04:00
{
if ( cpi->pass == 2 )
2010-05-18 19:58:33 +04:00
{
if (cpi->gfu_boost > 600)
cpi->active_best_quality = kf_low_motion_minq[Q];
else
cpi->active_best_quality = kf_high_motion_minq[Q];
// Special case for key frames forced because we have reached
// the maximum key frame interval. Here force the Q to a range
// based on the ambient Q to reduce the risk of popping
if ( cpi->this_key_frame_forced )
{
if ( cpi->active_best_quality > cpi->avg_frame_qindex * 7/8)
cpi->active_best_quality = cpi->avg_frame_qindex * 7/8;
else if ( cpi->active_best_quality < cpi->avg_frame_qindex >> 2 )
cpi->active_best_quality = cpi->avg_frame_qindex >> 2;
}
}
// One pass more conservative
else
cpi->active_best_quality = kf_high_motion_minq[Q];
}
2010-05-18 19:58:33 +04:00
else if (cm->refresh_golden_frame || cpi->common.refresh_alt_ref_frame)
{
// Use the lower of cpi->active_worst_quality and recent
// average Q as basis for GF/ARF Q limit unless last frame was
// a key frame.
if ( (cpi->frames_since_key > 1) &&
(cpi->avg_frame_qindex < cpi->active_worst_quality) )
{
Q = cpi->avg_frame_qindex;
if ( (cpi->oxcf.end_usage == USAGE_CONSTRAINED_QUALITY) &&
(Q < cpi->oxcf.cq_level) )
{
Q = cpi->oxcf.cq_level;
}
}
if ( cpi->pass == 2 )
{
if ( cpi->gfu_boost > 1000 )
cpi->active_best_quality = gf_low_motion_minq[Q];
else if ( cpi->gfu_boost < 400 )
cpi->active_best_quality = gf_high_motion_minq[Q];
2010-05-18 19:58:33 +04:00
else
cpi->active_best_quality = gf_mid_motion_minq[Q];
}
// One pass more conservative
else
cpi->active_best_quality = gf_high_motion_minq[Q];
2010-05-18 19:58:33 +04:00
}
else
{
if(cpi->pass != 2)
Q = cpi->auto_worst_q?
cpi->active_worst_quality:cpi->avg_frame_qindex;
cpi->active_best_quality = inter_minq[Q];
// For the constant/constrained quality mode we dont want
// the quality to rise above the cq level.
if ((cpi->oxcf.end_usage == USAGE_CONSTRAINED_QUALITY) &&
(cpi->active_best_quality < cpi->cq_target_quality) )
{
// If we are strongly undershooting the target rate in the last
// frames then use the user passed in cq value not the auto
// cq value.
if ( cpi->rolling_actual_bits < cpi->min_frame_bandwidth )
cpi->active_best_quality = cpi->oxcf.cq_level;
else
cpi->active_best_quality = cpi->cq_target_quality;
}
2010-05-18 19:58:33 +04:00
}
// If CBR and the buffer is as full then it is reasonable to allow
// higher quality on the frames to prevent bits just going to waste.
2010-05-18 19:58:33 +04:00
if (cpi->oxcf.end_usage == USAGE_STREAM_FROM_SERVER)
{
// Note that the use of >= here elliminates the risk of a devide
// by 0 error in the else if clause
2010-05-18 19:58:33 +04:00
if (cpi->buffer_level >= cpi->oxcf.maximum_buffer_size)
cpi->active_best_quality = cpi->best_quality;
else if (cpi->buffer_level > cpi->oxcf.optimal_buffer_level)
{
int Fraction = ((cpi->buffer_level - cpi->oxcf.optimal_buffer_level) * 128) / (cpi->oxcf.maximum_buffer_size - cpi->oxcf.optimal_buffer_level);
int min_qadjustment = ((cpi->active_best_quality - cpi->best_quality) * Fraction) / 128;
cpi->active_best_quality -= min_qadjustment;
}
}
}
// Make sure constrained quality mode limits are adhered to for the first
// few frames of one pass encodes
else if (cpi->oxcf.end_usage == USAGE_CONSTRAINED_QUALITY)
{
if ( (cm->frame_type == KEY_FRAME) ||
cm->refresh_golden_frame || cpi->common.refresh_alt_ref_frame )
{
cpi->active_best_quality = cpi->best_quality;
}
else if (cpi->active_best_quality < cpi->cq_target_quality)
{
cpi->active_best_quality = cpi->cq_target_quality;
}
}
2010-05-18 19:58:33 +04:00
// Clip the active best and worst quality values to limits
if (cpi->active_worst_quality > cpi->worst_quality)
cpi->active_worst_quality = cpi->worst_quality;
if (cpi->active_best_quality < cpi->best_quality)
cpi->active_best_quality = cpi->best_quality;
else if (cpi->active_best_quality > cpi->active_worst_quality)
cpi->active_best_quality = cpi->active_worst_quality;
// Determine initial Q to try
Q = vp8_regulate_q(cpi, cpi->this_frame_target);
last_zbin_oq = cpi->zbin_over_quant;
// Set highest allowed value for Zbin over quant
if (cm->frame_type == KEY_FRAME)
zbin_oq_high = 0; //ZBIN_OQ_MAX/16
else if (cm->refresh_alt_ref_frame || (cm->refresh_golden_frame && !cpi->source_alt_ref_active))
zbin_oq_high = 16;
else
zbin_oq_high = ZBIN_OQ_MAX;
// Setup background Q adjustment for error resilliant mode
if (cpi->cyclic_refresh_mode_enabled)
cyclic_background_refresh(cpi, Q, 0);
vp8_compute_frame_size_bounds(cpi, &frame_under_shoot_limit, &frame_over_shoot_limit);
// Limit Q range for the adaptive loop.
2010-05-18 19:58:33 +04:00
bottom_index = cpi->active_best_quality;
top_index = cpi->active_worst_quality;
q_low = cpi->active_best_quality;
q_high = cpi->active_worst_quality;
2010-05-18 19:58:33 +04:00
vp8_save_coding_context(cpi);
loop_count = 0;
scale_and_extend_source(cpi->un_scaled_source, cpi);
#if !(CONFIG_REALTIME_ONLY) && CONFIG_POSTPROC
if (cpi->oxcf.noise_sensitivity > 0)
{
unsigned char *src;
int l = 0;
switch (cpi->oxcf.noise_sensitivity)
{
case 1:
l = 20;
break;
case 2:
l = 40;
break;
case 3:
l = 60;
break;
case 4:
l = 80;
break;
case 5:
l = 100;
break;
case 6:
l = 150;
break;
}
if (cm->frame_type == KEY_FRAME)
{
vp8_de_noise(cpi->Source, cpi->Source, l , 1, 0, RTCD(postproc));
}
else
{
vp8_de_noise(cpi->Source, cpi->Source, l , 1, 0, RTCD(postproc));
src = cpi->Source->y_buffer;
if (cpi->Source->y_stride < 0)
{
src += cpi->Source->y_stride * (cpi->Source->y_height - 1);
}
}
}
#endif
#ifdef OUTPUT_YUV_SRC
vp8_write_yuv_frame(cpi->Source);
#endif
do
{
vp8_clear_system_state(); //__asm emms;
/*
if(cpi->is_src_frame_alt_ref)
Q = 127;
*/
vp8_set_quantizer(cpi, Q);
2010-05-18 19:58:33 +04:00
this_q = Q;
// setup skip prob for costing in mode/mv decision
if (cpi->common.mb_no_coeff_skip)
{
cpi->prob_skip_false = cpi->base_skip_false_prob[Q];
if (cm->frame_type != KEY_FRAME)
{
if (cpi->common.refresh_alt_ref_frame)
{
if (cpi->last_skip_false_probs[2] != 0)
cpi->prob_skip_false = cpi->last_skip_false_probs[2];
/*
if(cpi->last_skip_false_probs[2]!=0 && abs(Q- cpi->last_skip_probs_q[2])<=16 )
cpi->prob_skip_false = cpi->last_skip_false_probs[2];
else if (cpi->last_skip_false_probs[2]!=0)
cpi->prob_skip_false = (cpi->last_skip_false_probs[2] + cpi->prob_skip_false ) / 2;
*/
}
else if (cpi->common.refresh_golden_frame)
{
if (cpi->last_skip_false_probs[1] != 0)
cpi->prob_skip_false = cpi->last_skip_false_probs[1];
/*
if(cpi->last_skip_false_probs[1]!=0 && abs(Q- cpi->last_skip_probs_q[1])<=16 )
cpi->prob_skip_false = cpi->last_skip_false_probs[1];
else if (cpi->last_skip_false_probs[1]!=0)
cpi->prob_skip_false = (cpi->last_skip_false_probs[1] + cpi->prob_skip_false ) / 2;
*/
}
else
{
if (cpi->last_skip_false_probs[0] != 0)
cpi->prob_skip_false = cpi->last_skip_false_probs[0];
/*
if(cpi->last_skip_false_probs[0]!=0 && abs(Q- cpi->last_skip_probs_q[0])<=16 )
cpi->prob_skip_false = cpi->last_skip_false_probs[0];
else if(cpi->last_skip_false_probs[0]!=0)
cpi->prob_skip_false = (cpi->last_skip_false_probs[0] + cpi->prob_skip_false ) / 2;
*/
}
//as this is for cost estimate, let's make sure it does not go extreme eitehr way
if (cpi->prob_skip_false < 5)
cpi->prob_skip_false = 5;
if (cpi->prob_skip_false > 250)
cpi->prob_skip_false = 250;
if (cpi->is_src_frame_alt_ref)
cpi->prob_skip_false = 1;
}
#if 0
if (cpi->pass != 1)
{
FILE *f = fopen("skip.stt", "a");
fprintf(f, "%d, %d, %4d ", cpi->common.refresh_golden_frame, cpi->common.refresh_alt_ref_frame, cpi->prob_skip_false);
fclose(f);
}
#endif
}
if (cm->frame_type == KEY_FRAME)
{
resize_key_frame(cpi);
2010-05-18 19:58:33 +04:00
vp8_setup_key_frame(cpi);
}
2010-05-18 19:58:33 +04:00
// transform / motion compensation build reconstruction frame
vp8_encode_frame(cpi);
2010-05-18 19:58:33 +04:00
cpi->projected_frame_size -= vp8_estimate_entropy_savings(cpi);
cpi->projected_frame_size = (cpi->projected_frame_size > 0) ? cpi->projected_frame_size : 0;
vp8_clear_system_state(); //__asm emms;
// Test to see if the stats generated for this frame indicate that we should have coded a key frame
// (assuming that we didn't)!
