PC Welt 2004 March

home *** CD-ROM | disk | FTP | other *** search

/ PC Welt 2004 March / PCWELT_3_2004.ISO / pcwsoft / flaskmpeg_078_39_src.z.exe / flaskmpeg / FrameSource.cpp < prev next >

Wrap

C/C++ Source or Header | 2002-10-28 | 46.5 KB | 1,560 lines

/* * FrameSource.cpp * * Copyright (C) Alberto Vigata - July 2000 - ultraflask@yahoo.com * * This file is part of FlasKMPEG, a free MPEG to MPEG/AVI converter * * FlasKMPEG is 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, or (at your option) * any later version. * * FlasKMPEG 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 GNU Make; see the file COPYING. If not, write to * the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA. * */ #include <emmintrin.h> #include "FrameSource.h" #include "flmemcpy.h" #include "debug.h" // SSE2 Global consts static _MM_ALIGN16 const __int64 xmmmask_0001[2] = { 0x0001000100010001, 0x0001000100010001 }; static _MM_ALIGN16 const __int64 xmmmask_0002[2] = { 0x0002000200020002, 0x0002000200020002 }; static _MM_ALIGN16 const __int64 xmmmask_0003[2] = { 0x0003000300030003, 0x0003000300030003 }; static _MM_ALIGN16 const __int64 xmmmask_0004[2] = { 0x0004000400040004, 0x0004000400040004 }; static _MM_ALIGN16 const __int64 xmmmask_0005[2] = { 0x0005000500050005, 0x0005000500050005 }; static _MM_ALIGN16 const __int64 xmmmask_0007[2] = { 0x0007000700070007, 0x0007000700070007 }; static _MM_ALIGN16 const __int64 xmmmask_0016[2] = { 0x0010001000100010, 0x0010001000100010 }; static _MM_ALIGN16 const __int64 xmmmask_0040[2] = { 0x0040004000400040, 0x0040004000400040 }; static _MM_ALIGN16 const __int64 xmmmask_0128[2] = { 0x0080008000800080, 0x0080008000800080 }; static _MM_ALIGN16 const __int64 xmmmask_cbu[2] = { 0x0000408D0000408D, 0x0000408D0000408D }; static _MM_ALIGN16 const __int64 xmmmask_cgu_cgv[2] = { 0xF377E5FCF377E5FC, 0xF377E5FCF377E5FC }; static _MM_ALIGN16 const __int64 xmmmask_crv[2] = { 0x0000331300003313, 0x0000331300003313 }; static _MM_ALIGN16 const __int64 xYUVRGB_Scale[2] = { 0x1000254310002543, 0x1000254310002543 }; // MMX Global consts static _MM_ALIGN16 const __int64 mmmask_0001 = 0x0001000100010001; static _MM_ALIGN16 const __int64 mmmask_0002 = 0x0002000200020002; static _MM_ALIGN16 const __int64 mmmask_0003 = 0x0003000300030003; static _MM_ALIGN16 const __int64 mmmask_0004 = 0x0004000400040004; static _MM_ALIGN16 const __int64 mmmask_0005 = 0x0005000500050005; static _MM_ALIGN16 const __int64 mmmask_0007 = 0x0007000700070007; static _MM_ALIGN16 const __int64 mmmask_0016 = 0x0010001000100010; static _MM_ALIGN16 const __int64 mmmask_0040 = 0x0040004000400040; static _MM_ALIGN16 const __int64 mmmask_0128 = 0x0080008000800080; static _MM_ALIGN16 const __int64 mmmask_cbu = 0x0000408D0000408D; static _MM_ALIGN16 const __int64 mmmask_cgu_cgv = 0xF377E5FCF377E5FC; static _MM_ALIGN16 const __int64 mmmask_crv = 0x0000331300003313; static _MM_ALIGN16 const __int64 YUVRGB_Scale = 0x1000254310002543; void From420to422(unsigned char *src, unsigned char *dst, int width, int height, int frame_type) { int hwidth = width / 2; int dwidth = width * 2; int hheightd2 = height / 2 - 2; int qheightd2 = height / 4 - 2; if (frame_type) { __asm { mov eax, [src] mov ebx, [dst] mov ecx, ebx add ecx, [hwidth] mov esi, 0x00 movq mm3, [mmmask_0003] pxor mm0, mm0 movq mm4, [mmmask_0002] mov edx, eax add edx, [hwidth] convyuv422topp: movd mm1, [eax+esi] movd mm2, [edx+esi] movd [ebx+esi], mm1 punpcklbw mm1, mm0 pmullw mm1, mm3 paddusw mm1, mm4 punpcklbw mm2, mm0 paddusw mm2, mm1 psrlw mm2, 0x02 packuswb mm2, mm0 add esi, 0x04 cmp esi, [hwidth] movd [ecx+esi-4], mm2 jl convyuv422topp add eax, [hwidth] add ebx, [width] add ecx, [width] mov esi, 0x00 mov edi, [hheightd2] convyuv422p: movd mm1, [eax+esi] punpcklbw mm1, mm0 mov edx, eax pmullw mm1, mm3 sub edx, [hwidth] movd mm5, [edx+esi] movd mm2, [edx+esi] punpcklbw mm5, mm0 punpcklbw mm2, mm0 paddusw mm5, mm1 paddusw mm2, mm1 paddusw mm5, mm4 paddusw mm2, mm4 psrlw mm5, 0x02 psrlw mm2, 0x02 packuswb mm5, mm0 packuswb mm2, mm0 mov edx, eax add edx, [hwidth] add esi, 0x04 cmp esi, [hwidth] movd [ebx+esi-4], mm5 movd [ecx+esi-4], mm2 jl convyuv422p add eax, [hwidth] add ebx, [width] add ecx, [width] mov esi, 0x00 sub edi, 0x01 cmp edi, 0x00 jg convyuv422p mov edx, eax sub edx, [hwidth] convyuv422bottomp: movd mm1, [eax+esi] movd mm5, [edx+esi] punpcklbw mm5, mm0 movd [ecx+esi], mm1 