-static void vertical_compose53iL0(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2,
- int width)
-{
- int i;
-
- for (i = 0; i < width; i++)
- b1[i] -= (b0[i] + b2[i] + 2) >> 2;
-}
-
-
-static av_always_inline
-void interleave(IDWTELEM *dst, IDWTELEM *src0, IDWTELEM *src1, int w2, int add, int shift)
-{
- int i;
- for (i = 0; i < w2; i++) {
- dst[2*i ] = (src0[i] + add) >> shift;
- dst[2*i+1] = (src1[i] + add) >> shift;
- }
-}
-
-static void horizontal_compose_dirac53i(IDWTELEM *b, IDWTELEM *temp, int w)
-{
- const int w2 = w >> 1;
- int x;
-
- temp[0] = COMPOSE_53iL0(b[w2], b[0], b[w2]);
- for (x = 1; x < w2; x++) {
- temp[x ] = COMPOSE_53iL0 (b[x+w2-1], b[x ], b[x+w2]);
- temp[x+w2-1] = COMPOSE_DIRAC53iH0(temp[x-1], b[x+w2-1], temp[x]);
- }
- temp[w-1] = COMPOSE_DIRAC53iH0(temp[w2-1], b[w-1], temp[w2-1]);
-
- interleave(b, temp, temp+w2, w2, 1, 1);
-}
-
-static void horizontal_compose_dd97i(IDWTELEM *b, IDWTELEM *tmp, int w)
-{
- const int w2 = w >> 1;
- int x;
-
- tmp[0] = COMPOSE_53iL0(b[w2], b[0], b[w2]);
- for (x = 1; x < w2; x++)
- tmp[x] = COMPOSE_53iL0(b[x+w2-1], b[x], b[x+w2]);
-
- // extend the edges
- tmp[-1] = tmp[0];
- tmp[w2+1] = tmp[w2] = tmp[w2-1];
-
- for (x = 0; x < w2; x++) {
- b[2*x ] = (tmp[x] + 1)>>1;
- b[2*x+1] = (COMPOSE_DD97iH0(tmp[x-1], tmp[x], b[x+w2], tmp[x+1], tmp[x+2]) + 1)>>1;
- }
-}
-
-static void horizontal_compose_dd137i(IDWTELEM *b, IDWTELEM *tmp, int w)
-{
- const int w2 = w >> 1;
- int x;
-
- tmp[0] = COMPOSE_DD137iL0(b[w2], b[w2], b[0], b[w2 ], b[w2+1]);
- tmp[1] = COMPOSE_DD137iL0(b[w2], b[w2], b[1], b[w2+1], b[w2+2]);
- for (x = 2; x < w2-1; x++)
- tmp[x] = COMPOSE_DD137iL0(b[x+w2-2], b[x+w2-1], b[x], b[x+w2], b[x+w2+1]);
- tmp[w2-1] = COMPOSE_DD137iL0(b[w-3], b[w-2], b[w2-1], b[w-1], b[w-1]);
-
- // extend the edges
- tmp[-1] = tmp[0];
- tmp[w2+1] = tmp[w2] = tmp[w2-1];
-
- for (x = 0; x < w2; x++) {
- b[2*x ] = (tmp[x] + 1)>>1;
- b[2*x+1] = (COMPOSE_DD97iH0(tmp[x-1], tmp[x], b[x+w2], tmp[x+1], tmp[x+2]) + 1)>>1;
- }
-}
-
-static av_always_inline
-void horizontal_compose_haari(IDWTELEM *b, IDWTELEM *temp, int w, int shift)
-{
- const int w2 = w >> 1;
- int x;
-
- for (x = 0; x < w2; x++) {
- temp[x ] = COMPOSE_HAARiL0(b[x ], b[x+w2]);
- temp[x+w2] = COMPOSE_HAARiH0(b[x+w2], temp[x]);
- }
-
- interleave(b, temp, temp+w2, w2, shift, shift);
-}
-
-static void horizontal_compose_haar0i(IDWTELEM *b, IDWTELEM *temp, int w)
-{
- horizontal_compose_haari(b, temp, w, 0);
-}
-
-static void horizontal_compose_haar1i(IDWTELEM *b, IDWTELEM *temp, int w)
-{
- horizontal_compose_haari(b, temp, w, 1);
-}
-
-static void horizontal_compose_fidelityi(IDWTELEM *b, IDWTELEM *tmp, int w)
-{
- const int w2 = w >> 1;
- int i, x;
- IDWTELEM v[8];
-
- for (x = 0; x < w2; x++) {
- for (i = 0; i < 8; i++)
- v[i] = b[av_clip(x-3+i, 0, w2-1)];
