[Mplayerxp-cvslog] SF.net SVN: mplayerxp:[422] mplayerxp
Brought to you by:
olov
Menu
▾
▴
[Mplayerxp-cvslog] SF.net SVN: mplayerxp:[422] mplayerxp
|
From: <nic...@us...> - 2012-11-21 13:21:14
|
Revision: 422
http://mplayerxp.svn.sourceforge.net/mplayerxp/?rev=422&view=rev
Author: nickols_k
Date: 2012-11-21 13:21:00 +0000 (Wed, 21 Nov 2012)
Log Message:
-----------
more c++ sources + constantization + redesign query_format + redesign vosub_vidix
Modified Paths:
--------------
mplayerxp/libao2/Makefile
mplayerxp/libvo/Makefile
mplayerxp/libvo/img_format.h
mplayerxp/libvo/osd.cpp
mplayerxp/libvo/sub.h
mplayerxp/libvo/video_out.cpp
mplayerxp/libvo/video_out.h
mplayerxp/libvo/video_out_internal.h
mplayerxp/libvo/vosub_vidix.h
mplayerxp/libvo/x11_common.h
mplayerxp/postproc/vf_vo.cpp
mplayerxp/xmpcore/mp_image.h
Added Paths:
-----------
mplayerxp/libvo/aspect.cpp
mplayerxp/libvo/jpeg_enc.cpp
mplayerxp/libvo/osd_template.h
mplayerxp/libvo/vo_dga.cpp
mplayerxp/libvo/vo_fbdev.cpp
mplayerxp/libvo/vo_null.cpp
mplayerxp/libvo/vo_opengl.cpp
mplayerxp/libvo/vo_sdl.cpp
mplayerxp/libvo/vo_vesa.cpp
mplayerxp/libvo/vo_x11.cpp
mplayerxp/libvo/vo_xv.cpp
mplayerxp/libvo/vosub_vidix.cpp
mplayerxp/libvo/x11_common.cpp
Removed Paths:
-------------
mplayerxp/libvo/aspect.c
mplayerxp/libvo/jpeg_enc.c
mplayerxp/libvo/osd_template.c
mplayerxp/libvo/vo_dga.c
mplayerxp/libvo/vo_fbdev.c
mplayerxp/libvo/vo_null.c
mplayerxp/libvo/vo_opengl.c
mplayerxp/libvo/vo_sdl.c
mplayerxp/libvo/vo_vesa.c
mplayerxp/libvo/vo_x11.c
mplayerxp/libvo/vo_xv.c
mplayerxp/libvo/vosub_vidix.c
mplayerxp/libvo/x11_common.c
Modified: mplayerxp/libao2/Makefile
===================================================================
--- mplayerxp/libao2/Makefile 2012-11-21 12:54:43 UTC (rev 421)
+++ mplayerxp/libao2/Makefile 2012-11-21 13:21:00 UTC (rev 422)
@@ -60,7 +60,7 @@
dep: depend
depend:
- $(CC) -MM $(CFLAGS) $(SRCS) $(CXXSRCS) 1>.depend
+ $(CXX) -MM $(CXXFLAGS) $(SRCS) $(CXXSRCS) 1>.depend
#
# include dependency files if they exist
Modified: mplayerxp/libvo/Makefile
===================================================================
--- mplayerxp/libvo/Makefile 2012-11-21 12:54:43 UTC (rev 421)
+++ mplayerxp/libvo/Makefile 2012-11-21 13:21:00 UTC (rev 422)
@@ -3,30 +3,30 @@
LIBNAME = libvo.a
CXXSRCS=video_out.cpp font_load.cpp img_format.cpp screenshot.cpp sub.cpp osd.cpp
-SRCS=aspect.c vo_null.c
+CXXSRCS+=aspect.cpp vo_null.cpp
ifeq ($(HAVE_SDL),yes)
-SRCS+=vo_sdl.c
+CXXSRCS+=vo_sdl.cpp
endif
ifeq ($(HAVE_OPENGL),yes)
-SRCS+=vo_opengl.c
+CXXSRCS+=vo_opengl.cpp
endif
ifeq ($(HAVE_X11),yes)
-SRCS+=vo_x11.c x11_common.c
+CXXSRCS+=vo_x11.cpp x11_common.cpp
endif
ifeq ($(HAVE_XV),yes)
-SRCS+=vo_xv.c
+CXXSRCS+=vo_xv.cpp
endif
ifeq ($(HAVE_DGA),yes)
-SRCS+=vo_dga.c
+CXXSRCS+=vo_dga.cpp
endif
ifeq ($(HAVE_FBDEV),yes)
-SRCS+=vo_fbdev.c
+CXXSRCS+=vo_fbdev.cpp
endif
ifeq ($(HAVE_VESA),yes)
-SRCS += vo_vesa.c
+CXXSRCS += vo_vesa.cpp
endif
ifeq ($(CONFIG_VIDIX),yes)
-SRCS += vosub_vidix.c
+CXXSRCS += vosub_vidix.cpp
endif
CFLAGS = $(OPTFLAGS) -I. -I.. -DMPG12PLAY -Wall -W
@@ -63,7 +63,7 @@
dep: depend
depend:
- $(CC) -MM $(CFLAGS) $(SRCS) $(CXXSRCS) 1>.depend
+ $(CXX) -MM $(CXXFLAGS) $(SRCS) $(CXXSRCS) 1>.depend
#
# include dependency files if they exist
Deleted: mplayerxp/libvo/aspect.c
===================================================================
--- mplayerxp/libvo/aspect.c 2012-11-21 12:54:43 UTC (rev 421)
+++ mplayerxp/libvo/aspect.c 2012-11-21 13:21:00 UTC (rev 422)
@@ -1,90 +0,0 @@
-/* Stuff for correct aspect scaling. */
-#include "aspect.h"
-
-//#define ASPECT_DEBUG
-
-#ifdef ASPECT_DEBUG
-#include <stdio.h>
-#endif
-#include "vo_msg.h"
-
-float monitor_aspect=0;
-
-static struct {
- uint32_t orgw; // real width
- uint32_t orgh; // real height
- uint32_t prew; // prescaled width
- uint32_t preh; // prescaled height
- uint32_t scrw; // horizontal resolution
- uint32_t scrh; // vertical resolution
-} aspdat;
-
-void __FASTCALL__ aspect_save_orig(uint32_t orgw, uint32_t orgh){
- aspdat.orgw = orgw;
- aspdat.orgh = orgh;
-}
-
-void __FASTCALL__ aspect_save_prescale(uint32_t prew, uint32_t preh){
- aspdat.prew = prew;
- aspdat.preh = preh;
-}
-
-void __FASTCALL__ aspect_save_screenres(uint32_t scrw, uint32_t scrh){
- aspdat.scrw = scrw;
- aspdat.scrh = scrh;
- monitor_aspect = mp_conf.monitor_pixel_aspect * scrw / scrh;
-}
-
-/* aspect is called with the source resolution and the
- * resolution, that the scaled image should fit into
- */
-
-void __FASTCALL__ aspect(uint32_t *srcw, uint32_t *srch, int zoom){
- uint32_t tmpw;
-
-#ifdef ASPECT_DEBUG
- MSG_DBG2("aspect(0) fitin: %dx%d zoom: %d \n",aspdat.scrw,aspdat.scrh,zoom);
- MSG_DBG2("aspect(1) wh: %dx%d (org: %dx%d)\n",*srcw,*srch,aspdat.prew,aspdat.preh);
-#endif
- if(zoom){
- if(aspdat.prew >= aspdat.preh)
- {
- /* Landscape mode */
- *srcw = aspdat.scrw;
- *srch = (uint32_t)(((float)aspdat.scrw / (float)aspdat.prew * (float)aspdat.preh)
- * ((float)aspdat.scrh / ((float)aspdat.scrw / monitor_aspect)));
- }
- else
- {
- /* Portrait mode */
- *srch = aspdat.scrh;
- *srcw = (uint32_t)(((float)aspdat.scrh / (float)aspdat.preh * (float)aspdat.prew)
- * ((float)aspdat.scrw / ((float)aspdat.scrh * monitor_aspect)));
- }
- }else{
- *srcw = aspdat.prew;
- *srch = (uint32_t)((float)aspdat.preh
- * ((float)aspdat.scrh / ((float)aspdat.scrw / monitor_aspect)));
- }
- (*srch)+= (*srch)%2; // round
-#ifdef ASPECT_DEBUG
- MSG_DBG2("aspect(2) wh: %dx%d (org: %dx%d)\n",*srcw,*srch,aspdat.prew,aspdat.preh);
-#endif
- if(*srch>aspdat.scrh || *srch<aspdat.orgh){
- if(zoom)
- tmpw = (uint32_t)(((float)aspdat.scrh / (float)aspdat.preh * (float)aspdat.prew)
- * ((float)aspdat.scrw / ((float)aspdat.scrh / (1.0/monitor_aspect))));
- else
- tmpw = (uint32_t)((float)aspdat.prew
- * ((float)aspdat.scrw / ((float)aspdat.scrh / (1.0/monitor_aspect))));
- if(tmpw<=aspdat.scrw && tmpw>=aspdat.orgw){
- *srch = zoom?aspdat.scrh:aspdat.preh;
- *srcw = tmpw;
- *srcw+= *srcw%2; // round
- }
- }
-#ifdef ASPECT_DEBUG
- MSG_DBG2("aspect(3) wh: %dx%d (org: %dx%d)\n",*srcw,*srch,aspdat.prew,aspdat.preh);
-#endif
-}
-
Copied: mplayerxp/libvo/aspect.cpp (from rev 417, mplayerxp/libvo/aspect.c)
===================================================================
--- mplayerxp/libvo/aspect.cpp (rev 0)
+++ mplayerxp/libvo/aspect.cpp 2012-11-21 13:21:00 UTC (rev 422)
@@ -0,0 +1,90 @@
+/* Stuff for correct aspect scaling. */
+#include "aspect.h"
+
+//#define ASPECT_DEBUG
+
+#ifdef ASPECT_DEBUG
+#include <stdio.h>
+#endif
+#include "vo_msg.h"
+
+float monitor_aspect=0;
+
+static struct {
+ uint32_t orgw; // real width
+ uint32_t orgh; // real height
+ uint32_t prew; // prescaled width
+ uint32_t preh; // prescaled height
+ uint32_t scrw; // horizontal resolution
+ uint32_t scrh; // vertical resolution
+} aspdat;
+
+void __FASTCALL__ aspect_save_orig(uint32_t orgw, uint32_t orgh){
+ aspdat.orgw = orgw;
+ aspdat.orgh = orgh;
+}
+
+void __FASTCALL__ aspect_save_prescale(uint32_t prew, uint32_t preh){
+ aspdat.prew = prew;
+ aspdat.preh = preh;
+}
+
+void __FASTCALL__ aspect_save_screenres(uint32_t scrw, uint32_t scrh){
+ aspdat.scrw = scrw;
+ aspdat.scrh = scrh;
+ monitor_aspect = mp_conf.monitor_pixel_aspect * scrw / scrh;
+}
+
+/* aspect is called with the source resolution and the
+ * resolution, that the scaled image should fit into
+ */
+
+void __FASTCALL__ aspect(uint32_t *srcw, uint32_t *srch, int zoom){
+ uint32_t tmpw;
+
+#ifdef ASPECT_DEBUG
+ MSG_DBG2("aspect(0) fitin: %dx%d zoom: %d \n",aspdat.scrw,aspdat.scrh,zoom);
+ MSG_DBG2("aspect(1) wh: %dx%d (org: %dx%d)\n",*srcw,*srch,aspdat.prew,aspdat.preh);
+#endif
+ if(zoom){
+ if(aspdat.prew >= aspdat.preh)
+ {
+ /* Landscape mode */
+ *srcw = aspdat.scrw;
+ *srch = (uint32_t)(((float)aspdat.scrw / (float)aspdat.prew * (float)aspdat.preh)
+ * ((float)aspdat.scrh / ((float)aspdat.scrw / monitor_aspect)));
+ }
+ else
+ {
+ /* Portrait mode */
+ *srch = aspdat.scrh;
+ *srcw = (uint32_t)(((float)aspdat.scrh / (float)aspdat.preh * (float)aspdat.prew)
+ * ((float)aspdat.scrw / ((float)aspdat.scrh * monitor_aspect)));
+ }
+ }else{
+ *srcw = aspdat.prew;
+ *srch = (uint32_t)((float)aspdat.preh
+ * ((float)aspdat.scrh / ((float)aspdat.scrw / monitor_aspect)));
+ }
+ (*srch)+= (*srch)%2; // round
+#ifdef ASPECT_DEBUG
+ MSG_DBG2("aspect(2) wh: %dx%d (org: %dx%d)\n",*srcw,*srch,aspdat.prew,aspdat.preh);
+#endif
+ if(*srch>aspdat.scrh || *srch<aspdat.orgh){
+ if(zoom)
+ tmpw = (uint32_t)(((float)aspdat.scrh / (float)aspdat.preh * (float)aspdat.prew)
