An error occurred while loading the file. Please try again.
-
Ronald S. Bultje authored
This commit adds a pick_sb_mode() function which selects the best 32x32 superblock coding mode. Then it selects the best per-MB modes, compares the two and encodes that in the bitstream. The bitstream coding is rather simplistic right now. At the SB level, we code a bit to indicate whether this block uses SB-coding (32x32 prediction) or MB-coding (anything else), and then we follow with the actual modes. This could and should be modified in the future, but is omitted from this commit because it will likely involve reorganizing much more code rather than just adding SB coding, so it's better to let that be judged on its own merits. Gains on derf: about even, YT/HD: +0.75%, STD/HD: +1.5%. Change-Id: Iae313a7cbd8f75b3c66d04a68b991cb096eaaba6
5d4cffb3
variance_c.c 15.54 KiB
/* * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * * Use of this source code is governed by a BSD-style license * that can be found in the LICENSE file in the root of the source * tree. An additional intellectual property rights grant can be found * in the file PATENTS. All contributing project authors may * be found in the AUTHORS file in the root of the source tree. */#include "variance.h"#include "vp8/common/filter.h"unsigned int vp8_get_mb_ss_c( const short *src_ptr) { unsigned int i = 0, sum = 0; do { sum += (src_ptr[i] * src_ptr[i]); i++; } while (i < 256); return sum;}static void variance( const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int recon_stride, int w, int h, unsigned int *sse, int *sum) { int i, j; int diff; *sum = 0; *sse = 0; for (i = 0; i < h; i++) { for (j = 0; j < w; j++) { diff = src_ptr[j] - ref_ptr[j]; *sum += diff; *sse += diff * diff; } src_ptr += source_stride; ref_ptr += recon_stride; }}#if CONFIG_SUPERBLOCKSunsigned int vp8_variance32x32_c(const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int recon_stride, unsigned int *sse) { unsigned int var; int avg; variance(src_ptr, source_stride, ref_ptr, recon_stride, 32, 32, &var, &avg); *sse = var; return (var - ((avg * avg) >> 10));}7172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140
#endif
unsigned int vp8_variance16x16_c(
const unsigned char *src_ptr,
int source_stride,
const unsigned char *ref_ptr,
int recon_stride,
unsigned int *sse) {
unsigned int var;
int avg;
variance(src_ptr, source_stride, ref_ptr, recon_stride, 16, 16, &var, &avg);
*sse = var;
return (var - ((avg * avg) >> 8));
}
unsigned int vp8_variance8x16_c(
const unsigned char *src_ptr,
int source_stride,
const unsigned char *ref_ptr,
int recon_stride,
unsigned int *sse) {
unsigned int var;
int avg;
variance(src_ptr, source_stride, ref_ptr, recon_stride, 8, 16, &var, &avg);
*sse = var;
return (var - ((avg * avg) >> 7));
}
unsigned int vp8_variance16x8_c(
const unsigned char *src_ptr,
int source_stride,
const unsigned char *ref_ptr,
int recon_stride,
unsigned int *sse) {
unsigned int var;
int avg;
variance(src_ptr, source_stride, ref_ptr, recon_stride, 16, 8, &var, &avg);
*sse = var;
return (var - ((avg * avg) >> 7));
}
unsigned int vp8_variance8x8_c(
const unsigned char *src_ptr,
int source_stride,
const unsigned char *ref_ptr,
int recon_stride,
unsigned int *sse) {
unsigned int var;
int avg;
variance(src_ptr, source_stride, ref_ptr, recon_stride, 8, 8, &var, &avg);
*sse = var;
return (var - ((avg * avg) >> 6));
}
unsigned int vp8_variance4x4_c(
const unsigned char *src_ptr,
int source_stride,
const unsigned char *ref_ptr,
int recon_stride,
unsigned int *sse) {
unsigned int var;
141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210
int avg;
variance(src_ptr, source_stride, ref_ptr, recon_stride, 4, 4, &var, &avg);
*sse = var;
return (var - ((avg * avg) >> 4));
}
unsigned int vp8_mse16x16_c(
const unsigned char *src_ptr,
int source_stride,
const unsigned char *ref_ptr,
int recon_stride,
unsigned int *sse) {
unsigned int var;
int avg;
variance(src_ptr, source_stride, ref_ptr, recon_stride, 16, 16, &var, &avg);
*sse = var;
return var;
}
/****************************************************************************
*
* ROUTINE : filter_block2d_bil_first_pass
*
* INPUTS : UINT8 *src_ptr : Pointer to source block.