if (cpi->pass != 2 && cpi->oxcf.auto_key && cm->frame_type != KEY_FRAME)
{
#if CONFIG_REALTIME_ONLY
{
/* we don't do re-encoding in realtime mode
* if key frame is decided than we force it on next frame */
cpi->force_next_frame_intra = decide_key_frame(cpi);
}
#else
2010-05-18 19:58:33 +04:00
if (decide_key_frame(cpi))
{
// Reset all our sizing numbers and recode
cm->frame_type = KEY_FRAME;
vp8_pick_frame_size(cpi);
2010-05-18 19:58:33 +04:00
// Clear the Alt reference frame active flag when we have a key frame
cpi->source_alt_ref_active = FALSE;
// Reset the loop filter deltas and segmentation map
setup_features(cpi);
2010-05-18 19:58:33 +04:00
// If segmentation is enabled force a map update for key frames
if (cpi->mb.e_mbd.segmentation_enabled)
{
cpi->mb.e_mbd.update_mb_segmentation_map = 1;
cpi->mb.e_mbd.update_mb_segmentation_data = 1;
}
vp8_restore_coding_context(cpi);
Q = vp8_regulate_q(cpi, cpi->this_frame_target);
vp8_compute_frame_size_bounds(cpi, &frame_under_shoot_limit, &frame_over_shoot_limit);
// Limit Q range for the adaptive loop.
2010-05-18 19:58:33 +04:00
bottom_index = cpi->active_best_quality;
top_index = cpi->active_worst_quality;
q_low = cpi->active_best_quality;
q_high = cpi->active_worst_quality;
2010-05-18 19:58:33 +04:00
loop_count++;
Loop = TRUE;
continue;
}
#endif
2010-05-18 19:58:33 +04:00
}
vp8_clear_system_state();
if (frame_over_shoot_limit == 0)
frame_over_shoot_limit = 1;
// Are we are overshooting and up against the limit of active max Q.
if (((cpi->pass != 2) || (cpi->oxcf.end_usage == USAGE_STREAM_FROM_SERVER)) &&
(Q == cpi->active_worst_quality) &&
(cpi->active_worst_quality < cpi->worst_quality) &&
(cpi->projected_frame_size > frame_over_shoot_limit))
{
/* step down active_worst_quality such that the corresponding
* active_best_quality will be equal to the current
* active_worst_quality + 1. Once the limit on active_best_quality
* is reached, active_worst_quality will equal worst_quality.
*/
int i;
2010-05-18 19:58:33 +04:00
for(i=cpi->active_worst_quality; i<cpi->worst_quality; i++)
if(inter_minq[i] >= cpi->active_worst_quality + 1)
break;
cpi->active_worst_quality = i;
2010-05-18 19:58:33 +04:00
// If we have updated the active max Q do not call vp8_update_rate_correction_factors() this loop.
active_worst_qchanged = TRUE;
}
else
active_worst_qchanged = FALSE;
#if !(CONFIG_REALTIME_ONLY)
// Special case handling for forced key frames
if ( (cm->frame_type == KEY_FRAME) && cpi->this_key_frame_forced )
{
int last_q = Q;
int kf_err = vp8_calc_ss_err(cpi->Source,
&cm->yv12_fb[cm->new_fb_idx],
IF_RTCD(&cpi->rtcd.variance));
// The key frame is not good enough
if ( kf_err > ((cpi->ambient_err * 7) >> 3) )
{
// Lower q_high
q_high = (Q > q_low) ? (Q - 1) : q_low;
// Adjust Q
Q = (q_high + q_low) >> 1;
}
// The key frame is much better than the previous frame
else if ( kf_err < (cpi->ambient_err >> 1) )
{
// Raise q_low
q_low = (Q < q_high) ? (Q + 1) : q_high;
// Adjust Q
Q = (q_high + q_low + 1) >> 1;
}
// Clamp Q to upper and lower limits:
if (Q > q_high)
Q = q_high;
else if (Q < q_low)
Q = q_low;
Loop = ((Q != last_q)) ? TRUE : FALSE;
}
2010-05-18 19:58:33 +04:00
// Is the projected frame size out of range and are we allowed to attempt to recode.
else if ( recode_loop_test( cpi,
frame_over_shoot_limit, frame_under_shoot_limit,
Q, top_index, bottom_index ) )
2010-05-18 19:58:33 +04:00
{
int last_q = Q;
int Retries = 0;
// Frame size out of permitted range:
// Update correction factor & compute new Q to try...
// Frame is too large
if (cpi->projected_frame_size > cpi->this_frame_target)
2010-05-18 19:58:33 +04:00
{
//if ( cpi->zbin_over_quant == 0 )
q_low = (Q < q_high) ? (Q + 1) : q_high; // Raise Qlow as to at least the current value
if (cpi->zbin_over_quant > 0) // If we are using over quant do the same for zbin_oq_low
2010-05-18 19:58:33 +04:00
zbin_oq_low = (cpi->zbin_over_quant < zbin_oq_high) ? (cpi->zbin_over_quant + 1) : zbin_oq_high;
//if ( undershoot_seen || (Q == MAXQ) )
if (undershoot_seen)
{
// Update rate_correction_factor unless cpi->active_worst_quality has changed.
if (!active_worst_qchanged)
vp8_update_rate_correction_factors(cpi, 1);
Q = (q_high + q_low + 1) / 2;
// Adjust cpi->zbin_over_quant (only allowed when Q is max)
if (Q < MAXQ)
cpi->zbin_over_quant = 0;
else
{
zbin_oq_low = (cpi->zbin_over_quant < zbin_oq_high) ? (cpi->zbin_over_quant + 1) : zbin_oq_high;
cpi->zbin_over_quant = (zbin_oq_high + zbin_oq_low) / 2;
}
}
else
{
// Update rate_correction_factor unless cpi->active_worst_quality has changed.
if (!active_worst_qchanged)
vp8_update_rate_correction_factors(cpi, 0);
Q = vp8_regulate_q(cpi, cpi->this_frame_target);
while (((Q < q_low) || (cpi->zbin_over_quant < zbin_oq_low)) && (Retries < 10))
{
vp8_update_rate_correction_factors(cpi, 0);
Q = vp8_regulate_q(cpi, cpi->this_frame_target);
Retries ++;
}
}
overshoot_seen = TRUE;
}
// Frame is too small
2010-05-18 19:58:33 +04:00
else
{
if (cpi->zbin_over_quant == 0)
q_high = (Q > q_low) ? (Q - 1) : q_low; // Lower q_high if not using over quant
else // else lower zbin_oq_high
zbin_oq_high = (cpi->zbin_over_quant > zbin_oq_low) ? (cpi->zbin_over_quant - 1) : zbin_oq_low;
if (overshoot_seen)
{
// Update rate_correction_factor unless cpi->active_worst_quality has changed.
if (!active_worst_qchanged)
vp8_update_rate_correction_factors(cpi, 1);
Q = (q_high + q_low) / 2;
// Adjust cpi->zbin_over_quant (only allowed when Q is max)
if (Q < MAXQ)
cpi->zbin_over_quant = 0;
else
cpi->zbin_over_quant = (zbin_oq_high + zbin_oq_low) / 2;
}
else
{
// Update rate_correction_factor unless cpi->active_worst_quality has changed.
if (!active_worst_qchanged)
vp8_update_rate_correction_factors(cpi, 0);
Q = vp8_regulate_q(cpi, cpi->this_frame_target);
// Special case reset for qlow for constrained quality.
// This should only trigger where there is very substantial
// undershoot on a frame and the auto cq level is above
// the user passsed in value.
if ( (cpi->oxcf.end_usage == USAGE_CONSTRAINED_QUALITY) &&
(Q < q_low) )
{
q_low = Q;
}
2010-05-18 19:58:33 +04:00
while (((Q > q_high) || (cpi->zbin_over_quant > zbin_oq_high)) && (Retries < 10))
{
vp8_update_rate_correction_factors(cpi, 0);
Q = vp8_regulate_q(cpi, cpi->this_frame_target);
Retries ++;
}
}
undershoot_seen = TRUE;
}
// Clamp Q to upper and lower limits:
if (Q > q_high)
Q = q_high;
else if (Q < q_low)
Q = q_low;
// Clamp cpi->zbin_over_quant
cpi->zbin_over_quant = (cpi->zbin_over_quant < zbin_oq_low) ? zbin_oq_low : (cpi->zbin_over_quant > zbin_oq_high) ? zbin_oq_high : cpi->zbin_over_quant;
//Loop = ((Q != last_q) || (last_zbin_oq != cpi->zbin_over_quant)) ? TRUE : FALSE;
Loop = ((Q != last_q)) ? TRUE : FALSE;
last_zbin_oq = cpi->zbin_over_quant;
}
else
#endif
Loop = FALSE;
if (cpi->is_src_frame_alt_ref)
Loop = FALSE;
if (Loop == TRUE)
{
vp8_restore_coding_context(cpi);
loop_count++;
#if CONFIG_INTERNAL_STATS
2010-05-18 19:58:33 +04:00
cpi->tot_recode_hits++;
#endif
}
}
while (Loop == TRUE);
#if 0
// Experimental code for lagged and one pass
// Update stats used for one pass GF selection
{
/*
int frames_so_far;
double frame_intra_error;
double frame_coded_error;
double frame_pcnt_inter;
double frame_pcnt_motion;
double frame_mvr;
double frame_mvr_abs;
double frame_mvc;
double frame_mvc_abs;
*/
cpi->one_pass_frame_stats[cpi->one_pass_frame_index].frame_coded_error = (double)cpi->prediction_error;
cpi->one_pass_frame_stats[cpi->one_pass_frame_index].frame_intra_error = (double)cpi->intra_error;
cpi->one_pass_frame_stats[cpi->one_pass_frame_index].frame_pcnt_inter = (double)(100 - cpi->this_frame_percent_intra) / 100.0;
}
#endif
// Special case code to reduce pulsing when key frames are forced at a
// fixed interval. Note the reconstruction error if it is the frame before
// the force key frame
if ( cpi->next_key_frame_forced && (cpi->twopass.frames_to_key == 0) )
{
cpi->ambient_err = vp8_calc_ss_err(cpi->Source,
&cm->yv12_fb[cm->new_fb_idx],
IF_RTCD(&cpi->rtcd.variance));
}
// This frame's MVs are saved and will be used in next frame's MV prediction.
