generic-library/ffmpeg
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

dwt: K&R prettyprinting cosmetics

Browse Source 
Signed-off-by: Diego Biurrun <diego@biurrun.de>
This commit is contained in:
Jordi Ortiz 2012-05-19 13:06:50 +02:00 committed by Diego Biurrun
parent db9e00f469
commit 13a867a63e
2 changed files with 675 additions and 515 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

534
libavcodec/dwt.c
View File

@ -22,7 +22,9 @@
#include "dsputil.h"
#include "dwt.h"
void ff_slice_buffer_init(slice_buffer * buf, int line_count, int max_allocated_lines, int line_width, IDWTELEM * base_buffer)
void ff_slice_buffer_init(slice_buffer *buf, int line_count,
int max_allocated_lines, int line_width,
IDWTELEM *base_buffer)
{
int i;
@ -33,9 +35,8 @@ void ff_slice_buffer_init(slice_buffer * buf, int line_count, int max_allocated_
buf->line = av_mallocz(sizeof(IDWTELEM *) * line_count);
buf->data_stack = av_malloc(sizeof(IDWTELEM *) * max_allocated_lines);
for(i = 0; i < max_allocated_lines; i++){
for (i = 0; i < max_allocated_lines; i++)
buf->data_stack[i] = av_malloc(sizeof(IDWTELEM) * line_width);
}
buf->data_stack_top = max_allocated_lines - 1;
}
@ -72,37 +73,37 @@ void ff_slice_buffer_release(slice_buffer * buf, int line)
void ff_slice_buffer_flush(slice_buffer *buf)
{
int i;
for(i = 0; i < buf->line_count; i++){
for (i = 0; i < buf->line_count; i++)
if (buf->line[i])
ff_slice_buffer_release(buf, i);
}
}
void ff_slice_buffer_destroy(slice_buffer *buf)
{
int i;
ff_slice_buffer_flush(buf);
for(i = buf->data_count - 1; i >= 0; i--){
for (i = buf->data_count - 1; i >= 0; i--)
av_freep(&buf->data_stack[i]);
}
av_freep(&buf->data_stack);
av_freep(&buf->line);
}
static inline int mirror(int v, int m){
static inline int mirror(int v, int m)
{
while ((unsigned)v > (unsigned)m) {
v = -v;
if(v<0) v+= 2*m;
if (v < 0)
v += 2 * m;
}
return v;
}
static av_always_inline void
lift(DWTELEM *dst, DWTELEM *src, DWTELEM *ref,
static av_always_inline void lift(DWTELEM *dst, DWTELEM *src, DWTELEM *ref,
int dst_step, int src_step, int ref_step,
int width, int mul, int add, int shift,
int highpass, int inverse){
int highpass, int inverse)
{
const int mirror_left = !highpass;
const int mirror_right = (width & 1) ^ highpass;
const int w = (width >> 1) - 1 + (highpass & width);
@ -115,26 +116,24 @@ lift(DWTELEM *dst, DWTELEM *src, DWTELEM *ref,
src += src_step;
}
for(i=0; i<w; i++){
dst[i*dst_step] =
LIFT(src[i*src_step],
((mul*(ref[i*ref_step] + ref[(i+1)*ref_step])+add)>>shift),
for (i = 0; i < w; i++)
dst[i * dst_step] = LIFT(src[i * src_step],
((mul * (ref[i * ref_step] +
ref[(i + 1) * ref_step]) +
add) >> shift),
inverse);
}
if(mirror_right){
dst[w*dst_step] =
LIFT(src[w*src_step],
if (mirror_right)
dst[w * dst_step] = LIFT(src[w * src_step],
((mul * 2 * ref[w * ref_step] + add) >> shift),
inverse);
}
}
static av_always_inline void
inv_lift(IDWTELEM *dst, IDWTELEM *src, IDWTELEM *ref,
static av_always_inline void inv_lift(IDWTELEM *dst, IDWTELEM *src, IDWTELEM *ref,
int dst_step, int src_step, int ref_step,
int width, int mul, int add, int shift,
int highpass, int inverse){
int highpass, int inverse)
{
const int mirror_left = !highpass;
const int mirror_right = (width & 1) ^ highpass;
const int w = (width >> 1) - 1 + (highpass & width);
@ -147,27 +146,26 @@ inv_lift(IDWTELEM *dst, IDWTELEM *src, IDWTELEM *ref,
src += src_step;
}
for(i=0; i<w; i++){
dst[i*dst_step] =
LIFT(src[i*src_step],
((mul*(ref[i*ref_step] + ref[(i+1)*ref_step])+add)>>shift),
for (i = 0; i < w; i++)
dst[i * dst_step] = LIFT(src[i * src_step],
((mul * (ref[i * ref_step] +
ref[(i + 1) * ref_step]) +
add) >> shift),
inverse);
}
if (mirror_right) {
dst[w*dst_step] =
LIFT(src[w*src_step],
dst[w * dst_step] = LIFT(src[w * src_step],
((mul * 2 * ref[w * ref_step] + add) >> shift),
inverse);
}
}
#ifndef liftS
static av_always_inline void
liftS(DWTELEM *dst, DWTELEM *src, DWTELEM *ref,
static av_always_inline void liftS(DWTELEM *dst, DWTELEM *src, DWTELEM *ref,
int dst_step, int src_step, int ref_step,
int width, int mul, int add, int shift,
int highpass, int inverse){
int highpass, int inverse)
{
const int mirror_left = !highpass;
const int mirror_right = (width & 1) ^ highpass;
const int w = (width >> 1) - 1 + (highpass & width);
@ -175,32 +173,33 @@ liftS(DWTELEM *dst, DWTELEM *src, DWTELEM *ref,
assert(shift == 4);
#define LIFTS(src, ref, inv) \
((inv) ? \
(src) + (((ref) + 4*(src))>>shift): \
