1 // Copyright (C) 2002-2012 Nikolaus Gebhardt / Thomas Alten
2 // This file is part of the "Irrlicht Engine".
3 // For conditions of distribution and use, see copyright notice in irrlicht.h
4 
5 #include "IrrCompileConfig.h"
6 #include "IBurningShader.h"
7 
8 #ifdef _IRR_COMPILE_WITH_BURNINGSVIDEO_
9 
10 // compile flag for this file
11 #undef USE_ZBUFFER
12 #undef IPOL_Z
13 #undef CMP_Z
14 #undef WRITE_Z
15 
16 #undef IPOL_W
17 #undef CMP_W
18 #undef WRITE_W
19 
20 #undef SUBTEXEL
21 #undef INVERSE_W
22 
23 #undef IPOL_C0
24 #undef IPOL_T0
25 #undef IPOL_T1
26 
27 // define render case
28 #define SUBTEXEL
29 #define INVERSE_W
30 
31 #define USE_ZBUFFER
32 #define IPOL_W
33 #define CMP_W
34 #define WRITE_W
35 
36 //#define IPOL_C0
37 #define IPOL_T0
38 //#define IPOL_T1
39 
40 // apply global override
41 #ifndef SOFTWARE_DRIVER_2_PERSPECTIVE_CORRECT
42 	#undef INVERSE_W
43 #endif
44 
45 #ifndef SOFTWARE_DRIVER_2_SUBTEXEL
46 	#undef SUBTEXEL
47 #endif
48 
49 #ifndef SOFTWARE_DRIVER_2_USE_VERTEX_COLOR
50 	#undef IPOL_C0
51 #endif
52 
53 #if !defined ( SOFTWARE_DRIVER_2_USE_WBUFFER ) && defined ( USE_ZBUFFER )
54 	#ifndef SOFTWARE_DRIVER_2_PERSPECTIVE_CORRECT
55 		#undef IPOL_W
56 	#endif
57 	#define IPOL_Z
58 
59 	#ifdef CMP_W
60 		#undef CMP_W
61 		#define CMP_Z
62 	#endif
63 
64 	#ifdef WRITE_W
65 		#undef WRITE_W
66 		#define WRITE_Z
67 	#endif
68 
69 #endif
70 
71 
72 
73 namespace irr
74 {
75 
76 namespace video
77 {
78 
79 class CTRTextureGouraudAdd2 : public IBurningShader
80 {
81 public:
82 
83 	//! constructor
84 	CTRTextureGouraudAdd2(CBurningVideoDriver* driver);
85 
86 	//! draws an indexed triangle list
87 	virtual void drawTriangle ( const s4DVertex *a,const s4DVertex *b,const s4DVertex *c );
88 
89 
90 private:
91 	void scanline_bilinear ();
92 	sScanLineData line;
93 
94 };
95 
96 //! constructor
CTRTextureGouraudAdd2(CBurningVideoDriver * driver)97 CTRTextureGouraudAdd2::CTRTextureGouraudAdd2(CBurningVideoDriver* driver)
98 : IBurningShader(driver)
99 {
100 	#ifdef _DEBUG
101 	setDebugName("CTRTextureGouraudAdd2");
102 	#endif
103 }
104 
105 
106 
107 /*!
