From 8d8a2b7939502b86a22e0ec00eecbc2298dcfac0 Mon Sep 17 00:00:00 2001 From: Gernot Ziegler Date: Tue, 28 Oct 2003 18:54:20 +0000 Subject: welcome to glsink for 0.7.x :) Original commit message from CVS: welcome to glsink for 0.7.x :) --- sys/glsink/regcomb_yuvrgb.c | 349 ++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 349 insertions(+) create mode 100644 sys/glsink/regcomb_yuvrgb.c (limited to 'sys/glsink/regcomb_yuvrgb.c') diff --git a/sys/glsink/regcomb_yuvrgb.c b/sys/glsink/regcomb_yuvrgb.c new file mode 100644 index 00000000..0c953c6f --- /dev/null +++ b/sys/glsink/regcomb_yuvrgb.c @@ -0,0 +1,349 @@ +// these includes don't do a lot in Linux, they are more needed in Win32 (for the OpenGL function call pointers) +// but they are used at least for checking if the necessary extensions are present +#include "ARB_multitexture.h" +#include "NV_register_combiners.h" +#include "EXT_paletted_texture.h" + +/*********************************************************************************************** + INTERESTING PART: * + handles initialization of the Nvidia register combiners for YUV->RGB conversion. + This code was created by Jens Schneider + ***********************************************************************************************/ + +GLuint Yhandle; +GLuint Uhandle; +GLuint Vhandle; +unsigned char *YPlane; +unsigned char *UPlane; +unsigned char *VPlane; + +// YUV 4:2:2 example +unsigned int Ywidth=512, Yheight=512; +unsigned int UVwidth=256, UVheight=512; + +int tex_xsize, tex_ysize; + +void GenerateRGBTables(unsigned char *Ytable, // Y-palette + unsigned char *Utable, // U-palette + unsigned char *Vtable, // V-palette + float *bias, // bias (fourth vector to be added) + float *Uscale, // scaling color for U + float *Vscale) // scaling color for V +{ + int i; + const float mat[9]= { // the modified YUV->RGB matrix + +1.130469478f,-0.058755723f,+1.596026304f, + +1.130469478f,-0.450515935f,-0.812967512f, + +1.130469478f,+1.958477882f, 0.0f + }; + #define COMPRESS(a)(0.5f*(a)+128.0f) // counter-piece to EXPAND_NORMAL + #define fCOMPRESS(a) (0.5f*(a)+0.5f); + #define XCLAMP(a) ((a)<0.0f ? 0.0f : ((a)>255.0f ? 