+ * ((float)aspdat.scrw / ((float)aspdat.scrh / (1.0/monitor_aspect))));
+ else
+ tmpw = (uint32_t)((float)aspdat.prew
+ * ((float)aspdat.scrw / ((float)aspdat.scrh / (1.0/monitor_aspect))));
+ if(tmpw<=aspdat.scrw && tmpw>=aspdat.orgw){
+ *srch = zoom?aspdat.scrh:aspdat.preh;
+ *srcw = tmpw;
+ *srcw+= *srcw%2; // round
+ }
+ }
+#ifdef ASPECT_DEBUG
+ MSG_DBG2("aspect(3) wh: %dx%d (org: %dx%d)\n",*srcw,*srch,aspdat.prew,aspdat.preh);
+#endif
+}
+
Modified: mplayerxp/libvo/img_format.h
===================================================================
--- mplayerxp/libvo/img_format.h 2012-11-21 12:54:43 UTC (rev 421)
+++ mplayerxp/libvo/img_format.h 2012-11-21 13:21:00 UTC (rev 422)
@@ -4,12 +4,8 @@
#include <inttypes.h>
#include <stdint.h>
#include "mp_config.h"
-#include "libavutil/pixfmt.h"
+#include "mp_conf_lavc.h"
-#ifdef __cplusplus
-extern "C" {
-#endif
-
/* RGB/BGR Formats */
enum {
IMGFMT_RGB_MASK =0xFFFFFF00,
@@ -161,8 +157,5 @@
enum PixelFormat pixfmt_from_fourcc(uint32_t fourcc);
uint32_t fourcc_from_pixfmt(enum PixelFormat pixfmt);
extern unsigned rgbfmt_depth(unsigned fmt);
-#ifdef __cplusplus
-}
-#endif
#endif
Deleted: mplayerxp/libvo/jpeg_enc.c
===================================================================
--- mplayerxp/libvo/jpeg_enc.c 2012-11-21 12:54:43 UTC (rev 421)
+++ mplayerxp/libvo/jpeg_enc.c 2012-11-21 13:21:00 UTC (rev 422)
@@ -1,451 +0,0 @@
-/* Straightforward (to be) optimized JPEG encoder for the YUV422 format
- * based on mjpeg code from ffmpeg.
- *
- * Copyright (c) 2002, Rik Snel
- * Parts from ffmpeg Copyright (c) 2000, 2001 Gerard Lantau
- *
- * This program is mp_free software; you can redistribute it and/or modify
- * it under the terms of the GNU General Public License as published by
- * the Free Software Foundation; either version 2 of the License, or
- * (at your option) any later version.
- *
- * This program is distributed in the hope that it will be useful,
- * but WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
- * GNU General Public License for more details.
- *
- * You should have received a copy of the GNU General Public License
- * along with this program; if not, write to the Free Software
- * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
- *
- * For an excellent introduction to the JPEG format, see:
- * http://www.ece.purdue.edu/~bouman/grad-labs/lab8/pdf/lab.pdf
- */
-
-
-
-#include <sys/types.h>
-#include <stdio.h>
-#include <stdlib.h>
-#include "mp_config.h"
-#ifdef USE_FASTMEMCPY
-#include "fastmemcpy.h"
-#endif
-/* We need this #define because we need ../libavcodec/common.h to #define
- * be2me_32, otherwise the linker will complain that it doesn't exist */
-#include "mp_conf_lavc.h"
-
-#include "jpeg_enc.h"
-#include "vo_msg.h"
-
-/* zr_mjpeg_encode_mb needs access to these tables for the black & white
- * option */
-typedef struct MJpegContext {
- UINT8 huff_size_dc_luminance[12];
- UINT16 huff_code_dc_luminance[12];
- UINT8 huff_size_dc_chrominance[12];
- UINT16 huff_code_dc_chrominance[12];
-
- UINT8 huff_size_ac_luminance[256];
- UINT16 huff_code_ac_luminance[256];
- UINT8 huff_size_ac_chrominance[256];
- UINT16 huff_code_ac_chrominance[256];
-} MJpegContext;
-
-
-/* A very important function pointer */
-extern int (*dct_quantize)(MpegEncContext *s,
- DCTELEM *block, int n, int qscale);
-
-
-/* Begin excessive code duplication ************************************/
-/* Code coming from mpegvideo.c and mjpeg.c in ../libavcodec ***********/
-
-static const unsigned short aanscales[64] = {
- /* precomputed values scaled up by 14 bits */
- 16384, 22725, 21407, 19266, 16384, 12873, 8867, 4520,
- 22725, 31521, 29692, 26722, 22725, 17855, 12299, 6270,
- 21407, 29692, 27969, 25172, 21407, 16819, 11585, 5906,
- 19266, 26722, 25172, 22654, 19266, 15137, 10426, 5315,
- 16384, 22725, 21407, 19266, 16384, 12873, 8867, 4520,
- 12873, 17855, 16819, 15137, 12873, 10114, 6967, 3552,
- 8867, 12299, 11585, 10426, 8867, 6967, 4799, 2446,
- 4520, 6270, 5906, 5315, 4520, 3552, 2446, 1247
-};
-
-static void convert_matrix(int *qmat, UINT16 *qmat16, const UINT16 *quant_matrix, int qscale)
-{
- int i;
-
- if (av_fdct == jpeg_fdct_ifast) {
- for(i=0;i<64;i++) {
- /* 16 <= qscale * quant_matrix[i] <= 7905 */
- /* 19952 <= aanscales[i] * qscale * quant_matrix[i] <= 249205026 */
- /* (1<<36)/19952 >= (1<<36)/(aanscales[i] * qscale * quant_matrix[i]) >= (1<<36)/249205026 */
- /* 3444240 >= (1<<36)/(aanscales[i] * qscale * quant_matrix[i]) >= 275 */
-
- qmat[block_permute_op(i)] = (int)((UINT64_C(1) << (QMAT_SHIFT + 11)) /
- (aanscales[i] * qscale * quant_matrix[block_permute_op(i)]));
- }
- } else {
- for(i=0;i<64;i++) {
- /* We can safely suppose that 16 <= quant_matrix[i] <= 255
- So 16 <= qscale * quant_matrix[i] <= 7905
- so (1<<19) / 16 >= (1<<19) / (qscale * quant_matrix[i]) >= (1<<19) / 7905
- so 32768 >= (1<<19) / (qscale * quant_matrix[i]) >= 67
- */
- qmat[i] = (1 << QMAT_SHIFT_MMX) / (qscale * quant_matrix[i]);
- qmat16[i] = (1 << QMAT_SHIFT_MMX) / (qscale * quant_matrix[block_permute_op(i)]);
- }
- }
-}
-
-static inline void encode_dc(MpegEncContext *s, int val,
- UINT8 *huff_size, UINT16 *huff_code)
-{
- int mant, nbits;
-
- if (val == 0) {
- jput_bits(&s->pb, huff_size[0], huff_code[0]);
- } else {
- mant = val;
- if (val < 0) {
- val = -val;
- mant--;
- }
-
- /* compute the log (XXX: optimize) */
- nbits = 0;
- while (val != 0) {
- val = val >> 1;
- nbits++;
- }
-
- jput_bits(&s->pb, huff_size[nbits], huff_code[nbits]);
-
- jput_bits(&s->pb, nbits, mant & ((1 << nbits) - 1));
- }
-}
-
-static void encode_block(MpegEncContext *s, DCTELEM *block, int n)
-{
- int mant, nbits, code, i, j;
- int component, dc, run, last_index, val;
- MJpegContext *m = s->mjpeg_ctx;
- UINT8 *huff_size_ac;
- UINT16 *huff_code_ac;
-
- /* DC coef */
- component = (n <= 3 ? 0 : n - 4 + 1);
- dc = block[0]; /* overflow is impossible */
- val = dc - s->last_dc[component];
- if (n < 4) {
- encode_dc(s, val, m->huff_size_dc_luminance, m->huff_code_dc_luminance);
- huff_size_ac = m->huff_size_ac_luminance;
- huff_code_ac = m->huff_code_ac_luminance;
- } else {
- encode_dc(s, val, m->huff_size_dc_chrominance, m->huff_code_dc_chrominance);
- huff_size_ac = m->huff_size_ac_chrominance;
- huff_code_ac = m->huff_code_ac_chrominance;
- }
- s->last_dc[component] = dc;
-
- /* AC coefs */
-
- run = 0;
- last_index = s->block_last_index[n];
- for(i=1;i<=last_index;i++) {
- j = zigzag_direct[i];
- val = block[j];
- if (val == 0) {
- run++;
- } else {
- while (run >= 16) {
- jput_bits(&s->pb, huff_size_ac[0xf0], huff_code_ac[0xf0]);
- run -= 16;
- }
- mant = val;
- if (val < 0) {
- val = -val;
- mant--;
- }
-
- /* compute the log (XXX: optimize) */
- nbits = 0;
- while (val != 0) {
- val = val >> 1;
- nbits++;
- }
- code = (run << 4) | nbits;
-
- jput_bits(&s->pb, huff_size_ac[code], huff_code_ac[code]);
-
- jput_bits(&s->pb, nbits, mant & ((1 << nbits) - 1));
- run = 0;
- }
- }
-
- /* output EOB only if not already 64 values */
- if (last_index < 63 || run != 0)
- jput_bits(&s->pb, huff_size_ac[0], huff_code_ac[0]);
-}
-
-/* End excessive code duplication **************************************/
-
-/* this function is a reproduction of the one in mjpeg, it includes two
- * changes, it allows for black&white encoding (it skips the U and V
- * macroblocks and it outputs the huffman code for 'no change' (dc) and
- * 'all zero' (ac)) and it takes 4 macroblocks (422) instead of 6 (420) */
-static void zr_mjpeg_encode_mb(jpeg_enc_t *j) {
-
- MJpegContext *m = j->s->mjpeg_ctx;
-
- encode_block(j->s, j->s->block[0], 0);
- encode_block(j->s, j->s->block[1], 1);
- if (j->bw) {
- /* U */
- jput_bits(&j->s->pb, m->huff_size_dc_chrominance[0],
- m->huff_code_dc_chrominance[0]);
- jput_bits(&j->s->pb, m->huff_size_ac_chrominance[0],
- m->huff_code_ac_chrominance[0]);
- /* V */
- jput_bits(&j->s->pb, m->huff_size_dc_chrominance[0],
- m->huff_code_dc_chrominance[0]);
- jput_bits(&j->s->pb, m->huff_size_ac_chrominance[0],
- m->huff_code_ac_chrominance[0]);
- } else {
- /* we trick encode_block here so that it uses
- * chrominance huffman tables instead of luminance ones
- * (see the effect of second argument of encode_block) */
- encode_block(j->s, j->s->block[2], 4);
- encode_block(j->s, j->s->block[3], 5);
- }
-}
-
-/* this function can take all kinds of YUV colorspaces
- * YV12, YVYU, UYVY. The necesary parameters must be set up by the caller
- * y_ps means "y pixel size", y_rs means "y row size".