108 */
scanline_bilinear()109 void CTRTextureGouraudAdd2::scanline_bilinear ()
110 {
111 	tVideoSample *dst;
112 
113 #ifdef USE_ZBUFFER
114 	fp24 *z;
115 #endif
116 
117 	s32 xStart;
118 	s32 xEnd;
119 	s32 dx;
120 
121 
122 #ifdef SUBTEXEL
123 	f32 subPixel;
124 #endif
125 
126 #ifdef IPOL_Z
127 	f32 slopeZ;
128 #endif
129 #ifdef IPOL_W
130 	fp24 slopeW;
131 #endif
132 #ifdef IPOL_C0
133 	sVec4 slopeC;
134 #endif
135 #ifdef IPOL_T0
136 	sVec2 slopeT[BURNING_MATERIAL_MAX_TEXTURES];
137 #endif
138 
139 	// apply top-left fill-convention, left
140 	xStart = core::ceil32( line.x[0] );
141 	xEnd = core::ceil32( line.x[1] ) - 1;
142 
143 	dx = xEnd - xStart;
144 
145 	if ( dx < 0 )
146 		return;
147 
148 	// slopes
149 	const f32 invDeltaX = core::reciprocal_approxim ( line.x[1] - line.x[0] );
150 
151 #ifdef IPOL_Z
152 	slopeZ = (line.z[1] - line.z[0]) * invDeltaX;
153 #endif
154 #ifdef IPOL_W
155 	slopeW = (line.w[1] - line.w[0]) * invDeltaX;
156 #endif
157 #ifdef IPOL_C0
158 	slopeC = (line.c[1] - line.c[0]) * invDeltaX;
159 #endif
160 #ifdef IPOL_T0
161 	slopeT[0] = (line.t[0][1] - line.t[0][0]) * invDeltaX;
162 #endif
163 #ifdef IPOL_T1
164 	slopeT[1] = (line.t[1][1] - line.t[1][0]) * invDeltaX;
165 #endif
166 
167 #ifdef SUBTEXEL
168 	subPixel = ( (f32) xStart ) - line.x[0];
169 #ifdef IPOL_Z
170 	line.z[0] += slopeZ * subPixel;
171 #endif
172 #ifdef IPOL_W
173 	line.w[0] += slopeW * subPixel;
174 #endif
175 #ifdef IPOL_C0
176 	line.c[0] += slopeC * subPixel;
177 #endif
178 #ifdef IPOL_T0
179 	line.t[0][0] += slopeT[0] * subPixel;
180 #endif
181 #ifdef IPOL_T1
182 	line.t[1][0] += slopeT[1] * subPixel;
183 #endif
184 #endif
185 
186 	dst = (tVideoSample*)RenderTarget->lock() + ( line.y * RenderTarget->getDimension().Width ) + xStart;
187 
188 #ifdef USE_ZBUFFER
189 	z = (fp24*) DepthBuffer->lock() + ( line.y * RenderTarget->getDimension().Width ) + xStart;
190 #endif
191 
192 
193 #ifdef INVERSE_W
194 	f32 inversew;
195 #endif
196 
197 
198 #ifdef BURNINGVIDEO_RENDERER_FAST
199 	u32 dIndex = ( line.y & 3 ) << 2;
200 
201 #else
202 	tFixPoint tx0;
203 	tFixPoint ty0;
204 
205 	tFixPoint r0, g0, b0;
206 	tFixPoint r1, g1, b1;
207 #endif
208 
209 
210 	for ( s32 i = 0; i <= dx; ++i )
211 	{
212 #ifdef CMP_Z
213 		if ( line.z[0] < z[i] )
214 #endif
215 #ifdef CMP_W
216 		if ( line.w[0] >= z[i] )
217 #endif
218 
219 		{
220 
221 
222 #ifdef BURNINGVIDEO_RENDERER_FAST
223 
224 		const tFixPointu d = dithermask [ dIndex | ( i ) & 3 ];
225 
226 
227 #ifdef INVERSE_W
228 			inversew = fix_inverse32 ( line.w[0] );
229 
230 			dst[i] = PixelAdd32 (
231 						dst[i],