255.0f : (a))) // should not be necessary, but what do you know. + bias[0]=fCOMPRESS(-0.842580964f); + bias[1]=fCOMPRESS(+0.563287723f); + bias[2]=fCOMPRESS(-1.0f); + bias[3]=0.0f; + Uscale[0]=8.0f/255.0f; + Uscale[1]=60.0f/255.0f; + Uscale[2]=250.0f/255.0f; + Uscale[3]=0.0f; + Vscale[0]=204.0f/255.0f; + Vscale[1]=105.0f/255.0f; + Vscale[2]=0.5f; + Vscale[3]=0.0f; + for (i=0; i<256; i++) { + // Y-table holds unsigned values + Ytable[3*i ]=(unsigned char)XCLAMP(mat[0]*(float)i); // R + Ytable[3*i+1]=(unsigned char)XCLAMP(mat[3]*(float)i); // G + Ytable[3*i+2]=(unsigned char)XCLAMP(mat[6]*(float)i); // B + // U-table holds signed values + Utable[3*i ]=(unsigned char)XCLAMP(COMPRESS(255.0f/ 16.0f*mat[1]*(float)i)); // R + Utable[3*i+1]=(unsigned char)XCLAMP(COMPRESS(255.0f/120.0f*mat[4]*(float)i)); // G + Utable[3*i+2]=(unsigned char)XCLAMP(COMPRESS(255.0f/500.0f*mat[7]*(float)i)); // B + // V-table holds signed values + Vtable[3*i ]=(unsigned char)XCLAMP(COMPRESS(255.0f/408.0f*mat[2]*(float)i)); // R + Vtable[3*i+1]=(unsigned char)XCLAMP(COMPRESS(255.0f/210.0f*mat[5]*(float)i)); // G + Vtable[3*i+2]=(unsigned char)(128.0f-14.0f); // G constant + } + #undef fCOMPRESS + #undef COMPRESS + #undef XCLAMP +} + + +// Sets the constants. Call once prior to rendering. +void SetConsts(float *bias, float *Uscale, float *Vscale) { + glEnable(GL_REGISTER_COMBINERS_NV); + glColor3fv(bias); + //printf("%f %f %f\n",bias[0],bias[1],bias[2]); + glCombinerParameterfvNV(GL_CONSTANT_COLOR0_NV,Uscale); + glCombinerParameterfvNV(GL_CONSTANT_COLOR1_NV,Vscale); +} + +/* + * SOFTWARE PATH + */ + +inline void map_EXPAND_NORMAL(float *v) { + v[0]=2.0f*v[0]-1.0f; + v[1]=2.0f*v[1]-1.0f; + v[2]=2.0f*v[2]-1.0f; +} + +inline void map_UNSIGNED_INVERT(float *v) { + v[0]=1.0f-v[0]; + v[1]=1.0f-v[1]; + v[2]=1.0f-v[2]; +} + +inline void map_UNSIGNED_IDENTITY(float *v) { + v[0]=(v[0]<0.0f ? 0.0f : v[0]); + v[1]=(v[1]<0.0f ? 0.0f : v[1]); + v[2]=(v[2]<0.0f ? 0.0f : v[2]); +} + +inline void map_SIGNED_IDENTITY(float *v) { +} + +inline void omap_SCALE_BY_TWO(float *v) { + v[0]*=2.0f; + v[1]*=2.0f; + v[2]*=2.0f; +} + +inline void omap_SCALE_BY_ONE_HALF(float *v) { + v[0]*=0.5f; + v[1]*=0.5f; + v[2]*=0.5f; +} + +inline void omap_RANGE(float *v) { + v[0]=(v[0]<-1.0f ? -1.0f : (v[0]>1.0f ? 1.0f : v[0])); + v[1]=(v[1]<-1.0f ? -1.0f : (v[1]>1.0f ? 1.0f : v[1])); + v[2]=(v[2]<-1.0f ? -1.0f : (v[2]>1.0f ? 1.0f : v[2])); +} + + +inline void omap_CLAMP_01(float *v) { + v[0]=(v[0]<0.0f ? 0.0f : (v[0]>1.0f ? 1.0f : v[0])); + v[1]=(v[1]<0.0f ? 0.0f : (v[1]>1.0f ? 1.0f : v[1])); + v[2]=(v[2]<0.0f ? 0.0f : (v[2]>1.0f ? 