1 //---------------------------------------------------------------------------------
2 //
3 // Little Color Management System
4 // Copyright (c) 1998-2023 Marti Maria Saguer
5 //
6 // Permission is hereby granted, free of charge, to any person obtaining
7 // a copy of this software and associated documentation files (the "Software"),
8 // to deal in the Software without restriction, including without limitation
9 // the rights to use, copy, modify, merge, publish, distribute, sublicense,
10 // and/or sell copies of the Software, and to permit persons to whom the Software
11 // is furnished to do so, subject to the following conditions:
12 //
13 // The above copyright notice and this permission notice shall be included in
14 // all copies or substantial portions of the Software.
15 //
16 // THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
17 // EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO
18 // THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
19 // NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
20 // LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
21 // OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
22 // WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
23 //
24 //---------------------------------------------------------------------------------
25 //
26
27 #include "lcms2_internal.h"
28
29
30 //----------------------------------------------------------------------------------
31
32 // Optimization for 8 bits, Shaper-CLUT (3 inputs only)
33 typedef struct {
34
35 cmsContext ContextID;
36
37 const cmsInterpParams* p; // Tetrahedrical interpolation parameters. This is a not-owned pointer.
38
39 cmsUInt16Number rx[256], ry[256], rz[256];
40 cmsUInt32Number X0[256], Y0[256], Z0[256]; // Precomputed nodes and offsets for 8-bit input data
41
42
43 } Prelin8Data;
44
45
46 // Generic optimization for 16 bits Shaper-CLUT-Shaper (any inputs)
47 typedef struct {
48
49 cmsContext ContextID;
50
51 // Number of channels
52 cmsUInt32Number nInputs;
53 cmsUInt32Number nOutputs;
54
55 _cmsInterpFn16 EvalCurveIn16[MAX_INPUT_DIMENSIONS]; // The maximum number of input channels is known in advance
56 cmsInterpParams* ParamsCurveIn16[MAX_INPUT_DIMENSIONS];
57
58 _cmsInterpFn16 EvalCLUT; // The evaluator for 3D grid
59 const cmsInterpParams* CLUTparams; // (not-owned pointer)
60
61
62 _cmsInterpFn16* EvalCurveOut16; // Points to an array of curve evaluators in 16 bits (not-owned pointer)
63 cmsInterpParams** ParamsCurveOut16; // Points to an array of references to interpolation params (not-owned pointer)
64
65
66 } Prelin16Data;
67
68
69 // Optimization for matrix-shaper in 8 bits. Numbers are operated in n.14 signed, tables are stored in 1.14 fixed
70
71 typedef cmsInt32Number cmsS1Fixed14Number; // Note that this may hold more than 16 bits!
72
73 #define DOUBLE_TO_1FIXED14(x) ((cmsS1Fixed14Number) floor((x) * 16384.0 + 0.5))
74
75 typedef struct {
76
77 cmsContext ContextID;
78
79 cmsS1Fixed14Number Shaper1R[256]; // from 0..255 to 1.14 (0.0...1.0)
80 cmsS1Fixed14Number Shaper1G[256];
81 cmsS1Fixed14Number Shaper1B[256];
82
83 cmsS1Fixed14Number Mat[3][3]; // n.14 to n.14 (needs a saturation after that)
84 cmsS1Fixed14Number Off[3];
85
86 cmsUInt16Number Shaper2R[16385]; // 1.14 to 0..255
87 cmsUInt16Number Shaper2G[16385];
88 cmsUInt16Number Shaper2B[16385];
89
90 } MatShaper8Data;
91
92 // Curves, optimization is shared between 8 and 16 bits
93 typedef struct {
94
95 cmsContext ContextID;
96
97 cmsUInt32Number nCurves; // Number of curves
98 cmsUInt32Number nElements; // Elements in curves
99 cmsUInt16Number** Curves; // Points to a dynamically allocated array
100
101 } Curves16Data;
102
103
104 // Simple optimizations ----------------------------------------------------------------------------------------------------------
105
106
107 // Clamp a fixed point integer to signed 28 bits to avoid overflow in
108 // calculations. Clamp is intended for use with colorants, requiring one bit
109 // for a colorant and another two bits to avoid overflow when combining the
110 // colors.
_FixedClamp(cmsS1Fixed14Number n)111 cmsINLINE cmsS1Fixed14Number _FixedClamp(cmsS1Fixed14Number n) {
112 const cmsS1Fixed14Number max_positive = 268435455; // 0x0FFFFFFF;
113 const cmsS1Fixed14Number max_negative = -268435456; // 0xF0000000;
114 // Normally expect the provided number to be in the range [0..1] (but in
115 // fixed 1.14 format), so can perform a quick check for this typical case
116 // to reduce number of compares.
117 const cmsS1Fixed14Number typical_range_mask = 0xFFFF8000;
118
119 if (!(n & typical_range_mask))
120 return n;
121 if (n < max_negative)
122 return max_negative;
123 if (n > max_positive)
124 return max_positive;
125 return n;
126 }
127
128 // Perform one row of matrix multiply with translation for MatShaperEval16().
_MatShaperEvaluateRow(cmsS1Fixed14Number * mat,cmsS1Fixed14Number off,cmsS1Fixed14Number r,cmsS1Fixed14Number g,cmsS1Fixed14Number b)129 cmsINLINE cmsInt64Number _MatShaperEvaluateRow(cmsS1Fixed14Number* mat,
130 cmsS1Fixed14Number off,
131 cmsS1Fixed14Number r,
132 cmsS1Fixed14Number g,
133 cmsS1Fixed14Number b) {
134 return ((cmsInt64Number)mat[0] * r +
135 (cmsInt64Number)mat[1] * g +
136 (cmsInt64Number)mat[2] * b +
137 off + 0x2000) >> 14;
138 }
139
140 // Remove an element in linked chain
141 static
_RemoveElement(cmsStage ** head)142 void _RemoveElement(cmsStage** head)
143 {
144 cmsStage* mpe = *head;
145 cmsStage* next = mpe ->Next;
146 *head = next;
147 cmsStageFree(mpe);
148 }
149
150 // Remove all identities in chain. Note that pt actually is a double pointer to the element that holds the pointer.
151 static
_Remove1Op(cmsPipeline * Lut,cmsStageSignature UnaryOp)152 cmsBool _Remove1Op(cmsPipeline* Lut, cmsStageSignature UnaryOp)
153 {
154 cmsStage** pt = &Lut ->Elements;
155 cmsBool AnyOpt = FALSE;
156
157 while (*pt != NULL) {
158
159 if ((*pt) ->Implements == UnaryOp) {
160 _RemoveElement(pt);
161 AnyOpt = TRUE;
162 }
163 else
164 pt = &((*pt) -> Next);
165 }
166
167 return AnyOpt;
168 }
169
170 // Same, but only if two adjacent elements are found
171 static
_Remove2Op(cmsPipeline * Lut,cmsStageSignature Op1,cmsStageSignature Op2)172 cmsBool _Remove2Op(cmsPipeline* Lut, cmsStageSignature Op1, cmsStageSignature Op2)
173 {
174 cmsStage** pt1;
175 cmsStage** pt2;
176 cmsBool AnyOpt = FALSE;
177
178 pt1 = &Lut ->Elements;
179 if (*pt1 == NULL) return AnyOpt;
180
181 while (*pt1 != NULL) {
182
183 pt2 = &((*pt1) -> Next);
184 if (*pt2 == NULL) return AnyOpt;
185
186 if ((*pt1) ->Implements == Op1 && (*pt2) ->Implements == Op2) {
187 _RemoveElement(pt2);
188 _RemoveElement(pt1);
189 AnyOpt = TRUE;
190 }
191 else
192 pt1 = &((*pt1) -> Next);
193 }
194
195 return AnyOpt;
196 }
197
198
199 static
CloseEnoughFloat(cmsFloat64Number a,cmsFloat64Number b)200 cmsBool CloseEnoughFloat(cmsFloat64Number a, cmsFloat64Number b)
201 {
202 return fabs(b - a) < 0.00001f;
203 }
204
205 static
isFloatMatrixIdentity(const cmsMAT3 * a)206 cmsBool isFloatMatrixIdentity(const cmsMAT3* a)
207 {
208 cmsMAT3 Identity;
209 int i, j;
210
211 _cmsMAT3identity(&Identity);
212
213 for (i = 0; i < 3; i++)
214 for (j = 0; j < 3; j++)
215 if (!CloseEnoughFloat(a->v[i].n[j], Identity.v[i].n[j])) return FALSE;
216
217 return TRUE;
218 }
219 // if two adjacent matrices are found, multiply them.
220 static
_MultiplyMatrix(cmsPipeline * Lut)221 cmsBool _MultiplyMatrix(cmsPipeline* Lut)
222 {
223 cmsStage** pt1;
224 cmsStage** pt2;
225 cmsStage* chain;
226 cmsBool AnyOpt = FALSE;
227
228 pt1 = &Lut->Elements;
229 if (*pt1 == NULL) return AnyOpt;
230
231 while (*pt1 != NULL) {
232
233 pt2 = &((*pt1)->Next);
234 if (*pt2 == NULL) return AnyOpt;
235
236 if ((*pt1)->Implements == cmsSigMatrixElemType && (*pt2)->Implements == cmsSigMatrixElemType) {
237
238 // Get both matrices
239 _cmsStageMatrixData* m1 = (_cmsStageMatrixData*) cmsStageData(*pt1);
240 _cmsStageMatrixData* m2 = (_cmsStageMatrixData*) cmsStageData(*pt2);
241 cmsMAT3 res;
242
243 // Input offset and output offset should be zero to use this optimization
244 if (m1->Offset != NULL || m2 ->Offset != NULL ||
245 cmsStageInputChannels(*pt1) != 3 || cmsStageOutputChannels(*pt1) != 3 ||
246 cmsStageInputChannels(*pt2) != 3 || cmsStageOutputChannels(*pt2) != 3)
247 return FALSE;
248
249 // Multiply both matrices to get the result
250 _cmsMAT3per(&res, (cmsMAT3*)m2->Double, (cmsMAT3*)m1->Double);
251
252 // Get the next in chain after the matrices
253 chain = (*pt2)->Next;
254
255 // Remove both matrices
256 _RemoveElement(pt2);
257 _RemoveElement(pt1);
258
259 // Now what if the result is a plain identity?
260 if (!isFloatMatrixIdentity(&res)) {
261
262 // We can not get rid of full matrix
263 cmsStage* Multmat = cmsStageAllocMatrix(Lut->ContextID, 3, 3, (const cmsFloat64Number*) &res, NULL);
264 if (Multmat == NULL) return FALSE; // Should never happen
265
266 // Recover the chain
267 Multmat->Next = chain;
268 *pt1 = Multmat;
269 }
270
271 AnyOpt = TRUE;
272 }
273 else
274 pt1 = &((*pt1)->Next);
275 }
276
277 return AnyOpt;
278 }
279
280
281 // Preoptimize just gets rif of no-ops coming paired. Conversion from v2 to v4 followed
282 // by a v4 to v2 and vice-versa. The elements are then discarded.
