📄 rgb2rgb.c

📁 从FFMPEG转换而来的H264解码程序,VC下编译..
💻 C
📖 第 1 页 / 共 2 页
字号:
12 下一页
/*
 *
 *  rgb2rgb.c, Software RGB to RGB convertor
 *  pluralize by Software PAL8 to RGB convertor
 *               Software YUV to YUV convertor
 *               Software YUV to RGB convertor
 *  Written by Nick Kurshev.
 *  palette & yuv & runtime cpu stuff by Michael (michaelni@gmx.at) (under GPL)
 */
#include <inttypes.h>
#include "../config.h"
#include "rgb2rgb.h"
#include "swscale.h"
#include "../cpudetect.h"
#include "../mangle.h"
#include "../ffmpeg/libavutil/bswap.h"
#include "../libvo/fastmemcpy.h"

#define FAST_BGR2YV12 // use 7 bit coeffs instead of 15bit

void (*rgb24to32)(const uint8_t *src,uint8_t *dst,stride_t src_size);
void (*rgb24to16)(const uint8_t *src,uint8_t *dst,stride_t src_size);
void (*rgb24to15)(const uint8_t *src,uint8_t *dst,stride_t src_size);
void (*rgb32to24)(const uint8_t *src,uint8_t *dst,stride_t src_size);
void (*rgb32to16)(const uint8_t *src,uint8_t *dst,stride_t src_size);
void (*rgb32to15)(const uint8_t *src,uint8_t *dst,stride_t src_size);
void (*rgb15to16)(const uint8_t *src,uint8_t *dst,stride_t src_size);
void (*rgb15to24)(const uint8_t *src,uint8_t *dst,stride_t src_size);
void (*rgb15to32)(const uint8_t *src,uint8_t *dst,stride_t src_size);
void (*rgb16to15)(const uint8_t *src,uint8_t *dst,stride_t src_size);
void (*rgb16to24)(const uint8_t *src,uint8_t *dst,stride_t src_size);
void (*rgb16to32)(const uint8_t *src,uint8_t *dst,stride_t src_size);
//void (*rgb24tobgr32)(const uint8_t *src, uint8_t *dst, stride_t src_size);
void (*rgb24tobgr24)(const uint8_t *src, uint8_t *dst, stride_t src_size);
void (*rgb24tobgr16)(const uint8_t *src, uint8_t *dst, stride_t src_size);
void (*rgb24tobgr15)(const uint8_t *src, uint8_t *dst, stride_t src_size);
void (*rgb32tobgr32)(const uint8_t *src, uint8_t *dst, stride_t src_size);
//void (*rgb32tobgr24)(const uint8_t *src, uint8_t *dst, stride_t src_size);
void (*rgb32tobgr16)(const uint8_t *src, uint8_t *dst, stride_t src_size);
void (*rgb32tobgr15)(const uint8_t *src, uint8_t *dst, stride_t src_size);

void (*yv12toyuy2)(const uint8_t *ysrc, const uint8_t *usrc, const uint8_t *vsrc, uint8_t *dst,
	long width, long height,
	stride_t lumStride, stride_t chromStride, stride_t dstStride);
void (*yv12touyvy)(const uint8_t *ysrc, const uint8_t *usrc, const uint8_t *vsrc, uint8_t *dst,
	long width, long height,
	stride_t lumStride, stride_t chromStride, stride_t dstStride);
void (*yv12tovyuy)(const uint8_t *ysrc, const uint8_t *usrc, const uint8_t *vsrc, uint8_t *dst,
	long width, long height,
	stride_t lumStride, stride_t chromStride, stride_t dstStride);
void (*yv12toyvyu)(const uint8_t *ysrc, const uint8_t *usrc, const uint8_t *vsrc, uint8_t *dst,
	long width, long height,
	stride_t lumStride, stride_t chromStride, stride_t dstStride);
void (*yuv422ptoyuy2)(const uint8_t *ysrc, const uint8_t *usrc, const uint8_t *vsrc, uint8_t *dst,
	long width, long height,
	stride_t lumStride, stride_t chromStride, stride_t dstStride);
void (*yuy2toyv12)(const uint8_t *src, uint8_t *ydst, uint8_t *udst, uint8_t *vdst,
	long width, long height,
	stride_t lumStride, stride_t chromStride, stride_t srcStride);
void (*rgb24toyv12)(const uint8_t *src, uint8_t *ydst, uint8_t *udst, uint8_t *vdst,
	long width, long height,
	stride_t lumStride, stride_t chromStride, stride_t srcStride);
void (*planar2x)(const uint8_t *src, uint8_t *dst, long width, long height,
	stride_t srcStride, stride_t dstStride);
void (*interleaveBytes)(uint8_t *src1, uint8_t *src2, uint8_t *dst,
			    long width, long height, stride_t src1Stride,
			    stride_t src2Stride, stride_t dstStride);
void (*vu9_to_vu12)(const uint8_t *src1, const uint8_t *src2,
			uint8_t *dst1, uint8_t *dst2,
			long width, long height,
			stride_t srcStride1, stride_t srcStride2,
			stride_t dstStride1, stride_t dstStride2);
void (*yvu9_to_yuy2)(const uint8_t *src1, const uint8_t *src2, const uint8_t *src3,
			uint8_t *dst,
			long width, long height,
			stride_t srcStride1, stride_t srcStride2,
			stride_t srcStride3, stride_t dstStride);