-((-16*(src) + (ref) + add/4 + 1 + (5<<25))/(5*4) - (1<<23)))
((inv) ? (src) + (((ref) + 4 * (src)) >> shift) \
: -((-16 * (src) + (ref) + add / \
4 + 1 + (5 << 25)) / (5 * 4) - (1 << 23)))
if (mirror_left) {
dst[0] = LIFTS(src[0], mul * 2 * ref[0] + add, inverse);
dst += dst_step;
src += src_step;
}
for(i=0; i<w; i++){
dst[i*dst_step] =
LIFTS(src[i*src_step],
mul*(ref[i*ref_step] + ref[(i+1)*ref_step])+add,
for (i = 0; i < w; i++)
dst[i * dst_step] = LIFTS(src[i * src_step],
mul * (ref[i * ref_step] +
ref[(i + 1) * ref_step]) + add,
inverse);
if (mirror_right)
dst[w * dst_step] = LIFTS(src[w * src_step],
mul * 2 * ref[w * ref_step] + add,
inverse);
}
if(mirror_right){
dst[w*dst_step] =
LIFTS(src[w*src_step], mul*2*ref[w*ref_step]+add, inverse);
}
}
static av_always_inline void
inv_liftS(IDWTELEM *dst, IDWTELEM *src, IDWTELEM *ref,
int dst_step, int src_step, int ref_step,
static av_always_inline void inv_liftS(IDWTELEM *dst, IDWTELEM *src,
IDWTELEM *ref, int dst_step,
int src_step, int ref_step,
int width, int mul, int add, int shift,
int highpass, int inverse){
int highpass, int inverse)
{
const int mirror_left = !highpass;
const int mirror_right = (width & 1) ^ highpass;
const int w = (width >> 1) - 1 + (highpass & width);
@ -208,30 +207,29 @@ inv_liftS(IDWTELEM *dst, IDWTELEM *src, IDWTELEM *ref,
assert(shift == 4);
#define LIFTS(src, ref, inv) \
((inv) ? \
(src) + (((ref) + 4*(src))>>shift): \
-((-16*(src) + (ref) + add/4 + 1 + (5<<25))/(5*4) - (1<<23)))
((inv) ? (src) + (((ref) + 4 * (src)) >> shift) \
: -((-16 * (src) + (ref) + add / \
4 + 1 + (5 << 25)) / (5 * 4) - (1 << 23)))
if (mirror_left) {
dst[0] = LIFTS(src[0], mul * 2 * ref[0] + add, inverse);
dst += dst_step;
src += src_step;
}
for(i=0; i<w; i++){
dst[i*dst_step] =
LIFTS(src[i*src_step],
mul*(ref[i*ref_step] + ref[(i+1)*ref_step])+add,
for (i = 0; i < w; i++)
dst[i * dst_step] = LIFTS(src[i * src_step],
mul * (ref[i * ref_step] +
ref[(i + 1) * ref_step]) + add,
inverse);
}
if(mirror_right){
dst[w*dst_step] =
LIFTS(src[w*src_step], mul*2*ref[w*ref_step]+add, inverse);
}
if (mirror_right)
dst[w * dst_step] = LIFTS(src[w * src_step],
mul * 2 * ref[w * ref_step] + add, inverse);
}
#endif /* ! liftS */
static void horizontal_decompose53i(DWTELEM *b, int width){
static void horizontal_decompose53i(DWTELEM *b, int width)
{
DWTELEM temp[width];
const int width2 = width >> 1;
int x;
@ -280,23 +278,27 @@ static void horizontal_decompose53i(DWTELEM *b, int width){
#endif /* 0 */
}
static void vertical_decompose53iH0(DWTELEM *b0, DWTELEM *b1, DWTELEM *b2, int width){
static void vertical_decompose53iH0(DWTELEM *b0, DWTELEM *b1, DWTELEM *b2,
int width)
{
int i;
for(i=0; i<width; i++){
for (i = 0; i < width; i++)
b1[i] -= (b0[i] + b2[i]) >> 1;
}
}
static void vertical_decompose53iL0(DWTELEM *b0, DWTELEM *b1, DWTELEM *b2, int width){
static void vertical_decompose53iL0(DWTELEM *b0, DWTELEM *b1, DWTELEM *b2,
int width)
{
int i;
for(i=0; i<width; i++){
for (i = 0; i < width; i++)
b1[i] += (b0[i] + b2[i] + 2) >> 2;
}
}
static void spatial_decompose53i(DWTELEM *buffer, int width, int height, int stride){
static void spatial_decompose53i(DWTELEM *buffer, int width, int height,
int stride)
{
int y;
DWTELEM *b0 = buffer + mirror(-2 - 1, height - 1) * stride;
DWTELEM *b1 = buffer + mirror(-2, height - 1) * stride;
@ -305,18 +307,23 @@ static void spatial_decompose53i(DWTELEM *buffer, int width, int height, int str
DWTELEM *b2 = buffer + mirror(y + 1, height - 1) * stride;
DWTELEM *b3 = buffer + mirror(y + 2, height - 1) * stride;
if(y+1<(unsigned)height) horizontal_decompose53i(b2, width);
if(y+2<(unsigned)height) horizontal_decompose53i(b3, width);
if (y + 1 < (unsigned)height)
horizontal_decompose53i(b2, width);
if (y + 2 < (unsigned)height)
horizontal_decompose53i(b3, width);
if(y+1<(unsigned)height) vertical_decompose53iH0(b1, b2, b3, width);
if(y+0<(unsigned)height) vertical_decompose53iL0(b0, b1, b2, width);
if (y + 1 < (unsigned)height)
vertical_decompose53iH0(b1, b2, b3, width);
if (y + 0 < (unsigned)height)
vertical_decompose53iL0(b0, b1, b2, width);
b0 = b2;
b1 = b3;
}
}
static void horizontal_decompose97i(DWTELEM *b, int width){
static void horizontal_decompose97i(DWTELEM *b, int width)
{
DWTELEM temp[width];
const int w2 = (width + 1) >> 1;
@ -326,44 +333,50 @@ static void horizontal_decompose97i(DWTELEM *b, int width){
lift(b, temp, b + w2, 1, 1, 1, width, W_DM, W_DO, W_DS, 0, 0);
}