283 static
PreOptimize(cmsPipeline * Lut)284 cmsBool PreOptimize(cmsPipeline* Lut)
285 {
286 cmsBool AnyOpt = FALSE, Opt;
287
288 do {
289
290 Opt = FALSE;
291
292 // Remove all identities
293 Opt |= _Remove1Op(Lut, cmsSigIdentityElemType);
294
295 // Remove XYZ2Lab followed by Lab2XYZ
296 Opt |= _Remove2Op(Lut, cmsSigXYZ2LabElemType, cmsSigLab2XYZElemType);
297
298 // Remove Lab2XYZ followed by XYZ2Lab
299 Opt |= _Remove2Op(Lut, cmsSigLab2XYZElemType, cmsSigXYZ2LabElemType);
300
301 // Remove V4 to V2 followed by V2 to V4
302 Opt |= _Remove2Op(Lut, cmsSigLabV4toV2, cmsSigLabV2toV4);
303
304 // Remove V2 to V4 followed by V4 to V2
305 Opt |= _Remove2Op(Lut, cmsSigLabV2toV4, cmsSigLabV4toV2);
306
307 // Remove float pcs Lab conversions
308 Opt |= _Remove2Op(Lut, cmsSigLab2FloatPCS, cmsSigFloatPCS2Lab);
309
310 // Remove float pcs Lab conversions
311 Opt |= _Remove2Op(Lut, cmsSigXYZ2FloatPCS, cmsSigFloatPCS2XYZ);
312
313 // Simplify matrix.
314 Opt |= _MultiplyMatrix(Lut);
315
316 if (Opt) AnyOpt = TRUE;
317
318 } while (Opt);
319
320 return AnyOpt;
321 }
322
323 static
Eval16nop1D(CMSREGISTER const cmsUInt16Number Input[],CMSREGISTER cmsUInt16Number Output[],CMSREGISTER const struct _cms_interp_struc * p)324 void Eval16nop1D(CMSREGISTER const cmsUInt16Number Input[],
325 CMSREGISTER cmsUInt16Number Output[],
326 CMSREGISTER const struct _cms_interp_struc* p)
327 {
328 Output[0] = Input[0];
329
330 cmsUNUSED_PARAMETER(p);
331 }
332
333 static
PrelinEval16(CMSREGISTER const cmsUInt16Number Input[],CMSREGISTER cmsUInt16Number Output[],CMSREGISTER const void * D)334 void PrelinEval16(CMSREGISTER const cmsUInt16Number Input[],
335 CMSREGISTER cmsUInt16Number Output[],
336 CMSREGISTER const void* D)
337 {
338 Prelin16Data* p16 = (Prelin16Data*) D;
339 cmsUInt16Number StageABC[MAX_INPUT_DIMENSIONS];
340 cmsUInt16Number StageDEF[cmsMAXCHANNELS];
341 cmsUInt32Number i;
342
343 for (i=0; i < p16 ->nInputs; i++) {
344
345 p16 ->EvalCurveIn16[i](&Input[i], &StageABC[i], p16 ->ParamsCurveIn16[i]);
346 }
347
348 p16 ->EvalCLUT(StageABC, StageDEF, p16 ->CLUTparams);
349
350 for (i=0; i < p16 ->nOutputs; i++) {
351
352 p16 ->EvalCurveOut16[i](&StageDEF[i], &Output[i], p16 ->ParamsCurveOut16[i]);
353 }
354 }
355
356
357 static
PrelinOpt16free(cmsContext ContextID,void * ptr)358 void PrelinOpt16free(cmsContext ContextID, void* ptr)
359 {
360 Prelin16Data* p16 = (Prelin16Data*) ptr;
361
362 _cmsFree(ContextID, p16 ->EvalCurveOut16);
363 _cmsFree(ContextID, p16 ->ParamsCurveOut16);
364
365 _cmsFree(ContextID, p16);
366 }
367
368 static
Prelin16dup(cmsContext ContextID,const void * ptr)369 void* Prelin16dup(cmsContext ContextID, const void* ptr)
370 {
371 Prelin16Data* p16 = (Prelin16Data*) ptr;
372 Prelin16Data* Duped = (Prelin16Data*) _cmsDupMem(ContextID, p16, sizeof(Prelin16Data));
373
374 if (Duped == NULL) return NULL;
375
376 Duped->EvalCurveOut16 = (_cmsInterpFn16*) _cmsDupMem(ContextID, p16->EvalCurveOut16, p16->nOutputs * sizeof(_cmsInterpFn16));
377 Duped->ParamsCurveOut16 = (cmsInterpParams**)_cmsDupMem(ContextID, p16->ParamsCurveOut16, p16->nOutputs * sizeof(cmsInterpParams*));
378
379 return Duped;
380 }
381
382
383 static
PrelinOpt16alloc(cmsContext ContextID,const cmsInterpParams * ColorMap,cmsUInt32Number nInputs,cmsToneCurve ** In,cmsUInt32Number nOutputs,cmsToneCurve ** Out)384 Prelin16Data* PrelinOpt16alloc(cmsContext ContextID,
385 const cmsInterpParams* ColorMap,
386 cmsUInt32Number nInputs, cmsToneCurve** In,
387 cmsUInt32Number nOutputs, cmsToneCurve** Out )
388 {
389 cmsUInt32Number i;
390 Prelin16Data* p16 = (Prelin16Data*)_cmsMallocZero(ContextID, sizeof(Prelin16Data));
391 if (p16 == NULL) return NULL;
392
393 p16 ->nInputs = nInputs;
394 p16 ->nOutputs = nOutputs;
395
396
397 for (i=0; i < nInputs; i++) {
398
399 if (In == NULL) {
400 p16 -> ParamsCurveIn16[i] = NULL;
401 p16 -> EvalCurveIn16[i] = Eval16nop1D;
402
403 }
404 else {
405 p16 -> ParamsCurveIn16[i] = In[i] ->InterpParams;
406 p16 -> EvalCurveIn16[i] = p16 ->ParamsCurveIn16[i]->Interpolation.Lerp16;
407 }
408 }
409
410 p16 ->CLUTparams = ColorMap;
411 p16 ->EvalCLUT = ColorMap ->Interpolation.Lerp16;
412
413
414 p16 -> EvalCurveOut16 = (_cmsInterpFn16*) _cmsCalloc(ContextID, nOutputs, sizeof(_cmsInterpFn16));
415 if (p16->EvalCurveOut16 == NULL)
416 {
417 _cmsFree(ContextID, p16);
418 return NULL;
419 }
420
421 p16 -> ParamsCurveOut16 = (cmsInterpParams**) _cmsCalloc(ContextID, nOutputs, sizeof(cmsInterpParams* ));
422 if (p16->ParamsCurveOut16 == NULL)
423 {
424
425 _cmsFree(ContextID, p16->EvalCurveOut16);
426 _cmsFree(ContextID, p16);
427 return NULL;
428 }
429
430 for (i=0; i < nOutputs; i++) {
431
432 if (Out == NULL) {
433 p16 ->ParamsCurveOut16[i] = NULL;
434 p16 -> EvalCurveOut16[i] = Eval16nop1D;
435 }
436 else {
437
438 p16 ->ParamsCurveOut16[i] = Out[i] ->InterpParams;
439 p16 -> EvalCurveOut16[i] = p16 ->ParamsCurveOut16[i]->Interpolation.Lerp16;
440 }
441 }
442
443 return p16;
444 }
445
446
447
448 // Resampling ---------------------------------------------------------------------------------
449
450 #define PRELINEARIZATION_POINTS 4096
451
452 // Sampler implemented by another LUT. This is a clean way to precalculate the devicelink 3D CLUT for
453 // almost any transform. We use floating point precision and then convert from floating point to 16 bits.
454 static
XFormSampler16(CMSREGISTER const cmsUInt16Number In[],CMSREGISTER cmsUInt16Number Out[],CMSREGISTER void * Cargo)455 cmsInt32Number XFormSampler16(CMSREGISTER const cmsUInt16Number In[],
456 CMSREGISTER cmsUInt16Number Out[],
457 CMSREGISTER void* Cargo)
458 {
459 cmsPipeline* Lut = (cmsPipeline*) Cargo;
460 cmsFloat32Number InFloat[cmsMAXCHANNELS], OutFloat[cmsMAXCHANNELS];
461 cmsUInt32Number i;
462
463 _cmsAssert(Lut -> InputChannels < cmsMAXCHANNELS);
464 _cmsAssert(Lut -> OutputChannels < cmsMAXCHANNELS);
465
466 // From 16 bit to floating point
467 for (i=0; i < Lut ->InputChannels; i++)
468 InFloat[i] = (cmsFloat32Number) (In[i] / 65535.0);
469
470 // Evaluate in floating point
471 cmsPipelineEvalFloat(InFloat, OutFloat, Lut);
472
473 // Back to 16 bits representation
474 for (i=0; i < Lut ->OutputChannels; i++)
475 Out[i] = _cmsQuickSaturateWord(OutFloat[i] * 65535.0);
476
477 // Always succeed
478 return TRUE;
479 }
480
481 // Try to see if the curves of a given MPE are linear
482 static
AllCurvesAreLinear(cmsStage * mpe)483 cmsBool AllCurvesAreLinear(cmsStage* mpe)
484 {
485 cmsToneCurve** Curves;
486 cmsUInt32Number i, n;
487
488 Curves = _cmsStageGetPtrToCurveSet(mpe);
489 if (Curves == NULL) return FALSE;
490
491 n = cmsStageOutputChannels(mpe);
492
493 for (i=0; i < n; i++) {
494 if (!cmsIsToneCurveLinear(Curves[i])) return FALSE;
495 }
496
497 return TRUE;
498 }
499
500 // This function replaces a specific node placed in "At" by the "Value" numbers. Its purpose
501 // is to fix scum dot on broken profiles/transforms. Works on 1, 3 and 4 channels
502 static
PatchLUT(cmsStage * CLUT,cmsUInt16Number At[],cmsUInt16Number Value[],cmsUInt32Number nChannelsOut,cmsUInt32Number nChannelsIn)503 cmsBool PatchLUT(cmsStage* CLUT, cmsUInt16Number At[], cmsUInt16Number Value[],
504 cmsUInt32Number nChannelsOut, cmsUInt32Number nChannelsIn)
505 {
506 _cmsStageCLutData* Grid = (_cmsStageCLutData*) CLUT ->Data;
507 cmsInterpParams* p16 = Grid ->Params;
508 cmsFloat64Number px, py, pz, pw;
509 int x0, y0, z0, w0;
510 int i, index;
511
512 if (CLUT -> Type != cmsSigCLutElemType) {
513 cmsSignalError(CLUT->ContextID, cmsERROR_INTERNAL, "(internal) Attempt to PatchLUT on non-lut stage");
514 return FALSE;
515 }
516
517 if (nChannelsIn == 4) {
518
519 px = ((cmsFloat64Number) At[0] * (p16->Domain[0])) / 65535.0;
520 py = ((cmsFloat64Number) At[1] * (p16->Domain[1])) / 65535.0;
521 pz = ((cmsFloat64Number) At[2] * (p16->Domain[2])) / 65535.0;
522 pw = ((cmsFloat64Number) At[3] * (p16->Domain[3])) / 65535.0;
523
524 x0 = (int) floor(px);
525 y0 = (int) floor(py);
526 z0 = (int) floor(pz);
527 w0 = (int) floor(pw);
528
529 if (((px - x0) != 0) ||
530 ((py - y0) != 0) ||
531 ((pz - z0) != 0) ||
532 ((pw - w0) != 0)) return FALSE; // Not on exact node
533
534 index = (int) p16 -> opta[3] * x0 +
535 (int) p16 -> opta[2] * y0 +
536 (int) p16 -> opta[1] * z0 +
537 (int) p16 -> opta[0] * w0;
538 }
539 else
540 if (nChannelsIn == 3) {
541
542 px = ((cmsFloat64Number) At[0] * (p16->Domain[0])) / 65535.0;
543 py = ((cmsFloat64Number) At[1] * (p16->Domain[1])) / 65535.0;
544 pz = ((cmsFloat64Number) At[2] * (p16->Domain[2])) / 65535.0;
545
546 x0 = (int) floor(px);
547 y0 = (int) floor(py);
548 z0 = (int) floor(pz);
549
550 if (((px - x0) != 0) ||
551 ((py - y0) != 0) ||
552 ((pz - z0) != 0)) return FALSE; // Not on exact node
553
554 index = (int) p16 -> opta[2] * x0 +
555 (int) p16 -> opta[1] * y0 +
556 (int) p16 -> opta[0] * z0;
557 }
558 else
559 if (nChannelsIn == 1) {
560
561 px = ((cmsFloat64Number) At[0] * (p16->Domain[0])) / 65535.0;
562
563 x0 = (int) floor(px);
564
565 if (((px - x0) != 0)) return FALSE; // Not on exact node
566
567 index = (int) p16 -> opta[0] * x0;
568 }
569 else {
570 cmsSignalError(CLUT->ContextID, cmsERROR_INTERNAL, "(internal) %d Channels are not supported on PatchLUT", nChannelsIn);
571 return FALSE;
572 }
573
574 for (i = 0; i < (int) nChannelsOut; i++)
575 Grid->Tab.T[index + i] = Value[i];
576
577 return TRUE;
578 }
579
580 // Auxiliary, to see if two values are equal or very different
581 static
WhitesAreEqual(cmsUInt32Number n,cmsUInt16Number White1[],cmsUInt16Number White2[])582 cmsBool WhitesAreEqual(cmsUInt32Number n, cmsUInt16Number White1[], cmsUInt16Number White2[] )
583 {
584 cmsUInt32Number i;
585
586 for (i=0; i < n; i++) {
587
588 if (abs(White1[i] - White2[i]) > 0xf000) return TRUE; // Values are so extremely different that the fixup should be avoided
589 if (White1[i] != White2[i]) return FALSE;
590 }
591 return TRUE;
592 }
593
594
595 // Locate the node for the white point and fix it to pure white in order to avoid scum dot.
