/* $Xorg: mprgb.c,v 1.3 2000/08/17 19:47:54 cpqbld Exp $ */ /**** module mprgb.c ****/ /****************************************************************************** Copyright 1993, 1994, 1998 The Open Group All Rights Reserved. The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software. THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE OPEN GROUP BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. Except as contained in this notice, the name of The Open Group shall not be used in advertising or otherwise to promote the sale, use or other dealings in this Software without prior written authorization from The Open Group. NOTICE This software is being provided by AGE Logic, Inc. under the following license. By obtaining, using and/or copying this software, you agree that you have read, understood, and will comply with these terms and conditions: Permission to use, copy, modify, distribute and sell this software and its documentation for any purpose and without fee or royalty and to grant others any or all rights granted herein is hereby granted, provided that you agree to comply with the following copyright notice and statements, including the disclaimer, and that the same appears on all copies and derivative works of the software and documentation you make. "Copyright 1993, 1994 by AGE Logic, Inc." THIS SOFTWARE IS PROVIDED "AS IS". AGE LOGIC MAKES NO REPRESENTATIONS OR WARRANTIES, EXPRESS OR IMPLIED. By way of example, but not limitation, AGE LOGIC MAKE NO REPRESENTATIONS OR WARRANTIES OF MERCHANTABILITY OR FITNESS FOR ANY PARTICULAR PURPOSE OR THAT THE SOFTWARE DOES NOT INFRINGE THIRD-PARTY PROPRIETARY RIGHTS. AGE LOGIC SHALL BEAR NO LIABILITY FOR ANY USE OF THIS SOFTWARE. IN NO EVENT SHALL EITHER PARTY BE LIABLE FOR ANY INDIRECT, INCIDENTAL, SPECIAL, OR CONSEQUENTIAL DAMAGES, INCLUDING LOSS OF PROFITS, REVENUE, DATA OR USE, INCURRED BY EITHER PARTY OR ANY THIRD PARTY, WHETHER IN AN ACTION IN CONTRACT OR TORT OR BASED ON A WARRANTY, EVEN IF AGE LOGIC LICENSEES HEREUNDER HAVE BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES. The name of AGE Logic, Inc. may not be used in advertising or publicity pertaining to this software without specific, written prior permission from AGE Logic. Title to this software shall at all times remain with AGE Logic, Inc. ***************************************************************************** mprgb.c -- DDXIE Convert To/From RGB elements Larry Hare -- AGE Logic, Inc. August, 1993 *****************************************************************************/ /* $XFree86: xc/programs/Xserver/XIE/mixie/process/mprgb.c,v 3.4 2001/01/17 22:13:12 dawes Exp $ */ #define _XIEC_MPRGB #define _XIEC_PCFRGB #define _XIEC_PCTRGB /* * Include files */ /* * Core X Includes */ #include #include /* * XIE Includes */ #include #include /* * more X server includes. */ #include #include /* * Server XIE Includes */ #include #include #include #include #include #include #include /* * routines referenced by other DDXIE modules */ int miAnalyzeToRGB(); int miAnalyzeFromRGB(); /* * routines used internal to this module */ static int CreateToRGB(); static int InitializeToRGB(); static int SetupToRGB(); static int CreateFromRGB(); static int InitializeFromRGB(); static int SetupFromRGB(); static int ResetRGB(); static int DestroyRGB(); static void ClearRGB(); static void CheckRGB(); static int ActivateRGB(); /* * DDXIE Convert To/From RGB element entry points. */ static ddElemVecRec ToRGBVec = { CreateToRGB, InitializeToRGB, ActivateRGB, (xieIntProc) NULL, ResetRGB, DestroyRGB }; static ddElemVecRec FromRGBVec = { CreateFromRGB, InitializeFromRGB, ActivateRGB, (xieIntProc) NULL, ResetRGB, DestroyRGB }; /* ** Local Declarations. ** NOTE: use #define EARLY_SETUP to do setup at create time ** NOTE: uses RGBFloat's for matrix multiply. */ #define EARLY_SETUP typedef RealPixel RGBFloat; typedef struct _mprgbdef { void (*action) (); /* every one needs these */ void (*post) (); /* clipping, or CIELab, for floats */ CARD32 iclip[3]; /* clip values for integers */ pointer (*cvt_in[3])(); /* !0 if need to expand input */ pointer (*cvt_out[3])(); /* !0 if need to compress