// Last frame has one more line(add to bottom) and one more column(add to right) than cm->mip. The edge elements are initialized to 0.
if(cm->show_frame) //do not save for altref frame
{
int mb_row;
int mb_col;
MODE_INFO *tmp = cm->mip; //point to beginning of allocated MODE_INFO arrays.
if(cm->frame_type != KEY_FRAME)
{
for (mb_row = 0; mb_row < cm->mb_rows+1; mb_row ++)
{
for (mb_col = 0; mb_col < cm->mb_cols+1; mb_col ++)
{
if(tmp->mbmi.ref_frame != INTRA_FRAME)
cpi->lfmv[mb_col + mb_row*(cm->mode_info_stride+1)].as_int = tmp->mbmi.mv.as_int;
cpi->lf_ref_frame_sign_bias[mb_col + mb_row*(cm->mode_info_stride+1)] = cm->ref_frame_sign_bias[tmp->mbmi.ref_frame];
cpi->lf_ref_frame[mb_col + mb_row*(cm->mode_info_stride+1)] = tmp->mbmi.ref_frame;
tmp++;
}
}
}
}
// Update the GF useage maps.
2010-05-18 19:58:33 +04:00
// This is done after completing the compression of a frame when all modes etc. are finalized but before loop filter
// This is done after completing the compression of a frame when all modes etc. are finalized but before loop filter
vp8_update_gf_useage_maps(cpi, cm, &cpi->mb);
2010-05-18 19:58:33 +04:00
if (cm->frame_type == KEY_FRAME)
cm->refresh_last_frame = 1;
#if 0
2010-05-18 19:58:33 +04:00
{
FILE *f = fopen("gfactive.stt", "a");
fprintf(f, "%8d %8d %8d %8d %8d\n", cm->current_video_frame, (100 * cpi->gf_active_count) / (cpi->common.mb_rows * cpi->common.mb_cols), cpi->this_iiratio, cpi->next_iiratio, cm->refresh_golden_frame);
2010-05-18 19:58:33 +04:00
fclose(f);
}
#endif
2010-05-18 19:58:33 +04:00
// For inter frames the current default behavior is that when
// cm->refresh_golden_frame is set we copy the old GF over to the ARF buffer
// This is purely an encoder decision at present.
2010-05-18 19:58:33 +04:00
if (!cpi->oxcf.error_resilient_mode && cm->refresh_golden_frame)
cm->copy_buffer_to_arf = 2;
else
cm->copy_buffer_to_arf = 0;
cm->frame_to_show = &cm->yv12_fb[cm->new_fb_idx];
#if CONFIG_MULTITHREAD
if (cpi->b_multi_threaded)
2010-05-18 19:58:33 +04:00
{
sem_post(&cpi->h_event_start_lpf); /* start loopfilter in separate thread */
}
else
#endif
{
loopfilter_frame(cpi, cm);
}
2010-05-18 19:58:33 +04:00
update_reference_frames(cm);
if (cpi->oxcf.error_resilient_mode)
{
cm->refresh_entropy_probs = 0;
2010-05-18 19:58:33 +04:00
}
#if CONFIG_MULTITHREAD
/* wait that filter_level is picked so that we can continue with stream packing */
if (cpi->b_multi_threaded)
sem_wait(&cpi->h_event_end_lpf);
#endif
// build the bitstream
vp8_pack_bitstream(cpi, dest, size);
#if CONFIG_MULTITHREAD
/* wait for loopfilter thread done */
if (cpi->b_multi_threaded)
2010-05-18 19:58:33 +04:00
{
sem_wait(&cpi->h_event_end_lpf);
2010-05-18 19:58:33 +04:00
}
#endif
/* Move storing frame_type out of the above loop since it is also
* needed in motion search besides loopfilter */
cm->last_frame_type = cm->frame_type;
2010-05-18 19:58:33 +04:00
// Update rate control heuristics
cpi->total_byte_count += (*size);
cpi->projected_frame_size = (*size) << 3;
if (!active_worst_qchanged)
vp8_update_rate_correction_factors(cpi, 2);
cpi->last_q[cm->frame_type] = cm->base_qindex;
if (cm->frame_type == KEY_FRAME)
{
vp8_adjust_key_frame_context(cpi);
}
// Keep a record of ambient average Q.
if (cm->frame_type != KEY_FRAME)
2010-05-18 19:58:33 +04:00
cpi->avg_frame_qindex = (2 + 3 * cpi->avg_frame_qindex + cm->base_qindex) >> 2;
// Keep a record from which we can calculate the average Q excluding GF updates and key frames
if ((cm->frame_type != KEY_FRAME) && !cm->refresh_golden_frame && !cm->refresh_alt_ref_frame)
{
cpi->ni_frames++;
// Calculate the average Q for normal inter frames (not key or GFU
// frames).
if ( cpi->pass == 2 )
2010-05-18 19:58:33 +04:00
{
cpi->ni_tot_qi += Q;
cpi->ni_av_qi = (cpi->ni_tot_qi / cpi->ni_frames);
}
else
{
// Damp value for first few frames
if (cpi->ni_frames > 150 )
{
cpi->ni_tot_qi += Q;
cpi->ni_av_qi = (cpi->ni_tot_qi / cpi->ni_frames);
}
// For one pass, early in the clip ... average the current frame Q
// value with the worstq entered by the user as a dampening measure
else
{
cpi->ni_tot_qi += Q;
cpi->ni_av_qi = ((cpi->ni_tot_qi / cpi->ni_frames) + cpi->worst_quality + 1) / 2;
}
2010-05-18 19:58:33 +04:00
// If the average Q is higher than what was used in the last frame
// (after going through the recode loop to keep the frame size within range)
// then use the last frame value - 1.
// The -1 is designed to stop Q and hence the data rate, from progressively
// falling away during difficult sections, but at the same time reduce the number of
// itterations around the recode loop.
if (Q > cpi->ni_av_qi)
cpi->ni_av_qi = Q - 1;
}
2010-05-18 19:58:33 +04:00
}
#if 0
// If the frame was massively oversize and we are below optimal buffer level drop next frame
if ((cpi->drop_frames_allowed) &&
(cpi->oxcf.end_usage == USAGE_STREAM_FROM_SERVER) &&
(cpi->buffer_level < cpi->oxcf.drop_frames_water_mark * cpi->oxcf.optimal_buffer_level / 100) &&
(cpi->projected_frame_size > (4 * cpi->this_frame_target)))
{
cpi->drop_frame = TRUE;
}
#endif
// Set the count for maximum consequative dropped frames based upon the ratio of
// this frame size to the target average per frame bandwidth.
// (cpi->av_per_frame_bandwidth > 0) is just a sanity check to prevent / 0.
if (cpi->drop_frames_allowed && (cpi->av_per_frame_bandwidth > 0))
{
cpi->max_drop_count = cpi->projected_frame_size / cpi->av_per_frame_bandwidth;
if (cpi->max_drop_count > cpi->max_consec_dropped_frames)
cpi->max_drop_count = cpi->max_consec_dropped_frames;
}
// Update the buffer level variable.
// Non-viewable frames are a special case and are treated as pure overhead.
if ( cm->show_frame )
cpi->bits_off_target += cpi->av_per_frame_bandwidth;
cpi->bits_off_target -= cpi->projected_frame_size;
2010-05-18 19:58:33 +04:00
// Rolling monitors of whether we are over or underspending used to help regulate min and Max Q in two pass.