596 static
FixWhiteMisalignment(cmsPipeline * Lut,cmsColorSpaceSignature EntryColorSpace,cmsColorSpaceSignature ExitColorSpace)597 cmsBool FixWhiteMisalignment(cmsPipeline* Lut, cmsColorSpaceSignature EntryColorSpace, cmsColorSpaceSignature ExitColorSpace)
598 {
599 cmsUInt16Number *WhitePointIn, *WhitePointOut;
600 cmsUInt16Number WhiteIn[cmsMAXCHANNELS], WhiteOut[cmsMAXCHANNELS], ObtainedOut[cmsMAXCHANNELS];
601 cmsUInt32Number i, nOuts, nIns;
602 cmsStage *PreLin = NULL, *CLUT = NULL, *PostLin = NULL;
603
604 if (!_cmsEndPointsBySpace(EntryColorSpace,
605 &WhitePointIn, NULL, &nIns)) return FALSE;
606
607 if (!_cmsEndPointsBySpace(ExitColorSpace,
608 &WhitePointOut, NULL, &nOuts)) return FALSE;
609
610 // It needs to be fixed?
611 if (Lut ->InputChannels != nIns) return FALSE;
612 if (Lut ->OutputChannels != nOuts) return FALSE;
613
614 cmsPipelineEval16(WhitePointIn, ObtainedOut, Lut);
615
616 if (WhitesAreEqual(nOuts, WhitePointOut, ObtainedOut)) return TRUE; // whites already match
617
618 // Check if the LUT comes as Prelin, CLUT or Postlin. We allow all combinations
619 if (!cmsPipelineCheckAndRetreiveStages(Lut, 3, cmsSigCurveSetElemType, cmsSigCLutElemType, cmsSigCurveSetElemType, &PreLin, &CLUT, &PostLin))
620 if (!cmsPipelineCheckAndRetreiveStages(Lut, 2, cmsSigCurveSetElemType, cmsSigCLutElemType, &PreLin, &CLUT))
621 if (!cmsPipelineCheckAndRetreiveStages(Lut, 2, cmsSigCLutElemType, cmsSigCurveSetElemType, &CLUT, &PostLin))
622 if (!cmsPipelineCheckAndRetreiveStages(Lut, 1, cmsSigCLutElemType, &CLUT))
623 return FALSE;
624
625 // We need to interpolate white points of both, pre and post curves
626 if (PreLin) {
627
628 cmsToneCurve** Curves = _cmsStageGetPtrToCurveSet(PreLin);
629
630 for (i=0; i < nIns; i++) {
631 WhiteIn[i] = cmsEvalToneCurve16(Curves[i], WhitePointIn[i]);
632 }
633 }
634 else {
635 for (i=0; i < nIns; i++)
636 WhiteIn[i] = WhitePointIn[i];
637 }
638
639 // If any post-linearization, we need to find how is represented white before the curve, do
640 // a reverse interpolation in this case.
641 if (PostLin) {
642
643 cmsToneCurve** Curves = _cmsStageGetPtrToCurveSet(PostLin);
644
645 for (i=0; i < nOuts; i++) {
646
647 cmsToneCurve* InversePostLin = cmsReverseToneCurve(Curves[i]);
648 if (InversePostLin == NULL) {
649 WhiteOut[i] = WhitePointOut[i];
650
651 } else {
652
653 WhiteOut[i] = cmsEvalToneCurve16(InversePostLin, WhitePointOut[i]);
654 cmsFreeToneCurve(InversePostLin);
655 }
656 }
657 }
658 else {
659 for (i=0; i < nOuts; i++)
660 WhiteOut[i] = WhitePointOut[i];
661 }
662
663 // Ok, proceed with patching. May fail and we don't care if it fails
664 PatchLUT(CLUT, WhiteIn, WhiteOut, nOuts, nIns);
665
666 return TRUE;
667 }
668
669 // -----------------------------------------------------------------------------------------------------------------------------------------------
670 // This function creates simple LUT from complex ones. The generated LUT has an optional set of
671 // prelinearization curves, a CLUT of nGridPoints and optional postlinearization tables.
672 // These curves have to exist in the original LUT in order to be used in the simplified output.
673 // Caller may also use the flags to allow this feature.
674 // LUTS with all curves will be simplified to a single curve. Parametric curves are lost.
675 // This function should be used on 16-bits LUTS only, as floating point losses precision when simplified
676 // -----------------------------------------------------------------------------------------------------------------------------------------------
677
678 static
OptimizeByResampling(cmsPipeline ** Lut,cmsUInt32Number Intent,cmsUInt32Number * InputFormat,cmsUInt32Number * OutputFormat,cmsUInt32Number * dwFlags)679 cmsBool OptimizeByResampling(cmsPipeline** Lut, cmsUInt32Number Intent, cmsUInt32Number* InputFormat, cmsUInt32Number* OutputFormat, cmsUInt32Number* dwFlags)
680 {
681 cmsPipeline* Src = NULL;
682 cmsPipeline* Dest = NULL;
683 cmsStage* CLUT;
684 cmsStage *KeepPreLin = NULL, *KeepPostLin = NULL;
685 cmsUInt32Number nGridPoints;
686 cmsColorSpaceSignature ColorSpace, OutputColorSpace;
687 cmsStage *NewPreLin = NULL;
688 cmsStage *NewPostLin = NULL;
689 _cmsStageCLutData* DataCLUT;
690 cmsToneCurve** DataSetIn;
691 cmsToneCurve** DataSetOut;
692 Prelin16Data* p16;
693
694 // This is a lossy optimization! does not apply in floating-point cases
695 if (_cmsFormatterIsFloat(*InputFormat) || _cmsFormatterIsFloat(*OutputFormat)) return FALSE;
696
697 ColorSpace = _cmsICCcolorSpace((int) T_COLORSPACE(*InputFormat));
698 OutputColorSpace = _cmsICCcolorSpace((int) T_COLORSPACE(*OutputFormat));
699
700 // Color space must be specified
701 if (ColorSpace == (cmsColorSpaceSignature)0 ||
702 OutputColorSpace == (cmsColorSpaceSignature)0) return FALSE;
703
704 nGridPoints = _cmsReasonableGridpointsByColorspace(ColorSpace, *dwFlags);
705
706 // For empty LUTs, 2 points are enough
707 if (cmsPipelineStageCount(*Lut) == 0)
708 nGridPoints = 2;
709
710 Src = *Lut;
711
712 // Allocate an empty LUT
713 Dest = cmsPipelineAlloc(Src ->ContextID, Src ->InputChannels, Src ->OutputChannels);
714 if (!Dest) return FALSE;
715
716 // Prelinearization tables are kept unless indicated by flags
717 if (*dwFlags & cmsFLAGS_CLUT_PRE_LINEARIZATION) {
718
719 // Get a pointer to the prelinearization element
720 cmsStage* PreLin = cmsPipelineGetPtrToFirstStage(Src);
721
722 // Check if suitable
723 if (PreLin && PreLin ->Type == cmsSigCurveSetElemType) {
724
725 // Maybe this is a linear tram, so we can avoid the whole stuff
726 if (!AllCurvesAreLinear(PreLin)) {
727
728 // All seems ok, proceed.
729 NewPreLin = cmsStageDup(PreLin);
730 if(!cmsPipelineInsertStage(Dest, cmsAT_BEGIN, NewPreLin))
731 goto Error;
732
733 // Remove prelinearization. Since we have duplicated the curve
734 // in destination LUT, the sampling should be applied after this stage.
735 cmsPipelineUnlinkStage(Src, cmsAT_BEGIN, &KeepPreLin);
736 }
737 }
738 }
739
740 // Allocate the CLUT
741 CLUT = cmsStageAllocCLut16bit(Src ->ContextID, nGridPoints, Src ->InputChannels, Src->OutputChannels, NULL);
742 if (CLUT == NULL) goto Error;
743
744 // Add the CLUT to the destination LUT
745 if (!cmsPipelineInsertStage(Dest, cmsAT_END, CLUT)) {
746 goto Error;
747 }
748
749 // Postlinearization tables are kept unless indicated by flags
750 if (*dwFlags & cmsFLAGS_CLUT_POST_LINEARIZATION) {
751
752 // Get a pointer to the postlinearization if present
753 cmsStage* PostLin = cmsPipelineGetPtrToLastStage(Src);
754
755 // Check if suitable
756 if (PostLin && cmsStageType(PostLin) == cmsSigCurveSetElemType) {
757
758 // Maybe this is a linear tram, so we can avoid the whole stuff
759 if (!AllCurvesAreLinear(PostLin)) {
760
761 // All seems ok, proceed.
762 NewPostLin = cmsStageDup(PostLin);
763 if (!cmsPipelineInsertStage(Dest, cmsAT_END, NewPostLin))
764 goto Error;
765
766 // In destination LUT, the sampling should be applied after this stage.
767 cmsPipelineUnlinkStage(Src, cmsAT_END, &KeepPostLin);
768 }
769 }
770 }
771
772 // Now its time to do the sampling. We have to ignore pre/post linearization
773 // The source LUT without pre/post curves is passed as parameter.
774 if (!cmsStageSampleCLut16bit(CLUT, XFormSampler16, (void*) Src, 0)) {
775 Error:
776 // Ops, something went wrong, Restore stages
777 if (KeepPreLin != NULL) {
778 if (!cmsPipelineInsertStage(Src, cmsAT_BEGIN, KeepPreLin)) {
779 _cmsAssert(0); // This never happens
780 }
781 }
782 if (KeepPostLin != NULL) {
783 if (!cmsPipelineInsertStage(Src, cmsAT_END, KeepPostLin)) {
784 _cmsAssert(0); // This never happens
785 }
786 }
787 cmsPipelineFree(Dest);
788 return FALSE;
789 }
790
791 // Done.