static void vertical_decompose97iH0(DWTELEM *b0, DWTELEM *b1, DWTELEM *b2, int width){
static void vertical_decompose97iH0(DWTELEM *b0, DWTELEM *b1, DWTELEM *b2,
int width)
{
int i;
for(i=0; i<width; i++){
for (i = 0; i < width; i++)
b1[i] -= (W_AM * (b0[i] + b2[i]) + W_AO) >> W_AS;
}
}
static void vertical_decompose97iH1(DWTELEM *b0, DWTELEM *b1, DWTELEM *b2, int width){
static void vertical_decompose97iH1(DWTELEM *b0, DWTELEM *b1, DWTELEM *b2,
int width)
{
int i;
for(i=0; i<width; i++){
for (i = 0; i < width; i++)
b1[i] += (W_CM * (b0[i] + b2[i]) + W_CO) >> W_CS;
}
}
static void vertical_decompose97iL0(DWTELEM *b0, DWTELEM *b1, DWTELEM *b2, int width){
static void vertical_decompose97iL0(DWTELEM *b0, DWTELEM *b1, DWTELEM *b2,
int width)
{
int i;
for(i=0; i<width; i++){
for (i = 0; i < width; i++)
#ifdef liftS
b1[i] -= (W_BM * (b0[i] + b2[i]) + W_BO) >> W_BS;
#else
b1[i] = (16*4*b1[i] - 4*(b0[i] + b2[i]) + W_BO*5 + (5<<27)) / (5*16) - (1<<23);
b1[i] = (16 * 4 * b1[i] - 4 * (b0[i] + b2[i]) + W_BO * 5 + (5 << 27)) /
(5 * 16) - (1 << 23);
#endif
}
}
static void vertical_decompose97iL1(DWTELEM *b0, DWTELEM *b1, DWTELEM *b2, int width){
static void vertical_decompose97iL1(DWTELEM *b0, DWTELEM *b1, DWTELEM *b2,
int width)
{
int i;
for(i=0; i<width; i++){
for (i = 0; i < width; i++)
b1[i] += (W_DM * (b0[i] + b2[i]) + W_DO) >> W_DS;
}
}
static void spatial_decompose97i(DWTELEM *buffer, int width, int height, int stride){
static void spatial_decompose97i(DWTELEM *buffer, int width, int height,
int stride)
{
int y;
DWTELEM *b0 = buffer + mirror(-4 - 1, height - 1) * stride;
DWTELEM *b1 = buffer + mirror(-4, height - 1) * stride;
@ -374,13 +387,19 @@ static void spatial_decompose97i(DWTELEM *buffer, int width, int height, int str
DWTELEM *b4 = buffer + mirror(y + 3, height - 1) * stride;
DWTELEM *b5 = buffer + mirror(y + 4, height - 1) * stride;
if(y+3<(unsigned)height) horizontal_decompose97i(b4, width);
if(y+4<(unsigned)height) horizontal_decompose97i(b5, width);
if (y + 3 < (unsigned)height)
horizontal_decompose97i(b4, width);
if (y + 4 < (unsigned)height)
horizontal_decompose97i(b5, width);
if(y+3<(unsigned)height) vertical_decompose97iH0(b3, b4, b5, width);
if(y+2<(unsigned)height) vertical_decompose97iL0(b2, b3, b4, width);
if(y+1<(unsigned)height) vertical_decompose97iH1(b1, b2, b3, width);
if(y+0<(unsigned)height) vertical_decompose97iL1(b0, b1, b2, width);
if (y + 3 < (unsigned)height)
vertical_decompose97iH0(b3, b4, b5, width);
if (y + 2 < (unsigned)height)
vertical_decompose97iL0(b2, b3, b4, width);
if (y + 1 < (unsigned)height)
vertical_decompose97iH1(b1, b2, b3, width);
if (y + 0 < (unsigned)height)
vertical_decompose97iL1(b0, b1, b2, width);
b0 = b2;
b1 = b3;
@ -389,18 +408,29 @@ static void spatial_decompose97i(DWTELEM *buffer, int width, int height, int str
}
}
void ff_spatial_dwt(DWTELEM *buffer, int width, int height, int stride, int type, int decomposition_count){
void ff_spatial_dwt(DWTELEM *buffer, int width, int height, int stride,
int type, int decomposition_count)
{
int level;
for (level = 0; level < decomposition_count; level++) {
switch (type) {
case DWT_97: spatial_decompose97i(buffer, width>>level, height>>level, stride<<level); break;
case DWT_53: spatial_decompose53i(buffer, width>>level, height>>level, stride<<level); break;
case DWT_97:
spatial_decompose97i(buffer,
width >> level, height >> level,
stride << level);
break;
case DWT_53:
spatial_decompose53i(buffer,
width >> level, height >> level,
stride << level);
break;
}
}
}
static void horizontal_compose53i(IDWTELEM *b, int width){
static void horizontal_compose53i(IDWTELEM *b, int width)
{
IDWTELEM temp[width];
const int width2 = width >> 1;
const int w2 = (width + 1) >> 1;
@ -425,41 +455,55 @@ static void horizontal_compose53i(IDWTELEM *b, int width){
b[x - 1] = temp[x - 1] + b[x - 2];
}
static void vertical_compose53iH0(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, int width){
static void vertical_compose53iH0(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2,
int width)
{
int i;
for(i=0; i<width; i++){
for (i = 0; i < width; i++)
b1[i] += (b0[i] + b2[i]) >> 1;
}
}
static void vertical_compose53iL0(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, int width){
static void vertical_compose53iL0(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2,
int width)
{
int i;
for(i=0; i<width; i++){
for (i = 0; i < width; i++)
b1[i] -= (b0[i] + b2[i] + 2) >> 2;
}
}
static void spatial_compose53i_buffered_init(DWTCompose *cs, slice_buffer * sb, int height, int stride_line){