cpi->rolling_target_bits = ((cpi->rolling_target_bits * 3) + cpi->this_frame_target + 2) / 4;
cpi->rolling_actual_bits = ((cpi->rolling_actual_bits * 3) + cpi->projected_frame_size + 2) / 4;
cpi->long_rolling_target_bits = ((cpi->long_rolling_target_bits * 31) + cpi->this_frame_target + 16) / 32;
cpi->long_rolling_actual_bits = ((cpi->long_rolling_actual_bits * 31) + cpi->projected_frame_size + 16) / 32;
// Actual bits spent
cpi->total_actual_bits += cpi->projected_frame_size;
// Debug stats
cpi->total_target_vs_actual += (cpi->this_frame_target - cpi->projected_frame_size);
update_buffer_level(cpi);
2010-05-18 19:58:33 +04:00
// Update bits left to the kf and gf groups to account for overshoot or undershoot on these frames
if (cm->frame_type == KEY_FRAME)
{
cpi->twopass.kf_group_bits += cpi->this_frame_target - cpi->projected_frame_size;
2010-05-18 19:58:33 +04:00
if (cpi->twopass.kf_group_bits < 0)
cpi->twopass.kf_group_bits = 0 ;
2010-05-18 19:58:33 +04:00
}
else if (cm->refresh_golden_frame || cm->refresh_alt_ref_frame)
{
cpi->twopass.gf_group_bits += cpi->this_frame_target - cpi->projected_frame_size;
2010-05-18 19:58:33 +04:00
if (cpi->twopass.gf_group_bits < 0)
cpi->twopass.gf_group_bits = 0 ;
2010-05-18 19:58:33 +04:00
}
if (cm->frame_type != KEY_FRAME)
{
if (cpi->common.refresh_alt_ref_frame)
{
cpi->last_skip_false_probs[2] = cpi->prob_skip_false;
cpi->last_skip_probs_q[2] = cm->base_qindex;
}
else if (cpi->common.refresh_golden_frame)
{
cpi->last_skip_false_probs[1] = cpi->prob_skip_false;
cpi->last_skip_probs_q[1] = cm->base_qindex;
}
else
{
cpi->last_skip_false_probs[0] = cpi->prob_skip_false;
cpi->last_skip_probs_q[0] = cm->base_qindex;
//update the baseline
cpi->base_skip_false_prob[cm->base_qindex] = cpi->prob_skip_false;
}
}
#if 0 && CONFIG_INTERNAL_STATS
2010-05-18 19:58:33 +04:00
{
FILE *f = fopen("tmp.stt", "a");
vp8_clear_system_state(); //__asm emms;
if (cpi->twopass.total_coded_error_left != 0.0)
fprintf(f, "%10d %10d %10d %10d %10d %10d %10d %10d %6d %6d"
"%6d %6d %6d %5d %5d %5d %8d %8.2f %10d %10.3f"
"%10.3f %8d\n",
cpi->common.current_video_frame, cpi->this_frame_target,
cpi->projected_frame_size,
(cpi->projected_frame_size - cpi->this_frame_target),
(int)cpi->total_target_vs_actual,
(cpi->oxcf.starting_buffer_level-cpi->bits_off_target),
(int)cpi->total_actual_bits, cm->base_qindex,
cpi->active_best_quality, cpi->active_worst_quality,
cpi->ni_av_qi, cpi->cq_target_quality, cpi->zbin_over_quant,
//cpi->avg_frame_qindex, cpi->zbin_over_quant,
cm->refresh_golden_frame, cm->refresh_alt_ref_frame,
cm->frame_type, cpi->gfu_boost,
cpi->twopass.est_max_qcorrection_factor, (int)cpi->twopass.bits_left,
cpi->twopass.total_coded_error_left,
(double)cpi->twopass.bits_left / cpi->twopass.total_coded_error_left,
cpi->tot_recode_hits);
2010-05-18 19:58:33 +04:00
else
fprintf(f, "%10d %10d %10d %10d %10d %10d %10d %10d %6d %6d"
"%6d %6d %6d %5d %5d %5d %8d %8.2f %10d %10.3f"
"%8d\n",
cpi->common.current_video_frame,
cpi->this_frame_target, cpi->projected_frame_size,
(cpi->projected_frame_size - cpi->this_frame_target),
(int)cpi->total_target_vs_actual,
(cpi->oxcf.starting_buffer_level-cpi->bits_off_target),
(int)cpi->total_actual_bits, cm->base_qindex,
cpi->active_best_quality, cpi->active_worst_quality,
cpi->ni_av_qi, cpi->cq_target_quality, cpi->zbin_over_quant,
//cpi->avg_frame_qindex, cpi->zbin_over_quant,
cm->refresh_golden_frame, cm->refresh_alt_ref_frame,
cm->frame_type, cpi->gfu_boost,
cpi->twopass.est_max_qcorrection_factor, (int)cpi->twopass.bits_left,
cpi->twopass.total_coded_error_left, cpi->tot_recode_hits);
2010-05-18 19:58:33 +04:00
fclose(f);
{
FILE *fmodes = fopen("Modes.stt", "a");
int i;
fprintf(fmodes, "%6d:%1d:%1d:%1d ",
cpi->common.current_video_frame,
cm->frame_type, cm->refresh_golden_frame,
cm->refresh_alt_ref_frame);
2010-05-18 19:58:33 +04:00
for (i = 0; i < MAX_MODES; i++)
fprintf(fmodes, "%5d ", cpi->mode_chosen_counts[i]);
fprintf(fmodes, "\n");
fclose(fmodes);
}
}
#endif
// If this was a kf or Gf note the Q
if ((cm->frame_type == KEY_FRAME) || cm->refresh_golden_frame || cm->refresh_alt_ref_frame)
cm->last_kf_gf_q = cm->base_qindex;
if (cm->refresh_golden_frame == 1)
cm->frame_flags = cm->frame_flags | FRAMEFLAGS_GOLDEN;
else
cm->frame_flags = cm->frame_flags&~FRAMEFLAGS_GOLDEN;
if (cm->refresh_alt_ref_frame == 1)
cm->frame_flags = cm->frame_flags | FRAMEFLAGS_ALTREF;
else
cm->frame_flags = cm->frame_flags&~FRAMEFLAGS_ALTREF;
if (cm->refresh_last_frame & cm->refresh_golden_frame) // both refreshed
cpi->gold_is_last = 1;
else if (cm->refresh_last_frame ^ cm->refresh_golden_frame) // 1 refreshed but not the other
cpi->gold_is_last = 0;
if (cm->refresh_last_frame & cm->refresh_alt_ref_frame) // both refreshed
cpi->alt_is_last = 1;
else if (cm->refresh_last_frame ^ cm->refresh_alt_ref_frame) // 1 refreshed but not the other
cpi->alt_is_last = 0;
if (cm->refresh_alt_ref_frame & cm->refresh_golden_frame) // both refreshed
cpi->gold_is_alt = 1;
else if (cm->refresh_alt_ref_frame ^ cm->refresh_golden_frame) // 1 refreshed but not the other
cpi->gold_is_alt = 0;
cpi->ref_frame_flags = VP8_ALT_FLAG | VP8_GOLD_FLAG | VP8_LAST_FLAG;
if (cpi->gold_is_last)
cpi->ref_frame_flags &= ~VP8_GOLD_FLAG;
2010-05-18 19:58:33 +04:00
if (cpi->alt_is_last)
cpi->ref_frame_flags &= ~VP8_ALT_FLAG;
2010-05-18 19:58:33 +04:00
if (cpi->gold_is_alt)
cpi->ref_frame_flags &= ~VP8_ALT_FLAG;
2010-05-18 19:58:33 +04:00
if (!cpi->oxcf.error_resilient_mode)
2010-05-18 19:58:33 +04:00
{
if (cpi->oxcf.play_alternate && cm->refresh_alt_ref_frame && (cm->frame_type != KEY_FRAME))
// Update the alternate reference frame stats as appropriate.
update_alt_ref_frame_stats(cpi);
2010-05-18 19:58:33 +04:00
else
// Update the Golden frame stats as appropriate.
update_golden_frame_stats(cpi);
2010-05-18 19:58:33 +04:00
}
if (cm->frame_type == KEY_FRAME)
{
// Tell the caller that the frame was coded as a key frame
*frame_flags = cm->frame_flags | FRAMEFLAGS_KEY;
// As this frame is a key frame the next defaults to an inter frame.
cm->frame_type = INTER_FRAME;
cpi->last_frame_percent_intra = 100;
}
else
{
*frame_flags = cm->frame_flags&~FRAMEFLAGS_KEY;
cpi->last_frame_percent_intra = cpi->this_frame_percent_intra;
}
// Clear the one shot update flags for segmentation map and mode/ref loop filter deltas.
cpi->mb.e_mbd.update_mb_segmentation_map = 0;
cpi->mb.e_mbd.update_mb_segmentation_data = 0;
cpi->mb.e_mbd.mode_ref_lf_delta_update = 0;
// Dont increment frame counters if this was an altref buffer update not a real frame
if (cm->show_frame)
{
cm->current_video_frame++;
cpi->frames_since_key++;
}
// reset to normal state now that we are done.
#if 0
2010-05-18 19:58:33 +04:00
{
char filename[512];
FILE *recon_file;
sprintf(filename, "enc%04d.yuv", (int) cm->current_video_frame);
recon_file = fopen(filename, "wb");
fwrite(cm->yv12_fb[cm->lst_fb_idx].buffer_alloc,
cm->yv12_fb[cm->lst_fb_idx].frame_size, 1, recon_file);
2010-05-18 19:58:33 +04:00
fclose(recon_file);
}
#endif
2010-05-18 19:58:33 +04:00
// DEBUG
//vp8_write_yuv_frame("encoder_recon.yuv", cm->frame_to_show);
}
static void check_gf_quality(VP8_COMP *cpi)
2010-05-18 19:58:33 +04:00
{
VP8_COMMON *cm = &cpi->common;
int gf_active_pct = (100 * cpi->gf_active_count) / (cm->mb_rows * cm->mb_cols);
2010-05-18 19:58:33 +04:00
int gf_ref_usage_pct = (cpi->count_mb_ref_frame_usage[GOLDEN_FRAME] * 100) / (cm->mb_rows * cm->mb_cols);
int last_ref_zz_useage = (cpi->inter_zz_count * 100) / (cm->mb_rows * cm->mb_cols);
// Gf refresh is not currently being signalled
if (cpi->gf_update_recommended == 0)
{
if (cpi->common.frames_since_golden > 7)
{
// Low use of gf
if ((gf_active_pct < 10) || ((gf_active_pct + gf_ref_usage_pct) < 15))
{
// ...but last frame zero zero usage is reasonbable so a new gf might be appropriate
if (last_ref_zz_useage >= 25)
{
cpi->gf_bad_count ++;
if (cpi->gf_bad_count >= 8) // Check that the condition is stable
{
cpi->gf_update_recommended = 1;
cpi->gf_bad_count = 0;
}
}
else
cpi->gf_bad_count = 0; // Restart count as the background is not stable enough
}
else
cpi->gf_bad_count = 0; // Gf useage has picked up so reset count
}
}
// If the signal is set but has not been read should we cancel it.