* UINT32 src_pixels_per_line : Stride of input block.
* UINT32 pixel_step : Offset between filter input samples (see notes).
* UINT32 output_height : Input block height.
* UINT32 output_width : Input block width.
* INT32 *vp8_filter : Array of 2 bi-linear filter taps.
*
* OUTPUTS : INT32 *output_ptr : Pointer to filtered block.
*
* RETURNS : void
*
* FUNCTION : Applies a 1-D 2-tap bi-linear filter to the source block in
* either horizontal or vertical direction to produce the
* filtered output block. Used to implement first-pass
* of 2-D separable filter.
*
* SPECIAL NOTES : Produces INT32 output to retain precision for next pass.
* Two filter taps should sum to VP8_FILTER_WEIGHT.
* pixel_step defines whether the filter is applied
* horizontally (pixel_step=1) or vertically (pixel_step=stride).
* It defines the offset required to move from one input
* to the next.
*
****************************************************************************/
static void var_filter_block2d_bil_first_pass
(
const unsigned char *src_ptr,
unsigned short *output_ptr,
unsigned int src_pixels_per_line,
int pixel_step,
unsigned int output_height,
unsigned int output_width,
const short *vp8_filter
) {
unsigned int i, j;
for (i = 0; i < output_height; i++) {
for (j = 0; j < output_width; j++) {
// Apply bilinear filter
output_ptr[j] = (((int)src_ptr[0] * vp8_filter[0]) +
((int)src_ptr[pixel_step] * vp8_filter[1]) +
(VP8_FILTER_WEIGHT / 2)) >> VP8_FILTER_SHIFT;
211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280
src_ptr++;
}
// Next row...
src_ptr += src_pixels_per_line - output_width;
output_ptr += output_width;
}
}
/****************************************************************************
*
* ROUTINE : filter_block2d_bil_second_pass
*
* INPUTS : INT32 *src_ptr : Pointer to source block.
* UINT32 src_pixels_per_line : Stride of input block.
* UINT32 pixel_step : Offset between filter input samples (see notes).
* UINT32 output_height : Input block height.
* UINT32 output_width : Input block width.
* INT32 *vp8_filter : Array of 2 bi-linear filter taps.
*
* OUTPUTS : UINT16 *output_ptr : Pointer to filtered block.
*
* RETURNS : void
*
* FUNCTION : Applies a 1-D 2-tap bi-linear filter to the source block in
* either horizontal or vertical direction to produce the
* filtered output block. Used to implement second-pass
* of 2-D separable filter.
*
* SPECIAL NOTES : Requires 32-bit input as produced by filter_block2d_bil_first_pass.
* Two filter taps should sum to VP8_FILTER_WEIGHT.
* pixel_step defines whether the filter is applied
* horizontally (pixel_step=1) or vertically (pixel_step=stride).
* It defines the offset required to move from one input
* to the next.
*
****************************************************************************/
static void var_filter_block2d_bil_second_pass
(
const unsigned short *src_ptr,
unsigned char *output_ptr,
unsigned int src_pixels_per_line,
unsigned int pixel_step,
unsigned int output_height,
unsigned int output_width,
const short *vp8_filter
) {
unsigned int i, j;
int Temp;
for (i = 0; i < output_height; i++) {
for (j = 0; j < output_width; j++) {
// Apply filter
Temp = ((int)src_ptr[0] * vp8_filter[0]) +
((int)src_ptr[pixel_step] * vp8_filter[1]) +
(VP8_FILTER_WEIGHT / 2);
output_ptr[j] = (unsigned int)(Temp >> VP8_FILTER_SHIFT);
src_ptr++;
}
// Next row...