#if defined(ARCH_X86) || defined(ARCH_X86_64)
static const uint64_t mmx_null  __attribute__((aligned(8))) = 0x0000000000000000ULL;
static const uint64_t mmx_one   __attribute__((aligned(8))) = 0xFFFFFFFFFFFFFFFFULL;
static const uint64_t mask32b  attribute_used __attribute__((aligned(8))) = 0x000000FF000000FFULL;
static const uint64_t mask32g  attribute_used __attribute__((aligned(8))) = 0x0000FF000000FF00ULL;
static const uint64_t mask32r  attribute_used __attribute__((aligned(8))) = 0x00FF000000FF0000ULL;
static const uint64_t mask32   __attribute__((aligned(8))) = 0x00FFFFFF00FFFFFFULL;
static const uint64_t mask3216br __attribute__((aligned(8)))=0x00F800F800F800F8ULL;
static const uint64_t mask3216g  __attribute__((aligned(8)))=0x0000FC000000FC00ULL;
static const uint64_t mask3215g  __attribute__((aligned(8)))=0x0000F8000000F800ULL;
static const uint64_t mul3216  __attribute__((aligned(8))) = 0x2000000420000004ULL;
static const uint64_t mul3215  __attribute__((aligned(8))) = 0x2000000820000008ULL;
static const uint64_t mask24b  attribute_used __attribute__((aligned(8))) = 0x00FF0000FF0000FFULL;
static const uint64_t mask24g  attribute_used __attribute__((aligned(8))) = 0xFF0000FF0000FF00ULL;
static const uint64_t mask24r  attribute_used __attribute__((aligned(8))) = 0x0000FF0000FF0000ULL;
static const uint64_t mask24l  __attribute__((aligned(8))) = 0x0000000000FFFFFFULL;
static const uint64_t mask24h  __attribute__((aligned(8))) = 0x0000FFFFFF000000ULL;
static const uint64_t mask24hh  __attribute__((aligned(8))) = 0xffff000000000000ULL;
static const uint64_t mask24hhh  __attribute__((aligned(8))) = 0xffffffff00000000ULL;
static const uint64_t mask24hhhh  __attribute__((aligned(8))) = 0xffffffffffff0000ULL;
static const uint64_t mask15b  __attribute__((aligned(8))) = 0x001F001F001F001FULL; /* 00000000 00011111  xxB */
static const uint64_t mask15rg __attribute__((aligned(8))) = 0x7FE07FE07FE07FE0ULL; /* 01111111 11100000  RGx */
static const uint64_t mask15s  __attribute__((aligned(8))) = 0xFFE0FFE0FFE0FFE0ULL;
static const uint64_t mask15g  __attribute__((aligned(8))) = 0x03E003E003E003E0ULL;
static const uint64_t mask15r  __attribute__((aligned(8))) = 0x7C007C007C007C00ULL;
#define mask16b mask15b
static const uint64_t mask16g  __attribute__((aligned(8))) = 0x07E007E007E007E0ULL;
static const uint64_t mask16r  __attribute__((aligned(8))) = 0xF800F800F800F800ULL;
static const uint64_t red_16mask  __attribute__((aligned(8))) = 0x0000f8000000f800ULL;
static const uint64_t green_16mask __attribute__((aligned(8)))= 0x000007e0000007e0ULL;
static const uint64_t blue_16mask __attribute__((aligned(8))) = 0x0000001f0000001fULL;
static const uint64_t red_15mask  __attribute__((aligned(8))) = 0x00007c000000f800ULL;
static const uint64_t green_15mask __attribute__((aligned(8)))= 0x000003e0000007e0ULL;
static const uint64_t blue_15mask __attribute__((aligned(8))) = 0x0000001f0000001fULL;