else if (last_ref_zz_useage < 15)
{
cpi->gf_update_recommended = 0;
cpi->gf_bad_count = 0;
}
#if 0
{
FILE *f = fopen("gfneeded.stt", "a");
fprintf(f, "%10d %10d %10d %10d %10ld \n",
cm->current_video_frame,
cpi->common.frames_since_golden,
gf_active_pct, gf_ref_usage_pct,
cpi->gf_update_recommended);
fclose(f);
}
#endif
}
#if !(CONFIG_REALTIME_ONLY)
static void Pass2Encode(VP8_COMP *cpi, unsigned long *size, unsigned char *dest, unsigned int *frame_flags)
{
if (!cpi->common.refresh_alt_ref_frame)
vp8_second_pass(cpi);
encode_frame_to_data_rate(cpi, size, dest, frame_flags);
cpi->twopass.bits_left -= 8 * *size;
2010-05-18 19:58:33 +04:00
if (!cpi->common.refresh_alt_ref_frame)
{
double two_pass_min_rate = (double)(cpi->oxcf.target_bandwidth
*cpi->oxcf.two_pass_vbrmin_section / 100);
cpi->twopass.bits_left += (long long)(two_pass_min_rate / cpi->oxcf.frame_rate);
}
2010-05-18 19:58:33 +04:00
}
#endif
//For ARM NEON, d8-d15 are callee-saved registers, and need to be saved by us.
#if HAVE_ARMV7
extern void vp8_push_neon(INT64 *store);
extern void vp8_pop_neon(INT64 *store);
#endif
2010-05-18 19:58:33 +04:00
int vp8_receive_raw_frame(VP8_PTR ptr, unsigned int frame_flags, YV12_BUFFER_CONFIG *sd, INT64 time_stamp, INT64 end_time)
{
#if HAVE_ARMV7
Add runtime CPU detection support for ARM. The primary goal is to allow a binary to be built which supports NEON, but can fall back to non-NEON routines, since some Android devices do not have NEON, even if they are otherwise ARMv7 (e.g., Tegra). The configure-generated flags HAVE_ARMV7, etc., are used to decide which versions of each function to build, and when CONFIG_RUNTIME_CPU_DETECT is enabled, the correct version is chosen at run time. In order for this to work, the CFLAGS must be set to something appropriate (e.g., without -mfpu=neon for ARMv7, and with appropriate -march and -mcpu for even earlier configurations), or the native C code will not be able to run. The ASFLAGS must remain set for the most advanced instruction set required at build time, since the ARM assembler will refuse to emit them otherwise. I have not attempted to make any changes to configure to do this automatically. Doing so will probably require the addition of new configure options. Many of the hooks for RTCD on ARM were already there, but a lot of the code had bit-rotted, and a good deal of the ARM-specific code is not integrated into the RTCD structs at all. I did not try to resolve the latter, merely to add the minimal amount of protection around them to allow RTCD to work. Those functions that were called based on an ifdef at the calling site were expanded to check the RTCD flags at that site, but they should be added to an RTCD struct somewhere in the future. The functions invoked with global function pointers still are, but these should be moved into an RTCD struct for thread safety (I believe every platform currently supported has atomic pointer stores, but this is not guaranteed). The encoder's boolhuff functions did not even have _c and armv7 suffixes, and the correct version was resolved at link time. The token packing functions did have appropriate suffixes, but the version was selected with a define, with no associated RTCD struct. However, for both of these, the only armv7 instruction they actually used was rbit, and this was completely superfluous, so I reworked them to avoid it. The only non-ARMv4 instruction remaining in them is clz, which is ARMv5 (not even ARMv5TE is required). Considering that there are no ARM-specific configs which are not at least ARMv5TE, I did not try to detect these at runtime, and simply enable them for ARMv5 and above. Finally, the NEON register saving code was completely non-reentrant, since it saved the registers to a global, static variable. I moved the storage for this onto the stack. A single binary built with this code was tested on an ARM11 (ARMv6) and a Cortex A8 (ARMv7 w/NEON), for both the encoder and decoder, and produced identical output, while using the correct accelerated functions on each. I did not test on any earlier processors. Change-Id: I45cbd63a614f4554c3b325c45d46c0806f009eaa
2010-10-21 02:39:11 +04:00
INT64 store_reg[8];
#endif
VP8_COMP *cpi = (VP8_COMP *) ptr;
VP8_COMMON *cm = &cpi->common;
2010-05-18 19:58:33 +04:00
struct vpx_usec_timer timer;
int res = 0;
2010-05-18 19:58:33 +04:00
#if HAVE_ARMV7
Add runtime CPU detection support for ARM. The primary goal is to allow a binary to be built which supports NEON, but can fall back to non-NEON routines, since some Android devices do not have NEON, even if they are otherwise ARMv7 (e.g., Tegra). The configure-generated flags HAVE_ARMV7, etc., are used to decide which versions of each function to build, and when CONFIG_RUNTIME_CPU_DETECT is enabled, the correct version is chosen at run time. In order for this to work, the CFLAGS must be set to something appropriate (e.g., without -mfpu=neon for ARMv7, and with appropriate -march and -mcpu for even earlier configurations), or the native C code will not be able to run. The ASFLAGS must remain set for the most advanced instruction set required at build time, since the ARM assembler will refuse to emit them otherwise. I have not attempted to make any changes to configure to do this automatically. Doing so will probably require the addition of new configure options. Many of the hooks for RTCD on ARM were already there, but a lot of the code had bit-rotted, and a good deal of the ARM-specific code is not integrated into the RTCD structs at all. I did not try to resolve the latter, merely to add the minimal amount of protection around them to allow RTCD to work. Those functions that were called based on an ifdef at the calling site were expanded to check the RTCD flags at that site, but they should be added to an RTCD struct somewhere in the future. The functions invoked with global function pointers still are, but these should be moved into an RTCD struct for thread safety (I believe every platform currently supported has atomic pointer stores, but this is not guaranteed). The encoder's boolhuff functions did not even have _c and armv7 suffixes, and the correct version was resolved at link time. The token packing functions did have appropriate suffixes, but the version was selected with a define, with no associated RTCD struct. However, for both of these, the only armv7 instruction they actually used was rbit, and this was completely superfluous, so I reworked them to avoid it. The only non-ARMv4 instruction remaining in them is clz, which is ARMv5 (not even ARMv5TE is required). Considering that there are no ARM-specific configs which are not at least ARMv5TE, I did not try to detect these at runtime, and simply enable them for ARMv5 and above. Finally, the NEON register saving code was completely non-reentrant, since it saved the registers to a global, static variable. I moved the storage for this onto the stack. A single binary built with this code was tested on an ARM11 (ARMv6) and a Cortex A8 (ARMv7 w/NEON), for both the encoder and decoder, and produced identical output, while using the correct accelerated functions on each. I did not test on any earlier processors. Change-Id: I45cbd63a614f4554c3b325c45d46c0806f009eaa
2010-10-21 02:39:11 +04:00
#if CONFIG_RUNTIME_CPU_DETECT
if (cm->rtcd.flags & HAS_NEON)
#endif
{
vp8_push_neon(store_reg);
}
2010-05-18 19:58:33 +04:00
#endif
vpx_usec_timer_start(&timer);
if(vp8_lookahead_push(cpi->lookahead, sd, time_stamp, end_time,
frame_flags))
res = -1;
2010-05-18 19:58:33 +04:00
cm->clr_type = sd->clrtype;
vpx_usec_timer_mark(&timer);
cpi->time_receive_data += vpx_usec_timer_elapsed(&timer);
#if HAVE_ARMV7
Add runtime CPU detection support for ARM. The primary goal is to allow a binary to be built which supports NEON, but can fall back to non-NEON routines, since some Android devices do not have NEON, even if they are otherwise ARMv7 (e.g., Tegra). The configure-generated flags HAVE_ARMV7, etc., are used to decide which versions of each function to build, and when CONFIG_RUNTIME_CPU_DETECT is enabled, the correct version is chosen at run time. In order for this to work, the CFLAGS must be set to something appropriate (e.g., without -mfpu=neon for ARMv7, and with appropriate -march and -mcpu for even earlier configurations), or the native C code will not be able to run. The ASFLAGS must remain set for the most advanced instruction set required at build time, since the ARM assembler will refuse to emit them otherwise. I have not attempted to make any changes to configure to do this automatically. Doing so will probably require the addition of new configure options. Many of the hooks for RTCD on ARM were already there, but a lot of the code had bit-rotted, and a good deal of the ARM-specific code is not integrated into the RTCD structs at all. I did not try to resolve the latter, merely to add the minimal amount of protection around them to allow RTCD to work. Those functions that were called based on an ifdef at the calling site were expanded to check the RTCD flags at that site, but they should be added to an RTCD struct somewhere in the future. The functions invoked with global function pointers still are, but these should be moved into an RTCD struct for thread safety (I believe every platform currently supported has atomic pointer stores, but this is not guaranteed). The encoder's boolhuff functions did not even have _c and armv7 suffixes, and the correct version was resolved at link time. The token packing functions did have appropriate suffixes, but the version was selected with a define, with no associated RTCD struct. However, for both of these, the only armv7 instruction they actually used was rbit, and this was completely superfluous, so I reworked them to avoid it. The only non-ARMv4 instruction remaining in them is clz, which is ARMv5 (not even ARMv5TE is required). Considering that there are no ARM-specific configs which are not at least ARMv5TE, I did not try to detect these at runtime, and simply enable them for ARMv5 and above. Finally, the NEON register saving code was completely non-reentrant, since it saved the registers to a global, static variable. I moved the storage for this onto the stack. A single binary built with this code was tested on an ARM11 (ARMv6) and a Cortex A8 (ARMv7 w/NEON), for both the encoder and decoder, and produced identical output, while using the correct accelerated functions on each. I did not test on any earlier processors. Change-Id: I45cbd63a614f4554c3b325c45d46c0806f009eaa