punpcklbw mm1, mm0 pmullw mm1, mm3 paddusw mm5, mm1 paddusw mm5, mm4 psrlw mm5, 0x02 packuswb mm5, mm0 add esi, 0x04 cmp esi, [hwidth] movd [ebx+esi-4], mm5 jl convyuv422bottomp emms } } else { __asm { mov eax, [src] mov ecx, [dst] mov esi, 0x00 pxor mm0, mm0 movq mm3, [mmmask_0003] movq mm4, [mmmask_0004] movq mm5, [mmmask_0005] convyuv422topi: movd mm1, [eax+esi] mov ebx, eax add ebx, [hwidth] movd mm2, [ebx+esi] movd [ecx+esi], mm1 punpcklbw mm1, mm0 movq mm6, mm1 pmullw mm1, mm3 punpcklbw mm2, mm0 movq mm7, mm2 pmullw mm2, mm5 paddusw mm2, mm1 paddusw mm2, mm4 psrlw mm2, 0x03 packuswb mm2, mm0 mov edx, ecx add edx, [hwidth] pmullw mm6, mm5 movd [edx+esi], mm2 add ebx, [hwidth] movd mm2, [ebx+esi] punpcklbw mm2, mm0 pmullw mm2, mm3 paddusw mm2, mm6 paddusw mm2, mm4 psrlw mm2, 0x03 packuswb mm2, mm0 add edx, [hwidth] add ebx, [hwidth] pmullw mm7, [mmmask_0007] movd [edx+esi], mm2 movd mm2, [ebx+esi] punpcklbw mm2, mm0 paddusw mm2, mm7 paddusw mm2, mm4 psrlw mm2, 0x03 packuswb mm2, mm0 add edx, [hwidth] add esi, 0x04 cmp esi, [hwidth] movd [edx+esi-4], mm2 jl convyuv422topi add eax, [width] add ecx, [dwidth] mov esi, 0x00 mov edi, [qheightd2] convyuv422i: movd mm1, [eax+esi] punpcklbw mm1, mm0 movq mm6, mm1 mov ebx, eax sub ebx, [width] movd mm3, [ebx+esi] pmullw mm1, [mmmask_0007] punpcklbw mm3, mm0 paddusw mm3, mm1 paddusw mm3, mm4 psrlw mm3, 0x03 packuswb mm3, mm0 add ebx, [hwidth] movq mm1, [ebx+esi] add ebx, [width] movd [ecx+esi], mm3 movq mm3, [mmmask_0003] movd mm2, [ebx+esi] punpcklbw mm1, mm0 pmullw mm1, mm3 punpcklbw mm2, mm0 movq mm7, mm2 pmullw mm2, mm5 paddusw mm2, mm1 paddusw mm2, mm4 psrlw mm2, 0x03 packuswb mm2, mm0 pmullw mm6, mm5 mov edx, ecx add edx, [hwidth] movd [edx+esi], mm2 add ebx, [hwidth] movd mm2, [ebx+esi] punpcklbw mm2, mm0 pmullw mm2, mm3 paddusw mm2, mm6 paddusw mm2, mm4 psrlw mm2, 0x03 packuswb mm2, mm0 pmullw mm7, [mmmask_0007] add edx, [hwidth] add ebx, [hwidth] movd [edx+esi], mm2 movd mm2, [ebx+esi] punpcklbw mm2, mm0 paddusw mm2, mm7 paddusw mm2, mm4 psrlw mm2, 0x03 packuswb mm2, mm0 add edx, [hwidth] add esi, 0x04 cmp esi, [hwidth] movd [edx+esi-4], mm2 jl convyuv422i add eax, [width] add ecx, [dwidth] mov esi, 0x00 sub edi, 0x01 cmp edi, 0x00 jg convyuv422i convyuv422bottomi: movd mm1, [eax+esi] movq mm6, mm1 punpcklbw mm1, mm0 mov ebx, eax sub ebx, [width] movd mm3, [ebx+esi] punpcklbw mm3, mm0 pmullw mm1, [mmmask_0007] paddusw mm3, mm1 paddusw mm3, mm4 psrlw mm3, 0x03 packuswb mm3, mm0 add ebx, [hwidth] movq mm1, [ebx+esi] punpcklbw mm1, mm0 movd [ecx+esi], mm3 pmullw mm1, [mmmask_0003] add ebx, [width] movd mm2, [ebx+esi] punpcklbw mm2, mm0 movq mm7, mm2 pmullw mm2, mm5 paddusw mm2, mm1 paddusw mm2, mm4 psrlw mm2, 0x03 packuswb mm2, mm0 mov edx, ecx add edx, [hwidth] pmullw mm7, [mmmask_0007] movd [edx+esi], mm2 add edx, [hwidth] movd [edx+esi], mm6 punpcklbw mm6, mm0 paddusw mm6, mm7 paddusw mm6, mm4 psrlw mm6, 0x03 packuswb mm6, mm0 add edx, [hwidth] add esi, 0x04 cmp esi, [hwidth] movd [edx+esi-4], mm6 jl convyuv422bottomi emms } } } void From422to444(unsigned char *src, unsigned char *dst, int width, int height) { int hwidthd8 = width / 2 - 8; int hwidth = width / 2; __asm { mov eax, [src] mov ebx, [dst] mov edi, [height] movq mm1, [mmmask_0001] pxor mm0, mm0 convyuv444init: movq mm7, [eax] mov esi, 0x00 convyuv444: movq mm2, mm7 movq mm7, [eax+esi+8] movq mm3, mm2 movq mm4, mm7 psrlq mm3, 8 psllq mm4, 56 por mm3, mm4 movq mm4, mm2 movq mm5, mm3 punpcklbw mm4, mm0 punpcklbw mm5, mm0 movq mm6, mm4 paddusw mm4, mm1 paddusw mm4, mm5 psrlw mm4, 1 psllq mm4, 8 por mm4, mm6 punpckhbw mm2, mm0 punpckhbw mm3, mm0 movq mm6, mm2 paddusw mm2, mm1 paddusw mm2, mm3 movq [ebx+esi*2], mm4 psrlw mm2, 1 psllq mm2, 8 por mm2, mm6 add esi, 0x08 cmp esi, [hwidthd8] movq [ebx+esi*2-8], mm2 jl convyuv444 movq mm2, mm7 punpcklbw mm2, mm0 movq mm3, mm2 psllq mm2, 8 por mm2, mm3 movq [ebx+esi*2], mm2 punpckhbw mm7, mm0 movq mm6, mm7 psllq mm6, 8 por mm6, mm7 movq [ebx+esi*2+8], mm6 add eax, [hwidth] add ebx, [width] sub edi, 0x01 cmp edi, 0x00 jg convyuv444init emms } } static inline void From422toYUY2odd(unsigned char *py, unsigned char *pu, unsigned char *pv, unsigned char *dst, int width, int height) { int dwidth = width * 2; int qwidth = width * 4; int hwidth = width / 2; __asm { mov eax, [py] mov ebx, [pu] mov ecx, [pv] mov edx, [dst] mov esi, 0x00 mov edi, [height] yuy2conv: movd mm2, [ebx+esi] movd mm3, [ecx+esi] punpcklbw mm2, mm3 movq mm1, [eax+esi*2] movq mm4, mm1 punpcklbw mm1, mm2 punpckhbw mm4, mm2 add esi, 0x04 cmp esi, [hwidth] movq [edx+esi*4-16], mm1 movq [edx+esi*4-8], mm4 jl yuy2conv add eax, [dwidth] add ebx, [width] add ecx, [width] add edx, [qwidth] sub edi, 0x02 mov esi, 0x00 cmp edi, 0x00 jg yuy2conv emms } } static inline void From422toYUY2even(unsigned char *py, unsigned char *pu, unsigned char *pv, unsigned char *dst, int width, int height) { int hwidth = width / 2; int dwidth = width * 2; int qwidth = width * 4; py += width; pu += hwidth; pv += hwidth; dst += dwidth; __asm { mov eax, [py] mov ebx, [pu] mov ecx, [pv] mov edx, [dst] mov esi, 0x00 mov edi, [height] yuy2conv: movd mm2, [ebx+esi] movd mm3, [ecx+esi] punpcklbw mm2, mm3 movq mm1, [eax+esi*2] movq mm4, mm1 punpcklbw mm1, mm2 punpckhbw mm4, mm2 add esi, 0x04 cmp esi, [hwidth] movq [edx+esi*4-16], mm1 movq [edx+esi*4-8], mm4 jl yuy2conv add eax, [dwidth] add ebx, [width] add ecx, [width] add edx, [qwidth] sub edi, 0x02 mov esi, 0x00 cmp edi, 0x00 jg yuy2conv emms } } #if 1 //----------------------------------------------------------------------------- // From RGB32 //----------------------------------------------------------------------------- // YUV impacts from RGB #define Y(x) ( (66*((x)>>16) + 129*(((x)&0xFF00)>>8) + 25*((x)&0xFF))/256 + 16 ) #define U(x) ( (112*((x)>>16) - 94*(((x)&0xFF00)>>8) - 18*((x)&0xFF))/256 + 128 ) #define V(x) ( (-38*((x)>>16) - 75*(((x)&0xFF00)>>8) + 112*((x)&0xFF))/256 + 128 ) static inline void FromRGB32toYV12(Pixel32* src, Pixel8 *dst, int w, int h) { int row, col, a, b, c, d; int srcpitch = w * 4; int dstpitch = w; int hw = w>>1; int hh = h>>1; int cralign = 0; Pixel8 *YPtr1, *YPtr2, *UPtr, *VPtr; YPtr1 = dst; YPtr2 = YPtr1 + dstpitch; UPtr = dst + w * h; VPtr = dst + w * h + hw * hh; // The routines computes // the chroma values by averaging // the 4 neibours for(row = 0; row<hh; row++ ) { for(col=0; col<hw; col++) { a = *src; *YPtr1++ = Y(a); b = *(src+1); *YPtr1++ = Y(b); c = *(src + w); *YPtr2++ = Y(c); d = *(src + w + 1); src+= 2; *YPtr2++ = Y(d); *UPtr++ = (U(a) + U(b) + U(c) + U(d))>>2; *VPtr++ = (V(a) + V(b) + V(c) + V(d))>>2; } src += w; UPtr += cralign; VPtr += cralign; YPtr1 += dstpitch; YPtr2 += dstpitch; } } #endif static inline void From420toYV12(unsigned char *py, unsigned char *pu, unsigned char *pv, unsigned char *dst, int width, int height) { int hwidth = width >>1; int hlfheight = height >>1; int dest_addr = 0, src_addr = 0; int size = height * width; int hsize = hlfheight * hwidth; // Y plane flmemcpy( dst, py, size ); // U & V planes for ( int j=0; j< hlfheight; j++ ) // convert U & V planes { src_addr = j * hwidth; dest_addr = size + src_addr; flmemcpy( dst + dest_addr, pv + src_addr, hwidth); dest_addr += hsize; flmemcpy( dst + dest_addr, pu + src_addr, hwidth ); } // endfor j } static inline void FromYV12toYV12int(int top, unsigned char *py, unsigned char *pu, unsigned char *pv, unsigned char *dst, int width, int height) { int hwidth = width >>1; int hlfheight = height >>1; int dest_addr = 0, src_addr = 0; int size = height * width; int hsize = hlfheight * hwidth; int start_row = top ? 0 : 1; for (int j=start_row; j<height ; j+=2 ) // convert y-plane flmemcpy( dst + j*width, py + j*width, width); for ( j=start_row; j< hlfheight; j+=2 ) // convert U & V planes { src_addr = j * hwidth; dest_addr = size + src_addr; flmemcpy( dst + dest_addr, pv + src_addr, hwidth); dest_addr += hsize; flmemcpy( dst + dest_addr, pu + src_addr, hwidth ); } // endfor j } static inline void From444toRGB32odd(unsigned char *py, unsigned char *pu, unsigned char *pv, unsigned char *dst,int width, int height) { dst += width * (height-2) * 4; py += width; pu += width; pv += width; int nwidth = width * 12; int dwidth = width * 2; __asm { mov eax, [py] mov ebx, [pu] mov ecx, [pv] mov edx, [dst] mov edi, [height] mov esi, 0x00 pxor mm0, mm0 convRGB24: movd mm1, [eax+esi] ;mm1: [00][00][00][00][ y][ y][ y][ y] movd mm3, [ebx+esi] ;mm3: [00][00][00][00][ u][ u][ u][ u] punpcklbw mm1, mm0 ;mm1: [00][ y][00][ y][00][ y][00][ y] punpcklbw mm3, mm0 ;mm3: [00][ u][00][ u][00][ u][00][ u] movd mm5, [ecx+esi] ;mm5: [00][00][00][00][ v][ v][ v][ v] punpcklbw mm5, mm0 ;mm5: [00][ v][00][ v][00][ v][00][ v] movq mm7, [mmmask_0128] ;mm7: [ -128][ -128][ -128][ -128] psubw