#ifdef FAST_BGR2YV12
static const uint64_t bgr2YCoeff  attribute_used __attribute__((aligned(8))) = 0x000000210041000DULL;
static const uint64_t bgr2UCoeff  attribute_used __attribute__((aligned(8))) = 0x0000FFEEFFDC0038ULL;
static const uint64_t bgr2VCoeff  attribute_used __attribute__((aligned(8))) = 0x00000038FFD2FFF8ULL;
#else
static const uint64_t bgr2YCoeff  attribute_used __attribute__((aligned(8))) = 0x000020E540830C8BULL;
static const uint64_t bgr2UCoeff  attribute_used __attribute__((aligned(8))) = 0x0000ED0FDAC23831ULL;
static const uint64_t bgr2VCoeff  attribute_used __attribute__((aligned(8))) = 0x00003831D0E6F6EAULL;
#endif
static const uint64_t bgr2YOffset attribute_used __attribute__((aligned(8))) = 0x1010101010101010ULL;
static const uint64_t bgr2UVOffset attribute_used __attribute__((aligned(8)))= 0x8080808080808080ULL;
static const uint64_t w1111       attribute_used __attribute__((aligned(8))) = 0x0001000100010001ULL;

#if 0
static volatile uint64_t __attribute__((aligned(8))) b5Dither;
static volatile uint64_t __attribute__((aligned(8))) g5Dither;
static volatile uint64_t __attribute__((aligned(8))) g6Dither;
static volatile uint64_t __attribute__((aligned(8))) r5Dither;

static uint64_t __attribute__((aligned(8))) dither4[2]={
        0x0103010301030103LL,
        0x0200020002000200LL,};

static uint64_t __attribute__((aligned(8))) dither8[2]={
        0x0602060206020602LL,
        0x0004000400040004LL,};
#endif
#endif /* defined(ARCH_X86) || defined(ARCH_X86_64) */

#define RGB2YUV_SHIFT 8
#define BY ((int)( 0.098*(1<<RGB2YUV_SHIFT)+0.5))
#define BV ((int)(-0.071*(1<<RGB2YUV_SHIFT)+0.5))
#define BU ((int)( 0.439*(1<<RGB2YUV_SHIFT)+0.5))
#define GY ((int)( 0.504*(1<<RGB2YUV_SHIFT)+0.5))
#define GV ((int)(-0.368*(1<<RGB2YUV_SHIFT)+0.5))
#define GU ((int)(-0.291*(1<<RGB2YUV_SHIFT)+0.5))
#define RY ((int)( 0.257*(1<<RGB2YUV_SHIFT)+0.5))
#define RV ((int)( 0.439*(1<<RGB2YUV_SHIFT)+0.5))
#define RU ((int)(-0.148*(1<<RGB2YUV_SHIFT)+0.5))

//Note: we have C, MMX, MMX2, 3DNOW version therse no 3DNOW+MMX2 one
//Plain C versions
#undef HAVE_MMX
#undef HAVE_MMX2
#undef HAVE_3DNOW
#undef HAVE_SSE2
#define RENAME(a) a ## _C
#include "rgb2rgb_template.c"

#if defined(ARCH_X86) || defined(ARCH_X86_64)

//MMX versions
#undef RENAME
#define HAVE_MMX
#undef HAVE_MMX2
#undef HAVE_3DNOW
#undef HAVE_SSE2
#define RENAME(a) a ## _MMX
#include "rgb2rgb_template.c"