cs->b0 = slice_buffer_get_line(sb, mirror(-1-1, height-1) * stride_line);
static void spatial_compose53i_buffered_init(DWTCompose *cs, slice_buffer *sb,
int height, int stride_line)
{
cs->b0 = slice_buffer_get_line(sb,
mirror(-1 - 1, height - 1) * stride_line);
cs->b1 = slice_buffer_get_line(sb, mirror(-1, height - 1) * stride_line);
cs->y = -1;
}
static void spatial_compose53i_init(DWTCompose *cs, IDWTELEM *buffer, int height, int stride){
static void spatial_compose53i_init(DWTCompose *cs, IDWTELEM *buffer,
int height, int stride)
{
cs->b0 = buffer + mirror(-1 - 1, height - 1) * stride;
cs->b1 = buffer + mirror(-1, height - 1) * stride;
cs->y = -1;
}
static void spatial_compose53i_dy_buffered(DWTCompose *cs, slice_buffer * sb, int width, int height, int stride_line){
static void spatial_compose53i_dy_buffered(DWTCompose *cs, slice_buffer *sb,
int width, int height,
int stride_line)
{
int y = cs->y;
IDWTELEM *b0 = cs->b0;
IDWTELEM *b1 = cs->b1;
IDWTELEM *b2= slice_buffer_get_line(sb, mirror(y+1, height-1) * stride_line);
IDWTELEM *b3= slice_buffer_get_line(sb, mirror(y+2, height-1) * stride_line);
IDWTELEM *b2 = slice_buffer_get_line(sb,
mirror(y + 1, height - 1) *
stride_line);
IDWTELEM *b3 = slice_buffer_get_line(sb,
mirror(y + 2, height - 1) *
stride_line);
if (y + 1 < (unsigned)height && y < (unsigned)height) {
int x;
@ -469,45 +513,57 @@ static void spatial_compose53i_dy_buffered(DWTCompose *cs, slice_buffer * sb, in
b1[x] += (b0[x] + b2[x]) >> 1;
}
} else {
if(y+1<(unsigned)height) vertical_compose53iL0(b1, b2, b3, width);
if(y+0<(unsigned)height) vertical_compose53iH0(b0, b1, b2, width);
if (y + 1 < (unsigned)height)
vertical_compose53iL0(b1, b2, b3, width);
if (y + 0 < (unsigned)height)
vertical_compose53iH0(b0, b1, b2, width);
}
if(y-1<(unsigned)height) horizontal_compose53i(b0, width);
if(y+0<(unsigned)height) horizontal_compose53i(b1, width);
if (y - 1 < (unsigned)height)
horizontal_compose53i(b0, width);
if (y + 0 < (unsigned)height)
horizontal_compose53i(b1, width);
cs->b0 = b2;
cs->b1 = b3;
cs->y += 2;
}
static void spatial_compose53i_dy(DWTCompose *cs, IDWTELEM *buffer, int width, int height, int stride){
static void spatial_compose53i_dy(DWTCompose *cs, IDWTELEM *buffer, int width,
int height, int stride)
{
int y = cs->y;
IDWTELEM *b0 = cs->b0;
IDWTELEM *b1 = cs->b1;
IDWTELEM *b2 = buffer + mirror(y + 1, height - 1) * stride;
IDWTELEM *b3 = buffer + mirror(y + 2, height - 1) * stride;
if(y+1<(unsigned)height) vertical_compose53iL0(b1, b2, b3, width);
if(y+0<(unsigned)height) vertical_compose53iH0(b0, b1, b2, width);
if (y + 1 < (unsigned)height)
vertical_compose53iL0(b1, b2, b3, width);
if (y + 0 < (unsigned)height)
vertical_compose53iH0(b0, b1, b2, width);
if(y-1<(unsigned)height) horizontal_compose53i(b0, width);
if(y+0<(unsigned)height) horizontal_compose53i(b1, width);
if (y - 1 < (unsigned)height)
horizontal_compose53i(b0, width);
if (y + 0 < (unsigned)height)
horizontal_compose53i(b1, width);
cs->b0 = b2;
cs->b1 = b3;
cs->y += 2;
}
static void av_unused spatial_compose53i(IDWTELEM *buffer, int width, int height, int stride){
static void av_unused spatial_compose53i(IDWTELEM *buffer, int width,
int height, int stride)
{
DWTCompose cs;
spatial_compose53i_init(&cs, buffer, height, stride);
while (cs.y <= height)
spatial_compose53i_dy(&cs, buffer, width, height, stride);
}
void ff_snow_horizontal_compose97i(IDWTELEM *b, int width){
void ff_snow_horizontal_compose97i(IDWTELEM *b, int width)
{
IDWTELEM temp[width];
const int w2 = (width + 1) >> 1;
@ -543,43 +599,50 @@ void ff_snow_horizontal_compose97i(IDWTELEM *b, int width){
#endif
}
static void vertical_compose97iH0(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, int width){
static void vertical_compose97iH0(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2,
int width)
{
int i;
for(i=0; i<width; i++){
for (i = 0; i < width; i++)
b1[i] += (W_AM * (b0[i] + b2[i]) + W_AO) >> W_AS;
}
}
static void vertical_compose97iH1(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, int width){
static void vertical_compose97iH1(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2,
int width)
{
int i;
for(i=0; i<width; i++){
for (i = 0; i < width; i++)
b1[i] -= (W_CM * (b0[i] + b2[i]) + W_CO) >> W_CS;
}
}
static void vertical_compose97iL0(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, int width){