792
793 if (KeepPreLin != NULL) cmsStageFree(KeepPreLin);
794 if (KeepPostLin != NULL) cmsStageFree(KeepPostLin);
795 cmsPipelineFree(Src);
796
797 DataCLUT = (_cmsStageCLutData*) CLUT ->Data;
798
799 if (NewPreLin == NULL) DataSetIn = NULL;
800 else DataSetIn = ((_cmsStageToneCurvesData*) NewPreLin ->Data) ->TheCurves;
801
802 if (NewPostLin == NULL) DataSetOut = NULL;
803 else DataSetOut = ((_cmsStageToneCurvesData*) NewPostLin ->Data) ->TheCurves;
804
805
806 if (DataSetIn == NULL && DataSetOut == NULL) {
807
808 _cmsPipelineSetOptimizationParameters(Dest, (_cmsPipelineEval16Fn) DataCLUT->Params->Interpolation.Lerp16, DataCLUT->Params, NULL, NULL);
809 }
810 else {
811
812 p16 = PrelinOpt16alloc(Dest ->ContextID,
813 DataCLUT ->Params,
814 Dest ->InputChannels,
815 DataSetIn,
816 Dest ->OutputChannels,
817 DataSetOut);
818
819 _cmsPipelineSetOptimizationParameters(Dest, PrelinEval16, (void*) p16, PrelinOpt16free, Prelin16dup);
820 }
821
822
823 // Don't fix white on absolute colorimetric
824 if (Intent == INTENT_ABSOLUTE_COLORIMETRIC)
825 *dwFlags |= cmsFLAGS_NOWHITEONWHITEFIXUP;
826
827 if (!(*dwFlags & cmsFLAGS_NOWHITEONWHITEFIXUP)) {
828
829 FixWhiteMisalignment(Dest, ColorSpace, OutputColorSpace);
830 }
831
832 *Lut = Dest;
833 return TRUE;
834
835 cmsUNUSED_PARAMETER(Intent);
836 }
837
838
839 // -----------------------------------------------------------------------------------------------------------------------------------------------
840 // Fixes the gamma balancing of transform. This is described in my paper "Prelinearization Stages on
841 // Color-Management Application-Specific Integrated Circuits (ASICs)" presented at NIP24. It only works
842 // for RGB transforms. See the paper for more details
843 // -----------------------------------------------------------------------------------------------------------------------------------------------
844
845
846 // Normalize endpoints by slope limiting max and min. This assures endpoints as well.
847 // Descending curves are handled as well.
848 static
SlopeLimiting(cmsToneCurve * g)849 void SlopeLimiting(cmsToneCurve* g)
850 {
851 int BeginVal, EndVal;
852 int AtBegin = (int) floor((cmsFloat64Number) g ->nEntries * 0.02 + 0.5); // Cutoff at 2%
853 int AtEnd = (int) g ->nEntries - AtBegin - 1; // And 98%
854 cmsFloat64Number Val, Slope, beta;
855 int i;
856
857 if (cmsIsToneCurveDescending(g)) {
858 BeginVal = 0xffff; EndVal = 0;
859 }
860 else {
861 BeginVal = 0; EndVal = 0xffff;
862 }
863
864 // Compute slope and offset for begin of curve
865 Val = g ->Table16[AtBegin];
866 Slope = (Val - BeginVal) / AtBegin;
867 beta = Val - Slope * AtBegin;
868
869 for (i=0; i < AtBegin; i++)
870 g ->Table16[i] = _cmsQuickSaturateWord(i * Slope + beta);
871
872 // Compute slope and offset for the end
873 Val = g ->Table16[AtEnd];
874 Slope = (EndVal - Val) / AtBegin; // AtBegin holds the X interval, which is same in both cases
875 beta = Val - Slope * AtEnd;
876
877 for (i = AtEnd; i < (int) g ->nEntries; i++)
878 g ->Table16[i] = _cmsQuickSaturateWord(i * Slope + beta);
879 }
880
881
882 // Precomputes tables for 8-bit on input devicelink.
883 static
PrelinOpt8alloc(cmsContext ContextID,const cmsInterpParams * p,cmsToneCurve * G[3])884 Prelin8Data* PrelinOpt8alloc(cmsContext ContextID, const cmsInterpParams* p, cmsToneCurve* G[3])
885 {
886 int i;
887 cmsUInt16Number Input[3];
888 cmsS15Fixed16Number v1, v2, v3;
889 Prelin8Data* p8;
890
891 p8 = (Prelin8Data*)_cmsMallocZero(ContextID, sizeof(Prelin8Data));
892 if (p8 == NULL) return NULL;
893
894 // Since this only works for 8 bit input, values comes always as x * 257,
895 // we can safely take msb byte (x << 8 + x)
896
897 for (i=0; i < 256; i++) {
898
899 if (G != NULL) {
900
901 // Get 16-bit representation
902 Input[0] = cmsEvalToneCurve16(G[0], FROM_8_TO_16(i));
903 Input[1] = cmsEvalToneCurve16(G[1], FROM_8_TO_16(i));
904 Input[2] = cmsEvalToneCurve16(G[2], FROM_8_TO_16(i));
905 }
906 else {
907 Input[0] = FROM_8_TO_16(i);
908 Input[1] = FROM_8_TO_16(i);
909 Input[2] = FROM_8_TO_16(i);
910 }
911
912
913 // Move to 0..1.0 in fixed domain
914 v1 = _cmsToFixedDomain((int) (Input[0] * p -> Domain[0]));
915 v2 = _cmsToFixedDomain((int) (Input[1] * p -> Domain[1]));
916 v3 = _cmsToFixedDomain((int) (Input[2] * p -> Domain[2]));
917
918 // Store the precalculated table of nodes
919 p8 ->X0[i] = (p->opta[2] * FIXED_TO_INT(v1));
920 p8 ->Y0[i] = (p->opta[1] * FIXED_TO_INT(v2));
921 p8 ->Z0[i] = (p->opta[0] * FIXED_TO_INT(v3));
922
923 // Store the precalculated table of offsets
924 p8 ->rx[i] = (cmsUInt16Number) FIXED_REST_TO_INT(v1);
925 p8 ->ry[i] = (cmsUInt16Number) FIXED_REST_TO_INT(v2);
926 p8 ->rz[i] = (cmsUInt16Number) FIXED_REST_TO_INT(v3);
927 }
928
929 p8 ->ContextID = ContextID;
930 p8 ->p = p;
931
932 return p8;
933 }
934
935 static
Prelin8free(cmsContext ContextID,void * ptr)936 void Prelin8free(cmsContext ContextID, void* ptr)
937 {
938 _cmsFree(ContextID, ptr);
939 }
940
941 static
Prelin8dup(cmsContext ContextID,const void * ptr)942 void* Prelin8dup(cmsContext ContextID, const void* ptr)
943 {
944 return _cmsDupMem(ContextID, ptr, sizeof(Prelin8Data));
945 }
946
947
948
949 // A optimized interpolation for 8-bit input.
950 #define DENS(i,j,k) (LutTable[(i)+(j)+(k)+OutChan])
951 static CMS_NO_SANITIZE
PrelinEval8(CMSREGISTER const cmsUInt16Number Input[],CMSREGISTER cmsUInt16Number Output[],CMSREGISTER const void * D)952 void PrelinEval8(CMSREGISTER const cmsUInt16Number Input[],
953 CMSREGISTER cmsUInt16Number Output[],
954 CMSREGISTER const void* D)
955 {
956
957 cmsUInt8Number r, g, b;
958 cmsS15Fixed16Number rx, ry, rz;
959 cmsS15Fixed16Number c0, c1, c2, c3, Rest;
960 int OutChan;
961 CMSREGISTER cmsS15Fixed16Number X0, X1, Y0, Y1, Z0, Z1;
962 Prelin8Data* p8 = (Prelin8Data*) D;
963 CMSREGISTER const cmsInterpParams* p = p8 ->p;
964 int TotalOut = (int) p -> nOutputs;
965 const cmsUInt16Number* LutTable = (const cmsUInt16Number*) p->Table;
966
967 r = (cmsUInt8Number) (Input[0] >> 8);
968 g = (cmsUInt8Number) (Input[1] >> 8);
969 b = (cmsUInt8Number) (Input[2] >> 8);
970
971 X0 = (cmsS15Fixed16Number) p8->X0[r];
972 Y0 = (cmsS15Fixed16Number) p8->Y0[g];
973 Z0 = (cmsS15Fixed16Number) p8->Z0[b];
974
975 rx = p8 ->rx[r];
976 ry = p8 ->ry[g];
977 rz = p8 ->rz[b];
978
979 X1 = X0 + (cmsS15Fixed16Number)((rx == 0) ? 0 : p ->opta[2]);
980 Y1 = Y0 + (cmsS15Fixed16Number)((ry == 0) ? 0 : p ->opta[1]);
981 Z1 = Z0 + (cmsS15Fixed16Number)((rz == 0) ? 0 : p ->opta[0]);
982
983
984 // These are the 6 Tetrahedral
985 for (OutChan=0; OutChan < TotalOut; OutChan++) {
986
987 c0 = DENS(X0, Y0, Z0);
988
989 if (rx >= ry && ry >= rz)
990 {
991 c1 = DENS(X1, Y0, Z0) - c0;
992 c2 = DENS(X1, Y1, Z0) - DENS(X1, Y0, Z0);
993 c3 = DENS(X1, Y1, Z1) - DENS(X1, Y1, Z0);
994 }
995 else
996 if (rx >= rz && rz >= ry)
997 {
998 c1 = DENS(X1, Y0, Z0) - c0;
999 c2 = DENS(X1, Y1, Z1) - DENS(X1, Y0, Z1);
1000 c3 = DENS(X1, Y0, Z1) - DENS(X1, Y0, Z0);
1001 }
1002 else
1003 if (rz >= rx && rx >= ry)
1004 {
1005 c1 = DENS(X1, Y0, Z1) - DENS(X0, Y0, Z1);
1006 c2 = DENS(X1, Y1, Z1) - DENS(X1, Y0, Z1);
1007 c3 = DENS(X0, Y0, Z1) - c0;
1008 }
1009 else
1010 if (ry >= rx && rx >= rz)
1011 {
1012 c1 = DENS(X1, Y1, Z0) - DENS(X0, Y1, Z0);
1013 c2 = DENS(X0, Y1, Z0) - c0;
1014 c3 = DENS(X1, Y1, Z1) - DENS(X1, Y1, Z0);
1015 }
1016 else
1017 if (ry >= rz && rz >= rx)
1018 {
1019 c1 = DENS(X1, Y1, Z1) - DENS(X0, Y1, Z1);
1020 c2 = DENS(X0, Y1, Z0) - c0;
1021 c3 = DENS(X0, Y1, Z1) - DENS(X0, Y1, Z0);
1022 }
1023 else
1024 if (rz >= ry && ry >= rx)
1025 {
1026 c1 = DENS(X1, Y1, Z1) - DENS(X0, Y1, Z1);
1027 c2 = DENS(X0, Y1, Z1) - DENS(X0, Y0, Z1);
1028 c3 = DENS(X0, Y0, Z1) - c0;
1029 }
1030 else {
1031 c1 = c2 = c3 = 0;
1032 }
1033
1034 Rest = c1 * rx + c2 * ry + c3 * rz + 0x8001;
1035 Output[OutChan] = (cmsUInt16Number) (c0 + ((Rest + (Rest >> 16)) >> 16));
1036
1037 }
1038 }
1039
1040 #undef DENS
1041
1042
1043 // Curves that contain wide empty areas are not optimizeable
1044 static
IsDegenerated(const cmsToneCurve * g)1045 cmsBool IsDegenerated(const cmsToneCurve* g)
1046 {
1047 cmsUInt32Number i, Zeros = 0, Poles = 0;
1048 cmsUInt32Number nEntries = g ->nEntries;
1049
1050 for (i=0; i < nEntries; i++) {
1051
1052 if (g ->Table16[i] == 0x0000) Zeros++;
1053 if (g ->Table16[i] == 0xffff) Poles++;
1054 }
1055
1056 if (Zeros == 1 && Poles == 1) return FALSE; // For linear tables
1057 if (Zeros > (nEntries / 20)) return TRUE; // Degenerated, many zeros
1058 if (Poles > (nEntries / 20)) return TRUE; // Degenerated, many poles
1059
1060 return FALSE;
1061 }
1062
1063 // --------------------------------------------------------------------------------------------------------------
1064 // We need xput over here
1065
1066 static
