virtualenv/FFmpeg
1
0
Fork 0
mirror of https://github.com/FFmpeg/FFmpeg.git synced 2024-12-23 12:43:46 +02:00
Code Issues Releases Activity

avfilter/vf_convolve: use shorter variants for pointers

Browse Source 
Signed-off-by: Paul B Mahol <onemda@gmail.com>
This commit is contained in:
Paul B Mahol 2017-12-25 10:19:47 +01:00
parent ec6608f51b
commit 86e3eebfb7
1 changed files with 21 additions and 16 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

37
libavfilter/vf_convolve.c
View File

@ -266,6 +266,7 @@ static void ifft_vertical(ConvolveContext *s, int n, int plane)
static void ifft_horizontal(ConvolveContext *s, AVFrame *out, static void ifft_horizontal(ConvolveContext *s, AVFrame *out,
int w, int h, int n, int plane) int w, int h, int n, int plane)
{ {
FFTComplex *input = s->fft_hdata[plane];
const float scale = 1.f / (n * n); const float scale = 1.f / (n * n);
const int max = (1 << s->depth) - 1; const int max = (1 << s->depth) - 1;
const int hh = h / 2; const int hh = h / 2;
@ -273,51 +274,51 @@ static void ifft_horizontal(ConvolveContext *s, AVFrame *out,
int y, x; int y, x;
for (y = 0; y < n; y++) { for (y = 0; y < n; y++) {
av_fft_permute(s->ifft[plane], s->fft_hdata[plane] + y * n); av_fft_permute(s->ifft[plane], input + y * n);
av_fft_calc(s->ifft[plane], s->fft_hdata[plane] + y * n); av_fft_calc(s->ifft[plane], input + y * n);
} }
if (s->depth == 8) { if (s->depth == 8) {
for (y = 0; y < hh; y++) { for (y = 0; y < hh; y++) {
uint8_t *dst = out->data[plane] + (y + hh) * out->linesize[plane] + hw; uint8_t *dst = out->data[plane] + (y + hh) * out->linesize[plane] + hw;
for (x = 0; x < hw; x++) for (x = 0; x < hw; x++)
dst[x] = av_clip_uint8(s->fft_hdata[plane][y * n + x].re * scale); dst[x] = av_clip_uint8(input[y * n + x].re * scale);
} }
for (y = 0; y < hh; y++) { for (y = 0; y < hh; y++) {
uint8_t *dst = out->data[plane] + (y + hh) * out->linesize[plane]; uint8_t *dst = out->data[plane] + (y + hh) * out->linesize[plane];
for (x = 0; x < hw; x++) for (x = 0; x < hw; x++)
dst[x] = av_clip_uint8(s->fft_hdata[plane][y * n + n - hw + x].re * scale); dst[x] = av_clip_uint8(input[y * n + n - hw + x].re * scale);
} }
for (y = 0; y < hh; y++) { for (y = 0; y < hh; y++) {
uint8_t *dst = out->data[plane] + y * out->linesize[plane] + hw; uint8_t *dst = out->data[plane] + y * out->linesize[plane] + hw;
for (x = 0; x < hw; x++) for (x = 0; x < hw; x++)
dst[x] = av_clip_uint8(s->fft_hdata[plane][(n - hh + y) * n + x].re * scale); dst[x] = av_clip_uint8(input[(n - hh + y) * n + x].re * scale);
} }
for (y = 0; y < hh; y++) { for (y = 0; y < hh; y++) {
uint8_t *dst = out->data[plane] + y * out->linesize[plane]; uint8_t *dst = out->data[plane] + y * out->linesize[plane];
for (x = 0; x < hw; x++) for (x = 0; x < hw; x++)
dst[x] = av_clip_uint8(s->fft_hdata[plane][(n - hh + y) * n + n - hw + x].re * scale); dst[x] = av_clip_uint8(input[(n - hh + y) * n + n - hw + x].re * scale);
} }