static void vertical_compose97iL0(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2,
int width)
{
int i;
for(i=0; i<width; i++){
for (i = 0; i < width; i++)
#ifdef liftS
b1[i] += (W_BM * (b0[i] + b2[i]) + W_BO) >> W_BS;
#else
b1[i] += (W_BM * (b0[i] + b2[i]) + 4 * b1[i] + W_BO) >> W_BS;
#endif
}
}
static void vertical_compose97iL1(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, int width){
static void vertical_compose97iL1(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2,
int width)
{
int i;
for(i=0; i<width; i++){
for (i = 0; i < width; i++)
b1[i] -= (W_DM * (b0[i] + b2[i]) + W_DO) >> W_DS;
}
}
void ff_snow_vertical_compose97i(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, IDWTELEM *b3, IDWTELEM *b4, IDWTELEM *b5, int width){
void ff_snow_vertical_compose97i(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2,
IDWTELEM *b3, IDWTELEM *b4, IDWTELEM *b5,
int width)
{
int i;
for (i = 0; i < width; i++) {
@ -594,7 +657,9 @@ void ff_snow_vertical_compose97i(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, IDWTE
}
}
static void spatial_compose97i_buffered_init(DWTCompose *cs, slice_buffer * sb, int height, int stride_line){
static void spatial_compose97i_buffered_init(DWTCompose *cs, slice_buffer *sb,
int height, int stride_line)
{
cs->b0 = slice_buffer_get_line(sb, mirror(-3 - 1, height - 1) * stride_line);
cs->b1 = slice_buffer_get_line(sb, mirror(-3, height - 1) * stride_line);
cs->b2 = slice_buffer_get_line(sb, mirror(-3 + 1, height - 1) * stride_line);
@ -602,7 +667,9 @@ static void spatial_compose97i_buffered_init(DWTCompose *cs, slice_buffer * sb,
cs->y = -3;
}
static void spatial_compose97i_init(DWTCompose *cs, IDWTELEM *buffer, int height, int stride){
static void spatial_compose97i_init(DWTCompose *cs, IDWTELEM *buffer, int height,
int stride)
{
cs->b0 = buffer + mirror(-3 - 1, height - 1) * stride;
cs->b1 = buffer + mirror(-3, height - 1) * stride;
cs->b2 = buffer + mirror(-3 + 1, height - 1) * stride;
@ -610,27 +677,40 @@ static void spatial_compose97i_init(DWTCompose *cs, IDWTELEM *buffer, int height
cs->y = -3;
}
static void spatial_compose97i_dy_buffered(DWTContext *dsp, DWTCompose *cs, slice_buffer * sb, int width, int height, int stride_line){
static void spatial_compose97i_dy_buffered(DWTContext *dsp, DWTCompose *cs,
slice_buffer *sb, int width,
int height, int stride_line)
{
int y = cs->y;
IDWTELEM *b0 = cs->b0;
IDWTELEM *b1 = cs->b1;
IDWTELEM *b2 = cs->b2;
IDWTELEM *b3 = cs->b3;
IDWTELEM *b4= slice_buffer_get_line(sb, mirror(y + 3, height - 1) * stride_line);
IDWTELEM *b5= slice_buffer_get_line(sb, mirror(y + 4, height - 1) * stride_line);
IDWTELEM *b4 = slice_buffer_get_line(sb,
mirror(y + 3, height - 1) *
stride_line);
IDWTELEM *b5 = slice_buffer_get_line(sb,
mirror(y + 4, height - 1) *
stride_line);
if (y > 0 && y + 4 < height) {
dsp->vertical_compose97i(b0, b1, b2, b3, b4, b5, width);
} else {
if(y+3<(unsigned)height) vertical_compose97iL1(b3, b4, b5, width);
if(y+2<(unsigned)height) vertical_compose97iH1(b2, b3, b4, width);
if(y+1<(unsigned)height) vertical_compose97iL0(b1, b2, b3, width);
if(y+0<(unsigned)height) vertical_compose97iH0(b0, b1, b2, width);
if (y + 3 < (unsigned)height)
vertical_compose97iL1(b3, b4, b5, width);
if (y + 2 < (unsigned)height)
vertical_compose97iH1(b2, b3, b4, width);
if (y + 1 < (unsigned)height)
vertical_compose97iL0(b1, b2, b3, width);
if (y + 0 < (unsigned)height)
vertical_compose97iH0(b0, b1, b2, width);
}
if(y-1<(unsigned)height) dsp->horizontal_compose97i(b0, width);
if(y+0<(unsigned)height) dsp->horizontal_compose97i(b1, width);
if (y - 1 < (unsigned)height)
dsp->horizontal_compose97i(b0, width);
if (y + 0 < (unsigned)height)
dsp->horizontal_compose97i(b1, width);
cs->b0 = b2;
cs->b1 = b3;
@ -639,7 +719,9 @@ static void spatial_compose97i_dy_buffered(DWTContext *dsp, DWTCompose *cs, slic
cs->y += 2;
}
static void spatial_compose97i_dy(DWTCompose *cs, IDWTELEM *buffer, int width, int height, int stride){
static void spatial_compose97i_dy(DWTCompose *cs, IDWTELEM *buffer, int width,
int height, int stride)
{
int y = cs->y;
IDWTELEM *b0 = cs->b0;
IDWTELEM *b1 = cs->b1;
@ -648,13 +730,19 @@ static void spatial_compose97i_dy(DWTCompose *cs, IDWTELEM *buffer, int width, i
IDWTELEM *b4 = buffer + mirror(y + 3, height - 1) * stride;
IDWTELEM *b5 = buffer + mirror(y + 4, height - 1) * stride;
if(y+3<(unsigned)height) vertical_compose97iL1(b3, b4, b5, width);
if(y+2<(unsigned)height) vertical_compose97iH1(b2, b3, b4, width);