OptimizeByComputingLinearization(cmsPipeline ** Lut,cmsUInt32Number Intent,cmsUInt32Number * InputFormat,cmsUInt32Number * OutputFormat,cmsUInt32Number * dwFlags)1067 cmsBool OptimizeByComputingLinearization(cmsPipeline** Lut, cmsUInt32Number Intent, cmsUInt32Number* InputFormat, cmsUInt32Number* OutputFormat, cmsUInt32Number* dwFlags)
1068 {
1069 cmsPipeline* OriginalLut;
1070 cmsUInt32Number nGridPoints;
1071 cmsToneCurve *Trans[cmsMAXCHANNELS], *TransReverse[cmsMAXCHANNELS];
1072 cmsUInt32Number t, i;
1073 cmsFloat32Number v, In[cmsMAXCHANNELS], Out[cmsMAXCHANNELS];
1074 cmsBool lIsSuitable, lIsLinear;
1075 cmsPipeline* OptimizedLUT = NULL, *LutPlusCurves = NULL;
1076 cmsStage* OptimizedCLUTmpe;
1077 cmsColorSpaceSignature ColorSpace, OutputColorSpace;
1078 cmsStage* OptimizedPrelinMpe;
1079 cmsToneCurve** OptimizedPrelinCurves;
1080 _cmsStageCLutData* OptimizedPrelinCLUT;
1081
1082
1083 // This is a lossy optimization! does not apply in floating-point cases
1084 if (_cmsFormatterIsFloat(*InputFormat) || _cmsFormatterIsFloat(*OutputFormat)) return FALSE;
1085
1086 // Only on chunky RGB
1087 if (T_COLORSPACE(*InputFormat) != PT_RGB) return FALSE;
1088 if (T_PLANAR(*InputFormat)) return FALSE;
1089
1090 if (T_COLORSPACE(*OutputFormat) != PT_RGB) return FALSE;
1091 if (T_PLANAR(*OutputFormat)) return FALSE;
1092
1093 // On 16 bits, user has to specify the feature
1094 if (!_cmsFormatterIs8bit(*InputFormat)) {
1095 if (!(*dwFlags & cmsFLAGS_CLUT_PRE_LINEARIZATION)) return FALSE;
1096 }
1097
1098 OriginalLut = *Lut;
1099
1100 ColorSpace = _cmsICCcolorSpace((int) T_COLORSPACE(*InputFormat));
1101 OutputColorSpace = _cmsICCcolorSpace((int) T_COLORSPACE(*OutputFormat));
1102
1103 // Color space must be specified
1104 if (ColorSpace == (cmsColorSpaceSignature)0 ||
1105 OutputColorSpace == (cmsColorSpaceSignature)0) return FALSE;
1106
1107 nGridPoints = _cmsReasonableGridpointsByColorspace(ColorSpace, *dwFlags);
1108
1109 // Empty gamma containers
1110 memset(Trans, 0, sizeof(Trans));
1111 memset(TransReverse, 0, sizeof(TransReverse));
1112
1113 // If the last stage of the original lut are curves, and those curves are
1114 // degenerated, it is likely the transform is squeezing and clipping
1115 // the output from previous CLUT. We cannot optimize this case
1116 {
1117 cmsStage* last = cmsPipelineGetPtrToLastStage(OriginalLut);
1118
1119 if (last == NULL) goto Error;
1120 if (cmsStageType(last) == cmsSigCurveSetElemType) {
1121
1122 _cmsStageToneCurvesData* Data = (_cmsStageToneCurvesData*)cmsStageData(last);
1123 for (i = 0; i < Data->nCurves; i++) {
1124 if (IsDegenerated(Data->TheCurves[i]))
1125 goto Error;
1126 }
1127 }
1128 }
1129
1130 for (t = 0; t < OriginalLut ->InputChannels; t++) {
1131 Trans[t] = cmsBuildTabulatedToneCurve16(OriginalLut ->ContextID, PRELINEARIZATION_POINTS, NULL);
1132 if (Trans[t] == NULL) goto Error;
1133 }
1134
1135 // Populate the curves
1136 for (i=0; i < PRELINEARIZATION_POINTS; i++) {
1137
1138 v = (cmsFloat32Number) ((cmsFloat64Number) i / (PRELINEARIZATION_POINTS - 1));
1139
1140 // Feed input with a gray ramp
1141 for (t=0; t < OriginalLut ->InputChannels; t++)
1142 In[t] = v;
1143
1144 // Evaluate the gray value
1145 cmsPipelineEvalFloat(In, Out, OriginalLut);
1146
1147 // Store result in curve
1148 for (t=0; t < OriginalLut ->InputChannels; t++)
1149 Trans[t] ->Table16[i] = _cmsQuickSaturateWord(Out[t] * 65535.0);
1150 }
1151
1152 // Slope-limit the obtained curves
1153 for (t = 0; t < OriginalLut ->InputChannels; t++)
1154 SlopeLimiting(Trans[t]);
1155
1156 // Check for validity
1157 lIsSuitable = TRUE;
1158 lIsLinear = TRUE;
1159 for (t=0; (lIsSuitable && (t < OriginalLut ->InputChannels)); t++) {
1160
1161 // Exclude if already linear
1162 if (!cmsIsToneCurveLinear(Trans[t]))
1163 lIsLinear = FALSE;
1164
1165 // Exclude if non-monotonic
1166 if (!cmsIsToneCurveMonotonic(Trans[t]))
1167 lIsSuitable = FALSE;
1168
1169 if (IsDegenerated(Trans[t]))
1170 lIsSuitable = FALSE;
1171 }
1172
1173 // If it is not suitable, just quit
1174 if (!lIsSuitable) goto Error;
1175
1176 // Invert curves if possible
1177 for (t = 0; t < OriginalLut ->InputChannels; t++) {
1178 TransReverse[t] = cmsReverseToneCurveEx(PRELINEARIZATION_POINTS, Trans[t]);
1179 if (TransReverse[t] == NULL) goto Error;
1180 }
1181
1182 // Now inset the reversed curves at the begin of transform
1183 LutPlusCurves = cmsPipelineDup(OriginalLut);
1184 if (LutPlusCurves == NULL) goto Error;
1185
1186 if (!cmsPipelineInsertStage(LutPlusCurves, cmsAT_BEGIN, cmsStageAllocToneCurves(OriginalLut ->ContextID, OriginalLut ->InputChannels, TransReverse)))
1187 goto Error;
1188
1189 // Create the result LUT
1190 OptimizedLUT = cmsPipelineAlloc(OriginalLut ->ContextID, OriginalLut ->InputChannels, OriginalLut ->OutputChannels);
1191 if (OptimizedLUT == NULL) goto Error;
1192
1193 OptimizedPrelinMpe = cmsStageAllocToneCurves(OriginalLut ->ContextID, OriginalLut ->InputChannels, Trans);
1194
1195 // Create and insert the curves at the beginning
1196 if (!cmsPipelineInsertStage(OptimizedLUT, cmsAT_BEGIN, OptimizedPrelinMpe))
1197 goto Error;
1198
1199 // Allocate the CLUT for result
1200 OptimizedCLUTmpe = cmsStageAllocCLut16bit(OriginalLut ->ContextID, nGridPoints, OriginalLut ->InputChannels, OriginalLut ->OutputChannels, NULL);
1201
1202 // Add the CLUT to the destination LUT
1203 if (!cmsPipelineInsertStage(OptimizedLUT, cmsAT_END, OptimizedCLUTmpe))
1204 goto Error;
1205
1206 // Resample the LUT
1207 if (!cmsStageSampleCLut16bit(OptimizedCLUTmpe, XFormSampler16, (void*) LutPlusCurves, 0)) goto Error;
1208
1209 // Free resources
1210 for (t = 0; t < OriginalLut ->InputChannels; t++) {
1211
1212 if (Trans[t]) cmsFreeToneCurve(Trans[t]);
1213 if (TransReverse[t]) cmsFreeToneCurve(TransReverse[t]);
1214 }
1215
1216 cmsPipelineFree(LutPlusCurves);
1217
1218
1219 OptimizedPrelinCurves = _cmsStageGetPtrToCurveSet(OptimizedPrelinMpe);
1220 OptimizedPrelinCLUT = (_cmsStageCLutData*) OptimizedCLUTmpe ->Data;
1221
1222 // Set the evaluator if 8-bit
1223 if (_cmsFormatterIs8bit(*InputFormat)) {
1224
1225 Prelin8Data* p8 = PrelinOpt8alloc(OptimizedLUT ->ContextID,
1226 OptimizedPrelinCLUT ->Params,
1227 OptimizedPrelinCurves);
1228 if (p8 == NULL) return FALSE;
1229
1230 _cmsPipelineSetOptimizationParameters(OptimizedLUT, PrelinEval8, (void*) p8, Prelin8free, Prelin8dup);
1231
1232 }
1233 else
1234 {
1235 Prelin16Data* p16 = PrelinOpt16alloc(OptimizedLUT ->ContextID,
1236 OptimizedPrelinCLUT ->Params,
1237 3, OptimizedPrelinCurves, 3, NULL);
1238 if (p16 == NULL) return FALSE;
1239
1240 _cmsPipelineSetOptimizationParameters(OptimizedLUT, PrelinEval16, (void*) p16, PrelinOpt16free, Prelin16dup);
1241
1242 }
1243
1244 // Don't fix white on absolute colorimetric
1245 if (Intent == INTENT_ABSOLUTE_COLORIMETRIC)
1246 *dwFlags |= cmsFLAGS_NOWHITEONWHITEFIXUP;
1247
1248 if (!(*dwFlags & cmsFLAGS_NOWHITEONWHITEFIXUP)) {
1249
1250 if (!FixWhiteMisalignment(OptimizedLUT, ColorSpace, OutputColorSpace)) {
1251
1252 return FALSE;
1253 }
1254 }
1255
1256 // And return the obtained LUT
1257
1258 cmsPipelineFree(OriginalLut);
1259 *Lut = OptimizedLUT;
1260 return TRUE;
1261
1262 Error:
1263
1264 for (t = 0; t < OriginalLut ->InputChannels; t++) {
1265
1266 if (Trans[t]) cmsFreeToneCurve(Trans[t]);
1267 if (TransReverse[t]) cmsFreeToneCurve(TransReverse[t]);
1268 }
1269
1270 if (LutPlusCurves != NULL) cmsPipelineFree(LutPlusCurves);
1271 if (OptimizedLUT != NULL) cmsPipelineFree(OptimizedLUT);
1272
1273 return FALSE;
1274
1275 cmsUNUSED_PARAMETER(Intent);
1276 cmsUNUSED_PARAMETER(lIsLinear);
1277 }
1278
1279
1280 // Curves optimizer ------------------------------------------------------------------------------------------------------------------
1281
1282 static
CurvesFree(cmsContext ContextID,void * ptr)1283 void CurvesFree(cmsContext ContextID, void* ptr)
1284 {
1285 Curves16Data* Data = (Curves16Data*) ptr;
1286 cmsUInt32Number i;
1287
1288 for (i=0; i < Data -> nCurves; i++) {
1289
1290 _cmsFree(ContextID, Data ->Curves[i]);
1291 }
1292
1293 _cmsFree(ContextID, Data ->Curves);
1294 _cmsFree(ContextID, ptr);
1295 }
1296
1297 static
CurvesDup(cmsContext ContextID,const void * ptr)1298 void* CurvesDup(cmsContext ContextID, const void* ptr)
1299 {
1300 Curves16Data* Data = (Curves16Data*)_cmsDupMem(ContextID, ptr, sizeof(Curves16Data));
1301 cmsUInt32Number i;
1302
1303 if (Data == NULL) return NULL;
1304
1305 Data->Curves = (cmsUInt16Number**) _cmsDupMem(ContextID, Data->Curves, Data->nCurves * sizeof(cmsUInt16Number*));
1306
1307 for (i=0; i < Data -> nCurves; i++) {
1308 Data->Curves[i] = (cmsUInt16Number*) _cmsDupMem(ContextID, Data->Curves[i], Data->nElements * sizeof(cmsUInt16Number));
1309 }
1310
1311 return (void*) Data;
1312 }
1313
1314 // Precomputes tables for 8-bit on input devicelink.