mm3, mm7 ;mm3: [ u-128][ u-128][ u-128][ u-128] psubw mm5, mm7 ;mm5: [ v-128][ v-128][ v-128][ v-128] psubw mm1, [mmmask_0016] ;mm1: [ y-16 ][ y-16 ][ y-16 ][ y-16 ] movq mm2, mm1 ;mm2: [ y-16 ][ y-16 ][ y-16 ][ y-16 ] movq mm7, [mmmask_0001] ;mm7: [0001][0001][0001][0001] punpcklwd mm1, mm7 ;mm1: [0001][ y-16 ][0001][ y-16 ] punpckhwd mm2, mm7 ;mm2: [0001][ y-16 ][0001][ y-16 ] movq mm7, [YUVRGB_Scale] ;mm7: [1000][2000][1000][2000] pmaddwd mm1, mm7 ;mm1: [ y][ y] pmaddwd mm2, mm7 ;mm2: [ y][ y] movq mm4, mm3 ;mm4: [ u-128][ u-128][ u-128][ u-128] punpcklwd mm3, mm0 ;mm3: [ 0000 ][ u-128][ 0000 ][ u-128] punpckhwd mm4, mm0 ;mm4: [ 0000 ][ u-128][ 0000 ][ u-128] movq mm7, [mmmask_cbu] ;mm7: [0000][408D][0000][408D] pmaddwd mm3, mm7 ;mm3: [ mult u ][ mult u ] pmaddwd mm4, mm7 ;mm4: [ mult u ][ mult u ] paddd mm3, mm1 ;mm3: [ mult u +y ][ mult u +y ] paddd mm4, mm2 ;mm4: [ mult u +y ][ mult u +y ] psrld mm3, 13 psrld mm4, 13 packuswb mm3, mm0 ;mm3: [00][00][00][00][00][ b][00][ b] packuswb mm4, mm0 ;mm4: [00][00][00][00][00][ b][00][ b] movq mm6, mm5 ;mm6: [ v-128][ v-128][ v-128][ v-128] punpcklwd mm5, mm0 punpckhwd mm6, mm0 movq mm7, [mmmask_crv] pmaddwd mm5, mm7 pmaddwd mm6, mm7 paddd mm5, mm1 paddd mm6, mm2 psrld mm5, 13 psrld mm6, 13 packuswb mm5, mm0 ; mm5: [00][00][00][00][00][ r][00][ r] packuswb mm6, mm0 ; mm6: [00][00][00][00][00][ r][00][ r] punpcklbw mm3, mm5 ; mm3: [00][00][ r][ b][ 0][ 0][ r][ b] punpcklbw mm4, mm6 ; mm4: [00][00][ r][ b][ 0][ 0][ r][ b] movq mm5, mm3 ; mm5: [00][00][ r][ b][ 0][ 0][ r][ b] movq mm6, mm4 ; mm6: [00][00][ r][ b][ 0][ 0][ r][ b] psrlq mm5, 16 ; mm5: [ r][ b][ 0][ 0][ r][ b][ 0][ 0] psrlq mm6, 16 ; mm6: [ r][ b][ 0][ 0][ r][ b][ 0][ 0] por mm3, mm5 ; mm3: [ r][ b][ r][ b][ r][ b][ r][ b] por mm4, mm6 ; mm4: [ r][ b][ r][ b][ r][ b][ r][ b] movd mm5, [ebx+esi] ; mm5: [00][00][00][00][ u][ u][ u][ u] movd mm6, [ecx+esi] ; mm6: [00][00][00][00][ v][ v][ v][ v] punpcklbw mm5, mm0 ; mm5: [00][ u][00][ u][00][ u][00][ u] punpcklbw mm6, mm0 ; mm6: [00][ v][00][ v][00][ v][00][ v] movq mm7, [mmmask_0128] ; mm7: [00 80][00 80][00 80][00 80] psubw mm5, mm7 ; mm5: [u - 80][u - 80][u - 80][u - 80] psubw mm6, mm7 ; mm6: [v - 80][v - 80][v - 80][v - 80] movq mm7, mm6 ; mm7: [v - 80][v - 80][v - 80][v - 80] punpcklwd mm6, mm5 ; mm6: [u - 80][v - 80][u - 80][v - 80] punpckhwd mm7, mm5 ; mm7: [u - 80][v - 80][u - 80][v - 80] movq mm5, [mmmask_cgu_cgv] ;mm5: [F377][E5FC][F377][E5FC] pmaddwd mm6, mm5 ; mm6: [ madd uv ][ madd uv ] pmaddwd mm7, mm5 ; mm7: [ madd uv ][ madd uv ] paddd mm6, mm1 ; mm6: [ madd uv+y ][ madd uv+y ] paddd mm7, mm2 ; mm6: [ madd uv+y ][ madd uv+y ] psrld mm6, 13 ; mm6: [0000][ g][0000][ g] psrld mm7, 13 ; mm7: [0000][ g][0000][ g] packuswb mm6, mm0 ; mm6: [00][00][00][00][ g][ 0][ g] packuswb mm7, mm0 ; mm7: [00][00][00][00][ g][ 0][ g] punpcklbw mm3, mm6 ; mm3: [ 0][ r][ g][ b][ 0][ r][ g][ b] punpcklbw mm4, mm7 ; mm4: [ 0][ r][ g][ b][ 0][ r][ g][ b] movq [edx], mm3 movq [edx+8], mm4 add edx, 0x10 add esi, 0x04 cmp esi, [width] jl convRGB24 add eax, [dwidth] add ebx, [dwidth] add ecx, [dwidth] sub edx, [nwidth] mov esi, 0x00 sub edi, 0x02 cmp edi, 0x00 jg convRGB24 emms } } static inline void From444toRGB32even(unsigned char *py, unsigned char *pu, unsigned char *pv, unsigned char *dst, int width, int height) { dst += width * (height-1) * 4; int nwidth = width * 12; int dwidth = width * 2; __asm { mov eax, [py] mov ebx, [pu] mov ecx, [pv] mov edx, [dst] mov edi, [height] mov esi, 0x00 pxor mm0, mm0 convRGB24: movd mm1, [eax+esi] movd mm3, [ebx+esi] punpcklbw mm1, mm0 punpcklbw mm3, mm0 movd mm5, [ecx+esi] punpcklbw mm5, mm0 movq mm7, [mmmask_0128] psubw mm3, mm7 psubw mm5, mm7 psubw mm1, [mmmask_0016] movq mm2, mm1 movq mm7, [mmmask_0001] punpcklwd mm1, mm7 punpckhwd mm2, mm7 movq mm7, [YUVRGB_Scale] pmaddwd mm1, mm7 pmaddwd mm2, mm7 movq mm4, mm3 punpcklwd mm3, mm0 punpckhwd mm4, mm0 movq mm7, [mmmask_cbu] pmaddwd mm3, mm7 pmaddwd mm4, mm7 paddd mm3, mm1 paddd mm4, mm2 psrld mm3, 13 psrld mm4, 13 packuswb mm3, mm0 packuswb mm4, mm0 movq mm6, mm5 punpcklwd mm5, mm0 punpckhwd mm6, mm0 movq mm7, [mmmask_crv] pmaddwd mm5, mm7 pmaddwd mm6, mm7 paddd mm5, mm1 paddd mm6, mm2 psrld mm5, 13 psrld mm6, 13 packuswb mm5, mm0 packuswb mm6, mm0 punpcklbw mm3, mm5 punpcklbw mm4, mm6 movq mm5, mm3 movq mm6, mm4 psrlq mm5, 16 psrlq mm6, 16 por mm3, mm5 por mm4, mm6 movd