- * For YUYV, for example, is u_buf = y_buf + 1, v_buf = y_buf + 3,
- * y_ps = 2, u_ps = 4, v_ps = 4, y_rs = u_rs = v_rs.
- *
- * The actual buffers must be passed with mjpeg_encode_frame, this is
- * to make it possible to call encode on the buffer provided by the
- * codec in draw_frame.
- *
- * The data is straightened out at the moment it is put in DCT
- * blocks, there are therefore no spurious memcopies involved */
-/* Notice that w must be a multiple of 16 and h must be a multiple of 8 */
-/* We produce YUV422 jpegs, the colors must be subsampled horizontally,
- * if the colors are also subsampled vertically, then this function
- * performs cheap upsampling (better solution will be: a DCT that is
- * optimized in the case that every two rows are the same) */
-/* cu = 0 means 'No cheap upsampling'
- * cu = 1 means 'perform cheap upsampling' */
-/* The encoder doesn't know anything about interlacing, the halve height
- * needs to be passed and the double rowstride. Which field gets encoded
- * is decided by what buffers are passed to mjpeg_encode_frame */
-jpeg_enc_t *jpeg_enc_init(int w, int h, int y_psize, int y_rsize,
- int u_psize, int u_rsize, int v_psize, int v_rsize,
- int cu, int q, int b) {
- jpeg_enc_t *j;
- int i = 0;
- MSGL_V( "JPEnc init: %dx%d %d %d %d %d %d %d\n",
- w, h, y_psize, y_rsize, u_psize,
- u_rsize, v_psize, v_rsize);
-
- j = mp_malloc(sizeof(jpeg_enc_t));
- if (j == NULL) return NULL;
-
- j->s = mp_malloc(sizeof(MpegEncContext));
- if (j->s == NULL) {
- mp_free(j);
- return NULL;
- }
-
- /* info on how to access the pixels */
- j->y_ps = y_psize;
- j->u_ps = u_psize;
- j->v_ps = v_psize;
- j->y_rs = y_rsize;
- j->u_rs = u_rsize;
- j->v_rs = v_rsize;
-
- j->s->width = w;
- j->s->height = h;
- j->s->qscale = q;
-
- j->s->out_format = FMT_MJPEG;
- j->s->intra_only = 1;
- j->s->encoding = 1;
- j->s->pict_type = I_TYPE;
- j->s->y_dc_scale = 8;
- j->s->c_dc_scale = 8;
-
- j->s->mjpeg_write_tables = 1;
- j->s->mjpeg_vsample[0] = 1;
- j->s->mjpeg_vsample[1] = 1;
- j->s->mjpeg_vsample[2] = 1;
- j->s->mjpeg_hsample[0] = 2;
- j->s->mjpeg_hsample[1] = 1;
- j->s->mjpeg_hsample[2] = 1;
-
- j->cheap_upsample = cu;
- j->bw = b;
-
- /* if libavcodec is used by the decoder then we must not
- * initialize again, but if it is not initialized then we must
- * initialize it here. There must be a better way to find out
- * if it is initialized */
- if (av_fdct != jpeg_fdct_ifast
-#ifdef CAN_COMPILE_MMX
- && av_fdct != fdct_mmx
-#endif
- ) {
- /* we need to initialize libavcodec */
- avcodec_init();
- }
-
- if (mjpeg_init(j->s) < 0) {
- mp_free(j->s);
- mp_free(j);
- return NULL;
- }
-
- if (MPV_common_init(j->s) < 0) {
- mp_free(j->s);
- mp_free(j);
- return NULL;
- }
-
- /* correct the value for sc->mb_height */
- j->s->mb_height = j->s->height/8;
- j->s->mb_intra = 1;
-
- j->s->intra_matrix[0] = default_intra_matrix[0];
- for (i = 1; i < 64; i++)
- j->s->intra_matrix[i] =
- (default_intra_matrix[i]*j->s->qscale) >> 3;
- convert_matrix(j->s->q_intra_matrix, j->s->q_intra_matrix16,
- j->s->intra_matrix, 8);
- return j;
-}
-
-int jpeg_enc_frame(jpeg_enc_t *j, unsigned char *y_data,
- unsigned char *u_data, unsigned char *v_data, char *bufr) {
- int i, k, mb_x, mb_y;
- short int *dest;
- unsigned char *source;
- /* initialize the buffer */
-
- init_put_bits(&j->s->pb, bufr, 1024*256, NULL, NULL);
-
- mjpeg_picture_header(j->s);
-
- j->s->last_dc[0] = 128;
- j->s->last_dc[1] = 128;
- j->s->last_dc[2] = 128;
-
- for (mb_y = 0; mb_y < j->s->mb_height; mb_y++) {
- for (mb_x = 0; mb_x < j->s->mb_width; mb_x++) {
- /* conversion 8 to 16 bit and filling of blocks
- * must be mmx optimized */
- /* fill 2 Y macroblocks and one U and one V */
- source = mb_y * 8 * j->y_rs +
- 16 * j->y_ps * mb_x + y_data;
- dest = j->s->block[0];
- for (i = 0; i < 8; i++) {
- for (k = 0; k < 8; k++) {
- dest[k] = source[k*j->y_ps];
- }
- dest += 8;
- source += j->y_rs;
- }
- source = mb_y * 8 * j->y_rs +
- (16*mb_x + 8)*j->y_ps + y_data;
- dest = j->s->block[1];
- for (i = 0; i < 8; i++) {
- for (k = 0; k < 8; k++) {
- dest[k] = source[k*j->y_ps];
- }
- dest += 8;
- source += j->y_rs;
- }
- if (!j->bw && j->cheap_upsample) {
- source = mb_y*4*j->u_rs +
- 8*mb_x*j->u_ps + u_data;
- dest = j->s->block[2];
- for (i = 0; i < 4; i++) {
- for (k = 0; k < 8; k++) {
- dest[k] = source[k*j->u_ps];
- dest[k+8] = source[k*j->u_ps];
- }
- dest += 16;
- source += j->u_rs;
- }
- source = mb_y*4*j->v_rs +
- 8*mb_x*j->v_ps + v_data;
- dest = j->s->block[3];
- for (i = 0; i < 4; i++) {
- for (k = 0; k < 8; k++) {
- dest[k] = source[k*j->v_ps];
- dest[k+8] = source[k*j->v_ps];
- }
- dest += 16;
- source += j->u_rs;
- }
- } else if (!j->bw && !j->cheap_upsample) {
- source = mb_y*8*j->u_rs +
- 8*mb_x*j->u_ps + u_data;
- dest = j->s->block[2];
- for (i = 0; i < 8; i++) {
- for (k = 0; k < 8; k++)
- dest[k] = source[k*j->u_ps];
- dest += 8;
- source += j->u_rs;
- }
- source = mb_y*8*j->v_rs +
- 8*mb_x*j->v_ps + v_data;
- dest = j->s->block[3];
- for (i = 0; i < 8; i++) {
- for (k = 0; k < 8; k++)
- dest[k] = source[k*j->v_ps];
- dest += 8;
- source += j->u_rs;
- }
- }
- emms_c(); /* is this really needed? */
-
- j->s->block_last_index[0] =
- dct_quantize(j->s, j->s->block[0],
- 0, j->s->qscale);
- j->s->block_last_index[1] =
- dct_quantize(j->s, j->s->block[1],
- 1, j->s->qscale);
-
- if (!j->bw) {
- j->s->block_last_index[4] =
- dct_quantize(j->s, j->s->block[2],
- 4, j->s->qscale);
- j->s->block_last_index[5] =
- dct_quantize(j->s, j->s->block[3],
- 5, j->s->qscale);
- }
- zr_mjpeg_encode_mb(j);
- }
- }
- emms_c();
- mjpeg_picture_trailer(j->s);
- flush_put_bits(&j->s->pb);
-
- if (j->s->mjpeg_write_tables == 1)
- j->s->mjpeg_write_tables = 0;
-
- return pbBufPtr(&(j->s->pb)) - j->s->pb.buf;
-}
-
-void jpeg_enc_uninit(jpeg_enc_t *j) {
- mjpeg_close(j->s);
- mp_free(j->s);
- mp_free(j);
-}
-
Copied: mplayerxp/libvo/jpeg_enc.cpp (from rev 417, mplayerxp/libvo/jpeg_enc.c)
===================================================================
--- mplayerxp/libvo/jpeg_enc.cpp (rev 0)
+++ mplayerxp/libvo/jpeg_enc.cpp 2012-11-21 13:21:00 UTC (rev 422)
@@ -0,0 +1,451 @@
+/* Straightforward (to be) optimized JPEG encoder for the YUV422 format
+ * based on mjpeg code from ffmpeg.