output */ pointer aux_buf[3]; /* used for cvt_in/cvt_out */ RGBFloat matrix[12]; INT32 imatrix[12]; } mpRGBPvtRec, *mpRGBPvtPtr; /* ** NOTES on BASIC ALGORITHM: ** 1) All transforms center on a matrix. Either a passed in XYZMAT ** or one synthesized from Luma values. ** 2) White Point adjustments, if specified are applied to this matrix. ** 3) Bias (YCbCr) or Implicit Biasing and Scaling (YCC) are applied. ** (Float to Float stops here, matmul is done as floats) ** 4) If inputs are integers, a scale is applied to reduce input ** input numbers to range 0.0 to 1.0. ** (Int to Float stops here, matmul is done as floats) ** 5) If outputs are integers, a scale is applied to expand values ** back to output range. For tripleband with all 256 levels ** for inputs and outputs, this cancels step 4. But allows ** for mapping say an (8,8,4) image to say (16,4,4). This ** final matrix is turned into a scaled fraction integer matrix. ** (Int to Int stops here, matmul is done as scaled fractions) */ /*------------------------------------------------------------------------ ------------------------ fill in the vector --------------------------- ------------------------------------------------------------------------*/ int miAnalyzeToRGB(flo,ped) floDefPtr flo; peDefPtr ped; { ped->ddVec = ToRGBVec; return TRUE; } int miAnalyzeFromRGB(flo,ped) floDefPtr flo; peDefPtr ped; { ped->ddVec = FromRGBVec; return TRUE; } /*------------------------------------------------------------------------ ---------------------------- create peTex . . . -------------------------- ------------------------------------------------------------------------*/ static int CreateToRGB(flo,ped) floDefPtr flo; peDefPtr ped; { if (!MakePETex(flo,ped,sizeof(mpRGBPvtRec), NO_SYNC, SYNC)) return FALSE; #if defined(EARLY_SETUP) if (!SetupToRGB(flo, ped, 0 /* modify ?? */)) return FALSE; #endif return TRUE; } static int CreateFromRGB(flo,ped) floDefPtr flo; peDefPtr ped; { /* always force syncing between inputs (is nop if only one input) */ if (!MakePETex(flo,ped,sizeof(mpRGBPvtRec), NO_SYNC, SYNC)) return FALSE; #if defined(EARLY_SETUP) if (!SetupFromRGB(flo, ped, 0 /* modify ?? */)) return FALSE; #endif return TRUE; } /*------------------------------------------------------------------------ ---------------------------- initialize peTex . . . ---------------------- ------------------------------------------------------------------------*/ static int InitializeToRGB(flo,ped) floDefPtr flo; peDefPtr ped; { #if !defined(EARLY_SETUP) if (!SetupToRGB(flo, ped, 0 /* modify ?? */)) return FALSE; #endif InitReceptors(flo, ped, NO_DATAMAP, 1); InitEmitter(flo, ped, NO_DATAMAP, NO_INPLACE); return !ferrCode(flo); } static int InitializeFromRGB(flo,ped) floDefPtr flo; peDefPtr ped; { #if !defined(EARLY_SETUP) if (!SetupFromRGB(flo, ped, 0 /* modify ?? */)) return FALSE; #endif InitReceptors(flo, ped, NO_DATAMAP, 1); InitEmitter(flo, ped, NO_DATAMAP, NO_INPLACE); return !ferrCode(flo); } /*------------------------------------------------------------------------ ----------------------------- crank some data ---------------------------- ------------------------------------------------------------------------*/ static int ActivateRGB(flo,ped,pet) floDefPtr flo; peDefPtr ped; peTexPtr pet; { mpRGBPvtPtr pvt = (mpRGBPvtPtr) pet->private; bandPtr sband = &(pet->receptor[SRCt1].band[0]); bandPtr dband = &(pet->emitter[0]); CARD32 npix = sband->format->width; pointer svoid[3], dvoid[3], stvoid[3], dtvoid[3]; CARD32 b; BOOL stop; for (b = 0; b < 3; b++, sband++, dband++) { if (!(svoid[b] = GetCurrentSrc(flo,pet,sband))) return TRUE; if (!(dvoid[b] = GetCurrentDst(flo,pet,dband))) return TRUE; stvoid[b] = pvt->cvt_in[b] ? (*pvt->cvt_in[b]) (pvt->aux_buf[b], svoid[b], pvt, npix) : svoid[b] ; dtvoid[b] = pvt->cvt_out[b] ? pvt->aux_buf[b] : dvoid[b]; } sband -= 3; dband -= 3; do { (*(pvt->action)) (dtvoid, stvoid, pvt, npix); if (pvt->post) (*(pvt->post)) (dtvoid, npix); for (b = 0, stop = FALSE; b < 3; b++, sband++, dband++) { if (pvt->cvt_out[b]) (*pvt->cvt_out[b]) (dvoid[b], dtvoid[b], pvt, npix); stop |= !(svoid[b] = GetNextSrc(flo,pet,sband,FLUSH)); stop |= !