src_ptr += src_pixels_per_line - output_width;
output_ptr += output_width;
}
}
unsigned int vp8_sub_pixel_variance4x4_c
(
const unsigned char *src_ptr,
281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350
int src_pixels_per_line,
int xoffset,
int yoffset,
const unsigned char *dst_ptr,
int dst_pixels_per_line,
unsigned int *sse
) {
unsigned char temp2[20 * 16];
const short *HFilter, *VFilter;
unsigned short FData3[5 * 4]; // Temp data bufffer used in filtering
HFilter = vp8_bilinear_filters[xoffset];
VFilter = vp8_bilinear_filters[yoffset];
// First filter 1d Horizontal
var_filter_block2d_bil_first_pass(src_ptr, FData3, src_pixels_per_line, 1, 5, 4, HFilter);
// Now filter Verticaly
var_filter_block2d_bil_second_pass(FData3, temp2, 4, 4, 4, 4, VFilter);
return vp8_variance4x4_c(temp2, 4, dst_ptr, dst_pixels_per_line, sse);
}
unsigned int vp8_sub_pixel_variance8x8_c
(
const unsigned char *src_ptr,
int src_pixels_per_line,
int xoffset,
int yoffset,
const unsigned char *dst_ptr,
int dst_pixels_per_line,
unsigned int *sse
) {
unsigned short FData3[9 * 8]; // Temp data bufffer used in filtering
unsigned char temp2[20 * 16];
const short *HFilter, *VFilter;
HFilter = vp8_bilinear_filters[xoffset];
VFilter = vp8_bilinear_filters[yoffset];
var_filter_block2d_bil_first_pass(src_ptr, FData3, src_pixels_per_line, 1, 9, 8, HFilter);
var_filter_block2d_bil_second_pass(FData3, temp2, 8, 8, 8, 8, VFilter);
return vp8_variance8x8_c(temp2, 8, dst_ptr, dst_pixels_per_line, sse);
}
unsigned int vp8_sub_pixel_variance16x16_c
(
const unsigned char *src_ptr,
int src_pixels_per_line,
int xoffset,
int yoffset,
const unsigned char *dst_ptr,
int dst_pixels_per_line,
unsigned int *sse
) {
unsigned short FData3[17 * 16]; // Temp data bufffer used in filtering
unsigned char temp2[20 * 16];
const short *HFilter, *VFilter;
HFilter = vp8_bilinear_filters[xoffset];
VFilter = vp8_bilinear_filters[yoffset];
var_filter_block2d_bil_first_pass(src_ptr, FData3, src_pixels_per_line, 1, 17, 16, HFilter);
var_filter_block2d_bil_second_pass(FData3, temp2, 16, 16, 16, 16, VFilter);
return vp8_variance16x16_c(temp2, 16, dst_ptr, dst_pixels_per_line, sse);
}
351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420
#if CONFIG_SUPERBLOCKS
unsigned int vp8_sub_pixel_variance32x32_c(const unsigned char *src_ptr,
int src_pixels_per_line,
int xoffset,
int yoffset,
const unsigned char *dst_ptr,
int dst_pixels_per_line,
unsigned int *sse) {
unsigned short FData3[33 * 32]; // Temp data bufffer used in filtering
unsigned char temp2[36 * 32];
const short *HFilter, *VFilter;
HFilter = vp8_bilinear_filters[xoffset];
VFilter = vp8_bilinear_filters[yoffset];
var_filter_block2d_bil_first_pass(src_ptr, FData3, src_pixels_per_line, 1, 33, 32, HFilter);
var_filter_block2d_bil_second_pass(FData3, temp2, 32, 32, 32, 32, VFilter);
return vp8_variance32x32_c(temp2, 32, dst_ptr, dst_pixels_per_line, sse);
}
#endif
unsigned int vp8_variance_halfpixvar16x16_h_c(
const unsigned char *src_ptr,
int source_stride,
const unsigned char *ref_ptr,
int recon_stride,
unsigned int *sse) {
return vp8_sub_pixel_variance16x16_c(src_ptr, source_stride, 8, 0,
ref_ptr, recon_stride, sse);
}
#if CONFIG_SUPERBLOCKS
unsigned int vp8_variance_halfpixvar32x32_h_c(const unsigned char *src_ptr,
int source_stride,
const unsigned char *ref_ptr,
int recon_stride,