+ *
+ * Copyright (c) 2002, Rik Snel
+ * Parts from ffmpeg Copyright (c) 2000, 2001 Gerard Lantau
+ *
+ * This program is mp_free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ *
+ * For an excellent introduction to the JPEG format, see:
+ * http://www.ece.purdue.edu/~bouman/grad-labs/lab8/pdf/lab.pdf
+ */
+
+
+
+#include <sys/types.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include "mp_config.h"
+#ifdef USE_FASTMEMCPY
+#include "fastmemcpy.h"
+#endif
+/* We need this #define because we need ../libavcodec/common.h to #define
+ * be2me_32, otherwise the linker will complain that it doesn't exist */
+#include "mp_conf_lavc.h"
+
+#include "jpeg_enc.h"
+#include "vo_msg.h"
+
+/* zr_mjpeg_encode_mb needs access to these tables for the black & white
+ * option */
+typedef struct MJpegContext {
+ UINT8 huff_size_dc_luminance[12];
+ UINT16 huff_code_dc_luminance[12];
+ UINT8 huff_size_dc_chrominance[12];
+ UINT16 huff_code_dc_chrominance[12];
+
+ UINT8 huff_size_ac_luminance[256];
+ UINT16 huff_code_ac_luminance[256];
+ UINT8 huff_size_ac_chrominance[256];
+ UINT16 huff_code_ac_chrominance[256];
+} MJpegContext;
+
+
+/* A very important function pointer */
+extern int (*dct_quantize)(MpegEncContext *s,
+ DCTELEM *block, int n, int qscale);
+
+
+/* Begin excessive code duplication ************************************/
+/* Code coming from mpegvideo.c and mjpeg.c in ../libavcodec ***********/
+
+static const unsigned short aanscales[64] = {
+ /* precomputed values scaled up by 14 bits */
+ 16384, 22725, 21407, 19266, 16384, 12873, 8867, 4520,
+ 22725, 31521, 29692, 26722, 22725, 17855, 12299, 6270,
+ 21407, 29692, 27969, 25172, 21407, 16819, 11585, 5906,
+ 19266, 26722, 25172, 22654, 19266, 15137, 10426, 5315,
+ 16384, 22725, 21407, 19266, 16384, 12873, 8867, 4520,
+ 12873, 17855, 16819, 15137, 12873, 10114, 6967, 3552,
+ 8867, 12299, 11585, 10426, 8867, 6967, 4799, 2446,
+ 4520, 6270, 5906, 5315, 4520, 3552, 2446, 1247
+};
+
+static void convert_matrix(int *qmat, UINT16 *qmat16, const UINT16 *quant_matrix, int qscale)
+{
+ int i;
+
+ if (av_fdct == jpeg_fdct_ifast) {
+ for(i=0;i<64;i++) {
+ /* 16 <= qscale * quant_matrix[i] <= 7905 */
+ /* 19952 <= aanscales[i] * qscale * quant_matrix[i] <= 249205026 */
+ /* (1<<36)/19952 >= (1<<36)/(aanscales[i] * qscale * quant_matrix[i]) >= (1<<36)/249205026 */
+ /* 3444240 >= (1<<36)/(aanscales[i] * qscale * quant_matrix[i]) >= 275 */
+
+ qmat[block_permute_op(i)] = (int)((UINT64_C(1) << (QMAT_SHIFT + 11)) /
+ (aanscales[i] * qscale * quant_matrix[block_permute_op(i)]));
+ }
+ } else {
+ for(i=0;i<64;i++) {
+ /* We can safely suppose that 16 <= quant_matrix[i] <= 255
+ So 16 <= qscale * quant_matrix[i] <= 7905
+ so (1<<19) / 16 >= (1<<19) / (qscale * quant_matrix[i]) >= (1<<19) / 7905
+ so 32768 >= (1<<19) / (qscale * quant_matrix[i]) >= 67
+ */
+ qmat[i] = (1 << QMAT_SHIFT_MMX) / (qscale * quant_matrix[i]);
+ qmat16[i] = (1 << QMAT_SHIFT_MMX) / (qscale * quant_matrix[block_permute_op(i)]);
+ }
+ }
+}
+
+static inline void encode_dc(MpegEncContext *s, int val,
+ UINT8 *huff_size, UINT16 *huff_code)
+{
+ int mant, nbits;
+
+ if (val == 0) {
+ jput_bits(&s->pb, huff_size[0], huff_code[0]);
+ } else {
+ mant = val;
+ if (val < 0) {
+ val = -val;
+ mant--;
+ }
+
+ /* compute the log (XXX: optimize) */
+ nbits = 0;
+ while (val != 0) {
+ val = val >> 1;
+ nbits++;
+ }
+
+ jput_bits(&s->pb, huff_size[nbits], huff_code[nbits]);
+
+ jput_bits(&s->pb, nbits, mant & ((1 << nbits) - 1));
+ }
+}
+
+static void encode_block(MpegEncContext *s, DCTELEM *block, int n)
+{
+ int mant, nbits, code, i, j;
+ int component, dc, run, last_index, val;
+ MJpegContext *m = s->mjpeg_ctx;
+ UINT8 *huff_size_ac;
+ UINT16 *huff_code_ac;
+
+ /* DC coef */
+ component = (n <= 3 ? 0 : n - 4 + 1);
+ dc = block[0]; /* overflow is impossible */
+ val = dc - s->last_dc[component];
+ if (n < 4) {
+ encode_dc(s, val, m->huff_size_dc_luminance, m->huff_code_dc_luminance);
+ huff_size_ac = m->huff_size_ac_luminance;
+ huff_code_ac = m->huff_code_ac_luminance;
+ } else {
+ encode_dc(s, val, m->huff_size_dc_chrominance, m->huff_code_dc_chrominance);
+ huff_size_ac = m->huff_size_ac_chrominance;
+ huff_code_ac = m->huff_code_ac_chrominance;
+ }
+ s->last_dc[component] = dc;
+
+ /* AC coefs */
+
+ run = 0;
+ last_index = s->block_last_index[n];
+ for(i=1;i<=last_index;i++) {
+ j = zigzag_direct[i];
+ val = block[j];
+ if (val == 0) {
+ run++;
+ } else {
+ while (run >= 16) {
+ jput_bits(&s->pb, huff_size_ac[0xf0], huff_code_ac[0xf0]);
+ run -= 16;
+ }
+ mant = val;
+ if (val < 0) {
+ val = -val;
+ mant--;
+ }
+
+ /* compute the log (XXX: optimize) */
+ nbits = 0;
+ while (val != 0) {
+ val = val >> 1;
+ nbits++;
+ }
+ code = (run << 4) | nbits;
+
+ jput_bits(&s->pb, huff_size_ac[code], huff_code_ac[code]);
+
+ jput_bits(&s->pb, nbits, mant & ((1 << nbits) - 1));
+ run = 0;
+ }
+ }
+
+ /* output EOB only if not already 64 values */
+ if (last_index < 63 || run != 0)
+ jput_bits(&s->pb, huff_size_ac[0], huff_code_ac[0]);
+}
+
+/* End excessive code duplication **************************************/
+
+/* this function is a reproduction of the one in mjpeg, it includes two
+ * changes, it allows for black&white encoding (it skips the U and V
+ * macroblocks and it outputs the huffman code for 'no change' (dc) and
+ * 'all zero' (ac)) and it takes 4 macroblocks (422) instead of 6 (420) */
+static void zr_mjpeg_encode_mb(jpeg_enc_t *j) {
+
+ MJpegContext *m = j->s->mjpeg_ctx;
+
+ encode_block(j->s, j->s->block[0], 0);
+ encode_block(j->s, j->s->block[1], 1);
+ if (j->bw) {
+ /* U */
+ jput_bits(&j->s->pb, m->huff_size_dc_chrominance[0],
+ m->huff_code_dc_chrominance[0]);
+ jput_bits(&j->s->pb, m->huff_size_ac_chrominance[0],
+ m->huff_code_ac_chrominance[0]);
+ /* V */
+ jput_bits(&j->s->pb, m->huff_size_dc_chrominance[0],
+ m->huff_code_dc_chrominance[0]);
+ jput_bits(&j->s->pb, m->huff_size_ac_chrominance[0],
+ m->huff_code_ac_chrominance[0]);
+ } else {
+ /* we trick encode_block here so that it uses
+ * chrominance huffman tables instead of luminance ones
+ * (see the effect of second argument of encode_block) */
+ encode_block(j->s, j->s->block[2], 4);
+ encode_block(j->s, j->s->block[3], 5);
+ }
+}
+
+/* this function can take all kinds of YUV colorspaces
+ * YV12, YVYU, UYVY. The necesary parameters must be set up by the caller
+ * y_ps means "y pixel size", y_rs means "y row size".
+ * For YUYV, for example, is u_buf = y_buf + 1, v_buf = y_buf + 3,
+ * y_ps = 2, u_ps = 4, v_ps = 4, y_rs = u_rs = v_rs.
+ *
+ * The actual buffers must be passed with mjpeg_encode_frame, this is
+ * to make it possible to call encode on the buffer provided by the
+ * codec in draw_frame.