//MMX2 versions
#undef RENAME
#define HAVE_MMX
#define HAVE_MMX2
#undef HAVE_3DNOW
#undef HAVE_SSE2
#define RENAME(a) a ## _MMX2
#include "rgb2rgb_template.c"

//3DNOW versions
#undef RENAME
#define HAVE_MMX
#undef HAVE_MMX2
#define HAVE_3DNOW
#undef HAVE_SSE2
#define RENAME(a) a ## _3DNOW
#include "rgb2rgb_template.c"

#endif //ARCH_X86 || ARCH_X86_64

/*
 rgb15->rgb16 Original by Strepto/Astral
 ported to gcc & bugfixed : A'rpi
 MMX2, 3DNOW optimization by Nick Kurshev
 32bit c version, and and&add trick by Michael Niedermayer
*/

void sws_rgb2rgb_init(SwsParams *params){
		yv12toyvyu= yv12toyvyu_C;
		yv12tovyuy= yv12tovyuy_C;
#if defined(ARCH_X86) || defined(ARCH_X86_64)
	if(params->cpu & SWS_CPU_CAPS_MMX2){
		rgb15to16= rgb15to16_MMX2;
		rgb15to24= rgb15to24_MMX2;
		rgb15to32= rgb15to32_MMX2;
		rgb16to24= rgb16to24_MMX2;
		rgb16to32= rgb16to32_MMX2;
		rgb16to15= rgb16to15_MMX2;
		rgb24to16= rgb24to16_MMX2;
		rgb24to15= rgb24to15_MMX2;
		rgb24to32= rgb24to32_MMX2;
		rgb32to16= rgb32to16_MMX2;
		rgb32to15= rgb32to15_MMX2;
		rgb32to24= rgb32to24_MMX2;
		rgb24tobgr15= rgb24tobgr15_MMX2;
		rgb24tobgr16= rgb24tobgr16_MMX2;
		rgb24tobgr24= rgb24tobgr24_MMX2;
		rgb32tobgr32= rgb32tobgr32_MMX2;
		rgb32tobgr16= rgb32tobgr16_MMX2;
		rgb32tobgr15= rgb32tobgr15_MMX2;
		yv12toyuy2= yv12toyuy2_MMX2;
		yv12touyvy= yv12touyvy_MMX2;
		yuv422ptoyuy2= yuv422ptoyuy2_MMX2;
		yuy2toyv12= yuy2toyv12_MMX2;
//		uyvytoyv12= uyvytoyv12_MMX2;
//		yvu9toyv12= yvu9toyv12_MMX2;
		planar2x= planar2x_MMX2;
		rgb24toyv12= rgb24toyv12_MMX2;
		interleaveBytes= interleaveBytes_MMX2;
		vu9_to_vu12= vu9_to_vu12_MMX2;
		yvu9_to_yuy2= yvu9_to_yuy2_MMX2;
	}else if(params->cpu & SWS_CPU_CAPS_3DNOW){
		rgb15to16= rgb15to16_3DNOW;
		rgb15to24= rgb15to24_3DNOW;
		rgb15to32= rgb15to32_3DNOW;
		rgb16to24= rgb16to24_3DNOW;
		rgb16to32= rgb16to32_3DNOW;
		rgb16to15= rgb16to15_3DNOW;
		rgb24to16= rgb24to16_3DNOW;
		rgb24to15= rgb24to15_3DNOW;
		rgb24to32= rgb24to32_3DNOW;
		rgb32to16= rgb32to16_3DNOW;
		rgb32to15= rgb32to15_3DNOW;
		rgb32to24= rgb32to24_3DNOW;
		rgb24tobgr15= rgb24tobgr15_3DNOW;
		rgb24tobgr16= rgb24tobgr16_3DNOW;
		rgb24tobgr24= rgb24tobgr24_3DNOW;
		rgb32tobgr32= rgb32tobgr32_3DNOW;
		rgb32tobgr16= rgb32tobgr16_3DNOW;
		rgb32tobgr15= rgb32tobgr15_3DNOW;
		yv12toyuy2= yv12toyuy2_3DNOW;