1315 static
CurvesAlloc(cmsContext ContextID,cmsUInt32Number nCurves,cmsUInt32Number nElements,cmsToneCurve ** G)1316 Curves16Data* CurvesAlloc(cmsContext ContextID, cmsUInt32Number nCurves, cmsUInt32Number nElements, cmsToneCurve** G)
1317 {
1318 cmsUInt32Number i, j;
1319 Curves16Data* c16;
1320
1321 c16 = (Curves16Data*)_cmsMallocZero(ContextID, sizeof(Curves16Data));
1322 if (c16 == NULL) return NULL;
1323
1324 c16 ->nCurves = nCurves;
1325 c16 ->nElements = nElements;
1326
1327 c16->Curves = (cmsUInt16Number**) _cmsCalloc(ContextID, nCurves, sizeof(cmsUInt16Number*));
1328 if (c16->Curves == NULL) {
1329 _cmsFree(ContextID, c16);
1330 return NULL;
1331 }
1332
1333 for (i=0; i < nCurves; i++) {
1334
1335 c16->Curves[i] = (cmsUInt16Number*) _cmsCalloc(ContextID, nElements, sizeof(cmsUInt16Number));
1336
1337 if (c16->Curves[i] == NULL) {
1338
1339 for (j=0; j < i; j++) {
1340 _cmsFree(ContextID, c16->Curves[j]);
1341 }
1342 _cmsFree(ContextID, c16->Curves);
1343 _cmsFree(ContextID, c16);
1344 return NULL;
1345 }
1346
1347 if (nElements == 256U) {
1348
1349 for (j=0; j < nElements; j++) {
1350
1351 c16 ->Curves[i][j] = cmsEvalToneCurve16(G[i], FROM_8_TO_16(j));
1352 }
1353 }
1354 else {
1355
1356 for (j=0; j < nElements; j++) {
1357 c16 ->Curves[i][j] = cmsEvalToneCurve16(G[i], (cmsUInt16Number) j);
1358 }
1359 }
1360 }
1361
1362 return c16;
1363 }
1364
1365 static
FastEvaluateCurves8(CMSREGISTER const cmsUInt16Number In[],CMSREGISTER cmsUInt16Number Out[],CMSREGISTER const void * D)1366 void FastEvaluateCurves8(CMSREGISTER const cmsUInt16Number In[],
1367 CMSREGISTER cmsUInt16Number Out[],
1368 CMSREGISTER const void* D)
1369 {
1370 Curves16Data* Data = (Curves16Data*) D;
1371 int x;
1372 cmsUInt32Number i;
1373
1374 for (i=0; i < Data ->nCurves; i++) {
1375
1376 x = (In[i] >> 8);
1377 Out[i] = Data -> Curves[i][x];
1378 }
1379 }
1380
1381
1382 static
FastEvaluateCurves16(CMSREGISTER const cmsUInt16Number In[],CMSREGISTER cmsUInt16Number Out[],CMSREGISTER const void * D)1383 void FastEvaluateCurves16(CMSREGISTER const cmsUInt16Number In[],
1384 CMSREGISTER cmsUInt16Number Out[],
1385 CMSREGISTER const void* D)
1386 {
1387 Curves16Data* Data = (Curves16Data*) D;
1388 cmsUInt32Number i;
1389
1390 for (i=0; i < Data ->nCurves; i++) {
1391 Out[i] = Data -> Curves[i][In[i]];
1392 }
1393 }
1394
1395
1396 static
FastIdentity16(CMSREGISTER const cmsUInt16Number In[],CMSREGISTER cmsUInt16Number Out[],CMSREGISTER const void * D)1397 void FastIdentity16(CMSREGISTER const cmsUInt16Number In[],
1398 CMSREGISTER cmsUInt16Number Out[],
1399 CMSREGISTER const void* D)
1400 {
1401 cmsPipeline* Lut = (cmsPipeline*) D;
1402 cmsUInt32Number i;
1403
1404 for (i=0; i < Lut ->InputChannels; i++) {
1405 Out[i] = In[i];
1406 }
1407 }
1408
1409
1410 // If the target LUT holds only curves, the optimization procedure is to join all those
1411 // curves together. That only works on curves and does not work on matrices.
1412 static
OptimizeByJoiningCurves(cmsPipeline ** Lut,cmsUInt32Number Intent,cmsUInt32Number * InputFormat,cmsUInt32Number * OutputFormat,cmsUInt32Number * dwFlags)1413 cmsBool OptimizeByJoiningCurves(cmsPipeline** Lut, cmsUInt32Number Intent, cmsUInt32Number* InputFormat, cmsUInt32Number* OutputFormat, cmsUInt32Number* dwFlags)
1414 {
1415 cmsToneCurve** GammaTables = NULL;
1416 cmsFloat32Number InFloat[cmsMAXCHANNELS], OutFloat[cmsMAXCHANNELS];
1417 cmsUInt32Number i, j;
1418 cmsPipeline* Src = *Lut;
1419 cmsPipeline* Dest = NULL;
1420 cmsStage* mpe;
1421 cmsStage* ObtainedCurves = NULL;
1422
1423
1424 // This is a lossy optimization! does not apply in floating-point cases
1425 if (_cmsFormatterIsFloat(*InputFormat) || _cmsFormatterIsFloat(*OutputFormat)) return FALSE;
1426
1427 // Only curves in this LUT?
1428 for (mpe = cmsPipelineGetPtrToFirstStage(Src);
1429 mpe != NULL;
1430 mpe = cmsStageNext(mpe)) {
1431 if (cmsStageType(mpe) != cmsSigCurveSetElemType) return FALSE;
1432 }
1433
1434 // Allocate an empty LUT
1435 Dest = cmsPipelineAlloc(Src ->ContextID, Src ->InputChannels, Src ->OutputChannels);
1436 if (Dest == NULL) return FALSE;
1437
1438 // Create target curves
1439 GammaTables = (cmsToneCurve**) _cmsCalloc(Src ->ContextID, Src ->InputChannels, sizeof(cmsToneCurve*));
1440 if (GammaTables == NULL) goto Error;
1441
1442 for (i=0; i < Src ->InputChannels; i++) {
1443 GammaTables[i] = cmsBuildTabulatedToneCurve16(Src ->ContextID, PRELINEARIZATION_POINTS, NULL);
1444 if (GammaTables[i] == NULL) goto Error;
1445 }
1446
1447 // Compute 16 bit result by using floating point
1448 for (i=0; i < PRELINEARIZATION_POINTS; i++) {
1449
1450 for (j=0; j < Src ->InputChannels; j++)
1451 InFloat[j] = (cmsFloat32Number) ((cmsFloat64Number) i / (PRELINEARIZATION_POINTS - 1));
1452
1453 cmsPipelineEvalFloat(InFloat, OutFloat, Src);
1454
1455 for (j=0; j < Src ->InputChannels; j++)
1456 GammaTables[j] -> Table16[i] = _cmsQuickSaturateWord(OutFloat[j] * 65535.0);
1457 }
1458
1459 ObtainedCurves = cmsStageAllocToneCurves(Src ->ContextID, Src ->InputChannels, GammaTables);
1460 if (ObtainedCurves == NULL) goto Error;
1461
1462 for (i=0; i < Src ->InputChannels; i++) {
1463 cmsFreeToneCurve(GammaTables[i]);
1464 GammaTables[i] = NULL;
1465 }
1466
1467 if (GammaTables != NULL) {
1468 _cmsFree(Src->ContextID, GammaTables);
1469 GammaTables = NULL;
1470 }
1471
1472 // Maybe the curves are linear at the end
1473 if (!AllCurvesAreLinear(ObtainedCurves)) {
1474 _cmsStageToneCurvesData* Data;
1475
1476 if (!cmsPipelineInsertStage(Dest, cmsAT_BEGIN, ObtainedCurves))
1477 goto Error;
1478 Data = (_cmsStageToneCurvesData*) cmsStageData(ObtainedCurves);
1479 ObtainedCurves = NULL;
1480
1481 // If the curves are to be applied in 8 bits, we can save memory
1482 if (_cmsFormatterIs8bit(*InputFormat)) {
1483 Curves16Data* c16 = CurvesAlloc(Dest ->ContextID, Data ->nCurves, 256, Data ->TheCurves);
1484
1485 if (c16 == NULL) goto Error;
1486 *dwFlags |= cmsFLAGS_NOCACHE;
1487 _cmsPipelineSetOptimizationParameters(Dest, FastEvaluateCurves8, c16, CurvesFree, CurvesDup);
1488
1489 }
1490 else {
1491 Curves16Data* c16 = CurvesAlloc(Dest ->ContextID, Data ->nCurves, 65536, Data ->TheCurves);
1492
1493 if (c16 == NULL) goto Error;
1494 *dwFlags |= cmsFLAGS_NOCACHE;
1495 _cmsPipelineSetOptimizationParameters(Dest, FastEvaluateCurves16, c16, CurvesFree, CurvesDup);
1496 }
1497 }
1498 else {
1499
1500 // LUT optimizes to nothing. Set the identity LUT
1501 cmsStageFree(ObtainedCurves);
1502 ObtainedCurves = NULL;
1503
1504 if (!cmsPipelineInsertStage(Dest, cmsAT_BEGIN, cmsStageAllocIdentity(Dest ->ContextID, Src ->InputChannels)))
1505 goto Error;
1506
1507 *dwFlags |= cmsFLAGS_NOCACHE;
1508 _cmsPipelineSetOptimizationParameters(Dest, FastIdentity16, (void*) Dest, NULL, NULL);
1509 }
1510
1511 // We are done.