if(y+1<(unsigned)height) vertical_compose97iL0(b1, b2, b3, width);
if(y+0<(unsigned)height) vertical_compose97iH0(b0, b1, b2, width);
if (y + 3 < (unsigned)height)
vertical_compose97iL1(b3, b4, b5, width);
if (y + 2 < (unsigned)height)
vertical_compose97iH1(b2, b3, b4, width);
if (y + 1 < (unsigned)height)
vertical_compose97iL0(b1, b2, b3, width);
if (y + 0 < (unsigned)height)
vertical_compose97iH0(b0, b1, b2, width);
if(y-1<(unsigned)height) ff_snow_horizontal_compose97i(b0, width);
if(y+0<(unsigned)height) ff_snow_horizontal_compose97i(b1, width);
if (y - 1 < (unsigned)height)
ff_snow_horizontal_compose97i(b0, width);
if (y + 0 < (unsigned)height)
ff_snow_horizontal_compose97i(b1, width);
cs->b0 = b2;
cs->b1 = b3;
@ -663,102 +751,146 @@ static void spatial_compose97i_dy(DWTCompose *cs, IDWTELEM *buffer, int width, i
cs->y += 2;
}
static void av_unused spatial_compose97i(IDWTELEM *buffer, int width, int height, int stride){
static void av_unused spatial_compose97i(IDWTELEM *buffer, int width,
int height, int stride)
{
DWTCompose cs;
spatial_compose97i_init(&cs, buffer, height, stride);
while (cs.y <= height)
spatial_compose97i_dy(&cs, buffer, width, height, stride);
}
void ff_spatial_idwt_buffered_init(DWTCompose *cs, slice_buffer * sb, int width, int height, int stride_line, int type, int decomposition_count){
void ff_spatial_idwt_buffered_init(DWTCompose *cs, slice_buffer *sb, int width,
int height, int stride_line, int type,
int decomposition_count)
{
int level;
for (level = decomposition_count - 1; level >= 0; level--) {
switch (type) {
case DWT_97: spatial_compose97i_buffered_init(cs+level, sb, height>>level, stride_line<<level); break;
case DWT_53: spatial_compose53i_buffered_init(cs+level, sb, height>>level, stride_line<<level); break;
case DWT_97:
spatial_compose97i_buffered_init(cs + level, sb, height >> level,
stride_line << level);
break;
case DWT_53:
spatial_compose53i_buffered_init(cs + level, sb, height >> level,
stride_line << level);
break;
}
}
}
void ff_spatial_idwt_buffered_slice(DWTContext *dsp, DWTCompose *cs, slice_buffer * slice_buf, int width, int height, int stride_line, int type, int decomposition_count, int y){
void ff_spatial_idwt_buffered_slice(DWTContext *dsp, DWTCompose *cs,
slice_buffer *slice_buf, int width,
int height, int stride_line, int type,
int decomposition_count, int y)
{
const int support = type == 1 ? 3 : 5;
int level;
if(type==2) return;
if (type == 2)
return;
for(level=decomposition_count-1; level>=0; level--){
for (level = decomposition_count - 1; level >= 0; level--)
while (cs[level].y <= FFMIN((y >> level) + support, height >> level)) {
switch (type) {
case DWT_97: spatial_compose97i_dy_buffered(dsp, cs+level, slice_buf, width>>level, height>>level, stride_line<<level);
case DWT_97:
spatial_compose97i_dy_buffered(dsp, cs + level, slice_buf,
width >> level,
height >> level,
stride_line << level);
break;
case DWT_53: spatial_compose53i_dy_buffered(cs+level, slice_buf, width>>level, height>>level, stride_line<<level);
case DWT_53:
spatial_compose53i_dy_buffered(cs + level, slice_buf,
width >> level,
height >> level,
stride_line << level);
break;
}
}
}
}
static void ff_spatial_idwt_init(DWTCompose *cs, IDWTELEM *buffer, int width, int height, int stride, int type, int decomposition_count){
static void ff_spatial_idwt_init(DWTCompose *cs, IDWTELEM *buffer, int width,
int height, int stride, int type,
int decomposition_count)
{
int level;
for (level = decomposition_count - 1; level >= 0; level--) {
switch (type) {
case DWT_97: spatial_compose97i_init(cs+level, buffer, height>>level, stride<<level); break;
case DWT_53: spatial_compose53i_init(cs+level, buffer, height>>level, stride<<level); break;
case DWT_97:
spatial_compose97i_init(cs + level, buffer, height >> level,
stride << level);
break;
case DWT_53:
spatial_compose53i_init(cs + level, buffer, height >> level,
stride << level);
break;
}
}
}
static void ff_spatial_idwt_slice(DWTCompose *cs, IDWTELEM *buffer, int width, int height, int stride, int type, int decomposition_count, int y){
static void ff_spatial_idwt_slice(DWTCompose *cs, IDWTELEM *buffer, int width,
int height, int stride, int type,
int decomposition_count, int y)
{
const int support = type == 1 ? 3 : 5;
int level;
if(type==2) return;
if (type == 2)
return;
for(level=decomposition_count-1; level>=0; level--){
for (level = decomposition_count - 1; level >= 0; level--)
while (cs[level].y <= FFMIN((y >> level) + support, height >> level)) {