		yv12touyvy= yv12touyvy_3DNOW;
		yuv422ptoyuy2= yuv422ptoyuy2_3DNOW;
		yuy2toyv12= yuy2toyv12_3DNOW;
//		uyvytoyv12= uyvytoyv12_3DNOW;
//		yvu9toyv12= yvu9toyv12_3DNOW;
		planar2x= planar2x_3DNOW;
		rgb24toyv12= rgb24toyv12_3DNOW;
		interleaveBytes= interleaveBytes_3DNOW;
		vu9_to_vu12= vu9_to_vu12_3DNOW;
		yvu9_to_yuy2= yvu9_to_yuy2_3DNOW;
	}else if(params->cpu & SWS_CPU_CAPS_MMX){
		rgb15to16= rgb15to16_MMX;
		rgb15to24= rgb15to24_MMX;
		rgb15to32= rgb15to32_MMX;
		rgb16to24= rgb16to24_MMX;
		rgb16to32= rgb16to32_MMX;
		rgb16to15= rgb16to15_MMX;
		rgb24to16= rgb24to16_MMX;
		rgb24to15= rgb24to15_MMX;
		rgb24to32= rgb24to32_MMX;
		rgb32to16= rgb32to16_MMX;
		rgb32to15= rgb32to15_MMX;
		rgb32to24= rgb32to24_MMX;
		rgb24tobgr15= rgb24tobgr15_MMX;
		rgb24tobgr16= rgb24tobgr16_MMX;
		rgb24tobgr24= rgb24tobgr24_MMX;
		rgb32tobgr32= rgb32tobgr32_MMX;
		rgb32tobgr16= rgb32tobgr16_MMX;
		rgb32tobgr15= rgb32tobgr15_MMX;
		yv12toyuy2= yv12toyuy2_MMX;
		yv12touyvy= yv12touyvy_MMX;
		yuv422ptoyuy2= yuv422ptoyuy2_MMX;
		yuy2toyv12= yuy2toyv12_MMX;
//		uyvytoyv12= uyvytoyv12_MMX;
//		yvu9toyv12= yvu9toyv12_MMX;
		planar2x= planar2x_MMX;
		rgb24toyv12= rgb24toyv12_MMX;
		interleaveBytes= interleaveBytes_MMX;
		vu9_to_vu12= vu9_to_vu12_MMX;
		yvu9_to_yuy2= yvu9_to_yuy2_MMX;
	}else
#endif /* defined(ARCH_X86) || defined(ARCH_X86_64) */
	{
		rgb15to16= rgb15to16_C;
		rgb15to24= rgb15to24_C;
		rgb15to32= rgb15to32_C;
		rgb16to24= rgb16to24_C;
		rgb16to32= rgb16to32_C;
		rgb16to15= rgb16to15_C;
		rgb24to16= rgb24to16_C;
		rgb24to15= rgb24to15_C;
		rgb24to32= rgb24to32_C;
		rgb32to16= rgb32to16_C;
		rgb32to15= rgb32to15_C;
		rgb32to24= rgb32to24_C;
		rgb24tobgr15= rgb24tobgr15_C;
		rgb24tobgr16= rgb24tobgr16_C;
		rgb24tobgr24= rgb24tobgr24_C;
		rgb32tobgr32= rgb32tobgr32_C;
		rgb32tobgr16= rgb32tobgr16_C;
		rgb32tobgr15= rgb32tobgr15_C;
		yv12toyuy2= yv12toyuy2_C;
		yv12touyvy= yv12touyvy_C;
		yuv422ptoyuy2= yuv422ptoyuy2_C;
		yuy2toyv12= yuy2toyv12_C;
//		uyvytoyv12= uyvytoyv12_C;
//		yvu9toyv12= yvu9toyv12_C;
		planar2x= planar2x_C;
		rgb24toyv12= rgb24toyv12_C;
		interleaveBytes= interleaveBytes_C;
		vu9_to_vu12= vu9_to_vu12_C;
		yvu9_to_yuy2= yvu9_to_yuy2_C;
	}
12 下一页
⌨️ 快捷键说明

复制代码 Ctrl + C
搜索代码 Ctrl + F
全屏模式 F11
切换主题 Ctrl + Shift + D
显示快捷键 ?
增大字号 Ctrl + =
减小字号 Ctrl + -