1512 cmsPipelineFree(Src);
1513 *Lut = Dest;
1514 return TRUE;
1515
1516 Error:
1517
1518 if (ObtainedCurves != NULL) cmsStageFree(ObtainedCurves);
1519 if (GammaTables != NULL) {
1520 for (i=0; i < Src ->InputChannels; i++) {
1521 if (GammaTables[i] != NULL) cmsFreeToneCurve(GammaTables[i]);
1522 }
1523
1524 _cmsFree(Src ->ContextID, GammaTables);
1525 }
1526
1527 if (Dest != NULL) cmsPipelineFree(Dest);
1528 return FALSE;
1529
1530 cmsUNUSED_PARAMETER(Intent);
1531 cmsUNUSED_PARAMETER(InputFormat);
1532 cmsUNUSED_PARAMETER(OutputFormat);
1533 cmsUNUSED_PARAMETER(dwFlags);
1534 }
1535
1536 // -------------------------------------------------------------------------------------------------------------------------------------
1537 // LUT is Shaper - Matrix - Matrix - Shaper, which is very frequent when combining two matrix-shaper profiles
1538
1539
1540 static
FreeMatShaper(cmsContext ContextID,void * Data)1541 void FreeMatShaper(cmsContext ContextID, void* Data)
1542 {
1543 if (Data != NULL) _cmsFree(ContextID, Data);
1544 }
1545
1546 static
DupMatShaper(cmsContext ContextID,const void * Data)1547 void* DupMatShaper(cmsContext ContextID, const void* Data)
1548 {
1549 return _cmsDupMem(ContextID, Data, sizeof(MatShaper8Data));
1550 }
1551
1552
1553 // A fast matrix-shaper evaluator for 8 bits. This is a bit tricky since I'm using 1.14 signed fixed point
1554 // to accomplish some performance. Actually it takes 256x3 16 bits tables and 16385 x 3 tables of 8 bits,
1555 // in total about 50K, and the performance boost is huge!
1556 static CMS_NO_SANITIZE
MatShaperEval16(CMSREGISTER const cmsUInt16Number In[],CMSREGISTER cmsUInt16Number Out[],CMSREGISTER const void * D)1557 void MatShaperEval16(CMSREGISTER const cmsUInt16Number In[],
1558 CMSREGISTER cmsUInt16Number Out[],
1559 CMSREGISTER const void* D)
1560 {
1561 MatShaper8Data* p = (MatShaper8Data*) D;
1562 cmsS1Fixed14Number r, g, b;
1563 cmsInt64Number l1, l2, l3;
1564 cmsUInt32Number ri, gi, bi;
1565
1566 // In this case (and only in this case!) we can use this simplification since
1567 // In[] is assured to come from a 8 bit number. (a << 8 | a)
1568 ri = In[0] & 0xFFU;
1569 gi = In[1] & 0xFFU;
1570 bi = In[2] & 0xFFU;
1571
1572 // Across first shaper, which also converts to 1.14 fixed point
1573 r = _FixedClamp(p->Shaper1R[ri]);
1574 g = _FixedClamp(p->Shaper1G[gi]);
1575 b = _FixedClamp(p->Shaper1B[bi]);
1576
1577 // Evaluate the matrix in 1.14 fixed point
1578 l1 = _MatShaperEvaluateRow(p->Mat[0], p->Off[0], r, g, b);
1579 l2 = _MatShaperEvaluateRow(p->Mat[1], p->Off[1], r, g, b);
1580 l3 = _MatShaperEvaluateRow(p->Mat[2], p->Off[2], r, g, b);
1581
1582 // Now we have to clip to 0..1.0 range
1583 ri = (l1 < 0) ? 0 : ((l1 > 16384) ? 16384U : (cmsUInt32Number) l1);
1584 gi = (l2 < 0) ? 0 : ((l2 > 16384) ? 16384U : (cmsUInt32Number) l2);
1585 bi = (l3 < 0) ? 0 : ((l3 > 16384) ? 16384U : (cmsUInt32Number) l3);
1586
1587 // And across second shaper,
1588 Out[0] = p->Shaper2R[ri];
1589 Out[1] = p->Shaper2G[gi];
1590 Out[2] = p->Shaper2B[bi];
1591
1592 }
1593
1594 // This table converts from 8 bits to 1.14 after applying the curve
1595 static
FillFirstShaper(cmsS1Fixed14Number * Table,cmsToneCurve * Curve)1596 void FillFirstShaper(cmsS1Fixed14Number* Table, cmsToneCurve* Curve)
1597 {
1598 int i;
1599 cmsFloat32Number R, y;
1600
1601 for (i=0; i < 256; i++) {
1602
1603 R = (cmsFloat32Number) (i / 255.0);
1604 y = cmsEvalToneCurveFloat(Curve, R);
1605
1606 if (y < 131072.0)
1607 Table[i] = DOUBLE_TO_1FIXED14(y);
1608 else
1609 Table[i] = 0x7fffffff;
1610 }
1611 }
1612
1613 // This table converts form 1.14 (being 0x4000 the last entry) to 8 bits after applying the curve
1614 static
FillSecondShaper(cmsUInt16Number * Table,cmsToneCurve * Curve,cmsBool Is8BitsOutput)1615 void FillSecondShaper(cmsUInt16Number* Table, cmsToneCurve* Curve, cmsBool Is8BitsOutput)
1616 {
1617 int i;
1618 cmsFloat32Number R, Val;
1619
1620 for (i=0; i < 16385; i++) {
1621
1622 R = (cmsFloat32Number) (i / 16384.0);
1623 Val = cmsEvalToneCurveFloat(Curve, R); // Val comes 0..1.0
1624
1625 if (Val < 0)
1626 Val = 0;
1627
1628 if (Val > 1.0)
1629 Val = 1.0;
1630
1631 if (Is8BitsOutput) {
1632
1633 // If 8 bits output, we can optimize further by computing the / 257 part.
1634 // first we compute the resulting byte and then we store the byte times
1635 // 257. This quantization allows to round very quick by doing a >> 8, but
1636 // since the low byte is always equal to msb, we can do a & 0xff and this works!
1637 cmsUInt16Number w = _cmsQuickSaturateWord(Val * 65535.0);
1638 cmsUInt8Number b = FROM_16_TO_8(w);
1639
1640 Table[i] = FROM_8_TO_16(b);
1641 }
1642 else Table[i] = _cmsQuickSaturateWord(Val * 65535.0);
1643 }
1644 }
1645
1646 // Compute the matrix-shaper structure
1647 static
SetMatShaper(cmsPipeline * Dest,cmsToneCurve * Curve1[3],cmsMAT3 * Mat,cmsVEC3 * Off,cmsToneCurve * Curve2[3],cmsUInt32Number * OutputFormat)1648 cmsBool SetMatShaper(cmsPipeline* Dest, cmsToneCurve* Curve1[3], cmsMAT3* Mat, cmsVEC3* Off, cmsToneCurve* Curve2[3], cmsUInt32Number* OutputFormat)
1649 {
1650 MatShaper8Data* p;
1651 int i, j;
1652 cmsBool Is8Bits = _cmsFormatterIs8bit(*OutputFormat);
1653
1654 // Allocate a big chuck of memory to store precomputed tables
1655 p = (MatShaper8Data*) _cmsMalloc(Dest ->ContextID, sizeof(MatShaper8Data));
1656 if (p == NULL) return FALSE;
1657
1658 p -> ContextID = Dest -> ContextID;
1659
1660 // Precompute tables
1661 FillFirstShaper(p ->Shaper1R, Curve1[0]);
1662 FillFirstShaper(p ->Shaper1G, Curve1[1]);
1663 FillFirstShaper(p ->Shaper1B, Curve1[2]);
1664
1665 FillSecondShaper(p ->Shaper2R, Curve2[0], Is8Bits);
1666 FillSecondShaper(p ->Shaper2G, Curve2[1], Is8Bits);
1667 FillSecondShaper(p ->Shaper2B, Curve2[2], Is8Bits);
1668
1669 // Convert matrix to nFixed14. Note that those values may take more than 16 bits
1670 for (i=0; i < 3; i++) {
1671 for (j=0; j < 3; j++) {
1672 p ->Mat[i][j] = DOUBLE_TO_1FIXED14(Mat->v[i].n[j]);
1673 }
1674 }
1675
1676 for (i=0; i < 3; i++) {
1677
1678 if (Off == NULL) {
1679 p ->Off[i] = 0;
1680 }
1681 else {
1682 p ->Off[i] = DOUBLE_TO_1FIXED14(Off->n[i]);
1683 }
1684 }
1685
1686 // Mark as optimized for faster formatter
1687 if (Is8Bits)
1688 *OutputFormat |= OPTIMIZED_SH(1);
1689
1690 // Fill function pointers
1691 _cmsPipelineSetOptimizationParameters(Dest, MatShaperEval16, (void*) p, FreeMatShaper, DupMatShaper);
1692 return TRUE;
1693 }
1694
1695 // 8 bits on input allows matrix-shaper boot up to 25 Mpixels per second on RGB. That's fast!
1696 static
OptimizeMatrixShaper(cmsPipeline ** Lut,cmsUInt32Number Intent,cmsUInt32Number * InputFormat,cmsUInt32Number * OutputFormat,cmsUInt32Number * dwFlags)1697 cmsBool OptimizeMatrixShaper(cmsPipeline** Lut, cmsUInt32Number Intent, cmsUInt32Number* InputFormat, cmsUInt32Number* OutputFormat, cmsUInt32Number* dwFlags)
1698 {
1699 cmsStage* Curve1, *Curve2;
1700 cmsStage* Matrix1, *Matrix2;
1701 cmsMAT3 res;
1702 cmsBool IdentityMat;
1703 cmsPipeline* Dest, *Src;
1704 cmsFloat64Number* Offset;
1705
1706 // Only works on RGB to RGB
1707 if (T_CHANNELS(*InputFormat) != 3 || T_CHANNELS(*OutputFormat) != 3) return FALSE;
1708
1709 // Only works on 8 bit input
1710 if (!_cmsFormatterIs8bit(*InputFormat)) return FALSE;
1711
1712 // Seems suitable, proceed
1713 Src = *Lut;
1714
1715 // Check for:
1716 //
1717 // shaper-matrix-matrix-shaper
1718 // shaper-matrix-shaper
1719 //
1720 // Both of those constructs are possible (first because abs. colorimetric).
1721 // additionally, In the first case, the input matrix offset should be zero.
1722
1723 IdentityMat = FALSE;
1724 if (cmsPipelineCheckAndRetreiveStages(Src, 4,
1725 cmsSigCurveSetElemType, cmsSigMatrixElemType, cmsSigMatrixElemType, cmsSigCurveSetElemType,
1726 &Curve1, &Matrix1, &Matrix2, &Curve2)) {
1727
1728 // Get both matrices
1729 _cmsStageMatrixData* Data1 = (_cmsStageMatrixData*)cmsStageData(Matrix1);
1730 _cmsStageMatrixData* Data2 = (_cmsStageMatrixData*)cmsStageData(Matrix2);
1731
1732 // Only RGB to RGB
1733 if (Matrix1->InputChannels != 3 || Matrix1->OutputChannels != 3 ||
1734 Matrix2->InputChannels != 3 || Matrix2->OutputChannels != 3) return FALSE;
1735
1736 // Input offset should be zero
1737 if (Data1->Offset != NULL) return FALSE;
1738
1739 // Multiply both matrices to get the result
1740 _cmsMAT3per(&res, (cmsMAT3*)Data2->Double, (cmsMAT3*)Data1->Double);
1741
1742 // Only 2nd matrix has offset, or it is zero
1743 Offset = Data2->Offset;
1744
1745 // Now the result is in res + Data2 -> Offset. Maybe is a plain identity?