mm5, [ebx+esi] movd mm6, [ecx+esi] punpcklbw mm5, mm0 punpcklbw mm6, mm0 movq mm7, [mmmask_0128] psubw mm5, mm7 psubw mm6, mm7 movq mm7, mm6 punpcklwd mm6, mm5 punpckhwd mm7, mm5 movq mm5, [mmmask_cgu_cgv] pmaddwd mm6, mm5 pmaddwd mm7, mm5 paddd mm6, mm1 paddd mm7, mm2 psrld mm6, 13 psrld mm7, 13 packuswb mm6, mm0 packuswb mm7, mm0 punpcklbw mm3, mm6 punpcklbw mm4, mm7 movq [edx], mm3 movq [edx+8], mm4 add edx, 0x10 add esi, 0x04 cmp esi, [width] jl convRGB24 add eax, [dwidth] add ebx, [dwidth] add ecx, [dwidth] sub edx, [nwidth] mov esi, 0x00 sub edi, 0x02 cmp edi, 0x00 jg convRGB24 emms } } #if 0 // SSE2 full frame version static inline void From444toRGB32_SSE2(unsigned char *py, unsigned char *pu, unsigned char *pv, unsigned char *dst,int width, int height) { dst += width * (height-1) * 4; int nwidth = width * 2 * 4; int wcorr0x10 = (width & 0x7)? 0x10 : 0x00; __asm { mov eax, [py] mov ebx, [pu] mov ecx, [pv] mov edx, [dst] // mov edi, [height] xor esi, esi pxor xmm0, xmm0 lea edi, [esp - 32] and edi, ~0xf convRGB24: movq xmm1, [eax+esi] ;xmm1: [00][00][00][00][00][00][00][00][ y][ y][ y][ y][ y][ y][ y][ y] movq xmm3, [ebx+esi] ;xmm3: [00][00][00][00][00][00][00][00][ u][ u][ u][ u][ u][ u][ u][ u] punpcklbw xmm1, xmm0 ;xmm1: [00][ y][00][ y][00][ y][00][ y][00][ y][00][ y][00][ y][00][ y] punpcklbw xmm3, xmm0 ;xmm3: [00][ u][00][ u][00][ u][00][ u][00][ u][00][ u][00][ u][00][ u] movq xmm5, [ecx+esi] ;xmm5: [00][00][00][00][00][00][00][00][ v][ v][ v][ v][ v][ v][ v][ v] punpcklbw xmm5, xmm0 ;xmm5: [00][ v][00][ v][00][ v][00][ v][00][ v][00][ v][00][ v][00][ v] movdqa xmm7, [xmmmask_0128] ;xmm7: [ +128][ +128][ +128][ +128][ +128][ +128][ +128][ +128] psubw xmm3, xmm7 ;xmm3: [ u-128][ u-128][ u-128][ u-128][ u-128][ u-128][ u-128][ u-128] psubw xmm5, xmm7 ;xmm5: [ v-128][ v-128][ v-128][ v-128][ v-128][ v-128][ v-128][ v-128] movdqa [edi], xmm3 movdqa [edi + 16], xmm5 psubw xmm1, [xmmmask_0016] ;xmm1: [ y-16 ][ y-16 ][ y-16 ][ y-16 ][ y-16 ][ y-16 ][ y-16 ][ y-16 ] movq xmm2, xmm1 ;xmm2: [ y-16 ][ y-16 ][ y-16 ][ y-16 ][ y-16 ][ y-16 ][ y-16 ][ y-16 ] movdqa xmm7, [xmmmask_0001] ;xmm7: [0001][0001][0001][0001][0001][0001][0001][0001] punpcklwd xmm2, xmm7 ;xmm1: [0001][ y-16 ][0001][ y-16 ][0001][ y-16 ][0001][ y-16 ] punpckhwd xmm1, xmm7 ;xmm2: [0001][ y-16 ][0001][ y-16 ][0001][ y-16 ][0001][ y-16 ] movdqa xmm7, [xYUVRGB_Scale] ;xmm7: [1000][2000][1000][2000][1000][2000][1000][2000] pmaddwd xmm2, xmm7 ;xmm2: [ y][ y][ y][ y] pmaddwd xmm1, xmm7 ;xmm1: [ y][ y][ y][ y] movq xmm4, xmm3 ;xmm4: [ u-128][ u-128][ u-128][ u-128][ u-128][ u-128][ u-128][ u-128] punpckhwd xmm3, xmm0 ;xmm3: [ 0000 ][ u-128][ 0000 ][ u-128][ 0000 ][ u-128][ 0000 ][ u-128] punpcklwd xmm4, xmm0 ;xmm4: [ 0000 ][ u-128][ 0000 ][ u-128][ 0000 ][ u-128][ 0000 ][ u-128] movdqa xmm7, [xmmmask_cbu] ;xmm7: [0000][408D][0000][408D][0000][408D][0000][408D] pmaddwd xmm4, xmm7 ;xmm4: [ mult u ][ mult u ][ mult u ][ mult u ] pmaddwd xmm3, xmm7 ;xmm3: [ mult u ][ mult u ][ mult u ][ mult u ] paddd xmm4, xmm2 ;xmm4: [ mult u +y ][ mult u +y ][ mult u +y ][ mult u +y ] paddd xmm3, xmm1 ;xmm3: [ mult u +y ][ mult u +y ][ mult u +y ][ mult u +y ] psrld xmm4, 13 psrld xmm3, 13 packuswb xmm4, xmm3 ;xmm4: [00][ b][00][ b][00][ b][00][ b][00][ b][00][ b][00][ b][00][ b] movq xmm6, xmm5 ;xmm6: [ v-128][ v-128][ v-128][ v-128][ v-128][ v-128][ v-128][ v-128] punpckhwd xmm5, xmm0 punpcklwd xmm6, xmm0 movdqa xmm7, [xmmmask_crv] pmaddwd xmm6, xmm7 pmaddwd xmm5, xmm7 paddd xmm6, xmm2 paddd xmm5, xmm1 psrld xmm6, 13 psrld xmm5, 13 packuswb xmm6, xmm5 ; xmm6: [00][ r][00][ r][00][ r][00][ r][00][ r][00][ r][00][ r][00][ r] movdqa xmm7, [edi + 16] // v... movdqa xmm3, [edi] // u... movq xmm5, xmm7 punpckhwd xmm7, xmm3 ; xmm7: [u - 80][v - 80][u - 80][v - 80][u - 80][v - 80][u - 80][v - 80] punpcklwd xmm5, xmm3 ; xmm5: [u - 80][v - 80][u - 80][v - 80][u - 80][v - 80][u - 80][v - 80] movdqa xmm3, [xmmmask_cgu_cgv] ;xmm6: [F377][E5FC][F377][E5FC][F377][E5FC][F377][E5FC] pmaddwd xmm5, xmm3 ; xmm5: [ madd uv ][ madd uv ][ madd uv ][ madd uv ] pmaddwd xmm7, xmm3 ; xmm7: [ madd uv ][ madd uv ][ madd uv ][ madd uv ] paddd xmm5, xmm2 ; xmm5: [ madd uv+y ][ madd uv+y ][ madd uv+y ][ madd uv+y ] paddd xmm7, xmm1 ; xmm5: [ madd uv+y ][ madd uv+y ][ madd uv+y ][ madd uv+y ] psrld xmm5, 13 ; xmm5: [0000][ g][0000][ g][0000][ g][0000][ g] psrld xmm7, 13 ; xmm7: [0000][ g][0000][ g][0000][ g][0000][ g] packuswb xmm5, xmm7 ; xmm5: [00][ g][00][ g][00][ g][00][ g][00][ g][00][ g][00][ g][00][ g] // xmm6: [00][ r][00][ r][00][ r][00][ r][00][ r][00][ r][00][ r][00][ r] // xmm5: [00][ g][00][ g][00][ g][00][ g][00][ g][00][ g][00][ g][00][ g] // xmm4: [00][ b][00][ b][00][ b][00][ b][00][ b][00][ b][00][ b][00][ b] movq xmm3, xmm4 pxor xmm7, xmm7 punpckhbw xmm4, xmm5 // [00][00][ g][ b][00][00][ g][ b][00][00][ g][ b][00][00][ g][ b] pxor xmm1, xmm1 punpckhwd xmm7, xmm6 // [00][ r][00][00][00][ r][00][00][00][ r][00][00][00][ r][00][00] punpcklbw xmm3, xmm5 // [00][00][ g][ b][00][00][ g][ b][00][00][ g][ b][00][00][ g][ b] punpcklwd xmm1, xmm6 // [00][ r][00][00][00][ r][00][00][00][ r][00][00][00][ r][00][00] por xmm4, xmm7 por xmm3, xmm1 // xmm4: [ 0][ r][ g][ b][ 0][ r][ g][ b][ 0][ r][ g][ b][ 0][ r][ g][ b] // xmm3: [ 0][ r][ g][ b][ 0][ r][ g][ b][ 0][ r][ g][ b][ 0][ r][ g][ b] movntdq [edx], xmm3 movntdq [edx+16], xmm4 add edx, 0x10*2 add esi, 0x04*2 cmp esi, [width] jb convRGB24 sub edx, wcorr0x10 mov esi, height add eax, [width] add ebx, [width] add ecx, [width] sub edx, [nwidth] sub esi, 0x01 //0x02 //cmp esi, 0x00 mov height, esi mov esi, 0 // do not xor here!! jg convRGB24 // emms } } #endif // horizontal filter and 2:1 subsampling // handle, half sample shifts for MPEG1 sampling grid static void conv444to422(ui8 *src, ui8 *dst, int width, int height, int halfsample_shift, ui8 *clp ) { int i, j, im5, im4, im3, im2, im1, ip1, ip2, ip3, ip4, ip5, ip6; if (halfsample_shift) { for (j=0; j<height; j++) { for (i=0; i<width; i+=2) { im5 = (i<5) ? 0 : i-5; im4 = (i<4) ? 0 : i-4; im3 = (i<3) ? 0 : i-3; im2 = (i<2) ? 0 : i-2; im1 = (i<1) ? 0 : i-1; ip1 = (i<width-1) ? i+1 : width-1; ip2 = (i<width-2) ? i+2 : width-1; ip3 = (i<width-3) ? i+3 : width-1; ip4 = (i<width-4) ? i+4 : width-1; ip5 = (i<width-5) ? i+5 : width-1; ip6 = (i<width-5) ? i+6 : width-1; /* FIR filter with 0.5 sample interval phase shift */ dst[i>>1] = clp[(int)(228*(src[i]+src[ip1]) +70*(src[im1]+src[ip2]) -37*(src[im2]+src[ip3]) -21*(src[im3]+src[ip4]) +11*(src[im4]+src[ip5]) + 5*(src[im5]+src[ip6])+256)>>9]; } src+= width; dst+= width>>1; } } else { for (j=0; j<height; j++) { for (i=0; i<width; i+=2) { im5 = (i<5) ? 0 : i-5; im3 = (i<3) ? 0 : i-3; im1 = (i<1) ? 0 : i-1; ip1 = (i<width-1) ? i+1 : width-1; ip3 = (i<width-3) ? i+3 : width-1; ip5 = (i<width-5) ? i+5 : width-1; /* FIR filter coefficients (*512): 22 0 -52 0 159 256 159 0 -52 0 22 */ dst[i>>1] = clp[(int)( 22*(src[im5]+src[ip5])-52*(src[im3]+src[ip3]) +159*(src[im1]+src[ip1])+256*src[i]+256)>>9]; } src+= width; dst+= width>>1; } } } /* vertical filter and 2:1 subsampling */ static void conv422to420(ui8 *src, ui8 *dst, int width, int height, ui8 *clp, int prog_frame ) { int w, i, j, jm6, jm5, jm4, jm3, jm2, jm1; int jp1, jp2, jp3, jp4, jp5, jp6; w = width>>1; height = (height>>1)<<1; if (prog_frame) { /* intra frame */ for (i=0; i<w; i++) { for (j=0; j<height; j+=2) { jm5 = (j<5) ? 0 : j-5; jm4 = (j<4) ? 0 : j-4; jm3 = (j<3) ? 0 : j-3; jm2 = (j<2) ? 0 : j-2; jm1 = (j<1) ? 0 : j-1; jp1 = (j<height-1) ? j+1 : height-1; jp2 = (j<height-2) ? j+2 : height-1; jp3 = (j<height-3) ? j+3 : height-1; jp4 = (j<height-4) ? j+4 : height-1; jp5 = (j<height-5) ? j+5 : height-1; jp6 = (j<height-6) ? j+6 : height-1; /* FIR filter with 0.5 sample interval phase shift */ dst[w*(j>>1)] = clp[(int)(228*(src[w*j]+src[w*jp1]) +70*(src[w*jm1]+src[w*jp2]) -37*(src[w*jm2]+src[w*jp3]) -21*(src[w*jm3]+src[w*jp4]) +11*(src[w*jm4]+src[w*jp5]) + 5*(src[w*jm5]+src[w*jp6])+256)>>9]; } src++; dst++; } } else { /* intra field */ for (i=0; i<w; i++) { for (j=0; j<height; j+=4) { /* top field */ jm5 = (j<10) ? 0 : j-10; jm4 = (j<8) ? 0 : j-8; jm3 = (j<6) ? 0 : j-6; jm2 = (j<4) ? 0 : j-4; jm1 = (j<2) ? 0 : j-2; jp1 = (j<height-2) ? j+2 : height-2; jp2 = (j<height-4) ? j+4 : height-2; jp3 = (j<height-6) ? j+6 : height-2; jp4 = (j<height-8) ? j+8 : height-2; jp5 = (j<height-10) ? j+10 : height-2; jp6 = (j<height-12) ? j+12 : height-2; /* FIR filter with 0.25 sample interval phase shift */ dst[w*(j>>1)] = clp[(int)(8*src[w*jm5] +5*src[w*jm4] -30*src[w*jm3] -18*src[w*jm2] +113*src[w*jm1] +242*src[w*j] +192*src[w*jp1] +35*src[w*jp2] -38*src[w*jp3] -10*src[w*jp4] +11*src[w*jp5] +2*src[w*jp6]+256)>>9]; /* bottom field */ jm6 = (j<9) ? 1 : j-9; jm5 = (j<7) ? 1 : j-7; jm4 = (j<5) ? 1 : j-5; jm3 = (j<3) ? 