} else { } else {
for (y = 0; y < hh; y++) { for (y = 0; y < hh; y++) {
uint16_t *dst = (uint16_t *)(out->data[plane] + (y + hh) * out->linesize[plane] + hw * 2); uint16_t *dst = (uint16_t *)(out->data[plane] + (y + hh) * out->linesize[plane] + hw * 2);
for (x = 0; x < hw; x++) for (x = 0; x < hw; x++)
dst[x] = av_clip(s->fft_hdata[plane][y * n + x].re * scale, 0, max); dst[x] = av_clip(input[y * n + x].re * scale, 0, max);
} }
for (y = 0; y < hh; y++) { for (y = 0; y < hh; y++) {
uint16_t *dst = (uint16_t *)(out->data[plane] + (y + hh) * out->linesize[plane]); uint16_t *dst = (uint16_t *)(out->data[plane] + (y + hh) * out->linesize[plane]);
for (x = 0; x < hw; x++) for (x = 0; x < hw; x++)
dst[x] = av_clip(s->fft_hdata[plane][y * n + n - hw + x].re * scale, 0, max); dst[x] = av_clip(input[y * n + n - hw + x].re * scale, 0, max);
} }
for (y = 0; y < hh; y++) { for (y = 0; y < hh; y++) {
uint16_t *dst = (uint16_t *)(out->data[plane] + y * out->linesize[plane] + hw * 2); uint16_t *dst = (uint16_t *)(out->data[plane] + y * out->linesize[plane] + hw * 2);
for (x = 0; x < hw; x++) for (x = 0; x < hw; x++)
dst[x] = av_clip(s->fft_hdata[plane][(n - hh + y) * n + x].re * scale, 0, max); dst[x] = av_clip(input[(n - hh + y) * n + x].re * scale, 0, max);
} }
for (y = 0; y < hh; y++) { for (y = 0; y < hh; y++) {
uint16_t *dst = (uint16_t *)(out->data[plane] + y * out->linesize[plane]); uint16_t *dst = (uint16_t *)(out->data[plane] + y * out->linesize[plane]);
for (x = 0; x < hw; x++) for (x = 0; x < hw; x++)
dst[x] = av_clip(s->fft_hdata[plane][(n - hh + y) * n + n - hw + x].re * scale, 0, max); dst[x] = av_clip(input[(n - hh + y) * n + n - hw + x].re * scale, 0, max);
} }
} }
} }
@ -337,6 +338,8 @@ static int do_convolve(FFFrameSync *fs)
return ff_filter_frame(outlink, mainpic); return ff_filter_frame(outlink, mainpic);
for (plane = 0; plane < s->nb_planes; plane++) { for (plane = 0; plane < s->nb_planes; plane++) {
FFTComplex *filter = s->fft_vdata_impulse[plane];
FFTComplex *input = s->fft_vdata[plane];
const int n = s->fft_len[plane]; const int n = s->fft_len[plane];
const int w = s->planewidth[plane]; const int w = s->planewidth[plane];
const int h = s->planeheight[plane]; const int h = s->planeheight[plane];
@ -376,16 +379,18 @@ static int do_convolve(FFFrameSync *fs)
} }
for (y = 0; y < n; y++) { for (y = 0; y < n; y++) {
int yn = y * n;
for (x = 0; x < n; x++) { for (x = 0; x < n; x++) {
FFTSample re, im, ire, iim; FFTSample re, im, ire, iim;
re = s->fft_vdata[plane][y*n + x].re; re = input[yn + x].re;
im = s->fft_vdata[plane][y*n + x].im; im = input[yn + x].im;
ire = s->fft_vdata_impulse[plane][y*n + x].re; ire = filter[yn + x].re;
iim = s->fft_vdata_impulse[plane][y*n + x].im; iim = filter[yn + x].im;
s->fft_vdata[plane][y*n + x].re = ire * re - iim * im; input[yn + x].re = ire * re - iim * im;
s->fft_vdata[plane][y*n + x].im = iim * re + ire * im; input[yn + x].im = iim * re + ire * im;
} }
} }