1.0f : v[2])); +} + +void PerformSWCombiner(unsigned char *Result, + unsigned char *tex0, + unsigned char *tex1, + unsigned char *tex2, + float *COLOR0, + float *CONST0, + float *CONST1 + ) +{ + float SPARE0[3]; + float SPARE1[3]; + float A[3],B[3],C[3],D[3]; + float TEX0[3],TEX1[3],TEX2[3]; + float ZERO[3]={0.0f,0.0f,0.0f}; + + TEX0[0]=(float)tex0[0]/255.0f; + TEX0[1]=(float)tex0[1]/255.0f; + TEX0[2]=(float)tex0[2]/255.0f; + + TEX1[0]=(float)tex1[0]/255.0f; + TEX1[1]=(float)tex1[1]/255.0f; + TEX1[2]=(float)tex1[2]/255.0f; + + TEX2[0]=(float)tex2[0]/255.0f; + TEX2[1]=(float)tex2[1]/255.0f; + TEX2[2]=(float)tex2[2]/255.0f; + + // Combiner Stage 0: + memcpy(A,TEX0,3*sizeof(float)); map_UNSIGNED_IDENTITY(A); + memcpy(B,ZERO,3*sizeof(float)); map_UNSIGNED_INVERT(B); + memcpy(C,COLOR0,3*sizeof(float)); map_EXPAND_NORMAL(C); + memcpy(D,ZERO,3*sizeof(float)); map_UNSIGNED_INVERT(D); + SPARE0[0]=A[0]*B[0]+C[0]*D[0]; + SPARE0[1]=A[1]*B[1]+C[1]*D[1]; + SPARE0[2]=A[2]*B[2]+C[2]*D[2]; + omap_SCALE_BY_ONE_HALF(SPARE0); + omap_RANGE(SPARE0); + + // Combiner Stage 1: + memcpy(A,TEX1,3*sizeof(float)); map_EXPAND_NORMAL(A); + memcpy(B,CONST0,3*sizeof(float)); map_UNSIGNED_IDENTITY(B); + memcpy(C,TEX2,3*sizeof(float)); map_EXPAND_NORMAL(C); + memcpy(D,CONST1,3*sizeof(float)); map_UNSIGNED_IDENTITY(D); + SPARE1[0]=A[0]*B[0]+C[0]*D[0]; + SPARE1[1]=A[1]*B[1]+C[1]*D[1]; + SPARE1[2]=A[2]*B[2]+C[2]*D[2]; + omap_RANGE(SPARE1); + + // Combiner Stage 2: + memcpy(A,SPARE0,3*sizeof(float)); map_SIGNED_IDENTITY(A); + memcpy(B,ZERO,3*sizeof(float)); map_UNSIGNED_INVERT(B); + memcpy(C,SPARE1,3*sizeof(float)); map_SIGNED_IDENTITY(C); + memcpy(D,ZERO,3*sizeof(float)); map_UNSIGNED_INVERT(D); + SPARE0[0]=A[0]*B[0]+C[0]*D[0]; + SPARE0[1]=A[1]*B[1]+C[1]*D[1]; + SPARE0[2]=A[2]*B[2]+C[2]*D[2]; + omap_SCALE_BY_TWO(SPARE0); + omap_RANGE(SPARE0); + + // Final Combiner Stage: + memcpy(A,ZERO,3*sizeof(float)); map_UNSIGNED_INVERT(A); + memcpy(B,SPARE0,3*sizeof(float)); map_UNSIGNED_IDENTITY(B); + memcpy(C,ZERO,3*sizeof(float)); map_UNSIGNED_IDENTITY(C); + memcpy(D,ZERO,3*sizeof(float)); map_UNSIGNED_IDENTITY(D); + SPARE0[0]=A[0]*B[0]+(1.0f-A[0])*C[0]+D[0]; + SPARE0[1]=A[1]*B[1]+(1.0f-A[1])*C[1]+D[1]; + SPARE0[2]=A[2]*B[2]+(1.0f-A[2])*C[2]+D[2]; + omap_CLAMP_01(SPARE0); + Result[0]=(unsigned char)(SPARE0[0]*255.0f); + Result[1]=(unsigned char)(SPARE0[1]*255.0f); + Result[2]=(unsigned char)(SPARE0[2]*255.0f); +} + +// Sets up the register combiners. Call once prior to rendering +void SetupCombiners(void) +{ + glCombinerParameteriNV(GL_NUM_GENERAL_COMBINERS_NV,3); + // Combiner Stage 0: th. OK + glCombinerInputNV (GL_COMBINER0_NV,GL_RGB,GL_VARIABLE_A_NV,GL_TEXTURE0_ARB, GL_UNSIGNED_IDENTITY_NV,GL_RGB); + glCombinerInputNV (GL_COMBINER0_NV,GL_RGB,GL_VARIABLE_B_NV,GL_ZERO, GL_UNSIGNED_INVERT_NV, GL_RGB); + glCombinerInputNV (GL_COMBINER0_NV,GL_RGB,GL_VARIABLE_C_NV,GL_PRIMARY_COLOR_NV,GL_EXPAND_NORMAL_NV, GL_RGB); + glCombinerInputNV (GL_COMBINER0_NV,GL_RGB,GL_VARIABLE_D_NV,GL_ZERO, GL_UNSIGNED_INVERT_NV, GL_RGB); + glCombinerOutputNV(GL_COMBINER0_NV,GL_RGB,GL_DISCARD_NV,GL_DISCARD_NV,GL_SPARE0_NV,GL_SCALE_BY_ONE_HALF_NV,GL_NONE,GL_FALSE,GL_FALSE,GL_FALSE); + // Combiner Stage 1: th. OK + glCombinerInputNV (GL_COMBINER1_NV,GL_RGB,GL_VARIABLE_A_NV,GL_TEXTURE1_ARB, GL_EXPAND_NORMAL_NV, GL_RGB); + glCombinerInputNV (GL_COMBINER1_NV,GL_RGB,GL_VARIABLE_B_NV,GL_CONSTANT_COLOR0_NV,GL_UNSIGNED_IDENTITY_NV,GL_RGB); + glCombinerInputNV (GL_COMBINER1_NV,GL_RGB,GL_VARIABLE_C_NV,GL_TEXTURE2_ARB, GL_EXPAND_NORMAL_NV, GL_RGB); + glCombinerInputNV (GL_COMBINER1_NV,GL_RGB,GL_VARIABLE_D_NV,GL_CONSTANT_COLOR1_NV,GL_UNSIGNED_IDENTITY_NV,GL_RGB); + glCombinerOutputNV(GL_COMBINER1_NV,GL_RGB,GL_DISCARD_NV,GL_DISCARD_NV,GL_SPARE1_NV,GL_NONE,GL_NONE,GL_FALSE,GL_FALSE,GL_FALSE); + // Combiner Stage 2: th. OK + glCombinerInputNV (GL_COMBINER2_NV,GL_RGB,GL_VARIABLE_A_NV,GL_SPARE0_NV, GL_SIGNED_IDENTITY_NV, GL_RGB); + glCombinerInputNV (GL_COMBINER2_NV,GL_RGB,GL_VARIABLE_B_NV,GL_ZERO, GL_UNSIGNED_INVERT_NV, GL_RGB); + glCombinerInputNV (GL_COMBINER2_NV,GL_RGB,GL_VARIABLE_C_NV,GL_SPARE1_NV, GL_SIGNED_IDENTITY_NV, GL_RGB); + glCombinerInputNV (GL_COMBINER2_NV,GL_RGB,GL_VARIABLE_D_NV,GL_ZERO, GL_UNSIGNED_INVERT_NV, GL_RGB); + glCombinerOutputNV(GL_COMBINER2_NV,GL_RGB,GL_DISCARD_NV,GL_DISCARD_NV,GL_SPARE0_NV,GL_SCALE_BY_TWO_NV,GL_NONE,GL_FALSE,GL_FALSE,GL_FALSE); + // Final Sage: th. OK + glFinalCombinerInputNV(GL_VARIABLE_A_NV,GL_ZERO, GL_UNSIGNED_INVERT_NV, GL_RGB); + glFinalCombinerInputNV(GL_VARIABLE_B_NV,GL_SPARE0_NV,GL_UNSIGNED_IDENTITY_NV,GL_RGB); + glFinalCombinerInputNV(GL_VARIABLE_C_NV,GL_ZERO, GL_UNSIGNED_IDENTITY_NV,GL_RGB); + glFinalCombinerInputNV(GL_VARIABLE_D_NV,GL_ZERO, GL_UNSIGNED_IDENTITY_NV,GL_RGB); + glFinalCombinerInputNV(GL_VARIABLE_G_NV,GL_ZERO, GL_UNSIGNED_INVERT_NV, GL_ALPHA); +} + + +unsigned int PowerOfTwo(unsigned int i) { + unsigned int bitsum=0; + unsigned int shifts=0; + unsigned int j=(unsigned int)i; + // Check wether i is a power of two - may contain at most one set bit + do { + bitsum+=j&1; + j=j>>1; + ++shifts; + } while (j>0); + if (bitsum==1) return i; + else return (1<