(dvoid[b] = GetNextDst(flo,pet,dband,FLUSH)); if (!stop) { dtvoid[b] = pvt->cvt_out[b] ? dtvoid[b] : dvoid[b]; stvoid[b] = pvt->cvt_in[b] ? (*pvt->cvt_in[b]) (stvoid[b], svoid[b], pvt, npix) : svoid[b] ; } } sband -= 3; dband -= 3; } while (!ferrCode(flo) && !stop) ; for (b = 0; b < 3; b++, sband++) { FreeData(flo, pet, sband, sband->current); } sband -= 3; return TRUE; } /*------------------------------------------------------------------------ ------------------------ get rid of run-time stuff ----------------------- ------------------------------------------------------------------------*/ static int ResetRGB(flo,ped) floDefPtr flo; peDefPtr ped; { #if !defined(EARLY_SETUP) ClearRGB(flo, ped); #endif ResetReceptors(ped); ResetProcDomain(ped); ResetEmitter(ped); return TRUE; } /*------------------------------------------------------------------------ -------------------------- get rid of this element ----------------------- ------------------------------------------------------------------------*/ static int DestroyRGB(flo,ped) floDefPtr flo; peDefPtr ped; { #if defined(EARLY_SETUP) ClearRGB(flo, ped); #endif /* get rid of the peTex structure and private structures */ ped->peTex = (peTexPtr) XieFree(ped->peTex); /* zap this element's entry point vector */ ped->ddVec.create = (xieIntProc)NULL; ped->ddVec.initialize = (xieIntProc)NULL; ped->ddVec.activate = (xieIntProc)NULL; ped->ddVec.reset = (xieIntProc)NULL; ped->ddVec.destroy = (xieIntProc)NULL; return TRUE; } /*------------------------------------------------------------------------ ------------------------ Action Utility Routines ----------------------- ------------------------------------------------------------------------*/ /* * Used for CIELab conversions. * * NOTE: Investigate use of stock cbrt() function in libc. * NOTE: Borrowed from lib/X/XcmsMath.h * NOTE: Copyright 1990 The Open Group * * NOTE: Investigate use of stock cbrt() function in libc. * NOTE: Used better choice of seed values with emphasis on * range of 0 to 1, and other smallish numbers. */ #if !defined(XCMS_CUBEROOT) #define XCMS_CUBEROOT _cmsCubeRoot /* Newton's Method: x_n+1 = x_n - ( f(x_n) / f'(x_n) ) */ /* for cube roots, x^3 - a = 0, x_new = x - 1/3 (x - a/x^2) */ #ifndef DBL_EPSILON #define DBL_EPSILON 1e-6 #endif double _cmsCubeRoot(a) double a; { register double abs_a, cur_guess, delta; if (a == 0.) return 0.; /* convert to positive to speed loop tests */ abs_a = a < 0. ? -a : a; /* arbitrary first guess */ cur_guess = abs_a < 1. ? 0.5 + .500 * abs_a : abs_a < 1000. ? 1.0 + .125 * abs_a : 10.0 + .0125 * abs_a ; do { delta = (cur_guess - abs_a/(cur_guess*cur_guess))/3.; cur_guess -= delta; if (delta < 0.) delta = -delta; } while (delta >= cur_guess*DBL_EPSILON); return a > 0. ? cur_guess : -cur_guess; } #endif /*------------------------------------------------------------------------ --------------------- Lotsa Little Action Routines --------------------- ------------------------------------------------------------------------*/ /* (*(pvt->cvt_out)) (dvoid, svoid, pvt, npix); */ /* (*(pvt->cvt_in)) (dvoid, svoid, pvt, npix); */ /* (*(pvt->action)) (dvoid, svoid, pvt, npix); */ /* (*(pvt->post)) (dvoid, npix); */ static pointer cvt_bit_to_pair(dvoid,svoid,pvt,npix) pointer dvoid, svoid; mpRGBPvtPtr pvt; CARD32 npix; { LogInt *bitp = (LogInt *) svoid, ival, M; PairPixel *pairp = (PairPixel *) dvoid; int nw = npix >> LOGSHIFT; for ( ; nw > 0; nw--) for (ival = *bitp++, M=LOGLEFT; M ; LOGRIGHT(M)) *pairp++ = (ival & M) ? (PairPixel) 1 : (PairPixel) 0; if ((npix &= LOGMASK)) for (ival = *bitp, M=LOGLEFT; npix > 0 ; npix--, LOGRIGHT(M)) *pairp++ = (ival & M) ? (PairPixel) 1 : (PairPixel) 0; return dvoid; } static pointer cvt_byte_to_pair(dvoid,svoid,pvt,npix) pointer dvoid, svoid; mpRGBPvtPtr pvt; CARD32 npix; { BytePixel *bytep = (BytePixel *) svoid; PairPixel *pairp = (PairPixel *) dvoid; for ( ; npix > 0; npix--) *pairp++ = *bytep++; return dvoid; } static pointer cvt_pair_to_byte(dvoid,svoid,pvt,npix) pointer dvoid, svoid; mpRGBPvtPtr pvt; CARD32 npix; { PairPixel *pairp = (PairPixel *) svoid; BytePixel *bytep = (BytePixel *) dvoid; for ( ; npix > 0; npix--) *bytep++ = *pairp++; /* already hard-clipped!! */ return dvoid; } static pointer cvt_pair_to_bit(dvoid,svoid,pvt,npix) pointer dvoid, svoid; mpRGBPvtPtr pvt; CARD32 npix; { PairPixel *pairp = (PairPixel *) svoid; LogInt *bitp = (LogInt *) dvoid, bitval, M; for ( ; npix >= LOGSIZE; *bitp++ = bitval, npix -= LOGSIZE) for (M=LOGLEFT, bitval = 0; M; LOGRIGHT(M)) if (*pairp++) bitval |= M; if (npix > 0) { for (M=LOGLEFT, bitval = 0; npix > 0; npix--, LOGRIGHT(M)) if (*pairp++) bitval |= M; *bitp = bitval; } return dvoid; } static void act_mmRR(dvoid, svoid, pvt, npix) RealPixel *svoid[3], *dvoid[3]; mpRGBPvtPtr pvt; CARD32 npix; { RGBFloat *mtrx = &(pvt->matrix[0]); RGBFloat r, g, b; CARD32 i; /* ** dvector = matrix * svector */ for (i = 0; i < npix; i++) { r = svoid[0][i]; g = svoid[1][i]; b = svoid[2][i]; dvoid[0][i] = mtrx[0] * r + mtrx[1] * g + mtrx[2] * b; dvoid[1][i] = mtrx[3] * r + mtrx[4] * g + mtrx[5] * b; dvoid[2][i] = mtrx[6] * r + mtrx[7] * g + mtrx[8] * b; } } #define MakeIntFltMatMul(name, itype) \ static void \ name(dvoid, svoid, pvt, npix) \ itype *svoid[3]; \ RealPixel *dvoid[3]; \ mpRGBPvtPtr pvt; \ CARD32 npix; \ { \ RealPixel *mtrx = &(pvt->matrix[0]); \ RealPixel r, g, b, rr, gg, bb; \ CARD32 i; \ \ for (i = 0; i < npix; i++) { \ r = (RealPixel) svoid[0][i]; \ g = (RealPixel) svoid[1][i]; \ b = (RealPixel) svoid[2][i]; \ rr = mtrx[0] * r + mtrx[1] * g + mtrx[2] * b; \ gg = mtrx[3] * r + mtrx[4] * g + mtrx[5] * b; \ bb = mtrx[6] * r + mtrx[7] * g + mtrx[8] * b; \ dvoid[0][i] = rr; \ dvoid[1][i] = gg; \ dvoid[2][i] = bb; \ } \ } MakeIntFltMatMul(act_mmBR, BytePixel) MakeIntFltMatMul(act_mmPR, PairPixel) #define MakeIntIntMatMul(name, iotype, shift) \ static void \ name(dvoid, svoid, pvt, npix) \ iotype *svoid[3], *dvoid[3]; \ mpRGBPvtPtr pvt; \ CARD32 npix; \ { \ INT32 *mtrx = &(pvt->imatrix[0]); \ INT32 r, g, b, rr, gg, bb; \ CARD32 iclip0 = pvt->iclip[0]; \ CARD32 iclip1 = pvt->iclip[1]; \ CARD32 iclip2 = pvt->iclip[2]; \ INT32 bias0 = mtrx[9] + (1 << (shift-1)); \ INT32 bias1 = mtrx[10] + (1 << (shift-1)); \ INT32 bias2 = mtrx[11] + (1 << (shift-1)); \ CARD32 i; \ \ for (i = 0; i < npix; i++) { \ r = svoid[0][i]; \ g = svoid[1][i]; \ b = svoid[2][i]; \ rr = mtrx[0] * r + mtrx[1] * g + mtrx[2] * b + bias0; \ gg = mtrx[3] * r + mtrx[4] * g + mtrx[5] * b + bias1; \ bb = mtrx[6] * r + mtrx[7] * g + mtrx[8] * b + bias2; \ if (rr < 0) rr = 0; \ if (gg < 0) gg = 0; \ if (bb < 0) bb = 0; \ if ((rr >>= shift) > iclip0) rr = iclip0; \ if ((gg >>= shift) > iclip1) gg = iclip1; \ if ((bb >>= shift) > iclip2) bb = iclip2; \ dvoid[0][i] = rr; \ dvoid[1][i] = gg; \ dvoid[2][i] = bb; \ } \ } #define SF_BYTESHIFT 20 #define SF_PAIRSHIFT 12 MakeIntIntMatMul(act_mmBB, BytePixel, SF_BYTESHIFT) MakeIntIntMatMul(act_mmPP, PairPixel, SF_PAIRSHIFT) static void act_postClipR(dvoid, npix) RealPixel *dvoid[3]; { RGBFloat *fp, f, zero = (RGBFloat) 0.0, one = (RGBFloat) 1.0; CARD32 ipix, band; for (band = 0; band < 3; band++) { fp = dvoid[band]; for (ipix = npix; ipix > 0; ipix--, fp++) { f = *fp; if ( f < zero) *fp = zero; if ( f > one) *fp = one; } } } static void act_postCIELab(dvoid, npix) RealPixel *dvoid[3]; { RGBFloat *xp, *yp, *zp, x3, y3, z3; CARD32 ipix; xp = dvoid[0]; yp = dvoid[1]; zp = dvoid[2]; for (ipix = npix; ipix > 0; ipix--) { /* XXX, if numbers are small, do some more work */ x3 = XCMS_CUBEROOT((double)*xp); y3 = XCMS_CUBEROOT((double)*yp); z3 = XCMS_CUBEROOT((double)*zp); *xp++ = 116.0 * y3 - 16.0; /* L* */ *yp++ = 500.0 * (x3 - y3); /* a* */ *zp++ = 200.0 * (y3 - z3); /* b* */ } } static void act_preCIELab(dvoid, svoid, pvt, npix) RealPixel *svoid[3], *dvoid[3]; mpRGBPvtPtr pvt; CARD32 npix; { RGBFloat *Lp, *ap, *bp, *xp, *yp, *zp, L, a, b; CARD32 ipix; /* must be careful not to smash the source */ Lp = svoid[0]; ap = svoid[1]; bp = svoid[2]; xp = dvoid[0]; yp = dvoid[1]; zp = dvoid[2]; for (ipix = npix; ipix > 0; ipix--) { /* XXX, if numbers are small, do some more work */ L = (*Lp++ + 16.0) * (1.0 / 116.0); a = L + (*ap++ * 0.002); b = L - (*bp++ * 0.005); *yp++ = (L * L * L); /* Y = ... */ *xp++ = (a * a * a); /* X = ... */ *zp++ = (b * b * b); /* Z = ... */ } act_mmRR(dvoid, dvoid, pvt, npix); } /*------------------------------------------------------------------------ -------------------- utility routines for initialization ---------------- ------------------------------------------------------------------------*/ /* Scale entire matrix by single scale factor */ static void scale_mtrx(mtx,imtx,iscl) RGBFloat *mtx; INT32 *imtx; CARD32 iscl; { RGBFloat fscl = iscl; int i; for (i = 0; i < 12; i++) *imtx++ = *mtx++ * fscl; } /* Scale each row of matrix by respective factors */ void scale_rows(mtrx,scale1,scale2,scale3) RGBFloat *mtrx; double scale1, scale2, scale3; { int b; for (b = 0; b < 3; b++) *mtrx++ *= scale1; for (b = 0; b < 3; b++) *mtrx++ *= scale2; for (b = 0; b < 3; b++) *mtrx++ *= scale3; } /* Scale each column of matrix by respective factors */ void scale_columns(mtrx,scale1,scale2,scale3) RGBFloat *mtrx; double scale1, scale2, scale3; { int b; for (b = 0; b < 3; b++) { *mtrx++ *= scale1; *mtrx++ *= scale2; *mtrx++ *= scale3; } } /* ** Our matrix routines for integers use a BIAS like so: ** YCC = MATRIX * RGB - BIAS ** But when we are going back to RGB we would have liked to do: ** RGB = MATRIX * (YCC - BIAS) ** So we multiply do transitive property of matrix arithmetic ** to get a new BIAS' (BIAS' = -MATRIX * BIAS) such that: ** RGB = MATRIX * YCC - MATRIX * BIAS = MATRIX * YCC - BIAS' */ static void flip_bias(mtx) RGBFloat *mtx; { RGBFloat mtx9, mtx10, mtx11; mtx9 = -(mtx[0] * mtx[9] + mtx[1] * mtx[10] + mtx[2] * mtx[11]); mtx10 = -(mtx[3] * mtx[9] + mtx[4] * mtx[10] + mtx[5] * mtx[11]); mtx11 = -(mtx[6] * mtx[9] + mtx[7] * mtx[10] + mtx[8] * mtx[11]); mtx[9] = mtx9; mtx[10] = mtx10; mtx[11] = mtx11; } /* Seed matrix for CIEXYZ and CIELab */ static void copymatrix(pvt,input) mpRGBPvtPtr pvt; double *input; { RGBFloat *mtrx = pvt->matrix; int i; for (i = 0; i < 9; i++) *mtrx++ = *input++; for (; i < 12; i++) *mtrx++ = 0.0; /* bias */ } static void copywhiteLABFromRGB(pvt,tec,vec) mpRGBPvtPtr pvt; xieTypWhiteAdjustTechnique tec; double *vec; { RGBFloat *mtrx = pvt->matrix; double a, b, c; switch (tec) { case xieValWhiteAdjustCIELabShift: a = vec[0]; b = vec[1]; c = vec[2]; break; case xieValWhiteAdjustDefault: case xieValWhiteAdjustNone: /* calculate from pvt->matrix */ a = mtrx[0] + mtrx[1] + mtrx[2]; b = mtrx[3] + mtrx[4] + mtrx[5]; c = mtrx[6] + mtrx[7] + mtrx[8]; break; default: return; } if (a < .0001) return; if (b < .0001) return; if (c < .0001) return; scale_rows (mtrx, 1.0 / a, 1.0 / b, 1.0 / c); return; } static void copywhiteLABToRGB(pvt,tec,vec) mpRGBPvtPtr pvt; xieTypWhiteAdjustTechnique tec; double *vec; { RGBFloat *mtrx = pvt->matrix; double a, b, c; switch (tec) { case xieValWhiteAdjustCIELabShift: a = vec[0]; b = vec[1]; c = vec[2]; break; case xieValWhiteAdjustDefault: case xieValWhiteAdjustNone: { /* White Point comes from XYZ Matrix. Must invert it first */ RGBFloat *m = mtrx - 1; /* number matrix from 1 to 9 */ double determinant = m[1] * (m[5]*m[9] - m[6]*m[8]) - m[2] * (m[4]*m[9] - m[6]*m[7]) + m[3] * (m[4]*m[8] - m[5]*m[7]); a = (m[5]*m[9]-m[6]*m[8]); /* inv[00] */ a += -(m[2]*m[9]-m[3]*m[8]); /* inv[01] */ a += (m[2]*m[6]-m[3]*m[5]); /* inv[02] */ b = -(m[4]*m[9]-m[6]*m[7]); /* inv[10] */ b += (m[1]*m[9]-m[3]*m[7]); /* inv[11] */ b += -(m[1]*m[6]-m[3]*m[4]); /* inv[12] */ c = (m[4]*m[8]-m[5]*m[7]); /* inv[20] */ c += -(m[1]*m[8]-m[2]*m[7]); /* inv[21] */ c += (m[1]*m[5]-m[2]*m[4]); /* inv[22] */ a /= (determinant ? determinant : 1.0); b /= (determinant ? determinant : 1.0); c /= (determinant ? determinant : 1.0); } break; default: return; } scale_columns (pvt->matrix, a, b, c); return; } static void copywhiteXYZFromRGB(pvt,tec,vec) mpRGBPvtPtr pvt; xieTypWhiteAdjustTechnique tec; double *vec; { switch (tec) { case xieValWhiteAdjustCIELabShift: break; case xieValWhiteAdjustDefault: return; case xieValWhiteAdjustNone: return; default: return; } if (vec[0] < .0001) return; if (vec[1] < .0001) return; if (vec[2] < .0001) return; scale_rows (pvt->matrix, 1.0 / vec[0], 1.0 / vec[1], 1.0 / vec[2]); } static void copywhiteXYZToRGB(pvt,tec,vec) mpRGBPvtPtr pvt; xieTypWhiteAdjustTechnique tec; double *vec; { switch (tec) { case xieValWhiteAdjustCIELabShift: break; case xieValWhiteAdjustDefault: return; case xieValWhiteAdjustNone: return; default: return; } scale_columns (pvt->matrix, vec[0], vec[1], vec[2]); } static void copygamut(pvt,tec) mpRGBPvtPtr pvt; xieTypGamutTechnique tec; { if (tec == xieValGamutClipRGB) pvt->post = act_postClipR; } static void copybiasYCbCr(pvt,bias0,bias1,bias2) mpRGBPvtPtr pvt; double bias0, bias1, bias2; { pvt->matrix[9] = bias0; pvt->matrix[10] = bias1; pvt->matrix[11] = bias2; } static void copylumaYCbCrfromRGB(pvt,LumaRed,LumaGreen,LumaBlue) mpRGBPvtPtr pvt; double LumaRed,LumaGreen,LumaBlue; { RGBFloat *mtrx = pvt->matrix; if (LumaRed < .01) LumaRed = .01; if (LumaGreen < .01) LumaGreen = .01; if (LumaBlue < .01) LumaBlue = .01; if (LumaRed > .99) LumaRed = .99; if (LumaGreen > .99) LumaGreen = .99; if (LumaBlue > .99) LumaBlue = .99; /* ** FROM TIFF 6.0 ** ** Y = (LumaRed*R + LumaGreen * G + LumaBlue * B) ** Cb = (B - Y) / (2 - 2*LumaBlue) ** Cr = (R - Y) / (2 - 2*LumaRed) ** ** Cb = .5 * (B - (Lr*R + Lg*G + Lb*B)) / (1 - Lb) ** Cb = .5B/(1-Lb) - .5(Lr*R + Lg*G + Lb*B)) / (1-Lb) ** Cb = .5B/(1-Lb) - .5 Lr*R/(1-Lb) -.5 Lg*G/(1-Lb) -.5 Lb*B/(1-Lb) ** Cb = .5B(1-Lb)/(1-Lb) -.5 Lr*R/(1-Lb) -.5 Lg*G/(1-Lb) ** Cb = .5B -.5 Lr*R/(1-Lb) -.5 Lg*G/(1-Lb) */ mtrx[0] = LumaRed; mtrx[1] = LumaGreen; mtrx[2] = LumaBlue; mtrx[3] = -0.5 * LumaRed / (1.0 - LumaBlue); mtrx[4] = -0.5 * LumaGreen / (1.0 - LumaBlue); mtrx[5] = 0.5; mtrx[6] = 0.5; mtrx[7] = -0.5 * LumaGreen / (1.0 - LumaRed); mtrx[8] = -0.5 * LumaBlue / (1.0 - LumaRed); mtrx[9] = mtrx[10] = mtrx[11] = 0.0; /* bias */ } static void copylumaYCbCrtoRGB(pvt,LumaRed,LumaGreen,LumaBlue) mpRGBPvtPtr pvt; double LumaRed,LumaGreen,LumaBlue; { RGBFloat *mtrx = pvt->matrix; /* should/could check that LumaRed + LumaGreen + LumaBlue approx == 1.0 */ if (LumaGreen < .01) LumaGreen = .01; if (LumaGreen > .99) LumaGreen = .99; /* ** FROM TIFF 6.0 ** ** R = Cr * (2 - 2 * LumaRed) + Y ** G = (Y - LumaBlue * B - LumaRed * R) / LumaGreen ** B = Cb * (2 - 2 * LumaBlue) + Y ** ** G = (Y - Lb*(Cb*(2-2*Lb) + Y) - Lr*(Cr*(2-2*Lr) + Y) ) / Lg ** G = (Y - Lb*Cb*(2-2*Lb) - Lb*Y - Lr*Cr*(2-2*Lr) - Lr*Y ) / Lg ** G = ((1 - Lb - Lr)*Y - 2*Lb*(1-Lb)*Cb - 2*Lr*(1-Lr)*Cr ) / Lg ** ** R = 1 * Y + 0 * Cb + (2 - 2 * LumaRed) * Cr ** G = 1 * Y + ... ** B = 1 * Y + (2 - 2 * LumaBlue) * Cb + 0 * Cr ** ** note: mtrx[3] == 1.0 when LR+LG+LB==1.0 */ mtrx[0] = 1.0; mtrx[1] = 0.0; mtrx[2] = 2.0 - 2.0 * LumaRed; mtrx[3] = (1.0 - LumaRed - LumaBlue) / LumaGreen; mtrx[4] = -2.0 * LumaBlue * (1.0 - LumaBlue) / LumaGreen; mtrx[5] = -2.0 * LumaRed * (1.0 - LumaRed ) / LumaGreen; mtrx[6] = 1.0; mtrx[7] = 2.0 - 2.0 * LumaBlue; mtrx[8] = 0.0; mtrx[9] = mtrx[10] = mtrx[11] = 0.0; /* bias */ } static void copylumaYCCfromRGB(pvt,LumaRed,LumaGreen,LumaBlue,Scale,oband) mpRGBPvtPtr pvt; double LumaRed,LumaGreen,LumaBlue,Scale; bandPtr oband; { RGBFloat *mtrx = pvt->matrix; if (LumaGreen < .01) LumaGreen = .01; /* ** See "A Planning Guide For Developers" from KODAK PhotoCD Products. ** Typical CCIR601.1 values of (.299, .587, .114) are used for ** (LumaRed,LumaGreen,LumaBlue). ** ** Y = ( LumaRed*R + LumaGreen * G + LumaBlue * B) ** C1 = ( -LumaRed*R - LumaGreen * G + (1-LumaBlue) * B) ** C2 = ((1-LumaRed)*R - LumaGreen * G + -LumaBlue * B) ** ** From the same paper we are treated with the following equations ** for packing 8bit pixels: ** ** PACK: UMPACK ** Y' = 255/1.402 * Y Y = 1.3584 * Y' ** C1' = 111.40 * C1 + 156 C1 = 2.2179 * (C1' - 156) ** C2' = 135.64 * C2 + 137 C2 = 1.8215 * (C2' - 137) ** ** Extrapolating a bit we can see that this is: ** ** Y' = 255/1.402 * Y ** C1' = 255/2.289 * C1 + (.612 * 255) <2.289 = 1.402 * 1.6327 > ** C2' = 255/1.879 * C2 + (.537 * 255) <1.879 = 1.402 * 1.3409 > ** ** Y = 1.3584 * Y' ** C1 = 1.3584 * 1.6327 * (C1' - (.612*255)) ** C2 = 1.3584 * 1.3409 * (C2' - (.537*255)) ** ** So we are left with a choice between the magic 1.402 and 1.3584 ** values. Hey, better idea. Its now an XIE protocol parameter ** called "scale". ** ** In future versions, its possible we will want to accept the 1.3584 ** number as a technique parameter in addition to the Luma. */ mtrx[0] = LumaRed; mtrx[1] = LumaGreen; mtrx[2] = LumaBlue; mtrx[3] = -LumaRed; mtrx[4] = -LumaGreen; mtrx[5] = 1.0 - LumaBlue; mtrx[6] = 1.0 - LumaRed; mtrx[7] = -LumaGreen; mtrx[8] = -LumaBlue; if (IsConstrained(oband->format->class)) { /* Scale each row of matrix by compression factors */ scale_rows (mtrx, 1.0 / (Scale ), 1.0 / (Scale * 1.6327), 1.0 / (Scale * 1.3409)); /* Bias expressed as fraction of pixel range */ mtrx[9] = 0.0; mtrx[10] = 0.612 * (RGBFloat) ((oband+1)->format->levels-1); mtrx[11] = 0.5373 * (RGBFloat) ((oband+2)->format->levels-1); } } static void copylumaYCCtoRGB(pvt,LumaRed,LumaGreen,LumaBlue,Scale,iband) mpRGBPvtPtr pvt; double LumaRed,LumaGreen,LumaBlue,Scale; bandPtr iband; { RGBFloat *mtrx = pvt->matrix; if (LumaGreen < .01) LumaGreen = .01; /* ** See "A Planning Guide For Developers" from KODAK PhotoCD Products. ** Typical values of (.299, .587, .114) are used for (LR,LG,LB). ** ** R = 1 * Y + 0 * C1 + 1 * C2 ** G = 1 * Y - .194 * C1 - .509 * C2 ** B = 1 * Y + 1 * C1 + 0 * C2 ** ** By inverting the matrix symbolically we can deduce that the ** constants .194 and .509 are not arbitrary: ** ** R = 1 * Y + 0 * C1 + 1 * C2 ** G = 1 * Y - LB/LG * C1 - LR/LG * C2 ** B = 1 * Y + 1 * C1 + 0 * C2 ** ** Note: mtrx[3] == 1.0 when LR+LG+LB==1.0 */ mtrx[0] = 1.0; mtrx[1] = 0.0; mtrx[2] = 1.0; mtrx[3] = (1.0 - LumaRed - LumaBlue) / LumaGreen; mtrx[4] = - LumaBlue/LumaGreen; mtrx[5] = - LumaRed/LumaGreen; mtrx[6] = 1.0; mtrx[7] = 1.0; mtrx[8] = 0.0; if (IsConstrained(iband->format->class)) { /* Scale each column of matrix by compression factors */ scale_columns (mtrx, Scale, (Scale * 1.6327), (Scale * 1.3409)); /* Bias expressed as fraction of pixel range */ mtrx[9] = 0.0; mtrx[10] = 0.612 * (RGBFloat) ((iband+1)->format->levels-1); mtrx[11] = 0.5373 * (RGBFloat) ((iband+2)->format->levels-1); } } /*------------------------------------------------------------------------ -------------------------- initialize element . . . ---------------------- ------------------------------------------------------------------------*/ static int SetupToRGB(flo,ped,modify) floDefPtr flo; peDefPtr ped; int modify; { peTexPtr pet = ped->peTex; mpRGBPvtPtr pvt = (mpRGBPvtPtr) pet->private; pTecCIEToRGBDefPtr pCIE = (pTecCIEToRGBDefPtr) ped->techPvt; pTecYCCToRGBDefPtr pYCC = (pTecYCCToRGBDefPtr) ped->techPvt; pTecYCbCrToRGBDefPtr pYCb = (pTecYCbCrToRGBDefPtr) ped->techPvt; pvt->action = act_mmRR; pvt->post = 0; switch(ped->techVec->number) { case xieValCIELabToRGB: /* if whiteAdjusted, or calculate from inverted matrix */ copymatrix(pvt, pCIE->matrix); copywhiteLABToRGB (pvt, pCIE->whiteAdjusted, pCIE->whitePoint); copygamut (pvt, pCIE->gamutCompress); pvt->action = act_preCIELab; /* calls act_mmRR */ break; case xieValCIEXYZToRGB: copymatrix(pvt, pCIE->matrix); copywhiteXYZToRGB (pvt, pCIE->whiteAdjusted, pCIE->whitePoint); copygamut (pvt, pCIE->gamutCompress); break; case xieValYCbCrToRGB: copylumaYCbCrtoRGB(pvt, pYCb->red, pYCb->green, pYCb->blue); copybiasYCbCr(pvt, pYCb->bias0, pYCb->bias1, pYCb->bias2); copygamut(pvt, pYCb->gamutCompress); break; case xieValYCCToRGB: copylumaYCCtoRGB(pvt, pYCC->red, pYCC->green, pYCC->blue, pYCC->scale, &(pet->receptor[SRCtag].band[0])); copygamut (pvt, pYCC->gamutCompress); break; } CheckRGB(flo,ped,FALSE); return TRUE; } static int SetupFromRGB(flo,ped,modify) floDefPtr flo; peDefPtr ped; int modify; { peTexPtr pet = ped->peTex; mpRGBPvtPtr pvt = (mpRGBPvtPtr) pet->private; pTecRGBToCIEDefPtr pCIE = (pTecRGBToCIEDefPtr) ped->techPvt; pTecRGBToYCCDefPtr pYCC = (pTecRGBToYCCDefPtr) ped->techPvt; pTecRGBToYCbCrDefPtr pYCb = (pTecRGBToYCbCrDefPtr) ped->techPvt; pvt->action = act_mmRR; pvt->post = (void (*)()) 0; switch(ped->techVec->number) { case xieValRGBToCIELab: copymatrix(pvt, pCIE->matrix); copywhiteLABFromRGB(pvt, pCIE->whiteAdjusted, pCIE->whitePoint); pvt->post = act_postCIELab; break; case xieValRGBToCIEXYZ: copymatrix(pvt, pCIE->matrix); copywhiteXYZFromRGB(pvt, pCIE->whiteAdjusted, pCIE->whitePoint); break; case xieValRGBToYCbCr: copylumaYCbCrfromRGB(pvt, pYCb->red, pYCb->green, pYCb->blue); copybiasYCbCr(pvt, pYCb->bias0, pYCb->bias1, pYCb->bias2); break; case