- tmp[x] = COMPOSE_FIDELITYiH0(v[0], v[1], v[2], v[3], b[x+w2], v[4], v[5], v[6], v[7]);
- }
-
- for (x = 0; x < w2; x++) {
- for (i = 0; i < 8; i++)
- v[i] = tmp[av_clip(x-4+i, 0, w2-1)];
- tmp[x+w2] = COMPOSE_FIDELITYiL0(v[0], v[1], v[2], v[3], b[x], v[4], v[5], v[6], v[7]);
- }
-
- interleave(b, tmp+w2, tmp, w2, 0, 0);
-}
-
-static void horizontal_compose_daub97i(IDWTELEM *b, IDWTELEM *temp, int w)
-{
- const int w2 = w >> 1;
- int x, b0, b1, b2;
-
- temp[0] = COMPOSE_DAUB97iL1(b[w2], b[0], b[w2]);
- for (x = 1; x < w2; x++) {
- temp[x ] = COMPOSE_DAUB97iL1(b[x+w2-1], b[x ], b[x+w2]);
- temp[x+w2-1] = COMPOSE_DAUB97iH1(temp[x-1], b[x+w2-1], temp[x]);
- }
- temp[w-1] = COMPOSE_DAUB97iH1(temp[w2-1], b[w-1], temp[w2-1]);
-
- // second stage combined with interleave and shift
- b0 = b2 = COMPOSE_DAUB97iL0(temp[w2], temp[0], temp[w2]);
- b[0] = (b0 + 1) >> 1;
- for (x = 1; x < w2; x++) {
- b2 = COMPOSE_DAUB97iL0(temp[x+w2-1], temp[x ], temp[x+w2]);
- b1 = COMPOSE_DAUB97iH0( b0, temp[x+w2-1], b2 );
- b[2*x-1] = (b1 + 1) >> 1;
- b[2*x ] = (b2 + 1) >> 1;
- b0 = b2;
- }
- b[w-1] = (COMPOSE_DAUB97iH0(b2, temp[w-1], b2) + 1) >> 1;
-}
-
-static void vertical_compose_dirac53iH0(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, int width)
-{
- int i;
-
- for(i=0; i<width; i++){
- b1[i] = COMPOSE_DIRAC53iH0(b0[i], b1[i], b2[i]);
- }
-}
-
-static void vertical_compose_dd97iH0(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2,
- IDWTELEM *b3, IDWTELEM *b4, int width)
-{
- int i;
-
- for(i=0; i<width; i++){
- b2[i] = COMPOSE_DD97iH0(b0[i], b1[i], b2[i], b3[i], b4[i]);
- }
-}
-
-static void vertical_compose_dd137iL0(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2,
- IDWTELEM *b3, IDWTELEM *b4, int width)
-{
- int i;
-
- for(i=0; i<width; i++){
- b2[i] = COMPOSE_DD137iL0(b0[i], b1[i], b2[i], b3[i], b4[i]);
- }
-}
-
-static void vertical_compose_haar(IDWTELEM *b0, IDWTELEM *b1, int width)
-{
- int i;
-
- for (i = 0; i < width; i++) {
- b0[i] = COMPOSE_HAARiL0(b0[i], b1[i]);
- b1[i] = COMPOSE_HAARiH0(b1[i], b0[i]);
- }
-}
-
-static void vertical_compose_fidelityiH0(IDWTELEM *dst, IDWTELEM *b[8], int width)
-{
- int i;
-
- for(i=0; i<width; i++){
- dst[i] = COMPOSE_FIDELITYiH0(b[0][i], b[1][i], b[2][i], b[3][i], dst[i], b[4][i], b[5][i], b[6][i], b[7][i]);
- }
-}
-
-static void vertical_compose_fidelityiL0(IDWTELEM *dst, IDWTELEM *b[8], int width)
-{
- int i;
-
- for(i=0; i<width; i++){
- dst[i] = COMPOSE_FIDELITYiL0(b[0][i], b[1][i], b[2][i], b[3][i], dst[i], b[4][i], b[5][i], b[6][i], b[7][i]);
- }
-}
-
-static void vertical_compose_daub97iH0(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, int width)
-{
- int i;
-
- for(i=0; i<width; i++){
- b1[i] = COMPOSE_DAUB97iH0(b0[i], b1[i], b2[i]);
- }
-}
-
-static void vertical_compose_daub97iH1(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, int width)
-{
- int i;