unsigned int *sse) {
return vp8_sub_pixel_variance32x32_c(src_ptr, source_stride, 8, 0,
ref_ptr, recon_stride, sse);
}
#endif
unsigned int vp8_variance_halfpixvar16x16_v_c(const unsigned char *src_ptr,
int source_stride,
const unsigned char *ref_ptr,
int recon_stride,
unsigned int *sse) {
return vp8_sub_pixel_variance16x16_c(src_ptr, source_stride, 0, 8,
ref_ptr, recon_stride, sse);
}
#if CONFIG_SUPERBLOCKS
unsigned int vp8_variance_halfpixvar32x32_v_c(
const unsigned char *src_ptr,
int source_stride,
const unsigned char *ref_ptr,
int recon_stride,
unsigned int *sse) {
return vp8_sub_pixel_variance32x32_c(src_ptr, source_stride, 0, 8,
ref_ptr, recon_stride, sse);
}
#endif
unsigned int vp8_variance_halfpixvar16x16_hv_c(
const unsigned char *src_ptr,
int source_stride,
const unsigned char *ref_ptr,
int recon_stride,
421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490
unsigned int *sse) {
return vp8_sub_pixel_variance16x16_c(src_ptr, source_stride, 8, 8,
ref_ptr, recon_stride, sse);
}
#if CONFIG_SUPERBLOCKS
unsigned int vp8_variance_halfpixvar32x32_hv_c(const unsigned char *src_ptr,
int source_stride,
const unsigned char *ref_ptr,
int recon_stride,
unsigned int *sse) {
return vp8_sub_pixel_variance32x32_c(src_ptr, source_stride, 8, 8,
ref_ptr, recon_stride, sse);
}
#endif
unsigned int vp8_sub_pixel_mse16x16_c
(
const unsigned char *src_ptr,
int src_pixels_per_line,
int xoffset,
int yoffset,
const unsigned char *dst_ptr,
int dst_pixels_per_line,
unsigned int *sse
) {
vp8_sub_pixel_variance16x16_c(src_ptr, src_pixels_per_line, xoffset, yoffset, dst_ptr, dst_pixels_per_line, sse);
return *sse;
}
#if CONFIG_SUPERBLOCKS
unsigned int vp8_sub_pixel_mse32x32_c(const unsigned char *src_ptr,
int src_pixels_per_line,
int xoffset,
int yoffset,
const unsigned char *dst_ptr,
int dst_pixels_per_line,
unsigned int *sse) {
vp8_sub_pixel_variance32x32_c(src_ptr, src_pixels_per_line, xoffset, yoffset, dst_ptr, dst_pixels_per_line, sse);
return *sse;
}
#endif
unsigned int vp8_sub_pixel_variance16x8_c
(
const unsigned char *src_ptr,
int src_pixels_per_line,
int xoffset,
int yoffset,
const unsigned char *dst_ptr,
int dst_pixels_per_line,
unsigned int *sse
) {
unsigned short FData3[16 * 9]; // Temp data bufffer used in filtering
unsigned char temp2[20 * 16];
const short *HFilter, *VFilter;
HFilter = vp8_bilinear_filters[xoffset];
VFilter = vp8_bilinear_filters[yoffset];
var_filter_block2d_bil_first_pass(src_ptr, FData3, src_pixels_per_line, 1, 9, 16, HFilter);
var_filter_block2d_bil_second_pass(FData3, temp2, 16, 16, 8, 16, VFilter);
return vp8_variance16x8_c(temp2, 16, dst_ptr, dst_pixels_per_line, sse);
}
unsigned int vp8_sub_pixel_variance8x16_c
(
const unsigned char *src_ptr,
int src_pixels_per_line,
491492493494495496497498499500501502503504505506507508509510511
int xoffset,
int yoffset,
const unsigned char *dst_ptr,
int dst_pixels_per_line,
unsigned int *sse
) {
unsigned short FData3[9 * 16]; // Temp data bufffer used in filtering
unsigned char temp2[20 * 16];
const short *HFilter, *VFilter;
HFilter = vp8_bilinear_filters[xoffset];
VFilter = vp8_bilinear_filters[yoffset];
var_filter_block2d_bil_first_pass(src_ptr, FData3, src_pixels_per_line, 1, 17, 8, HFilter);
var_filter_block2d_bil_second_pass(FData3, temp2, 8, 8, 16, 8, VFilter);
return vp8_variance8x16_c(temp2, 8, dst_ptr, dst_pixels_per_line, sse);
}