+ *
+ * The data is straightened out at the moment it is put in DCT
+ * blocks, there are therefore no spurious memcopies involved */
+/* Notice that w must be a multiple of 16 and h must be a multiple of 8 */
+/* We produce YUV422 jpegs, the colors must be subsampled horizontally,
+ * if the colors are also subsampled vertically, then this function
+ * performs cheap upsampling (better solution will be: a DCT that is
+ * optimized in the case that every two rows are the same) */
+/* cu = 0 means 'No cheap upsampling'
+ * cu = 1 means 'perform cheap upsampling' */
+/* The encoder doesn't know anything about interlacing, the halve height
+ * needs to be passed and the double rowstride. Which field gets encoded
+ * is decided by what buffers are passed to mjpeg_encode_frame */
+jpeg_enc_t *jpeg_enc_init(int w, int h, int y_psize, int y_rsize,
+ int u_psize, int u_rsize, int v_psize, int v_rsize,
+ int cu, int q, int b) {
+ jpeg_enc_t *j;
+ int i = 0;
+ MSGL_V( "JPEnc init: %dx%d %d %d %d %d %d %d\n",
+ w, h, y_psize, y_rsize, u_psize,
+ u_rsize, v_psize, v_rsize);
+
+ j = mp_malloc(sizeof(jpeg_enc_t));
+ if (j == NULL) return NULL;
+
+ j->s = mp_malloc(sizeof(MpegEncContext));
+ if (j->s == NULL) {
+ mp_free(j);
+ return NULL;
+ }
+
+ /* info on how to access the pixels */
+ j->y_ps = y_psize;
+ j->u_ps = u_psize;
+ j->v_ps = v_psize;
+ j->y_rs = y_rsize;
+ j->u_rs = u_rsize;
+ j->v_rs = v_rsize;
+
+ j->s->width = w;
+ j->s->height = h;
+ j->s->qscale = q;
+
+ j->s->out_format = FMT_MJPEG;
+ j->s->intra_only = 1;
+ j->s->encoding = 1;
+ j->s->pict_type = I_TYPE;
+ j->s->y_dc_scale = 8;
+ j->s->c_dc_scale = 8;
+
+ j->s->mjpeg_write_tables = 1;
+ j->s->mjpeg_vsample[0] = 1;
+ j->s->mjpeg_vsample[1] = 1;
+ j->s->mjpeg_vsample[2] = 1;
+ j->s->mjpeg_hsample[0] = 2;
+ j->s->mjpeg_hsample[1] = 1;
+ j->s->mjpeg_hsample[2] = 1;
+
+ j->cheap_upsample = cu;
+ j->bw = b;
+
+ /* if libavcodec is used by the decoder then we must not
+ * initialize again, but if it is not initialized then we must
+ * initialize it here. There must be a better way to find out
+ * if it is initialized */
+ if (av_fdct != jpeg_fdct_ifast
+#ifdef CAN_COMPILE_MMX
+ && av_fdct != fdct_mmx
+#endif
+ ) {
+ /* we need to initialize libavcodec */
+ avcodec_init();
+ }
+
+ if (mjpeg_init(j->s) < 0) {
+ mp_free(j->s);
+ mp_free(j);
+ return NULL;
+ }
+
+ if (MPV_common_init(j->s) < 0) {
+ mp_free(j->s);
+ mp_free(j);
+ return NULL;
+ }
+
+ /* correct the value for sc->mb_height */
+ j->s->mb_height = j->s->height/8;
+ j->s->mb_intra = 1;
+
+ j->s->intra_matrix[0] = default_intra_matrix[0];
+ for (i = 1; i < 64; i++)
+ j->s->intra_matrix[i] =
+ (default_intra_matrix[i]*j->s->qscale) >> 3;
+ convert_matrix(j->s->q_intra_matrix, j->s->q_intra_matrix16,
+ j->s->intra_matrix, 8);
+ return j;
+}
+
+int jpeg_enc_frame(jpeg_enc_t *j, unsigned char *y_data,
+ unsigned char *u_data, unsigned char *v_data, char *bufr) {
+ int i, k, mb_x, mb_y;
+ short int *dest;
+ unsigned char *source;
+ /* initialize the buffer */
+
+ init_put_bits(&j->s->pb, bufr, 1024*256, NULL, NULL);
+
+ mjpeg_picture_header(j->s);
+
+ j->s->last_dc[0] = 128;
+ j->s->last_dc[1] = 128;
+ j->s->last_dc[2] = 128;
+
+ for (mb_y = 0; mb_y < j->s->mb_height; mb_y++) {
+ for (mb_x = 0; mb_x < j->s->mb_width; mb_x++) {
+ /* conversion 8 to 16 bit and filling of blocks
+ * must be mmx optimized */
+ /* fill 2 Y macroblocks and one U and one V */
+ source = mb_y * 8 * j->y_rs +
+ 16 * j->y_ps * mb_x + y_data;
+ dest = j->s->block[0];
+ for (i = 0; i < 8; i++) {
+ for (k = 0; k < 8; k++) {
+ dest[k] = source[k*j->y_ps];
+ }
+ dest += 8;
+ source += j->y_rs;
+ }
+ source = mb_y * 8 * j->y_rs +
+ (16*mb_x + 8)*j->y_ps + y_data;
+ dest = j->s->block[1];
+ for (i = 0; i < 8; i++) {
+ for (k = 0; k < 8; k++) {
+ dest[k] = source[k*j->y_ps];
+ }
+ dest += 8;
+ source += j->y_rs;
+ }
+ if (!j->bw && j->cheap_upsample) {
+ source = mb_y*4*j->u_rs +
+ 8*mb_x*j->u_ps + u_data;
+ dest = j->s->block[2];
+ for (i = 0; i < 4; i++) {
+ for (k = 0; k < 8; k++) {
+ dest[k] = source[k*j->u_ps];
+ dest[k+8] = source[k*j->u_ps];
+ }
+ dest += 16;
+ source += j->u_rs;
+ }
+ source = mb_y*4*j->v_rs +
+ 8*mb_x*j->v_ps + v_data;
+ dest = j->s->block[3];
+ for (i = 0; i < 4; i++) {
+ for (k = 0; k < 8; k++) {
+ dest[k] = source[k*j->v_ps];
+ dest[k+8] = source[k*j->v_ps];
+ }
+ dest += 16;
+ source += j->u_rs;
+ }
+ } else if (!j->bw && !j->cheap_upsample) {
+ source = mb_y*8*j->u_rs +
+ 8*mb_x*j->u_ps + u_data;
+ dest = j->s->block[2];
+ for (i = 0; i < 8; i++) {
+ for (k = 0; k < 8; k++)
+ dest[k] = source[k*j->u_ps];
+ dest += 8;
+ source += j->u_rs;
+ }
+ source = mb_y*8*j->v_rs +
+ 8*mb_x*j->v_ps + v_data;
+ dest = j->s->block[3];
+ for (i = 0; i < 8; i++) {
+ for (k = 0; k < 8; k++)
+ dest[k] = source[k*j->v_ps];
+ dest += 8;
+ source += j->u_rs;
+ }
+ }
+ emms_c(); /* is this really needed? */
+
+ j->s->block_last_index[0] =
+ dct_quantize(j->s, j->s->block[0],
+ 0, j->s->qscale);
+ j->s->block_last_index[1] =
+ dct_quantize(j->s, j->s->block[1],
+ 1, j->s->qscale);
+
+ if (!j->bw) {
+ j->s->block_last_index[4] =
+ dct_quantize(j->s, j->s->block[2],
+ 4, j->s->qscale);
+ j->s->block_last_index[5] =
+ dct_quantize(j->s, j->s->block[3],
+ 5, j->s->qscale);
+ }
+ zr_mjpeg_encode_mb(j);
+ }
+ }
+ emms_c();
+ mjpeg_picture_trailer(j->s);
+ flush_put_bits(&j->s->pb);
+
+ if (j->s->mjpeg_write_tables == 1)
+ j->s->mjpeg_write_tables = 0;
+
+ return pbBufPtr(&(j->s->pb)) - j->s->pb.buf;
+}
+
+void jpeg_enc_uninit(jpeg_enc_t *j) {
+ mjpeg_close(j->s);
+ mp_free(j->s);
+ mp_free(j);
+}
+
Modified: mplayerxp/libvo/osd.cpp
===================================================================
--- mplayerxp/libvo/osd.cpp 2012-11-21 12:54:43 UTC (rev 421)
+++ mplayerxp/libvo/osd.cpp 2012-11-21 13:21:00 UTC (rev 422)
@@ -23,7 +23,7 @@
static const unsigned long long mask24hl __attribute__((used)) __attribute__((aligned(8))) = 0x0000FFFFFFFFFFFFULL;
#endif
-#define PVECTOR_ACCEL_H "osd_template.c"
+#define PVECTOR_ACCEL_H "osd_template.h"
#include "pvector/pvector_inc.h"
#ifdef FAST_OSD_TABLE
Deleted: mplayerxp/libvo/osd_template.c
===================================================================
--- mplayerxp/libvo/osd_template.c 2012-11-21 12:54:43 UTC (rev 421)
+++ mplayerxp/libvo/osd_template.c 2012-11-21 13:21:00 UTC (rev 422)
@@ -1,318 +0,0 @@
-// Generic alpha renderers for all YUV modes and RGB depths.