1 : j-3; jm2 = (j<1) ? 1 : j-1; jm1 = (j<height-1) ? j+1 : height-1; jp1 = (j<height-3) ? j+3 : height-1; jp2 = (j<height-5) ? j+5 : height-1; jp3 = (j<height-7) ? j+7 : height-1; jp4 = (j<height-9) ? j+9 : height-1; jp5 = (j<height-11) ? j+11 : height-1; jp6 = (j<height-13) ? j+13 : height-1; /* FIR filter with 0.25 sample interval phase shift */ dst[w*((j>>1)+1)] = clp[(int)(8*src[w*jp6] +5*src[w*jp5] -30*src[w*jp4] -18*src[w*jp3] +113*src[w*jp2] +242*src[w*jp1] +192*src[w*jm1] +35*src[w*jm2] -38*src[w*jm3] -10*src[w*jm4] +11*src[w*jm5] +2*src[w*jm6]+256)>>9]; } src++; dst++; } } } void OverlayYV12( ui8 *dst, int dstwidth, int dstheight, ui8 *src, ui8 *alphaluma, ui8 *alphachroma, ui32 x, ui32 y, ui32 srcwidth, ui32 srcheight ) { ui32 w, h, i, hw; ui8 *d; ui8 *a; int lumasize = dstwidth*dstheight; int crsize = lumasize>>2; /* Y plane */ h = srcheight; d = dst + y*dstwidth + x; a = alphaluma; int oa, da; /* overlay alpha, dest alpha */ do{ w = srcwidth; do{ oa = *a; da = 0xFF - oa; *d++ = ((*src++)*oa + (*d)*da + 128)>>8; a++; }while(--w); d += dstwidth - srcwidth; }while(--h); /* V U planes */ hw = srcwidth>>1; for( i=0; i<2; i++) { a = alphachroma; h = srcheight>>1; d = dst + lumasize + i*crsize + (y>>1)*(dstwidth>>1) + (x>>1); do{ w = hw; do{ oa = *a; da = 0xFF - oa; *d++ = ((*src++)*oa + (*d)*da + 128)>>8; a++; }while(--w); d += (dstwidth>>1) - hw; }while(--h); } } void CFrame::Overlay( ui32 nX, ui32 nY, CFrame *pOverlay ) { if( !pOverlay ) return; ui8 *a422, *a420; TYUVImage yuv; // Check limits int ow = pOverlay->GetWidth(), oh = pOverlay->GetHeight(); if( (nX+ow) > m_nWidth || (nY+oh) > m_nHeight ){ DBG_STR((str, "CFrame::Overlay - Overlay is too big")); return; } switch( m_nFormat ) { case FRAME_YV12: switch( pOverlay->GetFormat() ) { case FRAME_YV12A: // Create an additional alpha plane with half width and half height // We are using m_pUpsample422 as a memorry allocator // We are using this because CFrame::Set() intelligently allocates memory // only if necessary m_pUpsample422->Set( ow, oh, FRAME_YUV422 ); m_pUpsample422->GetYuvInfo( &yuv ); a422 = yuv.u; // a chroma plane in 422 has the dimensions we want m_pOverlayTemp->Set( ow, oh, FRAME_YV12 ); m_pOverlayTemp->GetYuvInfo( &yuv ); a420 = yuv.u; // a chorma plane in 420 has the dimensions we are looking for // Get the alpha plane pOverlay->GetYuvInfo( &yuv ); // Down convert conv444to422( yuv.a, a422, ow, oh, 0, m_pClp ); conv422to420( a422, a420, ow, oh, m_pClp, 1 ); // Do the overlay with the alpha info OverlayYV12( GetBuffer(), m_nWidth, m_nHeight, pOverlay->GetBuffer(), yuv.a, a420, nX, nY, ow, oh ); break; } break; } } // Set the contents of the frame for the given frame // A conversion in format will be done if needed void CFrame::SetField( CFrame *pFrame, bool bTopField ) { // Resolutions must match if( (m_nWidth != pFrame->GetWidth()) || ( m_nHeight != pFrame->GetHeight()) ) return; int nInputFormat = pFrame->GetFormat(); switch( m_nFormat ) { case FRAME_RGB32: switch( nInputFormat ) { case FRAME_YV12: DBG_STR((str, "SetField - From YV12 to RGB32 not supported")); break; } break; case FRAME_YV12: switch( nInputFormat ) { case FRAME_YV12: TYUVImage in; pFrame->GetYuvInfo( &in ); if(bTopField) FromYV12toYV12int( true, in.y, in.u, in.v, (unsigned char *)m_pData, m_nWidth, m_nHeight); else FromYV12toYV12int( false, in.y, in.u, in.v, (unsigned char *)m_pData, m_nWidth, m_nHeight); break; } break; } } void CFrame::SetFrame( CFrame *pFrame ) { TYUVImage *YUVImage=0; // Resolutions must match if( (m_nWidth != pFrame->GetWidth()) || ( m_nHeight != pFrame->GetHeight()) ) return; int nInputFormat = pFrame->GetFormat(); m_nFrameFlags= pFrame->GetFlags(); switch( m_nFormat ) { case FRAME_RGB32: switch( nInputFormat ) { case FRAME_RGB32: flmemcpy( m_pData, pFrame->GetBuffer(), GetBufferSize() ); break; case FRAME_YV12: // We are going to need temporary frames for the upconversion // Set them. This will allocate necessary memory if not done before. m_pUpsample422->Set( m_nWidth, m_nHeight, FRAME_YUV422 ); m_pUpsample444->Set( m_nWidth, m_nHeight, FRAME_YUV444 ); TYUVImage in, y422, y444; pFrame->GetYuvInfo(&in); m_pUpsample422->GetYuvInfo(&y422); m_pUpsample444->GetYuvInfo(&y444); From420to422(in.u , y422.u , m_nWidth, m_nHeight, pFrame->IsProgressive() ); From420to422(in.v , y422.v , m_nWidth, m_nHeight, pFrame->IsProgressive() ); From422to444(y422.u, y444.u, m_nWidth, m_nHeight); From422to444(y422.v, y444.v, m_nWidth, m_nHeight); From444toRGB32odd(in.y, y444.u ,y444.v, (unsigned char *)m_pData, m_nWidth, m_nHeight); From444toRGB32even(in.y, y444.u, y444.v, (unsigned char *)m_pData, m_nWidth, m_nHeight); break; } break; case FRAME_YV12: switch( nInputFormat ) { case FRAME_YV12: flmemcpy( m_pData, pFrame->GetBuffer(), GetBufferSize() ); break; case FRAME_RGB32: FromRGB32toYV12((Pixel32 *)pFrame->GetBuffer(), (Pixel8 *)GetBuffer(), m_nWidth, m_nHeight); break; } break; case FRAME_YV12A: switch( nInputFormat ) { case FRAME_YUV444A: TYUVImage y444, y422, y420; if( pFrame->GetWidth()%2 || pFrame->GetHeight()%2 ) { DBG_STR((str, "CFrame - Dimensions are not multiple of 2\n")); return; } m_pUpsample422->Set( m_nWidth, m_nHeight, FRAME_YUV422 ); if( !m_pUpsample422->GetBuffer() ) int aa = 0; // Grab pointers pFrame->GetYuvInfo(&y444); m_pUpsample422->GetYuvInfo(&y422); GetYuvInfo( &y420 ); // Down convert conv444to422( y444.u, y422.u, m_nWidth, m_nHeight, 0, m_pClp ); conv444to422( y444.v, y422.v, m_nWidth, m_nHeight, 0, m_pClp ); flmemcpy( y420.y, y444.y, m_nWidth*m_nHeight ); conv422to420( y422.u, y420.u, m_nWidth, m_nHeight, m_pClp, 1 ); conv422to420( y422.v, y420.v, m_nWidth, m_nHeight, m_pClp, 1 ); // copy alpha plane flmemcpy( y420.a, y444.a, m_nWidth*m_nHeight ); break; } break; case FRAME_YUY2: switch( nInputFormat ) { case FRAME_YV12: TYUVImage yuv422, yuv420; // We are going to need temporary frames for the upconversion // Set them. This will allocate necessary memory if not done before. m_pUpsample422->Set( m_nWidth, m_nHeight, FRAME_YUV422 ); pFrame->GetYuvInfo( &yuv420 ); m_pUpsample422->GetYuvInfo( &yuv422 ); From420to422( yuv420.u, yuv422.u, m_nWidth, m_nHeight, pFrame->IsProgressive() ); From420to422( yuv420.v, yuv422.v, m_nWidth, m_nHeight, pFrame->IsProgressive() ); From422toYUY2odd ( yuv420.y, yuv422.u, yuv422.v, m_pData, m_nWidth, m_nHeight ); From422toYUY2even( yuv420.y, yuv422.u, yuv422.v, m_pData, m_nWidth, m_nHeight ); break; } break; case FRAME_YUV444: switch( nInputFormat ) { case FRAME_YV12: // We are going to need temporary frames for the upconversion // Set them. This will allocate necessary memory if not done before. m_pUpsample422->Set( m_nWidth, m_nHeight, FRAME_YUV422 ); TYUVImage in, y422, y444; pFrame->GetYuvInfo(&in); m_pUpsample422->GetYuvInfo(&y422); GetYuvInfo(&y444); From420to422(in.u , y422.u , m_nWidth, m_nHeight, pFrame->IsProgressive() ); From420to422(in.v , y422.v , m_nWidth, m_nHeight, pFrame->IsProgressive() ); flmemcpy( y444.y, in.y, m_nWidth*m_nHeight ); From422to444(y422.u, y444.u, m_nWidth, m_nHeight); From422to444(y422.v, y444.v, m_nWidth, m_nHeight); break; } break; } } ui32 CFrame::GetRequiredMem() { int bufSize; int nChromaSize; int nLumaSize; switch(m_nFormat) { case FRAME_RGB32: bufSize = m_nWidth * m_nHeight * (m_nDepth/8); break; case FRAME_YV12: nChromaSize = (m_nWidth>>1) * (m_nHeight>>1); bufSize = nChromaSize * 6; break; case FRAME_YUY2: bufSize = m_nHeight * (m_nWidth * 2); break; case FRAME_YUV422: nLumaSize = m_nWidth * m_nHeight; nChromaSize = (m_nWidth>>1) * m_nHeight; bufSize = nLumaSize + nChromaSize * 2; break; case FRAME_YUV444: nLumaSize = m_nWidth * m_nHeight; bufSize = 3 * nLumaSize; break; case FRAME_YV12A: nLumaSize = m_nWidth * m_nHeight; nChromaSize = (m_nWidth>>1) * (m_nHeight>>1); bufSize = nChromaSize * 6 + nLumaSize; break; case FRAME_YUV444A: nLumaSize = m_nWidth * m_nHeight; bufSize = 4 * nLumaSize; break; default: bufSize = 0; break; } return bufSize; } bool CFrame::Alloc(ui32 nBufferSize) { m_pUpsample444 = new CFrame; m_pUpsample422 = new CFrame; m_pOverlayTemp = new CFrame; if(!nBufferSize) { DBG_STR((str, "CFrame::Alloc - nBuffersize is zero\n")); m_nAllocatedSize = 0; return false; } m_pData = aligned_new(nBufferSize); if( !m_pData ) { DBG_STR((str, "CFrame::Alloc - allocation of %d bytes failed!\n", nBufferSize)); m_nAllocatedSize = 0; return false; } m_bOwnBuffer = true; m_nAllocatedSize = m_pData ? nBufferSize : 0; return true; } void CFrame::DeAlloc() { if( m_pUpsample444 ){ delete m_pUpsample444; m_pUpsample444 = NULL; } if( m_pUpsample422 ){ delete m_pUpsample422; m_pUpsample422 = NULL; } if( m_pOverlayTemp ){ delete m_pOverlayTemp; m_pOverlayTemp = NULL; } if(m_pData && m_bOwnBuffer) { aligned_delete(m_pData); m_pData = NULL; } m_nAllocatedSize = 0; }