xieValRGBToYCC: copylumaYCCfromRGB(pvt, pYCC->red, pYCC->green, pYCC->blue, pYCC->scale, &(pet->emitter[0])); break; } CheckRGB(flo,ped,TRUE); return TRUE; } static void CheckRGB(flo,ped,fromrgb) floDefPtr flo; peDefPtr ped; Bool fromrgb; { peTexPtr pet = ped->peTex; bandPtr iband = &(pet->receptor[SRCtag].band[0]); bandPtr oband = &(pet->emitter[0]); CARD32 nbands = pet->receptor[SRCtag].inFlo->bands; mpRGBPvtPtr pvt = (mpRGBPvtPtr) pet->private; CARD32 c, cmin, cmax, l, lmin, lmax; CARD32 band; pvt->cvt_in[0] = pvt->cvt_in[1] = pvt->cvt_in[2] = (pointer (*)()) 0; pvt->cvt_out[0] = pvt->cvt_out[1] = pvt->cvt_out[2] = (pointer (*)()) 0; pvt->aux_buf[0] = pvt->aux_buf[1] = pvt->aux_buf[2] = (pointer) 0; if (IsntConstrained(iband->format->class)) { /* Already set up to activate floating routine */ return; } cmin = PAIR_PIXEL; cmax = BIT_PIXEL; lmin = (1<<24); lmax = 1; for (band = 0; band < nbands; band++, iband++, oband++) { c = iband->format->class; if (c < cmin) cmin = c; if (c > cmax) cmax = c; l = iband->format->levels; if (l < lmin) lmin = l; if (l > lmax) lmax = l; if (IsntConstrained(oband->format->class)) continue; c = oband->format->class; if (c < cmin) cmin = c; if (c > cmax) cmax = c; l = oband->format->levels; if (l < lmin) lmin = l; if (l > lmax) lmax = l; } iband -= 3; oband -= 3; /* Scale each column of matrix to scale from input levels to 0 ... 1 */ scale_columns (pvt->matrix, (double) (1.0 / ((iband+0)->format->levels - 1)), (double) (1.0 / ((iband+1)->format->levels - 1)), (double) (1.0 / ((iband+2)->format->levels - 1))); if (IsntConstrained(oband->format->class)) { /* Only RGBtoCIEXYZ and RGBtoCIELab will Unconstrain the data */ if (lmin > (1<<1) && lmax <= (1<<8)) { /* ALL Bytes */ pvt->action = act_mmBR; return; } pvt->action = act_mmPR; if (lmin > (1<<8)) /* ALL Pairs */ return; for (band = 0; band < nbands; band++, iband++) { /* SOME Pairs */ int levels = iband->format->levels; if (levels <= (1<<8)) { pvt->cvt_in[band] = levels < (1<<1) ? cvt_bit_to_pair : cvt_byte_to_pair; pvt->aux_buf[band] = (pointer) XieMalloc( iband->format->width * sizeof(PairPixel)); if (!pvt->aux_buf[band]) AllocError(flo,ped,return); } } iband -= 3; return; } /* If we got to here, we will implicitly HardClip the outputs */ pvt->post = (void (*)()) 0; pvt->iclip[0] = (oband+0)->format->levels - 1; pvt->iclip[1] = (oband+1)->format->levels - 1; pvt->iclip[2] = (oband+2)->format->levels - 1; /* Scale each row of matrix to map output from 0...1 to 0...nlev-1 */ scale_rows (pvt->matrix, (double) ((oband+0)->format->levels - 1), (double) ((oband+1)->format->levels - 1), (double) ((oband+2)->format->levels - 1)); /* Adjust bias for integer YCbCr and YCC */ if (!fromrgb) switch (ped->techVec->number) { case xieValRGBToYCC: /* fall thru */ case xieValRGBToYCbCr: flip_bias(pvt->matrix); break; default: break; } if (lmin > (1<<1) && lmax <= (1<<8)) { /* ALL Bytes */ pvt->action = act_mmBB; scale_mtrx(pvt->matrix,pvt->imatrix,(1<action = act_mmPP; scale_mtrx(pvt->matrix,pvt->imatrix,(1< (1<<8)) /* ALL Pairs */ return; for (band = 0; band < nbands; band++, iband++, oband++) { int levels; levels = iband->format->levels; pvt->cvt_in[band] = levels <= (1<<1) ? cvt_bit_to_pair : levels <= (1<<8) ? cvt_byte_to_pair : (pointer (*)()) 0; levels = oband->format->levels; pvt->cvt_out[band] = levels <= (1<<1) ? cvt_pair_to_bit : levels <= (1<<8) ? cvt_pair_to_byte : (pointer (*)()) 0; if ( pvt->cvt_in[band] || pvt->cvt_out[band]) { pvt->aux_buf[band] = (pointer) XieMalloc( iband->format->width * sizeof(PairPixel)); if (!pvt->aux_buf[band]) AllocError(flo,ped,return); } } iband -= 3; oband -= 3; return; } static void ClearRGB(flo,ped) floDefPtr flo; peDefPtr ped; { mpRGBPvtPtr pvt = (mpRGBPvtPtr) ped->peTex->private; if (!pvt) return; pvt->cvt_in[0] = pvt->cvt_in[1] = pvt->cvt_in[2] = (pointer (*)()) 0; pvt->cvt_out[0] = pvt->cvt_out[1] = pvt->cvt_out[2] = (pointer (*)()) 0; if (pvt->aux_buf[0]) pvt->aux_buf[0] = (pointer) XieFree(pvt->aux_buf[0]); if (pvt->aux_buf[1]) pvt->aux_buf[1] = (pointer) XieFree(pvt->aux_buf[1]); if (pvt->aux_buf[2]) pvt->aux_buf[2] = (pointer) XieFree(pvt->aux_buf[2]); } /* end module mprgb.c */