-// Optimized by Nick and Michael
-// Code from Michael Niedermayer (mic...@gm...) is under GPL
-#include "pvector/pvector.h"
-
-#ifdef HAVE_INT_PVECTOR
-static __inline __m64 __attribute__((__gnu_inline__, __always_inline__))
-PVECTOR_RENAME(_m_load)(const any_t*__P)
-{
- return *(const __m64 *)__P;
-}
-#undef _m_load
-#define _m_load PVECTOR_RENAME(_m_load)
-
-static __inline __m64 __attribute__((__gnu_inline__, __always_inline__))
-PVECTOR_RENAME(_m_load_half)(const any_t*__P)
-{
- return _mm_cvtsi32_si64 (*(const int *)__P);
-}
-#undef _m_load_half
-#define _m_load_half PVECTOR_RENAME(_m_load_half)
-
-static __inline void __attribute__((__gnu_inline__, __always_inline__))
-PVECTOR_RENAME(_m_store)(any_t*__P, __m64 src)
-{
- *(__m64 *)__P = src;
-}
-#undef _m_store
-#define _m_store PVECTOR_RENAME(_m_store)
-
-static __inline void __attribute__((__gnu_inline__, __always_inline__))
-PVECTOR_RENAME(_m_store_half)(any_t*__P, __m64 src)
-{
- *(int *)__P = _mm_cvtsi64_si32(src);
-}
-#undef _m_store_half
-#define _m_store_half PVECTOR_RENAME(_m_store_half)
-
-static __inline void __attribute__((__gnu_inline__, __always_inline__))
-PVECTOR_RENAME(_m_movntq)(any_t*__P, __m64 src)
-{
-#ifdef HAVE_MMX2
- _mm_stream_pi(__P,src);
-#else
- _m_store(__P,src);
-#endif
-}
-#undef _m_movntq
-#define _m_movntq PVECTOR_RENAME(_m_movntq)
-
-#endif
-
-static inline void PVECTOR_RENAME(vo_draw_alpha_yv12)(int w,int h,const unsigned char* src,const unsigned char *srca, int srcstride, unsigned char* dstbase,int dststride,int finalize){
- unsigned y;
-#ifdef HAVE_INT_PVECTOR
- __ivec vzero = _ivec_setzero();
-#endif
-PROFILE_START();
- for(y=0;y<(unsigned)h;y++){
- int x;
- x=0;
-#ifdef HAVE_INT_PVECTOR
- _ivec_prefetchw(&dstbase[x]);
- _ivec_prefetch(&src[x]);
- _ivec_prefetch(&srca[x]);
- /* MOVNTDQ: #GP(0) - If memory operand is not aligned on a 16-byte boundary */
- if(!_ivec_aligned(dstbase))
- for(;x<w;x++){
- unsigned char *dst=&dstbase[x];
- if(srca[x]) *dst=((dstbase[x]*srca[x])>>8)+src[x];
- if(_ivec_aligned(dst)) break; /* align on sizeof(MMREG) boundary */
- }
- if((w-x)>=_ivec_size())
- for(;x<w;x+=_ivec_size()){
- __ivec vmsk,vdest,vsrc,vsrca,vt[4];
- _ivec_prefetchw(&dstbase[x+_ivec_size()*4]);
- _ivec_prefetch(&src[x+_ivec_size()*4]);
- _ivec_prefetch(&srca[x+_ivec_size()*4]);
- vdest = _ivec_loada(&dstbase[x]);
- if(_ivec_aligned(&src[x]))
- vsrc = _ivec_loada(&src[x]);
- else
- vsrc = _ivec_loadu(&src[x]);
- if(_ivec_aligned(&srca[x]))
- vsrca = _ivec_loada(&srca[x]);
- else
- vsrca = _ivec_loadu(&srca[x]);
- vmsk = _ivec_not(_ivec_cmpeq_s8(vsrca,vzero));
- vt[0] = _ivec_u16_from_u8(vdest,&vt[1]);
- vt[2] = _ivec_u16_from_u8(vsrca,&vt[3]);
- vt[0] = _ivec_srl_s16_imm(_ivec_mullo_s16(vt[0],vt[2]),8);
- vt[1] = _ivec_srl_s16_imm(_ivec_mullo_s16(vt[1],vt[3]),8);
- vt[0] = _ivec_add_s8(_ivec_u8_from_u16(vt[0],vt[1]),vsrc);
- vt[1] = _ivec_blend_u8(vdest,vt[0],vmsk);
- if(finalize)
- _ivec_stream(&dstbase[x],vt[1]);
- else
- _ivec_storea(&dstbase[x],vt[1]);
- }
-#endif
- for(;x<w;x++){
- if(srca[x]) dstbase[x]=((dstbase[x]*srca[x])>>8)+src[x];
- }
- src+=srcstride;
- srca+=srcstride;
- dstbase+=dststride;
- }
-#ifdef HAVE_INT_PVECTOR
- _ivec_empty();
- if(finalize) _ivec_sfence();
-#endif
-PROFILE_END("vo_draw_alpha_yv12");
- return;
-}
-
-static inline void PVECTOR_RENAME(vo_draw_alpha_yuy2)(int w,int h,const unsigned char* src,const unsigned char *srca, int srcstride, unsigned char* dstbase,int dststride,int finalize){
- int y;
-#if defined(FAST_OSD) && !defined(HAVE_MMX)
- w=w>>1;
-#endif
-PROFILE_START();
- for(y=0;y<h;y++){
- register int x=0;
-#ifdef HAVE_INT_PVECTOR
- /* TODO: port this stuff on PVector */
- __m64 mm[8];
- _ivec_prefetchw(&dstbase[x]);
- _ivec_prefetch(&src[x]);
- _ivec_prefetch(&srca[x]);
- mm[7]=_mm_setzero_si64();
- mm[5]=_mm_set1_pi8(0xFF);
- mm[4]=mm[5];
- mm[5]=_m_psllwi(mm[5],8);
- mm[4]=_m_psrlwi(mm[4],8);
- for(;x<w;x+=4){
- _ivec_prefetchw(&dstbase[x+_ivec_size()*4]);
- _ivec_prefetch(&src[x+_ivec_size()*4]);
- _ivec_prefetch(&srca[x+_ivec_size()*4]);
- mm[0]=_m_load(&(((char *)dstbase)[x*2]));
- mm[1]=mm[0];
- mm[0]=_m_pand(mm[0],mm[4]);
- mm[2]=_m_load_half(&(((char *)srca)[x]));
- mm[2]=_m_paddb(mm[2],_m_load(&bFF));
- mm[2]=_m_punpcklbw(mm[2],mm[7]);
- mm[0]=_m_pmullw(mm[0],mm[2]);
- mm[0]=_m_psrlwi(mm[0],8);
- mm[1]=_m_pand(mm[1],mm[5]);
- mm[2]=_m_load_half(&(((char *)src)[x]));
- mm[2]=_m_punpcklbw(mm[2],mm[7]);
- mm[0]=_m_por(mm[0],mm[1]);
- mm[0]=_m_paddb(mm[0],mm[2]);
- if(finalize)
- _m_movntq(&(((char *)dstbase)[x*2]),mm[0]);
- else
- _m_store(&(((char *)dstbase)[x*2]),mm[0]);
- }
-#endif
- for(;x<w;x++){
-#ifdef FAST_OSD
- if(srca[2*x+0]) dstbase[4*x+0]=src[2*x+0];
- if(srca[2*x+1]) dstbase[4*x+2]=src[2*x+1];
-#else
- if(srca[x]) {
- dstbase[2*x]=((dstbase[2*x]*srca[x])>>8)+src[x];
- dstbase[2*x+1]=((((signed)dstbase[2*x+1]-128)*srca[x])>>8)+128;
- }
-#endif
- }
- src+=srcstride;
- srca+=srcstride;
- dstbase+=dststride;
- }
-#ifdef HAVE_INT_PVECTOR
- _ivec_empty();
- if(finalize) _ivec_sfence();
-#endif
-PROFILE_END("vo_draw_alpha_yuy2");
- return;
-}
-
-static inline void PVECTOR_RENAME(vo_draw_alpha_rgb24)(int w,int h,const unsigned char* src,const unsigned char *srca, int srcstride, unsigned char* dstbase,int dststride,int finalize){
- int y;
- for(y=0;y<h;y++){
- register unsigned char *dst = dstbase;
- register int x=0;
-#ifdef HAVE_INT_PVECTOR
- __m64 mm[8];
- _ivec_prefetchw(&dstbase[x]);
- _ivec_prefetch(&src[x]);
- _ivec_prefetch(&srca[x]);
- mm[7]=_mm_setzero_si64();
- mm[6]=_mm_set1_pi8(0xFF);
- for(;x<w;x+=2){
- if(srca[x] || srca[x+1]) {
- _ivec_prefetchw(&dstbase[x+_ivec_size()*4]);
- _ivec_prefetch(&src[x+_ivec_size()*4]);
- _ivec_prefetch(&srca[x+_ivec_size()*4]);
- mm[0]=_m_load(&dstbase[0]);
- mm[1]=mm[0];
- mm[5]=mm[0];
- mm[0]=_m_punpcklbw(mm[0],mm[7]);
- mm[1]=_m_punpckhbw(mm[1],mm[7]);
- mm[2]=_m_load_half(&srca[x]);
- mm[2]=_m_paddb(mm[2],mm[6]);
- mm[2]=_m_punpcklbw(mm[2],mm[2]);
- mm[2]=_m_punpcklbw(mm[2],mm[2]);
- mm[3]=mm[2];
- mm[2]=_m_punpcklbw(mm[2],mm[7]);
- mm[3]=_m_punpckhbw(mm[3],mm[7]);
- mm[0]=_m_pmullw(mm[0],mm[2]);
- mm[1]=_m_pmullw(mm[1],mm[3]);
- mm[0]=_m_psrlwi(mm[0],8);
- mm[1]=_m_psrlwi(mm[1],8);
- mm[0]=_m_packuswb(mm[0],mm[1]);
- mm[2]=_m_load_half(&src[x]);
- mm[2]=_m_punpcklbw(mm[2],mm[2]);
- mm[2]=_m_punpcklbw(mm[2],mm[2]);
- mm[0]=_m_paddb(mm[0],mm[2]);
- mm[5]=_m_pand(mm[5],_m_load(&mask24lh));
- mm[0]=_m_pand(mm[0],_m_load(&mask24hl));
- mm[5]=_m_por(mm[5],mm[0]);
- if(finalize)
- _m_movntq(&dstbase[0],mm[5]);
- else
- _m_store(&dstbase[0],mm[5]);
- }
- dst += 6;
- }
-#endif /* arch_x86 */
- for(;x<w;x++){
- if(srca[x]){
-#ifdef FAST_OSD
- dst[0]=dst[1]=dst[2]=src[x];
-#else
- dst[0]=((dst[0]*srca[x])>>8)+src[x];
- dst[1]=((dst[1]*srca[x])>>8)+src[x];
- dst[2]=((dst[2]*srca[x])>>8)+src[x];
-#endif
- }
- dst+=3; // 24bpp
- }
- src+=srcstride;
- srca+=srcstride;
- dstbase+=dststride;
- }
-#ifdef HAVE_INT_PVECTOR
- _ivec_empty();
- if(finalize) _ivec_sfence();
-#endif
- return;
-}
-
-static inline void PVECTOR_RENAME(vo_draw_alpha_rgb32)(int w,int h,const unsigned char* src,const unsigned char *srca, int srcstride, unsigned char* dstbase,int dststride,int finalize){
- int y;
-PROFILE_START();
- for(y=0;y<h;y++){
- register int x=0;
-#ifdef HAVE_INT_PVECTOR
- __m64 mm[8];
- _ivec_prefetchw(&dstbase[x]);
- _ivec_prefetch(&src[x]);
- _ivec_prefetch(&srca[x]);
- mm[7]=_mm_setzero_si64();
- mm[6]=_mm_set1_pi8(0xFF);
- for(;x<w;x+=2){
- if(srca[x] || srca[x+1]) {
- _ivec_prefetchw(&dstbase[x+_ivec_size()*4]);
- _ivec_prefetch(&src[x+_ivec_size()*4]);
- _ivec_prefetch(&srca[x+_ivec_size()*4]);
- mm[0]=_m_load(&dstbase[4*x]);
- mm[1]=mm[0];
- mm[0]=_m_punpcklbw(mm[0],mm[7]);
- mm[1]=_m_punpckhbw(mm[1],mm[7]);
- mm[2]=_m_load_half(&srca[x]);
- mm[2]=_m_paddb(mm[2],mm[6]);
- mm[2]=_m_punpcklbw(mm[2],mm[2]);
- mm[2]=_m_punpcklbw(mm[2],mm[2]);
- mm[3]=mm[2];
- mm[2]=_m_punpcklbw(mm[2],mm[7]);
- mm[3]=_m_punpckhbw(mm[3],mm[7]);
- mm[0]=_m_pmullw(mm[0],mm[2]);
- mm[1]=_m_pmullw(mm[1],mm[3]);
- mm[0]=_m_psrlwi(mm[0],8);
- mm[1]=_m_psrlwi(mm[1],8);
- mm[0]=_m_packuswb(mm[0],mm[1]);
- mm[2]=_m_load_half(&src[x]);
- mm[2]=_m_punpcklbw(mm[2],mm[2]);
- mm[2]=_m_punpcklbw(mm[2],mm[2]);
- mm[0]=_m_paddb(mm[0],mm[2]);
- if(finalize)
- _m_movntq(&dstbase[4*x],mm[0]);
- else
- _m_store(&dstbase[4*x],mm[0]);
- }
- }
-#endif /* arch_x86 */
- for(;x<w;x++){
- if(srca[x]){
-#ifdef FAST_OSD
- dstbase[4*x+0]=dstbase[4*x+1]=dstbase[4*x+2]=src[x];
-#else
- dstbase[4*x+0]=((dstbase[4*x+0]*srca[x])>>8)+src[x];
- dstbase[4*x+1]=((dstbase[4*x+1]*srca[x])>>8)+src[x];
- dstbase[4*x+2]=((dstbase[4*x+2]*srca[x])>>8)+src[x];
-#endif
- }
- }
- src+=srcstride;
- srca+=srcstride;
- dstbase+=dststride;
- }
-#ifdef HAVE_INT_PVECTOR
- _ivec_empty();
- if(finalize) _ivec_sfence();
-#endif
-PROFILE_END("vo_draw_alpha_rgb32");
- return;
-}
Copied: mplayerxp/libvo/osd_template.h (from rev 417, mplayerxp/libvo/osd_template.c)
===================================================================
--- mplayerxp/libvo/osd_template.h (rev 0)
+++ mplayerxp/libvo/osd_template.h 2012-11-21 13:21:00 UTC (rev 422)
@@ -0,0 +1,318 @@
+// Generic alpha renderers for all YUV modes and RGB depths.