1746 if (_cmsMAT3isIdentity(&res) && Offset == NULL) {
1747
1748 // We can get rid of full matrix
1749 IdentityMat = TRUE;
1750 }
1751
1752 }
1753 else {
1754
1755 if (cmsPipelineCheckAndRetreiveStages(Src, 3,
1756 cmsSigCurveSetElemType, cmsSigMatrixElemType, cmsSigCurveSetElemType,
1757 &Curve1, &Matrix1, &Curve2)) {
1758
1759 _cmsStageMatrixData* Data = (_cmsStageMatrixData*)cmsStageData(Matrix1);
1760
1761 // Copy the matrix to our result
1762 memcpy(&res, Data->Double, sizeof(res));
1763
1764 // Preserve the Odffset (may be NULL as a zero offset)
1765 Offset = Data->Offset;
1766
1767 if (_cmsMAT3isIdentity(&res) && Offset == NULL) {
1768
1769 // We can get rid of full matrix
1770 IdentityMat = TRUE;
1771 }
1772 }
1773 else
1774 return FALSE; // Not optimizeable this time
1775
1776 }
1777
1778 // Allocate an empty LUT
1779 Dest = cmsPipelineAlloc(Src ->ContextID, Src ->InputChannels, Src ->OutputChannels);
1780 if (!Dest) return FALSE;
1781
1782 // Assamble the new LUT
1783 if (!cmsPipelineInsertStage(Dest, cmsAT_BEGIN, cmsStageDup(Curve1)))
1784 goto Error;
1785
1786 if (!IdentityMat) {
1787
1788 if (!cmsPipelineInsertStage(Dest, cmsAT_END, cmsStageAllocMatrix(Dest->ContextID, 3, 3, (const cmsFloat64Number*)&res, Offset)))
1789 goto Error;
1790 }
1791
1792 if (!cmsPipelineInsertStage(Dest, cmsAT_END, cmsStageDup(Curve2)))
1793 goto Error;
1794
1795 // If identity on matrix, we can further optimize the curves, so call the join curves routine
1796 if (IdentityMat) {
1797
1798 OptimizeByJoiningCurves(&Dest, Intent, InputFormat, OutputFormat, dwFlags);
1799 }
1800 else {
1801 _cmsStageToneCurvesData* mpeC1 = (_cmsStageToneCurvesData*) cmsStageData(Curve1);
1802 _cmsStageToneCurvesData* mpeC2 = (_cmsStageToneCurvesData*) cmsStageData(Curve2);
1803
1804 // In this particular optimization, cache does not help as it takes more time to deal with
1805 // the cache that with the pixel handling
1806 *dwFlags |= cmsFLAGS_NOCACHE;
1807
1808 // Setup the optimizarion routines
1809 SetMatShaper(Dest, mpeC1 ->TheCurves, &res, (cmsVEC3*) Offset, mpeC2->TheCurves, OutputFormat);
1810 }
1811
1812 cmsPipelineFree(Src);
1813 *Lut = Dest;
1814 return TRUE;
1815 Error:
1816 // Leave Src unchanged
1817 cmsPipelineFree(Dest);
1818 return FALSE;
1819 }
1820
1821
1822 // -------------------------------------------------------------------------------------------------------------------------------------
1823 // Optimization plug-ins
1824
1825 // List of optimizations
1826 typedef struct _cmsOptimizationCollection_st {
1827
1828 _cmsOPToptimizeFn OptimizePtr;
1829
1830 struct _cmsOptimizationCollection_st *Next;
1831
1832 } _cmsOptimizationCollection;
1833
1834
1835 // The built-in list. We currently implement 4 types of optimizations. Joining of curves, matrix-shaper, linearization and resampling
1836 static _cmsOptimizationCollection DefaultOptimization[] = {
1837
1838 { OptimizeByJoiningCurves, &DefaultOptimization[1] },
1839 { OptimizeMatrixShaper, &DefaultOptimization[2] },
1840 { OptimizeByComputingLinearization, &DefaultOptimization[3] },
1841 { OptimizeByResampling, NULL }
1842 };
1843
1844 // The linked list head
1845 _cmsOptimizationPluginChunkType _cmsOptimizationPluginChunk = { NULL };
1846
1847
1848 // Duplicates the zone of memory used by the plug-in in the new context
1849 static
DupPluginOptimizationList(struct _cmsContext_struct * ctx,const struct _cmsContext_struct * src)1850 void DupPluginOptimizationList(struct _cmsContext_struct* ctx,
1851 const struct _cmsContext_struct* src)
1852 {
1853 _cmsOptimizationPluginChunkType newHead = { NULL };
1854 _cmsOptimizationCollection* entry;
1855 _cmsOptimizationCollection* Anterior = NULL;
1856 _cmsOptimizationPluginChunkType* head = (_cmsOptimizationPluginChunkType*) src->chunks[OptimizationPlugin];
1857
1858 _cmsAssert(ctx != NULL);
1859 _cmsAssert(head != NULL);
1860
1861 // Walk the list copying all nodes
1862 for (entry = head->OptimizationCollection;
1863 entry != NULL;
1864 entry = entry ->Next) {
1865
1866 _cmsOptimizationCollection *newEntry = ( _cmsOptimizationCollection *) _cmsSubAllocDup(ctx ->MemPool, entry, sizeof(_cmsOptimizationCollection));
1867
1868 if (newEntry == NULL)
1869 return;
1870
1871 // We want to keep the linked list order, so this is a little bit tricky
1872 newEntry -> Next = NULL;
1873 if (Anterior)
1874 Anterior -> Next = newEntry;
1875
1876 Anterior = newEntry;
1877
1878 if (newHead.OptimizationCollection == NULL)
1879 newHead.OptimizationCollection = newEntry;
1880 }
1881
1882 ctx ->chunks[OptimizationPlugin] = _cmsSubAllocDup(ctx->MemPool, &newHead, sizeof(_cmsOptimizationPluginChunkType));
1883 }
1884
_cmsAllocOptimizationPluginChunk(struct _cmsContext_struct * ctx,const struct _cmsContext_struct * src)1885 void _cmsAllocOptimizationPluginChunk(struct _cmsContext_struct* ctx,
1886 const struct _cmsContext_struct* src)
1887 {
1888 if (src != NULL) {
1889
1890 // Copy all linked list
1891 DupPluginOptimizationList(ctx, src);
1892 }
1893 else {
1894 static _cmsOptimizationPluginChunkType OptimizationPluginChunkType = { NULL };
1895 ctx ->chunks[OptimizationPlugin] = _cmsSubAllocDup(ctx ->MemPool, &OptimizationPluginChunkType, sizeof(_cmsOptimizationPluginChunkType));
1896 }
1897 }
1898
1899
1900 // Register new ways to optimize
_cmsRegisterOptimizationPlugin(cmsContext ContextID,cmsPluginBase * Data)1901 cmsBool _cmsRegisterOptimizationPlugin(cmsContext ContextID, cmsPluginBase* Data)
1902 {
1903 cmsPluginOptimization* Plugin = (cmsPluginOptimization*) Data;
1904 _cmsOptimizationPluginChunkType* ctx = ( _cmsOptimizationPluginChunkType*) _cmsContextGetClientChunk(ContextID, OptimizationPlugin);
1905 _cmsOptimizationCollection* fl;
1906
1907 if (Data == NULL) {
1908
1909 ctx->OptimizationCollection = NULL;
1910 return TRUE;
1911 }
1912
1913 // Optimizer callback is required
1914 if (Plugin ->OptimizePtr == NULL) return FALSE;
1915
1916 fl = (_cmsOptimizationCollection*) _cmsPluginMalloc(ContextID, sizeof(_cmsOptimizationCollection));
1917 if (fl == NULL) return FALSE;
1918
1919 // Copy the parameters
1920 fl ->OptimizePtr = Plugin ->OptimizePtr;
1921
1922 // Keep linked list
1923 fl ->Next = ctx->OptimizationCollection;
1924
1925 // Set the head
1926 ctx ->OptimizationCollection = fl;
1927
1928 // All is ok
1929 return TRUE;
1930 }
1931
1932 // The entry point for LUT optimization
_cmsOptimizePipeline(cmsContext ContextID,cmsPipeline ** PtrLut,cmsUInt32Number Intent,cmsUInt32Number * InputFormat,cmsUInt32Number * OutputFormat,cmsUInt32Number * dwFlags)1933 cmsBool CMSEXPORT _cmsOptimizePipeline(cmsContext ContextID,
1934 cmsPipeline** PtrLut,
1935 cmsUInt32Number Intent,
1936 cmsUInt32Number* InputFormat,
1937 cmsUInt32Number* OutputFormat,
1938 cmsUInt32Number* dwFlags)
1939 {
1940 _cmsOptimizationPluginChunkType* ctx = ( _cmsOptimizationPluginChunkType*) _cmsContextGetClientChunk(ContextID, OptimizationPlugin);
1941 _cmsOptimizationCollection* Opts;
1942 cmsBool AnySuccess = FALSE;
1943 cmsStage* mpe;
1944
1945 // A CLUT is being asked, so force this specific optimization
1946 if (*dwFlags & cmsFLAGS_FORCE_CLUT) {
1947
1948 PreOptimize(*PtrLut);
1949 return OptimizeByResampling(PtrLut, Intent, InputFormat, OutputFormat, dwFlags);
1950 }
1951
1952 // Anything to optimize?
1953 if ((*PtrLut) ->Elements == NULL) {
1954 _cmsPipelineSetOptimizationParameters(*PtrLut, FastIdentity16, (void*) *PtrLut, NULL, NULL);
1955 return TRUE;
1956 }
1957
1958 // Named color pipelines cannot be optimized
1959 for (mpe = cmsPipelineGetPtrToFirstStage(*PtrLut);
1960 mpe != NULL;
1961 mpe = cmsStageNext(mpe)) {
1962 if (cmsStageType(mpe) == cmsSigNamedColorElemType) return FALSE;
1963 }
1964
1965 // Try to get rid of identities and trivial conversions.
1966 AnySuccess = PreOptimize(*PtrLut);
1967
1968 // After removal do we end with an identity?
1969 if ((*PtrLut) ->Elements == NULL) {
1970 _cmsPipelineSetOptimizationParameters(*PtrLut, FastIdentity16, (void*) *PtrLut, NULL, NULL);
1971 return TRUE;
1972 }
1973
1974 // Do not optimize, keep all precision
1975 if (*dwFlags & cmsFLAGS_NOOPTIMIZE)
1976 return FALSE;
1977
1978 // Try plug-in optimizations
1979 for (Opts = ctx->OptimizationCollection;
1980 Opts != NULL;
1981 Opts = Opts ->Next) {
1982
1983 // If one schema succeeded, we are done
1984 if (Opts ->OptimizePtr(PtrLut, Intent, InputFormat, OutputFormat, dwFlags)) {
1985
1986 return TRUE; // Optimized!
1987 }
1988 }
1989
1990 // Try built-in optimizations
1991 for (Opts = DefaultOptimization;
1992 Opts != NULL;
1993 Opts = Opts ->Next) {
1994
1995 if (Opts ->OptimizePtr(PtrLut, Intent, InputFormat, OutputFormat, dwFlags)) {
1996
1997 return TRUE;
1998 }
1999 }
2000
2001 // Only simple optimizations succeeded
2002 return AnySuccess;
2003 }
2004
2005
2006
2007