2010-10-21 02:39:11 +04:00
#if CONFIG_RUNTIME_CPU_DETECT
if (cm->rtcd.flags & HAS_NEON)
#endif
{
vp8_pop_neon(store_reg);
}
2010-05-18 19:58:33 +04:00
#endif
return res;
2010-05-18 19:58:33 +04:00
}
static int frame_is_reference(const VP8_COMP *cpi)
{
const VP8_COMMON *cm = &cpi->common;
const MACROBLOCKD *xd = &cpi->mb.e_mbd;
return cm->frame_type == KEY_FRAME || cm->refresh_last_frame
|| cm->refresh_golden_frame || cm->refresh_alt_ref_frame
|| cm->copy_buffer_to_gf || cm->copy_buffer_to_arf
|| cm->refresh_entropy_probs
|| xd->mode_ref_lf_delta_update
|| xd->update_mb_segmentation_map || xd->update_mb_segmentation_data;
}
2010-05-18 19:58:33 +04:00
int vp8_get_compressed_data(VP8_PTR ptr, unsigned int *frame_flags, unsigned long *size, unsigned char *dest, INT64 *time_stamp, INT64 *time_end, int flush)
{
#if HAVE_ARMV7
Add runtime CPU detection support for ARM. The primary goal is to allow a binary to be built which supports NEON, but can fall back to non-NEON routines, since some Android devices do not have NEON, even if they are otherwise ARMv7 (e.g., Tegra). The configure-generated flags HAVE_ARMV7, etc., are used to decide which versions of each function to build, and when CONFIG_RUNTIME_CPU_DETECT is enabled, the correct version is chosen at run time. In order for this to work, the CFLAGS must be set to something appropriate (e.g., without -mfpu=neon for ARMv7, and with appropriate -march and -mcpu for even earlier configurations), or the native C code will not be able to run. The ASFLAGS must remain set for the most advanced instruction set required at build time, since the ARM assembler will refuse to emit them otherwise. I have not attempted to make any changes to configure to do this automatically. Doing so will probably require the addition of new configure options. Many of the hooks for RTCD on ARM were already there, but a lot of the code had bit-rotted, and a good deal of the ARM-specific code is not integrated into the RTCD structs at all. I did not try to resolve the latter, merely to add the minimal amount of protection around them to allow RTCD to work. Those functions that were called based on an ifdef at the calling site were expanded to check the RTCD flags at that site, but they should be added to an RTCD struct somewhere in the future. The functions invoked with global function pointers still are, but these should be moved into an RTCD struct for thread safety (I believe every platform currently supported has atomic pointer stores, but this is not guaranteed). The encoder's boolhuff functions did not even have _c and armv7 suffixes, and the correct version was resolved at link time. The token packing functions did have appropriate suffixes, but the version was selected with a define, with no associated RTCD struct. However, for both of these, the only armv7 instruction they actually used was rbit, and this was completely superfluous, so I reworked them to avoid it. The only non-ARMv4 instruction remaining in them is clz, which is ARMv5 (not even ARMv5TE is required). Considering that there are no ARM-specific configs which are not at least ARMv5TE, I did not try to detect these at runtime, and simply enable them for ARMv5 and above. Finally, the NEON register saving code was completely non-reentrant, since it saved the registers to a global, static variable. I moved the storage for this onto the stack. A single binary built with this code was tested on an ARM11 (ARMv6) and a Cortex A8 (ARMv7 w/NEON), for both the encoder and decoder, and produced identical output, while using the correct accelerated functions on each. I did not test on any earlier processors. Change-Id: I45cbd63a614f4554c3b325c45d46c0806f009eaa
2010-10-21 02:39:11 +04:00
INT64 store_reg[8];
#endif
2010-05-18 19:58:33 +04:00
VP8_COMP *cpi = (VP8_COMP *) ptr;
VP8_COMMON *cm = &cpi->common;
struct vpx_usec_timer tsctimer;
struct vpx_usec_timer ticktimer;
struct vpx_usec_timer cmptimer;
YV12_BUFFER_CONFIG *force_src_buffer = NULL;
2010-05-18 19:58:33 +04:00
if (!cpi)
return -1;
#if HAVE_ARMV7
Add runtime CPU detection support for ARM. The primary goal is to allow a binary to be built which supports NEON, but can fall back to non-NEON routines, since some Android devices do not have NEON, even if they are otherwise ARMv7 (e.g., Tegra). The configure-generated flags HAVE_ARMV7, etc., are used to decide which versions of each function to build, and when CONFIG_RUNTIME_CPU_DETECT is enabled, the correct version is chosen at run time. In order for this to work, the CFLAGS must be set to something appropriate (e.g., without -mfpu=neon for ARMv7, and with appropriate -march and -mcpu for even earlier configurations), or the native C code will not be able to run. The ASFLAGS must remain set for the most advanced instruction set required at build time, since the ARM assembler will refuse to emit them otherwise. I have not attempted to make any changes to configure to do this automatically. Doing so will probably require the addition of new configure options. Many of the hooks for RTCD on ARM were already there, but a lot of the code had bit-rotted, and a good deal of the ARM-specific code is not integrated into the RTCD structs at all. I did not try to resolve the latter, merely to add the minimal amount of protection around them to allow RTCD to work. Those functions that were called based on an ifdef at the calling site were expanded to check the RTCD flags at that site, but they should be added to an RTCD struct somewhere in the future. The functions invoked with global function pointers still are, but these should be moved into an RTCD struct for thread safety (I believe every platform currently supported has atomic pointer stores, but this is not guaranteed). The encoder's boolhuff functions did not even have _c and armv7 suffixes, and the correct version was resolved at link time. The token packing functions did have appropriate suffixes, but the version was selected with a define, with no associated RTCD struct. However, for both of these, the only armv7 instruction they actually used was rbit, and this was completely superfluous, so I reworked them to avoid it. The only non-ARMv4 instruction remaining in them is clz, which is ARMv5 (not even ARMv5TE is required). Considering that there are no ARM-specific configs which are not at least ARMv5TE, I did not try to detect these at runtime, and simply enable them for ARMv5 and above. Finally, the NEON register saving code was completely non-reentrant, since it saved the registers to a global, static variable. I moved the storage for this onto the stack. A single binary built with this code was tested on an ARM11 (ARMv6) and a Cortex A8 (ARMv7 w/NEON), for both the encoder and decoder, and produced identical output, while using the correct accelerated functions on each. I did not test on any earlier processors. Change-Id: I45cbd63a614f4554c3b325c45d46c0806f009eaa
2010-10-21 02:39:11 +04:00
#if CONFIG_RUNTIME_CPU_DETECT
if (cm->rtcd.flags & HAS_NEON)
#endif
{
vp8_push_neon(store_reg);
}
2010-05-18 19:58:33 +04:00
#endif
vpx_usec_timer_start(&cmptimer);
cpi->source = NULL;
2010-05-18 19:58:33 +04:00
#if !(CONFIG_REALTIME_ONLY)
// Should we code an alternate reference frame
if (cpi->oxcf.error_resilient_mode == 0 &&
cpi->oxcf.play_alternate &&
cpi->source_alt_ref_pending)
{
if ((cpi->source = vp8_lookahead_peek(cpi->lookahead,
cpi->frames_till_gf_update_due)))
2010-05-18 19:58:33 +04:00
{
cpi->alt_ref_source = cpi->source;
2010-05-18 19:58:33 +04:00
if (cpi->oxcf.arnr_max_frames > 0)
{
vp8_temporal_filter_prepare_c(cpi,
cpi->frames_till_gf_update_due);
force_src_buffer = &cpi->alt_ref_buffer;
2010-05-18 19:58:33 +04:00
}
cm->frames_till_alt_ref_frame = cpi->frames_till_gf_update_due;
cm->refresh_alt_ref_frame = 1;
cm->refresh_golden_frame = 0;
cm->refresh_last_frame = 0;
cm->show_frame = 0;
cpi->source_alt_ref_pending = FALSE; // Clear Pending altf Ref flag.