232 					getTexel_plain ( &IT[0],	d + tofix ( line.t[0][0].x,inversew),
233 												d + tofix ( line.t[0][0].y,inversew) )
234 												  );
235 #else
236 			dst[i] = PixelAdd32 (
237 						dst[i],
238 					getTexel_plain ( &IT[0],	d + tofix ( line.t[0][0].x),
239 												d + tofix ( line.t[0][0].y) )
240 												 );
241 
242 #endif
243 #else
244 
245 #ifdef INVERSE_W
246 			inversew = fix_inverse32 ( line.w[0] );
247 
248 			tx0 = tofix ( line.t[0][0].x,inversew);
249 			ty0 = tofix ( line.t[0][0].y,inversew);
250 #else
251 			tx0 = tofix ( line.t[0][0].x );
252 			ty0 = tofix ( line.t[0][0].y );
253 #endif
254 			getSample_texture ( r0, g0, b0, &IT[0], tx0,ty0 );
255 
256 			color_to_fix ( r1, g1, b1, dst[i] );
257 
258 			dst[i] = fix_to_color ( clampfix_maxcolor ( r1 + r0 ),
259 									clampfix_maxcolor ( g1 + g0 ),
260 									clampfix_maxcolor ( b1 + b0 )
261 								);
262 #endif
263 
264 #ifdef WRITE_Z
265 			z[i] = line.z[0];
266 #endif
267 #ifdef WRITE_W
268 			z[i] = line.w[0];
269 #endif
270 
271 		}
272 
273 #ifdef IPOL_Z
274 		line.z[0] += slopeZ;
275 #endif
276 #ifdef IPOL_W
277 		line.w[0] += slopeW;
278 #endif
279 #ifdef IPOL_C0
280 		line.c[0] += slopeC;
281 #endif
282 #ifdef IPOL_T0
283 		line.t[0][0] += slopeT[0];
284 #endif
285 #ifdef IPOL_T1
286 		line.t[1][0] += slopeT[1];
287 #endif
288 	}
289 
290 }
291 
drawTriangle(const s4DVertex * a,const s4DVertex * b,const s4DVertex * c)292 void CTRTextureGouraudAdd2::drawTriangle ( const s4DVertex *a,const s4DVertex *b,const s4DVertex *c )
293 {
294 	sScanConvertData scan;
295 
296 	// sort on height, y
297 	if ( F32_A_GREATER_B ( a->Pos.y , b->Pos.y ) ) swapVertexPointer(&a, &b);
298 	if ( F32_A_GREATER_B ( b->Pos.y , c->Pos.y ) ) swapVertexPointer(&b, &c);
299 	if ( F32_A_GREATER_B ( a->Pos.y , b->Pos.y ) ) swapVertexPointer(&a, &b);
300 
301 	const f32 ca = c->Pos.y - a->Pos.y;
302 	const f32 ba = b->Pos.y - a->Pos.y;
303 	const f32 cb = c->Pos.y - b->Pos.y;
304 	// calculate delta y of the edges
305 	scan.invDeltaY[0] = core::reciprocal( ca );
306 	scan.invDeltaY[1] = core::reciprocal( ba );
307 	scan.invDeltaY[2] = core::reciprocal( cb );
308 
309 	// find if the major edge is left or right aligned
310 	f32 temp[4];
311 
312 	temp[0] = a->Pos.x - c->Pos.x;
313 	temp[1] = -ca;
314 	temp[2] = b->Pos.x - a->Pos.x;
315 	temp[3] = ba;
316 
317 	scan.left = ( temp[0] * temp[3] - temp[1] * temp[2] ) > 0.f ? 0 : 1;
318 	scan.right = 1 - scan.left;
319 
320 	// calculate slopes for the major edge
321 	scan.slopeX[0] = (c->Pos.x - a->Pos.x) * scan.invDeltaY[0];
322 	scan.x[0] = a->Pos.x;
323 
324 #ifdef IPOL_Z