+// Optimized by Nick and Michael
+// Code from Michael Niedermayer (mic...@gm...) is under GPL
+#include "pvector/pvector.h"
+
+#ifdef HAVE_INT_PVECTOR
+static __inline __m64 __attribute__((__gnu_inline__, __always_inline__))
+PVECTOR_RENAME(_m_load)(const any_t*__P)
+{
+ return *(const __m64 *)__P;
+}
+#undef _m_load
+#define _m_load PVECTOR_RENAME(_m_load)
+
+static __inline __m64 __attribute__((__gnu_inline__, __always_inline__))
+PVECTOR_RENAME(_m_load_half)(const any_t*__P)
+{
+ return _mm_cvtsi32_si64 (*(const int *)__P);
+}
+#undef _m_load_half
+#define _m_load_half PVECTOR_RENAME(_m_load_half)
+
+static __inline void __attribute__((__gnu_inline__, __always_inline__))
+PVECTOR_RENAME(_m_store)(any_t*__P, __m64 src)
+{
+ *(__m64 *)__P = src;
+}
+#undef _m_store
+#define _m_store PVECTOR_RENAME(_m_store)
+
+static __inline void __attribute__((__gnu_inline__, __always_inline__))
+PVECTOR_RENAME(_m_store_half)(any_t*__P, __m64 src)
+{
+ *(int *)__P = _mm_cvtsi64_si32(src);
+}
+#undef _m_store_half
+#define _m_store_half PVECTOR_RENAME(_m_store_half)
+
+static __inline void __attribute__((__gnu_inline__, __always_inline__))
+PVECTOR_RENAME(_m_movntq)(any_t*__P, __m64 src)
+{
+#ifdef HAVE_MMX2
+ _mm_stream_pi(__P,src);
+#else
+ _m_store(__P,src);
+#endif
+}
+#undef _m_movntq
+#define _m_movntq PVECTOR_RENAME(_m_movntq)
+
+#endif
+
+static inline void PVECTOR_RENAME(vo_draw_alpha_yv12)(int w,int h,const unsigned char* src,const unsigned char *srca, int srcstride, unsigned char* dstbase,int dststride,int finalize){
+ unsigned y;
+#ifdef HAVE_INT_PVECTOR
+ __ivec vzero = _ivec_setzero();
+#endif
+PROFILE_START();
+ for(y=0;y<(unsigned)h;y++){
+ int x;
+ x=0;
+#ifdef HAVE_INT_PVECTOR
+ _ivec_prefetchw(&dstbase[x]);
+ _ivec_prefetch(&src[x]);
+ _ivec_prefetch(&srca[x]);
+ /* MOVNTDQ: #GP(0) - If memory operand is not aligned on a 16-byte boundary */
+ if(!_ivec_aligned(dstbase))
+ for(;x<w;x++){
+ unsigned char *dst=&dstbase[x];
+ if(srca[x]) *dst=((dstbase[x]*srca[x])>>8)+src[x];
+ if(_ivec_aligned(dst)) break; /* align on sizeof(MMREG) boundary */
+ }
+ if((w-x)>=_ivec_size())
+ for(;x<w;x+=_ivec_size()){
+ __ivec vmsk,vdest,vsrc,vsrca,vt[4];
+ _ivec_prefetchw(&dstbase[x+_ivec_size()*4]);
+ _ivec_prefetch(&src[x+_ivec_size()*4]);
+ _ivec_prefetch(&srca[x+_ivec_size()*4]);
+ vdest = _ivec_loada(&dstbase[x]);
+ if(_ivec_aligned(&src[x]))
+ vsrc = _ivec_loada(&src[x]);
+ else
+ vsrc = _ivec_loadu(&src[x]);
+ if(_ivec_aligned(&srca[x]))
+ vsrca = _ivec_loada(&srca[x]);
+ else
+ vsrca = _ivec_loadu(&srca[x]);
+ vmsk = _ivec_not(_ivec_cmpeq_s8(vsrca,vzero));
+ vt[0] = _ivec_u16_from_u8(vdest,&vt[1]);
+ vt[2] = _ivec_u16_from_u8(vsrca,&vt[3]);
+ vt[0] = _ivec_srl_s16_imm(_ivec_mullo_s16(vt[0],vt[2]),8);
+ vt[1] = _ivec_srl_s16_imm(_ivec_mullo_s16(vt[1],vt[3]),8);
+ vt[0] = _ivec_add_s8(_ivec_u8_from_u16(vt[0],vt[1]),vsrc);
+ vt[1] = _ivec_blend_u8(vdest,vt[0],vmsk);
+ if(finalize)
+ _ivec_stream(&dstbase[x],vt[1]);
+ else
+ _ivec_storea(&dstbase[x],vt[1]);
+ }
+#endif
+ for(;x<w;x++){
+ if(srca[x]) dstbase[x]=((dstbase[x]*srca[x])>>8)+src[x];
+ }
+ src+=srcstride;
+ srca+=srcstride;
+ dstbase+=dststride;
+ }
+#ifdef HAVE_INT_PVECTOR
+ _ivec_empty();
+ if(finalize) _ivec_sfence();
+#endif
+PROFILE_END("vo_draw_alpha_yv12");
+ return;
+}
+
+static inline void PVECTOR_RENAME(vo_draw_alpha_yuy2)(int w,int h,const unsigned char* src,const unsigned char *srca, int srcstride, unsigned char* dstbase,int dststride,int finalize){
+ int y;
+#if defined(FAST_OSD) && !defined(HAVE_MMX)
+ w=w>>1;
+#endif
+PROFILE_START();
+ for(y=0;y<h;y++){
+ register int x=0;
+#ifdef HAVE_INT_PVECTOR
+ /* TODO: port this stuff on PVector */
+ __m64 mm[8];
+ _ivec_prefetchw(&dstbase[x]);
+ _ivec_prefetch(&src[x]);
+ _ivec_prefetch(&srca[x]);
+ mm[7]=_mm_setzero_si64();
+ mm[5]=_mm_set1_pi8(0xFF);
+ mm[4]=mm[5];
+ mm[5]=_m_psllwi(mm[5],8);
+ mm[4]=_m_psrlwi(mm[4],8);
+ for(;x<w;x+=4){
+ _ivec_prefetchw(&dstbase[x+_ivec_size()*4]);
+ _ivec_prefetch(&src[x+_ivec_size()*4]);
+ _ivec_prefetch(&srca[x+_ivec_size()*4]);
+ mm[0]=_m_load(&(((char *)dstbase)[x*2]));
+ mm[1]=mm[0];
+ mm[0]=_m_pand(mm[0],mm[4]);
+ mm[2]=_m_load_half(&(((char *)srca)[x]));
+ mm[2]=_m_paddb(mm[2],_m_load(&bFF));
+ mm[2]=_m_punpcklbw(mm[2],mm[7]);
+ mm[0]=_m_pmullw(mm[0],mm[2]);
+ mm[0]=_m_psrlwi(mm[0],8);
+ mm[1]=_m_pand(mm[1],mm[5]);
+ mm[2]=_m_load_half(&(((char *)src)[x]));
+ mm[2]=_m_punpcklbw(mm[2],mm[7]);
+ mm[0]=_m_por(mm[0],mm[1]);
+ mm[0]=_m_paddb(mm[0],mm[2]);
+ if(finalize)
+ _m_movntq(&(((char *)dstbase)[x*2]),mm[0]);
+ else
+ _m_store(&(((char *)dstbase)[x*2]),mm[0]);
+ }
+#endif
+ for(;x<w;x++){
+#ifdef FAST_OSD
+ if(srca[2*x+0]) dstbase[4*x+0]=src[2*x+0];
+ if(srca[2*x+1]) dstbase[4*x+2]=src[2*x+1];
+#else
+ if(srca[x]) {
+ dstbase[2*x]=((dstbase[2*x]*srca[x])>>8)+src[x];
+ dstbase[2*x+1]=((((signed)dstbase[2*x+1]-128)*srca[x])>>8)+128;
+ }
+#endif
+ }
+ src+=srcstride;
+ srca+=srcstride;
+ dstbase+=dststride;
+ }
+#ifdef HAVE_INT_PVECTOR
+ _ivec_empty();
+ if(finalize) _ivec_sfence();
+#endif
+PROFILE_END("vo_draw_alpha_yuy2");
+ return;
+}
+
+static inline void PVECTOR_RENAME(vo_draw_alpha_rgb24)(int w,int h,const unsigned char* src,const unsigned char *srca, int srcstride, unsigned char* dstbase,int dststride,int finalize){
+ int y;
+ for(y=0;y<h;y++){
+ register unsigned char *dst = dstbase;
+ register int x=0;
+#ifdef HAVE_INT_PVECTOR
+ __m64 mm[8];
+ _ivec_prefetchw(&dstbase[x]);
+ _ivec_prefetch(&src[x]);
+ _ivec_prefetch(&srca[x]);
+ mm[7]=_mm_setzero_si64();
+ mm[6]=_mm_set1_pi8(0xFF);
+ for(;x<w;x+=2){
+ if(srca[x] || srca[x+1]) {
+ _ivec_prefetchw(&dstbase[x+_ivec_size()*4]);
+ _ivec_prefetch(&src[x+_ivec_size()*4]);
+ _ivec_prefetch(&srca[x+_ivec_size()*4]);
+ mm[0]=_m_load(&dstbase[0]);
+ mm[1]=mm[0];
+ mm[5]=mm[0];
+ mm[0]=_m_punpcklbw(mm[0],mm[7]);
