From b2ec5c4b66b1f0e4b0f8263d721d1ddc83157da5 Mon Sep 17 00:00:00 2001 From: wm4 Date: Sun, 16 Mar 2014 10:36:06 +0100 Subject: vf_gradfun: remove internal implementation This follows the (only slowly progressing) plan to replace all internal video filters with libavfilter. All what's left in vf_gradfun.c is the weird wrapper around vf_lavfi.c. --- video/filter/vf_gradfun.c | 344 +--------------------------------------------- 1 file changed, 5 insertions(+), 339 deletions(-) diff --git a/video/filter/vf_gradfun.c b/video/filter/vf_gradfun.c index 2ca211fb74..6b98f38437 100644 --- a/video/filter/vf_gradfun.c +++ b/video/filter/vf_gradfun.c @@ -18,31 +18,14 @@ * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA. */ -/* - * Debanding algorithm (from gradfun2db by prunedtree): - * Boxblur. - * Foreach pixel, if it's within threshold of the blurred value, make it closer. - * So now we have a smoothed and higher bitdepth version of all the shallow - * gradients, while leaving detailed areas untouched. - * Dither it back to 8bit. - */ - #include #include -#include #include #include -#include - -#include "config.h" -#include "common/cpudetect.h" -#include "video/img_format.h" -#include "video/mp_image.h" #include "vf.h" -#include "video/memcpy_pic.h" -#include "compat/x86_cpu.h" +#include "common/common.h" #include "options/m_option.h" #include "vf_lavfi.h" @@ -51,13 +34,8 @@ struct vf_priv_s { float cfg_thresh; int cfg_radius; float cfg_size; - int thresh; int radius; - uint16_t *buf; - void (*filter_line)(uint8_t *dst, uint8_t *src, uint16_t *dc, - int width, int thresh, const uint16_t *dithers); - void (*blur_line)(uint16_t *dc, uint16_t *buf, uint16_t *buf1, - uint8_t *src, int sstride, int width); + struct vf_lw_opts *lw_opts; } const vf_priv_dflt = { .cfg_thresh = 1.5, @@ -65,297 +43,6 @@ struct vf_priv_s { .cfg_size = -1, }; -static const uint16_t __attribute__((aligned(16))) pw_7f[8] = {127,127,127,127,127,127,127,127}; -static const uint16_t __attribute__((aligned(16))) pw_ff[8] = {255,255,255,255,255,255,255,255}; -static const uint16_t __attribute__((aligned(16))) dither[8][8] = { - { 0, 96, 24,120, 6,102, 30,126 }, - { 64, 32, 88, 56, 70, 38, 94, 62 }, - { 16,112, 8,104, 22,118, 14,110 }, - { 80, 48, 72, 40, 86, 54, 78, 46 }, - { 4,100, 28,124, 2, 98, 26,122 }, - { 68, 36, 92, 60, 66, 34, 90, 58 }, - { 20,116, 12,108, 18,114, 10,106 }, - { 84, 52, 76, 44, 82, 50, 74, 42 }, -}; - -static void filter_line_c(uint8_t *dst, uint8_t *src, uint16_t *dc, - int width, int thresh, const uint16_t *dithers) -{ - int x; - for (x=0; x> 16; - m = FFMAX(0, 127-m); - m = m*m*delta >> 14; - pix += m + dithers[x&7]; - dst[x] = av_clip_uint8(pix>>7); - } -} - -static void blur_line_c(uint16_t *dc, uint16_t *buf, uint16_t *buf1, - uint8_t *src, int sstride, int width) -{ - int x, v, old; - for (x=0; x> 16 - "psubw %%mm6, %%mm2 \n" - "pminsw %%mm7, %%mm2 \n" // m = -max(0, 127-m) - "pmullw %%mm2, %%mm2 \n" - "paddw %%mm4, %%mm0 \n" // pix += dither - "pmulhw %%mm2, %%mm1 \n" - "psllw $2, %%mm1 \n" // m = m*m*delta >> 14 - "paddw %%mm1, %%mm0 \n" // pix += m - "psraw $7, %%mm0 \n" - "packuswb %%mm0, %%mm0 \n" - "movd %%mm0, (%1,%0) \n" // dst = clip(pix>>7) - "add $4, %0 \n" - "jl 1b \n" - "emms \n" - :"+r"(x) - :"r"(dst+width), "r"(src+width), "r"(dc+width/2), - "rm"(thresh), "m"(*dithers), "m"(*pw_7f) - :"memory" - ); -} -#endif - -#if HAVE_SSSE3 -static void filter_line_ssse3(uint8_t *dst, uint8_t *src, uint16_t *dc, - int width, int thresh, const uint16_t *dithers) -{ - intptr_t x; - if (width&7) { - // could be 10% faster if I somehow eliminated this - x = width&~7; - filter_line_c(dst+x, src+x, dc+x/2, width-x, thresh, dithers); - width = x; - } - x = -width; - __asm__ volatile( - "movd %4, %%xmm5 \n" - "pxor %%xmm7, %%xmm7 \n" - "pshuflw $0,%%xmm5, %%xmm5 \n" - "movdqa %6, %%xmm6 \n" - "punpcklqdq %%xmm5, %%xmm5 \n" - "movdqa %5, %%xmm4 \n" - "1: \n" - "movq (%2,%0), %%xmm0 \n" - "movq (%3,%0), %%xmm1 \n" - "punpcklbw %%xmm7, %%xmm0 \n" - "punpcklwd %%xmm1, %%xmm1 \n" - "psllw $7, %%xmm0 \n" - "psubw %%xmm0, %%xmm1 \n" // delta = dc - pix - "pabsw %%xmm1, %%xmm2 \n" - "pmulhuw %%xmm5, %%xmm2 \n" // m = abs(delta) * thresh >> 16 - "psubw %%xmm6, %%xmm2 \n" - "pminsw %%xmm7, %%xmm2 \n" // m = -max(0, 127-m) - "pmullw %%xmm2, %%xmm2 \n" - "psllw $1, %%xmm2 \n" - "paddw %%xmm4, %%xmm0 \n" // pix += dither - "pmulhrsw %%xmm2, %%xmm1 \n" // m = m*m*delta >> 14 - "paddw %%xmm1, %%xmm0 \n" // pix += m - "psraw $7, %%xmm0 \n" - "packuswb %%xmm0, %%xmm0 \n" - "movq %%xmm0, (%1,%0) \n" // dst = clip(pix>>7) - "add $8, %0 \n" - "jl 1b \n" - :"+&r"(x) - :"r"(dst+width), "r"(src+width), "r"(dc+width/2), - "rm"(thresh), "m"(*dithers), "m"(*pw_7f) - :"memory" - ); -} -#endif // HAVE_SSSE3 - -#if HAVE_SSE2 && HAVE_6REGS -#define BLURV(load)\ - intptr_t x = -2*width;\ - __asm__ volatile(\ - "movdqa %6, %%xmm7 \n"\ - "1: \n"\ - load" (%4,%0), %%xmm0 \n"\ - load" (%5,%0), %%xmm1 \n"\ - "movdqa %%xmm0, %%xmm2 \n"\ - "movdqa %%xmm1, %%xmm3 \n"\ - "psrlw $8, %%xmm0 \n"\ - "psrlw $8, %%xmm1 \n"\ - "pand %%xmm7, %%xmm2 \n"\ - "pand %%xmm7, %%xmm3 \n"\ - "paddw %%xmm1, %%xmm0 \n"\ - "paddw %%xmm3, %%xmm2 \n"\ - "paddw %%xmm2, %%xmm0 \n"\ - "paddw (%2,%0), %%xmm0 \n"\ - "movdqa (%1,%0), %%xmm1 \n"\ - "movdqa %%xmm0, (%1,%0) \n"\ - "psubw %%xmm1, %%xmm0 \n"\ - "movdqa %%xmm0, (%3,%0) \n"\ - "add $16, %0 \n"\ - "jl 1b \n"\ - :"+&r"(x)\ - :"r"(buf+width),\ - "r"(buf1+width),\ - "r"(dc+width),\ - "r"(src+width*2),\ - "r"(src+width*2+sstride),\ - "m"(*pw_ff)\ - :"memory"\ - ); - -static void blur_line_sse2(uint16_t *dc, uint16_t *buf, uint16_t *buf1, - uint8_t *src, int sstride, int width) -{ - if (((intptr_t)src|sstride)&15) { - BLURV("movdqu"); - } else { - BLURV("movdqa"); - } -} -#endif // HAVE_6REGS && HAVE_SSE2 - -static void filter_plane(struct vf_priv_s *ctx, uint8_t *dst, uint8_t *src, - int width, int height, int dstride, int sstride, int r) -{ - int bstride = ((width+15)&~15)/2; - int y; - uint32_t dc_factor = (1<<21)/(r*r); - uint16_t *dc = ctx->buf+16; - uint16_t *buf = ctx->buf+bstride+32; - int thresh = ctx->thresh; - - memset(dc, 0, (bstride+16)*sizeof(*buf)); - for (y=0; yblur_line(dc, buf+y*bstride, buf+(y-1)*bstride, src+2*y*sstride, sstride, width/2); - for (;;) { - if (y < height-r) { - int mod = ((y+r)/2)%r; - uint16_t *buf0 = buf+mod*bstride; - uint16_t *buf1 = buf+(mod?mod-1:r-1)*bstride; - int x, v; - ctx->blur_line(dc, buf0, buf1, src+(y+r)*sstride, sstride, width/2); - for (x=v=0; x> 16; - } - for (; x<(width+r+1)/2; x++) - dc[x-r] = v * dc_factor >> 16; - for (x=-r/2; x<0; x++) - dc[x] = dc[0]; - } - if (y == r) { - for (y=0; yfilter_line(dst+y*dstride, src+y*sstride, dc-r/2, width, thresh, dither[y&7]); - } - ctx->filter_line(dst+y*dstride, src+y*sstride, dc-r/2, width, thresh, dither[y&7]); - if (++y >= height) break; - ctx->filter_line(dst+y*dstride, src+y*sstride, dc-r/2, width, thresh, dither[y&7]); - if (++y >= height) break; - } -} - -static struct mp_image *filter(struct vf_instance *vf, struct mp_image *mpi) -{ - struct mp_image *dmpi = mpi; - if (!mp_image_is_writeable(mpi)) { - dmpi = vf_alloc_out_image(vf); - mp_image_copy_attributes(dmpi, mpi); - } - - for (int p=0; p < mpi->num_planes; p++) { - int w = mpi->w; - int h = mpi->h; - int r = vf->priv->radius; - if (p) { - w >>= mpi->chroma_x_shift; - h >>= mpi->chroma_y_shift; - r = ((r>>mpi->chroma_x_shift) + (r>>mpi->chroma_y_shift)) / 2; - r = av_clip((r+1)&~1,4,32); - } - if (FFMIN(w,h) > 2*r) - filter_plane(vf->priv, dmpi->planes[p], mpi->planes[p], w, h, - dmpi->stride[p], mpi->stride[p], r); - else if (dmpi->planes[p] != mpi->planes[p]) - memcpy_pic(dmpi->planes[p], mpi->planes[p], w, h, - dmpi->stride[p], mpi->stride[p]); - } - - if (dmpi != mpi) - talloc_free(mpi); - return dmpi; -} - -static int query_format(struct vf_instance *vf, unsigned int fmt) -{ - switch (fmt){ - case IMGFMT_Y8: - case IMGFMT_444P: - case IMGFMT_422P: - case IMGFMT_440P: - case IMGFMT_420P: - case IMGFMT_411P: - case IMGFMT_410P: - return vf_next_query_format(vf,fmt); - } - return 0; -} - -static int config(struct vf_instance *vf, - int width, int height, int d_width, int d_height, - unsigned int flags, unsigned int outfmt) -{ - free(vf->priv->buf); - vf->priv->radius = vf->priv->cfg_radius; - if (vf->priv->cfg_size > -1) { - vf->priv->radius = (vf->priv->cfg_size / 100.0f) - * sqrtf(width * width + height * height); - } - vf->priv->radius = av_clip((vf->priv->radius+1)&~1, 4, 32); - vf->priv->buf = av_mallocz((((width+15)&~15)*(vf->priv->radius+1)/2+32)*sizeof(uint16_t)); - return vf_next_config(vf,width,height,d_width,d_height,flags,outfmt); -} - -static void uninit(struct vf_instance *vf) -{ - if (!vf->priv) return; - av_free(vf->priv->buf); -} - static void lavfi_recreate(struct vf_instance *vf) { struct vf_priv_s *p = vf_lw_old_priv(vf); @@ -364,17 +51,12 @@ static void lavfi_recreate(struct vf_instance *vf) p->radius = p->cfg_radius; if (p->cfg_size > -1) p->radius = (p->cfg_size / 100.0f) * sqrtf(w * w + h * h); - p->radius = av_clip((p->radius+1)&~1, 4, 32); + p->radius = MPCLAMP((p->radius+1)&~1, 4, 32); vf_lw_update_graph(vf, "gradfun", "%f:%d", p->cfg_thresh, p->radius); } static int vf_open(vf_instance_t *vf) { - vf->filter=filter; - vf->query_format=query_format; - vf->config=config; - vf->uninit=uninit; - bool have_radius = vf->priv->cfg_radius > -1; bool have_size = vf->priv->cfg_size > -1; @@ -394,24 +76,8 @@ static int vf_open(vf_instance_t *vf) return 1; } - vf->priv->thresh = (1<<15)/av_clipf(vf->priv->cfg_thresh,0.51,255); - - vf->priv->blur_line = blur_line_c; - vf->priv->filter_line = filter_line_c; -#if HAVE_SSE2 && HAVE_6REGS - if (gCpuCaps.hasSSE2) - vf->priv->blur_line = blur_line_sse2; -#endif -#if HAVE_MMX2 - if (gCpuCaps.hasMMX2) - vf->priv->filter_line = filter_line_mmx2; -#endif -#if HAVE_SSSE3 - if (gCpuCaps.hasSSSE3) - vf->priv->filter_line = filter_line_ssse3; -#endif - - return 1; + MP_FATAL(vf, "Requires libavfilter.\n"); + return 0; } #define OPT_BASE_STRUCT struct vf_priv_s -- cgit v1.2.3