325 	scan.slopeZ[0] = (c->Pos.z - a->Pos.z) * scan.invDeltaY[0];
326 	scan.z[0] = a->Pos.z;
327 #endif
328 
329 #ifdef IPOL_W
330 	scan.slopeW[0] = (c->Pos.w - a->Pos.w) * scan.invDeltaY[0];
331 	scan.w[0] = a->Pos.w;
332 #endif
333 
334 #ifdef IPOL_C0
335 	scan.slopeC[0] = (c->Color[0] - a->Color[0]) * scan.invDeltaY[0];
336 	scan.c[0] = a->Color[0];
337 #endif
338 
339 #ifdef IPOL_T0
340 	scan.slopeT[0][0] = (c->Tex[0] - a->Tex[0]) * scan.invDeltaY[0];
341 	scan.t[0][0] = a->Tex[0];
342 #endif
343 
344 #ifdef IPOL_T1
345 	scan.slopeT[1][0] = (c->Tex[1] - a->Tex[1]) * scan.invDeltaY[0];
346 	scan.t[1][0] = a->Tex[1];
347 #endif
348 
349 	// top left fill convention y run
350 	s32 yStart;
351 	s32 yEnd;
352 
353 #ifdef SUBTEXEL
354 	f32 subPixel;
355 #endif
356 
357 	// rasterize upper sub-triangle
358 	if ( F32_GREATER_0 ( scan.invDeltaY[1] )  )
359 	{
360 		// calculate slopes for top edge
361 		scan.slopeX[1] = (b->Pos.x - a->Pos.x) * scan.invDeltaY[1];
362 		scan.x[1] = a->Pos.x;
363 
364 #ifdef IPOL_Z
365 		scan.slopeZ[1] = (b->Pos.z - a->Pos.z) * scan.invDeltaY[1];
366 		scan.z[1] = a->Pos.z;
367 #endif
368 
369 #ifdef IPOL_W
370 		scan.slopeW[1] = (b->Pos.w - a->Pos.w) * scan.invDeltaY[1];
371 		scan.w[1] = a->Pos.w;
372 #endif
373 
374 #ifdef IPOL_C0
375 		scan.slopeC[1] = (b->Color[0] - a->Color[0]) * scan.invDeltaY[1];
376 		scan.c[1] = a->Color[0];
377 #endif
378 
379 #ifdef IPOL_T0
380 		scan.slopeT[0][1] = (b->Tex[0] - a->Tex[0]) * scan.invDeltaY[1];
381 		scan.t[0][1] = a->Tex[0];
382 #endif
383 
384 #ifdef IPOL_T1
385 		scan.slopeT[1][1] = (b->Tex[1] - a->Tex[1]) * scan.invDeltaY[1];
386 		scan.t[1][1] = a->Tex[1];
387 #endif
388 
389 		// apply top-left fill convention, top part
390 		yStart = core::ceil32( a->Pos.y );
391 		yEnd = core::ceil32( b->Pos.y ) - 1;
392 
393 #ifdef SUBTEXEL
394 		subPixel = ( (f32) yStart ) - a->Pos.y;
395 
396 		// correct to pixel center
397 		scan.x[0] += scan.slopeX[0] * subPixel;
398 		scan.x[1] += scan.slopeX[1] * subPixel;
399 
400 #ifdef IPOL_Z
401 		scan.z[0] += scan.slopeZ[0] * subPixel;
402 		scan.z[1] += scan.slopeZ[1] * subPixel;
403 #endif
404 
405 #ifdef IPOL_W
406 		scan.w[0] += scan.slopeW[0] * subPixel;
407 		scan.w[1] += scan.slopeW[1] * subPixel;
408 #endif
409 
410 #ifdef IPOL_C0
411 		scan.c[0] += scan.slopeC[0] * subPixel;
412 		scan.c[1] += scan.slopeC[1] * subPixel;
413 #endif
414 
415 #ifdef IPOL_T0
416 		scan.t[0][0] += scan.slopeT[0][0] * subPixel;
417 		scan.t[0][1] += scan.slopeT[0][1] * subPixel;
418 #endif
419 
420 #ifdef IPOL_T1
421 		scan.t[1][0] += scan.slopeT[1][0] * subPixel;