switch (type) {
case DWT_97: spatial_compose97i_dy(cs+level, buffer, width>>level, height>>level, stride<<level);
case DWT_97:
spatial_compose97i_dy(cs + level, buffer, width >> level,
height >> level, stride << level);
break;
case DWT_53: spatial_compose53i_dy(cs+level, buffer, width>>level, height>>level, stride<<level);
case DWT_53:
spatial_compose53i_dy(cs + level, buffer, width >> level,
height >> level, stride << level);
break;
}
}
}
}
void ff_spatial_idwt(IDWTELEM *buffer, int width, int height, int stride, int type, int decomposition_count){
void ff_spatial_idwt(IDWTELEM *buffer, int width, int height, int stride,
int type, int decomposition_count)
{
DWTCompose cs[MAX_DECOMPOSITIONS];
int y;
ff_spatial_idwt_init(cs, buffer, width, height, stride, type, decomposition_count);
ff_spatial_idwt_init(cs, buffer, width, height, stride, type,
decomposition_count);
for (y = 0; y < height; y += 4)
ff_spatial_idwt_slice(cs, buffer, width, height, stride, type, decomposition_count, y);
ff_spatial_idwt_slice(cs, buffer, width, height, stride, type,
decomposition_count, y);
}
static inline int w_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, int w, int h, int type){
static inline int w_c(void *v, uint8_t *pix1, uint8_t *pix2, int line_size,
int w, int h, int type)
{
int s, i, j;
const int dec_count = w == 8 ? 3 : 4;
int tmp[32 * 32];
int level, ori;
static const int scale[2][2][4][4] = {
{
{
// 9/7 8x8 dec=3
{ // 9/7 8x8 dec=3
{ 268, 239, 239, 213 },
{ 0, 224, 224, 152 },
{ 0, 135, 135, 110 },
},{
// 9/7 16x16 or 32x32 dec=4
},
{ // 9/7 16x16 or 32x32 dec=4
{ 344, 310, 310, 280 },
{ 0, 320, 320, 228 },
{ 0, 175, 175, 136 },
{ 0, 129, 129, 102 },
}
},{
},
{
// 5/3 8x8 dec=3
{ // 5/3 8x8 dec=3
{ 275, 245, 245, 218 },
{ 0, 230, 230, 156 },
{ 0, 138, 138, 113 },
},{
// 5/3 16x16 or 32x32 dec=4
},
{ // 5/3 16x16 or 32x32 dec=4
{ 352, 317, 317, 286 },
{ 0, 328, 328, 233 },
{ 0, 180, 180, 140 },
@ -782,46 +914,51 @@ static inline int w_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, in
s = 0;
assert(w == h);
for(level=0; level<dec_count; level++){
for (level = 0; level < dec_count; level++)
for (ori = level ? 1 : 0; ori < 4; ori++) {
int size = w >> (dec_count - level);
int sx = (ori & 1) ? size : 0;
int stride = 32 << (dec_count - level);
int sy = (ori & 2) ? stride >> 1 : 0;
for(i=0; i<size; i++){
for (i = 0; i < size; i++)
for (j = 0; j < size; j++) {
int v= tmp[sx + sy + i*stride + j] * scale[type][dec_count-3][level][ori];
int v = tmp[sx + sy + i * stride + j] *
scale[type][dec_count - 3][level][ori];
s += FFABS(v);
}
}
}
}
assert(s >= 0);
return s >> 9;
}
static int w53_8_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, int h){
static int w53_8_c(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
{
return w_c(v, pix1, pix2, line_size, 8, h, 1);
}
static int w97_8_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, int h){
static int w97_8_c(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
{
return w_c(v, pix1, pix2, line_size, 8, h, 0);
}
static int w53_16_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, int h){
static int w53_16_c(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
{
return w_c(v, pix1, pix2, line_size, 16, h, 1);
}
static int w97_16_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, int h){
static int w97_16_c(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
{
return w_c(v, pix1, pix2, line_size, 16, h, 0);
}
int ff_w53_32_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, int h){
int ff_w53_32_c(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
{
return w_c(v, pix1, pix2, line_size, 32, h, 1);
}
int ff_w97_32_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, int h){
int ff_w97_32_c(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
{
return w_c(v, pix1, pix2, line_size, 32, h, 0);
}
@ -839,5 +976,6 @@ void ff_dwt_init(DWTContext *c)
c->horizontal_compose97i = ff_snow_horizontal_compose97i;
c->inner_add_yblock = ff_snow_inner_add_yblock;
if (HAVE_MMX) ff_dwt_init_x86(c);
if (HAVE_MMX)
ff_dwt_init_x86(c);
}

46
libavcodec/dwt.h
View File

@ -34,7 +34,8 @@ typedef struct {
int y;
} DWTCompose;
/** Used to minimize the amount of memory used in order to optimize cache performance. **/
/** Used to minimize the amount of memory used in order to
* optimize cache performance. **/
typedef struct slice_buffer_s {
IDWTELEM **line; ///< For use by idwt and predict_slices.