+ mm[1]=_m_punpckhbw(mm[1],mm[7]);
+ mm[2]=_m_load_half(&srca[x]);
+ mm[2]=_m_paddb(mm[2],mm[6]);
+ mm[2]=_m_punpcklbw(mm[2],mm[2]);
+ mm[2]=_m_punpcklbw(mm[2],mm[2]);
+ mm[3]=mm[2];
+ mm[2]=_m_punpcklbw(mm[2],mm[7]);
+ mm[3]=_m_punpckhbw(mm[3],mm[7]);
+ mm[0]=_m_pmullw(mm[0],mm[2]);
+ mm[1]=_m_pmullw(mm[1],mm[3]);
+ mm[0]=_m_psrlwi(mm[0],8);
+ mm[1]=_m_psrlwi(mm[1],8);
+ mm[0]=_m_packuswb(mm[0],mm[1]);
+ mm[2]=_m_load_half(&src[x]);
+ mm[2]=_m_punpcklbw(mm[2],mm[2]);
+ mm[2]=_m_punpcklbw(mm[2],mm[2]);
+ mm[0]=_m_paddb(mm[0],mm[2]);
+ mm[5]=_m_pand(mm[5],_m_load(&mask24lh));
+ mm[0]=_m_pand(mm[0],_m_load(&mask24hl));
+ mm[5]=_m_por(mm[5],mm[0]);
+ if(finalize)
+ _m_movntq(&dstbase[0],mm[5]);
+ else
+ _m_store(&dstbase[0],mm[5]);
+ }
+ dst += 6;
+ }
+#endif /* arch_x86 */
+ for(;x<w;x++){
+ if(srca[x]){
+#ifdef FAST_OSD
+ dst[0]=dst[1]=dst[2]=src[x];
+#else
+ dst[0]=((dst[0]*srca[x])>>8)+src[x];
+ dst[1]=((dst[1]*srca[x])>>8)+src[x];
+ dst[2]=((dst[2]*srca[x])>>8)+src[x];
+#endif
+ }
+ dst+=3; // 24bpp
+ }
+ src+=srcstride;
+ srca+=srcstride;
+ dstbase+=dststride;
+ }
+#ifdef HAVE_INT_PVECTOR
+ _ivec_empty();
+ if(finalize) _ivec_sfence();
+#endif
+ return;
+}
+
+static inline void PVECTOR_RENAME(vo_draw_alpha_rgb32)(int w,int h,const unsigned char* src,const unsigned char *srca, int srcstride, unsigned char* dstbase,int dststride,int finalize){
+ int y;
+PROFILE_START();
+ for(y=0;y<h;y++){
+ register int x=0;
+#ifdef HAVE_INT_PVECTOR
+ __m64 mm[8];
+ _ivec_prefetchw(&dstbase[x]);
+ _ivec_prefetch(&src[x]);
+ _ivec_prefetch(&srca[x]);
+ mm[7]=_mm_setzero_si64();
+ mm[6]=_mm_set1_pi8(0xFF);
+ for(;x<w;x+=2){
+ if(srca[x] || srca[x+1]) {
+ _ivec_prefetchw(&dstbase[x+_ivec_size()*4]);
+ _ivec_prefetch(&src[x+_ivec_size()*4]);
+ _ivec_prefetch(&srca[x+_ivec_size()*4]);
+ mm[0]=_m_load(&dstbase[4*x]);
+ mm[1]=mm[0];
+ mm[0]=_m_punpcklbw(mm[0],mm[7]);
+ mm[1]=_m_punpckhbw(mm[1],mm[7]);
+ mm[2]=_m_load_half(&srca[x]);
+ mm[2]=_m_paddb(mm[2],mm[6]);
+ mm[2]=_m_punpcklbw(mm[2],mm[2]);
+ mm[2]=_m_punpcklbw(mm[2],mm[2]);
+ mm[3]=mm[2];
+ mm[2]=_m_punpcklbw(mm[2],mm[7]);
+ mm[3]=_m_punpckhbw(mm[3],mm[7]);
+ mm[0]=_m_pmullw(mm[0],mm[2]);
+ mm[1]=_m_pmullw(mm[1],mm[3]);
+ mm[0]=_m_psrlwi(mm[0],8);
+ mm[1]=_m_psrlwi(mm[1],8);
+ mm[0]=_m_packuswb(mm[0],mm[1]);
+ mm[2]=_m_load_half(&src[x]);
+ mm[2]=_m_punpcklbw(mm[2],mm[2]);
+ mm[2]=_m_punpcklbw(mm[2],mm[2]);
+ mm[0]=_m_paddb(mm[0],mm[2]);
+ if(finalize)
+ _m_movntq(&dstbase[4*x],mm[0]);
+ else
+ _m_store(&dstbase[4*x],mm[0]);
+ }
+ }
+#endif /* arch_x86 */
+ for(;x<w;x++){
+ if(srca[x]){
+#ifdef FAST_OSD
+ dstbase[4*x+0]=dstbase[4*x+1]=dstbase[4*x+2]=src[x];
+#else
+ dstbase[4*x+0]=((dstbase[4*x+0]*srca[x])>>8)+src[x];
+ dstbase[4*x+1]=((dstbase[4*x+1]*srca[x])>>8)+src[x];
+ dstbase[4*x+2]=((dstbase[4*x+2]*srca[x])>>8)+src[x];
+#endif
+ }
+ }
+ src+=srcstride;
+ srca+=srcstride;
+ dstbase+=dststride;
+ }
+#ifdef HAVE_INT_PVECTOR
+ _ivec_empty();
+ if(finalize) _ivec_sfence();
+#endif
+PROFILE_END("vo_draw_alpha_rgb32");
+ return;
+}
Modified: mplayerxp/libvo/sub.h
===================================================================
--- mplayerxp/libvo/sub.h 2012-11-21 12:54:43 UTC (rev 421)
+++ mplayerxp/libvo/sub.h 2012-11-21 13:21:00 UTC (rev 422)
@@ -3,10 +3,6 @@
#include <inttypes.h>
#include "osd.h"
-#ifdef __cplusplus
-extern "C" {
-#endif
-
typedef struct mp_osd_bbox_s {
int x1,y1,x2,y2;
} mp_osd_bbox_t;
@@ -104,8 +100,5 @@
int __FASTCALL__ vo_osd_changed(int new_value);
int __FASTCALL__ get_osd_height(any_t* vo,int c,int h);
void __FASTCALL__ osd_set_nav_box (uint16_t sx, uint16_t sy, uint16_t ex, uint16_t ey);
-#ifdef __cplusplus
-}
-#endif
#endif
Modified: mplayerxp/libvo/video_out.cpp
===================================================================
--- mplayerxp/libvo/video_out.cpp 2012-11-21 12:54:43 UTC (rev 421)
+++ mplayerxp/libvo/video_out.cpp 2012-11-21 13:21:00 UTC (rev 422)
@@ -51,18 +51,18 @@
//
// Externally visible list of all vo drivers
//
-extern vo_functions_t video_out_x11;
-extern vo_functions_t video_out_xv;
-extern vo_functions_t video_out_dga;
-extern vo_functions_t video_out_sdl;
-extern vo_functions_t video_out_null;
-extern vo_functions_t video_out_pgm;
-extern vo_functions_t video_out_md5;
-extern vo_functions_t video_out_fbdev;
-extern vo_functions_t video_out_png;
-extern vo_functions_t video_out_opengl;
+extern const vo_functions_t video_out_x11;
+extern const vo_functions_t video_out_xv;
+extern const vo_functions_t video_out_dga;
+extern const vo_functions_t video_out_sdl;
+extern const vo_functions_t video_out_null;
+extern const vo_functions_t video_out_pgm;
+extern const vo_functions_t video_out_md5;
+extern const vo_functions_t video_out_fbdev;
+extern const vo_functions_t video_out_png;
+extern const vo_functions_t video_out_opengl;
#ifdef HAVE_VESA
-extern vo_functions_t video_out_vesa;
+extern const vo_functions_t video_out_vesa;
#endif
static const vo_functions_t* video_out_drivers[] =
@@ -454,17 +454,18 @@
uint32_t __FASTCALL__ vo_query_format(vo_data_t*vo,uint32_t* fourcc, unsigned src_w, unsigned src_h)
{
vo_priv_t* priv=(vo_priv_t*)vo->vo_priv;
- uint32_t retval,dri_forced_fourcc;
+ uint32_t dri_forced_fourcc;
+ MPXP_Rc retval;
vo_query_fourcc_t qfourcc;
MSG_DBG3("dri_vo_dbg: vo_query_format(%08lX)\n",*fourcc);
qfourcc.fourcc = *fourcc;
qfourcc.w = src_w;
qfourcc.h = src_h;
- retval = priv->video_out->control(vo,VOCTRL_QUERY_FORMAT,&qfourcc);
- MSG_V("dri_vo: request for %s fourcc: %s\n",vo_format_name(*fourcc),retval?"OK":"False");
+ if(priv->video_out->control(vo,VOCTRL_QUERY_FORMAT,&qfourcc)==MPXP_False)
+ qfourcc.flags=VOCAP_NA;
+ MSG_V("dri_vo: request for %s fourcc: %i\n",vo_format_name(*fourcc),qfourcc.flags);
dri_forced_fourcc = *fourcc;
- if(retval) retval = 0x3; /...
[truncated message content] |
