diff options
author | rfelker <rfelker@b3059339-0415-0410-9bf9-f77b7e298cf2> | 2002-08-30 06:16:40 +0000 |
---|---|---|
committer | rfelker <rfelker@b3059339-0415-0410-9bf9-f77b7e298cf2> | 2002-08-30 06:16:40 +0000 |
commit | c66e62b6a53cbcd868f9bf8a8439d9b894da2625 (patch) | |
tree | 3122d7e632af65ea48df3ffe7592fca6afc183e5 /libmpcodecs/vf_halfpack.c | |
parent | 05253c1c9b7c089f799d48aab3fab62b9be17a6f (diff) | |
download | mpv-c66e62b6a53cbcd868f9bf8a8439d9b894da2625.tar.bz2 mpv-c66e62b6a53cbcd868f9bf8a8439d9b894da2625.tar.xz |
"halfpack" (yuv planar 4:2:0 -> packed 4:2:2, half height) video filter
(useful for downsampling luma for low-res output devices without
losing chroma samples, when hardware downscaling is poor quality or
unavailable)
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@7156 b3059339-0415-0410-9bf9-f77b7e298cf2
Diffstat (limited to 'libmpcodecs/vf_halfpack.c')
-rw-r--r-- | libmpcodecs/vf_halfpack.c | 200 |
1 files changed, 200 insertions, 0 deletions
diff --git a/libmpcodecs/vf_halfpack.c b/libmpcodecs/vf_halfpack.c new file mode 100644 index 0000000000..6f60045017 --- /dev/null +++ b/libmpcodecs/vf_halfpack.c @@ -0,0 +1,200 @@ +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <inttypes.h> + +#include "../config.h" +#include "../mp_msg.h" +#include "../cpudetect.h" + +#include "img_format.h" +#include "mp_image.h" +#include "vf.h" + +#include "../libvo/fastmemcpy.h" +#include "../postproc/rgb2rgb.h" + + +#ifdef HAVE_MMX +static void halfpack_MMX(unsigned char *dst, unsigned char *src[3], + unsigned int dststride, unsigned int srcstride[3], + int w, int h) +{ + int j; + unsigned char *y1, *y2, *u, *v; + unsigned int dstinc, yinc, uinc, vinc; + + y1 = src[0]; + y2 = src[0] + srcstride[0]; + u = src[1]; + v = src[2]; + + dstinc = dststride - 2*w; + yinc = 2*srcstride[0] - w; + uinc = srcstride[1] - w/2; + vinc = srcstride[2] - w/2; + + for (h/=2; h; h--) { + asm ( + "pxor %%mm0, %%mm0 \n\t" + ".align 16 \n\t" + "1: \n\t" + "movq (%0), %%mm1 \n\t" + "movq (%0), %%mm2 \n\t" + "movq (%1), %%mm3 \n\t" + "movq (%1), %%mm4 \n\t" + "punpcklbw %%mm0, %%mm1 \n\t" + "punpckhbw %%mm0, %%mm2 \n\t" + "punpcklbw %%mm0, %%mm3 \n\t" + "punpckhbw %%mm0, %%mm4 \n\t" + "paddw %%mm3, %%mm1 \n\t" + "paddw %%mm4, %%mm2 \n\t" + "psrlw $1, %%mm1 \n\t" + "psrlw $1, %%mm2 \n\t" + + "movq (%2), %%mm3 \n\t" + "movq (%3), %%mm5 \n\t" + "punpcklbw %%mm0, %%mm3 \n\t" + "punpcklbw %%mm0, %%mm5 \n\t" + "movq %%mm3, %%mm4 \n\t" + "movq %%mm5, %%mm6 \n\t" + "punpcklwd %%mm0, %%mm3 \n\t" + "punpckhwd %%mm0, %%mm4 \n\t" + "punpcklwd %%mm0, %%mm5 \n\t" + "punpckhwd %%mm0, %%mm6 \n\t" + "pslld $8, %%mm3 \n\t" + "pslld $8, %%mm4 \n\t" + "pslld $24, %%mm5 \n\t" + "pslld $24, %%mm6 \n\t" + + "por %%mm3, %%mm1 \n\t" + "por %%mm4, %%mm2 \n\t" + "por %%mm5, %%mm1 \n\t" + "por %%mm6, %%mm2 \n\t" + + "addl $8, %0 \n\t" + "addl $8, %1 \n\t" + "addl $4, %2 \n\t" + "addl $4, %3 \n\t" + "movq %%mm1, (%8) \n\t" + "movq %%mm2, 8(%8) \n\t" + "addl $16, %8 \n\t" + "decl %9 \n\t" + "jnz 1b \n\t" + : "=r" (y1), "=r" (y2), "=r" (u), "=r" (v) + : "0" (y1), "1" (y2), "2" (u), "3" (v), "r" (dst), "r" (w/8) + : "memory" + ); + for (j = (w&7)/2; j; j--) { + *dst++ = (*y1++ + *y2++)/2; + *dst++ = *u++; + *dst++ = (*y1++ + *y2++)/2; + *dst++ = *v++; + } + y1 += yinc; + y2 += yinc; + u += uinc; + v += vinc; + dst += dstinc; + } + asm volatile ( "emms \n\t" ::: "memory" ); +} +#endif + + + +static void halfpack_C(unsigned char *dst, unsigned char *src[3], + unsigned int dststride, unsigned int srcstride[3], + int w, int h) +{ + int i, j; + unsigned char *y1, *y2, *u, *v; + unsigned int dstinc, yinc, uinc, vinc; + + y1 = src[0]; + y2 = src[0] + srcstride[0]; + u = src[1]; + v = src[2]; + + dstinc = dststride - 2*w; + yinc = 2*srcstride[0] - w; + uinc = srcstride[1] - w/2; + vinc = srcstride[2] - w/2; + + for (i = h/2; i; i--) { + for (j = w/2; j; j--) { + *dst++ = (*y1++ + *y2++)>>1; + *dst++ = *u++; + *dst++ = (*y1++ + *y2++)>>1; + *dst++ = *v++; + } + y1 += yinc; + y2 += yinc; + u += uinc; + v += vinc; + dst += dstinc; + } +} + +static void (*halfpack)(unsigned char *dst, unsigned char *src[3], + unsigned int dststride, unsigned int srcstride[3], int w, int h); + + +static void put_image(struct vf_instance_s* vf, mp_image_t *mpi) +{ + mp_image_t *dmpi; + + // hope we'll get DR buffer: + dmpi=vf_get_image(vf->next, IMGFMT_YUY2, + MP_IMGTYPE_TEMP, MP_IMGFLAG_ACCEPT_STRIDE, + mpi->w, mpi->h/2); + + halfpack(dmpi->planes[0], mpi->planes, + dmpi->stride[0], mpi->stride, + mpi->w, mpi->h); + + vf_next_put_image(vf,dmpi); +} + +static int config(struct vf_instance_s* vf, + int width, int height, int d_width, int d_height, + unsigned int flags, unsigned int outfmt) +{ + /* FIXME - also support UYVY output? */ + return vf_next_config(vf, width, height/2, d_width, d_height, flags, IMGFMT_YUY2); +} + + +static int query_format(struct vf_instance_s* vf, unsigned int fmt) +{ + /* FIXME - really any YUV 4:2:0 input format should work */ + switch (fmt) { + case IMGFMT_YV12: + case IMGFMT_IYUV: + case IMGFMT_I420: + return vf_next_query_format(vf,IMGFMT_YUY2); + } + return 0; +} + + +static int open(vf_instance_t *vf, char* args) +{ + vf->config=config; + vf->query_format=query_format; + vf->put_image=put_image; + halfpack = halfpack_C; +#ifdef HAVE_MMX + if(gCpuCaps.hasMMX) halfpack = halfpack_MMX; +#endif + return 1; +} + +vf_info_t vf_info_halfpack = { + "yuv planar 4:2:0 -> packed 4:2:2, half height", + "halfpack", + "Richard Felker", + "", + open +}; + |