cpi->is_src_frame_alt_ref = 0;
}
}
2010-05-18 19:58:33 +04:00
#endif
if (!cpi->source)
{
if ((cpi->source = vp8_lookahead_pop(cpi->lookahead, flush)))
2010-05-18 19:58:33 +04:00
{
cm->show_frame = 1;
cpi->is_src_frame_alt_ref = cpi->alt_ref_source
&& (cpi->source == cpi->alt_ref_source);
if(cpi->is_src_frame_alt_ref)
cpi->alt_ref_source = NULL;
2010-05-18 19:58:33 +04:00
}
}
2010-05-18 19:58:33 +04:00
if (cpi->source)
{
cpi->un_scaled_source =
cpi->Source = force_src_buffer ? force_src_buffer : &cpi->source->img;
*time_stamp = cpi->source->ts_start;
*time_end = cpi->source->ts_end;
*frame_flags = cpi->source->flags;
2010-05-18 19:58:33 +04:00
}
else
{
*size = 0;
#if !(CONFIG_REALTIME_ONLY)
if (flush && cpi->pass == 1 && !cpi->twopass.first_pass_done)
2010-05-18 19:58:33 +04:00
{
vp8_end_first_pass(cpi); /* get last stats packet */
cpi->twopass.first_pass_done = 1;
2010-05-18 19:58:33 +04:00
}
#endif
#if HAVE_ARMV7
Add runtime CPU detection support for ARM. The primary goal is to allow a binary to be built which supports NEON, but can fall back to non-NEON routines, since some Android devices do not have NEON, even if they are otherwise ARMv7 (e.g., Tegra). The configure-generated flags HAVE_ARMV7, etc., are used to decide which versions of each function to build, and when CONFIG_RUNTIME_CPU_DETECT is enabled, the correct version is chosen at run time. In order for this to work, the CFLAGS must be set to something appropriate (e.g., without -mfpu=neon for ARMv7, and with appropriate -march and -mcpu for even earlier configurations), or the native C code will not be able to run. The ASFLAGS must remain set for the most advanced instruction set required at build time, since the ARM assembler will refuse to emit them otherwise. I have not attempted to make any changes to configure to do this automatically. Doing so will probably require the addition of new configure options. Many of the hooks for RTCD on ARM were already there, but a lot of the code had bit-rotted, and a good deal of the ARM-specific code is not integrated into the RTCD structs at all. I did not try to resolve the latter, merely to add the minimal amount of protection around them to allow RTCD to work. Those functions that were called based on an ifdef at the calling site were expanded to check the RTCD flags at that site, but they should be added to an RTCD struct somewhere in the future. The functions invoked with global function pointers still are, but these should be moved into an RTCD struct for thread safety (I believe every platform currently supported has atomic pointer stores, but this is not guaranteed). The encoder's boolhuff functions did not even have _c and armv7 suffixes, and the correct version was resolved at link time. The token packing functions did have appropriate suffixes, but the version was selected with a define, with no associated RTCD struct. However, for both of these, the only armv7 instruction they actually used was rbit, and this was completely superfluous, so I reworked them to avoid it. The only non-ARMv4 instruction remaining in them is clz, which is ARMv5 (not even ARMv5TE is required). Considering that there are no ARM-specific configs which are not at least ARMv5TE, I did not try to detect these at runtime, and simply enable them for ARMv5 and above. Finally, the NEON register saving code was completely non-reentrant, since it saved the registers to a global, static variable. I moved the storage for this onto the stack. A single binary built with this code was tested on an ARM11 (ARMv6) and a Cortex A8 (ARMv7 w/NEON), for both the encoder and decoder, and produced identical output, while using the correct accelerated functions on each. I did not test on any earlier processors. Change-Id: I45cbd63a614f4554c3b325c45d46c0806f009eaa
2010-10-21 02:39:11 +04:00
#if CONFIG_RUNTIME_CPU_DETECT
if (cm->rtcd.flags & HAS_NEON)
#endif
{
vp8_pop_neon(store_reg);
}
2010-05-18 19:58:33 +04:00
#endif
return -1;
}
if (cpi->source->ts_start < cpi->first_time_stamp_ever)
{
cpi->first_time_stamp_ever = cpi->source->ts_start;
cpi->last_end_time_stamp_seen = cpi->source->ts_start;
}
2010-05-18 19:58:33 +04:00
// adjust frame rates based on timestamps given
if (!cm->refresh_alt_ref_frame)
{
long long this_duration;
int step = 0;
if (cpi->source->ts_start == cpi->first_time_stamp_ever)
2010-05-18 19:58:33 +04:00
{
this_duration = cpi->source->ts_end - cpi->source->ts_start;
step = 1;
2010-05-18 19:58:33 +04:00
}
else
{
long long last_duration;
this_duration = cpi->source->ts_end - cpi->last_end_time_stamp_seen;
last_duration = cpi->last_end_time_stamp_seen
- cpi->last_time_stamp_seen;
// do a step update if the duration changes by 10%
if (last_duration)
step = ((this_duration - last_duration) * 10 / last_duration);
}
2010-05-18 19:58:33 +04:00
if (this_duration)
{
if (step)
vp8_new_frame_rate(cpi, 10000000.0 / this_duration);
else
{
double avg_duration, interval;
/* Average this frame's rate into the last second's average
* frame rate. If we haven't seen 1 second yet, then average
* over the whole interval seen.
*/
interval = cpi->source->ts_end - cpi->first_time_stamp_ever;
if(interval > 10000000.0)
interval = 10000000;
2010-05-18 19:58:33 +04:00
avg_duration = 10000000.0 / cpi->oxcf.frame_rate;
avg_duration *= (interval - avg_duration + this_duration);
avg_duration /= interval;
vp8_new_frame_rate(cpi, 10000000.0 / avg_duration);
}
2010-05-18 19:58:33 +04:00
}
cpi->last_time_stamp_seen = cpi->source->ts_start;
cpi->last_end_time_stamp_seen = cpi->source->ts_end;
2010-05-18 19:58:33 +04:00
}
if (cpi->compressor_speed == 2)
{
check_gf_quality(cpi);
2010-05-18 19:58:33 +04:00
vpx_usec_timer_start(&tsctimer);
vpx_usec_timer_start(&ticktimer);
}
// start with a 0 size frame
*size = 0;
// Clear down mmx registers
vp8_clear_system_state(); //__asm emms;
cm->frame_type = INTER_FRAME;
cm->frame_flags = *frame_flags;
#if 0
if (cm->refresh_alt_ref_frame)
{
//cm->refresh_golden_frame = 1;
cm->refresh_golden_frame = 0;
cm->refresh_last_frame = 0;
}
else
{
cm->refresh_golden_frame = 0;
cm->refresh_last_frame = 1;
}
#endif
/* find a free buffer for the new frame */
{
int i = 0;
for(; i < NUM_YV12_BUFFERS; i++)
{
if(!cm->yv12_fb[i].flags)
{
cm->new_fb_idx = i;
break;
}
}
2010-05-18 19:58:33 +04:00
assert(i < NUM_YV12_BUFFERS );
}
2010-05-18 19:58:33 +04:00
#if !(CONFIG_REALTIME_ONLY)
if (cpi->pass == 1)
{
Pass1Encode(cpi, size, dest, frame_flags);
}
else if (cpi->pass == 2)
{
Pass2Encode(cpi, size, dest, frame_flags);
}
else
#endif
encode_frame_to_data_rate(cpi, size, dest, frame_flags);
if (cpi->compressor_speed == 2)
{
unsigned int duration, duration2;
vpx_usec_timer_mark(&tsctimer);
vpx_usec_timer_mark(&ticktimer);
duration = vpx_usec_timer_elapsed(&ticktimer);
duration2 = (unsigned int)((double)duration / 2);
if (cm->frame_type != KEY_FRAME)
{
if (cpi->avg_encode_time == 0)
cpi->avg_encode_time = duration;
else
cpi->avg_encode_time = (7 * cpi->avg_encode_time + duration) >> 3;
}
if (duration2)
{
//if(*frame_flags!=1)
{
if (cpi->avg_pick_mode_time == 0)
cpi->avg_pick_mode_time = duration2;
else
cpi->avg_pick_mode_time = (7 * cpi->avg_pick_mode_time + duration2) >> 3;
}
}
}
if (cm->refresh_entropy_probs == 0)
{
vpx_memcpy(&cm->fc, &cm->lfc, sizeof(cm->fc));
}
// if its a dropped frame honor the requests on subsequent frames
if (*size > 0)
{
cpi->droppable = !frame_is_reference(cpi);
2010-05-18 19:58:33 +04:00
// return to normal state
cm->refresh_entropy_probs = 1;
cm->refresh_alt_ref_frame = 0;
cm->refresh_golden_frame = 0;
cm->refresh_last_frame = 1;
cm->frame_type = INTER_FRAME;
}
vpx_usec_timer_mark(&cmptimer);
cpi->time_compress_data += vpx_usec_timer_elapsed(&cmptimer);
if (cpi->b_calculate_psnr && cpi->pass != 1 && cm->show_frame)
{
2010-05-18 19:58:33 +04:00
generate_psnr_packet(cpi);
}
2010-05-18 19:58:33 +04:00
#if CONFIG_INTERNAL_STATS
2010-05-18 19:58:33 +04:00
if (cpi->pass != 1)
{
cpi->bytes += *size;
if (cm->show_frame)
{
cpi->count ++;
if (cpi->b_calculate_psnr)
{
double ye,ue,ve;
double frame_psnr;
YV12_BUFFER_CONFIG *orig = cpi->Source;
YV12_BUFFER_CONFIG *recon = cpi->common.frame_to_show;
YV12_BUFFER_CONFIG *pp = &cm->post_proc_buffer;
int y_samples = orig->y_height * orig->y_width ;
int uv_samples = orig->uv_height * orig->uv_width ;
int t_samples = y_samples + 2 * uv_samples;
long long sq_error;
ye = calc_plane_error(orig->y_buffer, orig->y_stride,
recon->y_buffer, recon->y_stride, orig->y_width, orig->y_height,
IF_RTCD(&cpi->rtcd.variance));
ue = calc_plane_error(orig->u_buffer, orig->uv_stride,
recon->u_buffer, recon->uv_stride, orig->uv_width, orig->uv_height,
IF_RTCD(&cpi->rtcd.variance));
ve = calc_plane_error(orig->v_buffer, orig->uv_stride,
recon->v_buffer, recon->uv_stride, orig->uv_width, orig->uv_height,
IF_RTCD(&cpi->rtcd.variance));