-
- for(i=0; i<width; i++){
- b1[i] = COMPOSE_DAUB97iH1(b0[i], b1[i], b2[i]);
- }
-}
-
-static void vertical_compose_daub97iL0(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, int width)
-{
- int i;
-
- for(i=0; i<width; i++){
- b1[i] = COMPOSE_DAUB97iL0(b0[i], b1[i], b2[i]);
- }
-}
-
-static void vertical_compose_daub97iL1(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, int width)
-{
- int i;
-
- for(i=0; i<width; i++){
- b1[i] = COMPOSE_DAUB97iL1(b0[i], b1[i], b2[i]);
- }
-}
-
-
-static void spatial_compose_dd97i_dy(DWTContext *d, int level, int width, int height, int stride)
-{
- vertical_compose_3tap vertical_compose_l0 = (void*)d->vertical_compose_l0;
- vertical_compose_5tap vertical_compose_h0 = (void*)d->vertical_compose_h0;
- DWTCompose *cs = d->cs + level;
-
- int i, y = cs->y;
- IDWTELEM *b[8];
- for (i = 0; i < 6; i++)
- b[i] = cs->b[i];
- b[6] = d->buffer + av_clip(y+5, 0, height-2)*stride;
- b[7] = d->buffer + av_clip(y+6, 1, height-1)*stride;
-
- if(y+5<(unsigned)height) vertical_compose_l0( b[5], b[6], b[7], width);
- if(y+1<(unsigned)height) vertical_compose_h0(b[0], b[2], b[3], b[4], b[6], width);
-
- if(y-1<(unsigned)height) d->horizontal_compose(b[0], d->temp, width);
- if(y+0<(unsigned)height) d->horizontal_compose(b[1], d->temp, width);
-
- for (i = 0; i < 6; i++)
- cs->b[i] = b[i+2];
- cs->y += 2;
-}
-
-static void spatial_compose_dirac53i_dy(DWTContext *d, int level, int width, int height, int stride)
-{
- vertical_compose_3tap vertical_compose_l0 = (void*)d->vertical_compose_l0;
- vertical_compose_3tap vertical_compose_h0 = (void*)d->vertical_compose_h0;
- DWTCompose *cs = d->cs + level;
-
- int y= cs->y;
- IDWTELEM *b[4] = { cs->b[0], cs->b[1] };
- b[2] = d->buffer + avpriv_mirror(y+1, height-1)*stride;
- b[3] = d->buffer + avpriv_mirror(y+2, height-1)*stride;
-
- if(y+1<(unsigned)height) vertical_compose_l0(b[1], b[2], b[3], width);
- if(y+0<(unsigned)height) vertical_compose_h0(b[0], b[1], b[2], width);
-
- if(y-1<(unsigned)height) d->horizontal_compose(b[0], d->temp, width);
- if(y+0<(unsigned)height) d->horizontal_compose(b[1], d->temp, width);
-
- cs->b[0] = b[2];
- cs->b[1] = b[3];
- cs->y += 2;
-}
-
-
-static void spatial_compose_dd137i_dy(DWTContext *d, int level, int width, int height, int stride)
-{
- vertical_compose_5tap vertical_compose_l0 = (void*)d->vertical_compose_l0;
- vertical_compose_5tap vertical_compose_h0 = (void*)d->vertical_compose_h0;
- DWTCompose *cs = d->cs + level;
-
- int i, y = cs->y;
- IDWTELEM *b[10];
- for (i = 0; i < 8; i++)
- b[i] = cs->b[i];
- b[8] = d->buffer + av_clip(y+7, 0, height-2)*stride;
- b[9] = d->buffer + av_clip(y+8, 1, height-1)*stride;
-
- if(y+5<(unsigned)height) vertical_compose_l0(b[3], b[5], b[6], b[7], b[9], width);
- if(y+1<(unsigned)height) vertical_compose_h0(b[0], b[2], b[3], b[4], b[6], width);