IDWTELEM **data_stack; ///< Used for internal purposes.
@ -46,9 +47,14 @@ typedef struct slice_buffer_s {
} slice_buffer;
typedef struct DWTContext {
void (*vertical_compose97i)(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, IDWTELEM *b3, IDWTELEM *b4, IDWTELEM *b5, int width);
void (*vertical_compose97i)(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2,
IDWTELEM *b3, IDWTELEM *b4, IDWTELEM *b5,
int width);
void (*horizontal_compose97i)(IDWTELEM *b, int width);
void (*inner_add_yblock)(const uint8_t *obmc, const int obmc_stride, uint8_t * * block, int b_w, int b_h, int src_x, int src_y, int src_stride, slice_buffer * sb, int add, uint8_t * dst8);
void (*inner_add_yblock)(const uint8_t *obmc, const int obmc_stride,
uint8_t **block, int b_w, int b_h, int src_x,
int src_y, int src_stride, slice_buffer *sb,
int add, uint8_t *dst8);
} DWTContext;
#define MAX_DECOMPOSITIONS 8
@ -126,27 +132,43 @@ typedef struct DWTContext {
#define W_DS 9
#endif
#define slice_buffer_get_line(slice_buf, line_num) ((slice_buf)->line[line_num] ? (slice_buf)->line[line_num] : ff_slice_buffer_load_line((slice_buf), (line_num)))
//#define slice_buffer_get_line(slice_buf, line_num) (ff_slice_buffer_load_line((slice_buf), (line_num)))
#define slice_buffer_get_line(slice_buf, line_num) \
((slice_buf)->line[line_num] ? (slice_buf)->line[line_num] \
: ff_slice_buffer_load_line((slice_buf), \
(line_num)))
void ff_slice_buffer_init(slice_buffer * buf, int line_count, int max_allocated_lines, int line_width, IDWTELEM * base_buffer);
void ff_slice_buffer_init(slice_buffer *buf, int line_count,
int max_allocated_lines, int line_width,
IDWTELEM *base_buffer);
void ff_slice_buffer_release(slice_buffer *buf, int line);
void ff_slice_buffer_flush(slice_buffer *buf);
void ff_slice_buffer_destroy(slice_buffer *buf);
IDWTELEM *ff_slice_buffer_load_line(slice_buffer *buf, int line);
void ff_snow_vertical_compose97i(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, IDWTELEM *b3, IDWTELEM *b4, IDWTELEM *b5, int width);
void ff_snow_vertical_compose97i(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2,
IDWTELEM *b3, IDWTELEM *b4, IDWTELEM *b5,
int width);
void ff_snow_horizontal_compose97i(IDWTELEM *b, int width);
void ff_snow_inner_add_yblock(const uint8_t *obmc, const int obmc_stride, uint8_t * * block, int b_w, int b_h, int src_x, int src_y, int src_stride, slice_buffer * sb, int add, uint8_t * dst8);
void ff_snow_inner_add_yblock(const uint8_t *obmc, const int obmc_stride,
uint8_t **block, int b_w, int b_h, int src_x,
int src_y, int src_stride, slice_buffer *sb,
int add, uint8_t *dst8);
int ff_w53_32_c(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h);
int ff_w97_32_c(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h);
void ff_spatial_dwt(int *buffer, int width, int height, int stride, int type, int decomposition_count);
void ff_spatial_dwt(int *buffer, int width, int height, int stride, int type,
int decomposition_count);
void ff_spatial_idwt_buffered_init(DWTCompose *cs, slice_buffer * sb, int width, int height, int stride_line, int type, int decomposition_count);
void ff_spatial_idwt_buffered_slice(DWTContext *dsp, DWTCompose *cs, slice_buffer * slice_buf, int width, int height, int stride_line, int type, int decomposition_count, int y);
void ff_spatial_idwt(IDWTELEM *buffer, int width, int height, int stride, int type, int decomposition_count);
void ff_spatial_idwt_buffered_init(DWTCompose *cs, slice_buffer *sb, int width,
int height, int stride_line, int type,
int decomposition_count);
void ff_spatial_idwt_buffered_slice(DWTContext *dsp, DWTCompose *cs,
slice_buffer *slice_buf, int width,
int height, int stride_line, int type,
int decomposition_count, int y);
void ff_spatial_idwt(IDWTELEM *buffer, int width, int height, int stride,
int type, int decomposition_count);
void ff_dwt_init(DWTContext *c);
void ff_dwt_init_x86(DWTContext *c);