sq_error = ye + ue + ve;
frame_psnr = vp8_mse2psnr(t_samples, 255.0, sq_error);
cpi->total_y += vp8_mse2psnr(y_samples, 255.0, ye);
cpi->total_u += vp8_mse2psnr(uv_samples, 255.0, ue);
cpi->total_v += vp8_mse2psnr(uv_samples, 255.0, ve);
2010-05-18 19:58:33 +04:00
cpi->total_sq_error += sq_error;
cpi->total += frame_psnr;
{
double frame_psnr2, frame_ssim2 = 0;
2010-05-18 19:58:33 +04:00
double weight = 0;
vp8_deblock(cm->frame_to_show, &cm->post_proc_buffer, cm->filter_level * 10 / 6, 1, 0, IF_RTCD(&cm->rtcd.postproc));
vp8_clear_system_state();
ye = calc_plane_error(orig->y_buffer, orig->y_stride,
pp->y_buffer, pp->y_stride, orig->y_width, orig->y_height,
IF_RTCD(&cpi->rtcd.variance));
ue = calc_plane_error(orig->u_buffer, orig->uv_stride,
pp->u_buffer, pp->uv_stride, orig->uv_width, orig->uv_height,
IF_RTCD(&cpi->rtcd.variance));
ve = calc_plane_error(orig->v_buffer, orig->uv_stride,
pp->v_buffer, pp->uv_stride, orig->uv_width, orig->uv_height,
IF_RTCD(&cpi->rtcd.variance));
sq_error = ye + ue + ve;
frame_psnr2 = vp8_mse2psnr(t_samples, 255.0, sq_error);
cpi->totalp_y += vp8_mse2psnr(y_samples, 255.0, ye);
cpi->totalp_u += vp8_mse2psnr(uv_samples, 255.0, ue);
cpi->totalp_v += vp8_mse2psnr(uv_samples, 255.0, ve);
cpi->total_sq_error2 += sq_error;
cpi->totalp += frame_psnr2;
frame_ssim2 = vp8_calc_ssim(cpi->Source,
&cm->post_proc_buffer, 1, &weight,
IF_RTCD(&cpi->rtcd.variance));
2010-05-18 19:58:33 +04:00
cpi->summed_quality += frame_ssim2 * weight;
cpi->summed_weights += weight;
}
}
if (cpi->b_calculate_ssimg)
{
double y, u, v, frame_all;
frame_all = vp8_calc_ssimg(cpi->Source, cm->frame_to_show,
&y, &u, &v, IF_RTCD(&cpi->rtcd.variance));
2010-05-18 19:58:33 +04:00
cpi->total_ssimg_y += y;
cpi->total_ssimg_u += u;
cpi->total_ssimg_v += v;
cpi->total_ssimg_all += frame_all;
}
}
}
#if 0
if (cpi->common.frame_type != 0 && cpi->common.base_qindex == cpi->oxcf.worst_allowed_q)
{
skiptruecount += cpi->skip_true_count;
skipfalsecount += cpi->skip_false_count;
}
#endif
#if 0
if (cpi->pass != 1)
{
FILE *f = fopen("skip.stt", "a");
fprintf(f, "frame:%4d flags:%4x Q:%4d P:%4d Size:%5d\n", cpi->common.current_video_frame, *frame_flags, cpi->common.base_qindex, cpi->prob_skip_false, *size);
if (cpi->is_src_frame_alt_ref == 1)
fprintf(f, "skipcount: %4d framesize: %d\n", cpi->skip_true_count , *size);
fclose(f);
}
#endif
#endif
#if HAVE_ARMV7
Add runtime CPU detection support for ARM. The primary goal is to allow a binary to be built which supports NEON, but can fall back to non-NEON routines, since some Android devices do not have NEON, even if they are otherwise ARMv7 (e.g., Tegra). The configure-generated flags HAVE_ARMV7, etc., are used to decide which versions of each function to build, and when CONFIG_RUNTIME_CPU_DETECT is enabled, the correct version is chosen at run time. In order for this to work, the CFLAGS must be set to something appropriate (e.g., without -mfpu=neon for ARMv7, and with appropriate -march and -mcpu for even earlier configurations), or the native C code will not be able to run. The ASFLAGS must remain set for the most advanced instruction set required at build time, since the ARM assembler will refuse to emit them otherwise. I have not attempted to make any changes to configure to do this automatically. Doing so will probably require the addition of new configure options. Many of the hooks for RTCD on ARM were already there, but a lot of the code had bit-rotted, and a good deal of the ARM-specific code is not integrated into the RTCD structs at all. I did not try to resolve the latter, merely to add the minimal amount of protection around them to allow RTCD to work. Those functions that were called based on an ifdef at the calling site were expanded to check the RTCD flags at that site, but they should be added to an RTCD struct somewhere in the future. The functions invoked with global function pointers still are, but these should be moved into an RTCD struct for thread safety (I believe every platform currently supported has atomic pointer stores, but this is not guaranteed). The encoder's boolhuff functions did not even have _c and armv7 suffixes, and the correct version was resolved at link time. The token packing functions did have appropriate suffixes, but the version was selected with a define, with no associated RTCD struct. However, for both of these, the only armv7 instruction they actually used was rbit, and this was completely superfluous, so I reworked them to avoid it. The only non-ARMv4 instruction remaining in them is clz, which is ARMv5 (not even ARMv5TE is required). Considering that there are no ARM-specific configs which are not at least ARMv5TE, I did not try to detect these at runtime, and simply enable them for ARMv5 and above. Finally, the NEON register saving code was completely non-reentrant, since it saved the registers to a global, static variable. I moved the storage for this onto the stack. A single binary built with this code was tested on an ARM11 (ARMv6) and a Cortex A8 (ARMv7 w/NEON), for both the encoder and decoder, and produced identical output, while using the correct accelerated functions on each. I did not test on any earlier processors. Change-Id: I45cbd63a614f4554c3b325c45d46c0806f009eaa
2010-10-21 02:39:11 +04:00
#if CONFIG_RUNTIME_CPU_DETECT
if (cm->rtcd.flags & HAS_NEON)
#endif
{
vp8_pop_neon(store_reg);
}
2010-05-18 19:58:33 +04:00
#endif
return 0;
}
int vp8_get_preview_raw_frame(VP8_PTR comp, YV12_BUFFER_CONFIG *dest, vp8_ppflags_t *flags)
2010-05-18 19:58:33 +04:00
{
VP8_COMP *cpi = (VP8_COMP *) comp;
if (cpi->common.refresh_alt_ref_frame)
return -1;
else
{
int ret;
#if CONFIG_POSTPROC
ret = vp8_post_proc_frame(&cpi->common, dest, flags);
2010-05-18 19:58:33 +04:00
#else
if (cpi->common.frame_to_show)
{
*dest = *cpi->common.frame_to_show;
dest->y_width = cpi->common.Width;
dest->y_height = cpi->common.Height;
dest->uv_height = cpi->common.Height / 2;
ret = 0;
}
else
{
ret = -1;
}
#endif //!CONFIG_POSTPROC
vp8_clear_system_state();
return ret;
}
}
int vp8_set_roimap(VP8_PTR comp, unsigned char *map, unsigned int rows, unsigned int cols, int delta_q[4], int delta_lf[4], unsigned int threshold[4])
{
VP8_COMP *cpi = (VP8_COMP *) comp;
signed char feature_data[MB_LVL_MAX][MAX_MB_SEGMENTS];
if (cpi->common.mb_rows != rows || cpi->common.mb_cols != cols)
return -1;
if (!map)
{
disable_segmentation((VP8_PTR)cpi);
return 0;
}
// Set the segmentation Map
set_segmentation_map((VP8_PTR)cpi, map);
// Activate segmentation.
enable_segmentation((VP8_PTR)cpi);
// Set up the quant segment data
feature_data[MB_LVL_ALT_Q][0] = delta_q[0];
feature_data[MB_LVL_ALT_Q][1] = delta_q[1];
feature_data[MB_LVL_ALT_Q][2] = delta_q[2];
feature_data[MB_LVL_ALT_Q][3] = delta_q[3];
// Set up the loop segment data s
feature_data[MB_LVL_ALT_LF][0] = delta_lf[0];
feature_data[MB_LVL_ALT_LF][1] = delta_lf[1];
feature_data[MB_LVL_ALT_LF][2] = delta_lf[2];
feature_data[MB_LVL_ALT_LF][3] = delta_lf[3];
cpi->segment_encode_breakout[0] = threshold[0];
cpi->segment_encode_breakout[1] = threshold[1];
cpi->segment_encode_breakout[2] = threshold[2];
cpi->segment_encode_breakout[3] = threshold[3];
// Initialise the feature data structure
// SEGMENT_DELTADATA 0, SEGMENT_ABSDATA 1
set_segment_data((VP8_PTR)cpi, &feature_data[0][0], SEGMENT_DELTADATA);
return 0;
}
int vp8_set_active_map(VP8_PTR comp, unsigned char *map, unsigned int rows, unsigned int cols)
{
VP8_COMP *cpi = (VP8_COMP *) comp;
if (rows == cpi->common.mb_rows && cols == cpi->common.mb_cols)
{
if (map)
{
vpx_memcpy(cpi->active_map, map, rows * cols);
cpi->active_map_enabled = 1;
}
else
cpi->active_map_enabled = 0;
return 0;
}
else
{
//cpi->active_map_enabled = 0;
return -1 ;
}
}
int vp8_set_internal_size(VP8_PTR comp, VPX_SCALING horiz_mode, VPX_SCALING vert_mode)
{
VP8_COMP *cpi = (VP8_COMP *) comp;
if (horiz_mode <= ONETWO)
2010-05-18 19:58:33 +04:00
cpi->common.horiz_scale = horiz_mode;
else
return -1;
if (vert_mode <= ONETWO)
2010-05-18 19:58:33 +04:00
cpi->common.vert_scale = vert_mode;
else
return -1;
return 0;
}
int vp8_calc_ss_err(YV12_BUFFER_CONFIG *source, YV12_BUFFER_CONFIG *dest, const vp8_variance_rtcd_vtable_t *rtcd)
{
int i, j;
int Total = 0;
unsigned char *src = source->y_buffer;
unsigned char *dst = dest->y_buffer;
(void)rtcd;
// Loop through the Y plane raw and reconstruction data summing (square differences)
for (i = 0; i < source->y_height; i += 16)
{
for (j = 0; j < source->y_width; j += 16)
{
unsigned int sse;
Total += VARIANCE_INVOKE(rtcd, mse16x16)(src + j, source->y_stride, dst + j, dest->y_stride, &sse);
}
src += 16 * source->y_stride;
dst += 16 * dest->y_stride;
}
return Total;
}
2010-05-18 19:58:33 +04:00
int vp8_get_quantizer(VP8_PTR c)
{
VP8_COMP *cpi = (VP8_COMP *) c;
return cpi->common.base_qindex;
}