422 		scan.t[1][1] += scan.slopeT[1][1] * subPixel;
423 #endif
424 
425 #endif
426 
427 		// rasterize the edge scanlines
428 		for( line.y = yStart; line.y <= yEnd; ++line.y)
429 		{
430 			line.x[scan.left] = scan.x[0];
431 			line.x[scan.right] = scan.x[1];
432 
433 #ifdef IPOL_Z
434 			line.z[scan.left] = scan.z[0];
435 			line.z[scan.right] = scan.z[1];
436 #endif
437 
438 #ifdef IPOL_W
439 			line.w[scan.left] = scan.w[0];
440 			line.w[scan.right] = scan.w[1];
441 #endif
442 
443 #ifdef IPOL_C0
444 			line.c[scan.left] = scan.c[0];
445 			line.c[scan.right] = scan.c[1];
446 #endif
447 
448 #ifdef IPOL_T0
449 			line.t[0][scan.left] = scan.t[0][0];
450 			line.t[0][scan.right] = scan.t[0][1];
451 #endif
452 
453 #ifdef IPOL_T1
454 			line.t[1][scan.left] = scan.t[1][0];
455 			line.t[1][scan.right] = scan.t[1][1];
456 #endif
457 
458 			// render a scanline
459 			scanline_bilinear ();
460 
461 			scan.x[0] += scan.slopeX[0];
462 			scan.x[1] += scan.slopeX[1];
463 
464 #ifdef IPOL_Z
465 			scan.z[0] += scan.slopeZ[0];
466 			scan.z[1] += scan.slopeZ[1];
467 #endif
468 
469 #ifdef IPOL_W
470 			scan.w[0] += scan.slopeW[0];
471 			scan.w[1] += scan.slopeW[1];
472 #endif
473 
474 #ifdef IPOL_C0
475 			scan.c[0] += scan.slopeC[0];
476 			scan.c[1] += scan.slopeC[1];
477 #endif
478 
479 #ifdef IPOL_T0
480 			scan.t[0][0] += scan.slopeT[0][0];
481 			scan.t[0][1] += scan.slopeT[0][1];
482 #endif
483 
484 #ifdef IPOL_T1
485 			scan.t[1][0] += scan.slopeT[1][0];
486 			scan.t[1][1] += scan.slopeT[1][1];
487 #endif
488 
489 		}
490 	}
491 
492 	// rasterize lower sub-triangle
493 	if ( (f32) 0.0 != scan.invDeltaY[2] )
494 	{
495 		// advance to middle point
496 		if( (f32) 0.0 != scan.invDeltaY[1] )
497 		{
498 			temp[0] = b->Pos.y - a->Pos.y;	// dy
499 
500 			scan.x[0] = a->Pos.x + scan.slopeX[0] * temp[0];
501 #ifdef IPOL_Z
502 			scan.z[0] = a->Pos.z + scan.slopeZ[0] * temp[0];
503 #endif
504 #ifdef IPOL_W
505 			scan.w[0] = a->Pos.w + scan.slopeW[0] * temp[0];
506 #endif
507 #ifdef IPOL_C0
508 			scan.c[0] = a->Color[0] + scan.slopeC[0] * temp[0];
509 #endif
510 #ifdef IPOL_T0
511 			scan.t[0][0] = a->Tex[0] + scan.slopeT[0][0] * temp[0];
512 #endif
513 #ifdef IPOL_T1
514 			scan.t[1][0] = a->Tex[1] + scan.slopeT[1][0] * temp[0];
515 #endif
516 
517 		}
518 
519 		// calculate slopes for bottom edge
520 		scan.slopeX[1] = (c->Pos.x - b->Pos.x) * scan.invDeltaY[2];
521 		scan.x[1] = b->Pos.x;
522 
523 #ifdef IPOL_Z
524 		scan.slopeZ[1] = (c->Pos.z - b->Pos.z) * scan.invDeltaY[2];
525 		scan.z[1] = b->Pos.z;
526 #endif
527 
528 #ifdef IPOL_W
529 		scan.slopeW[1] = (c->Pos.w - b->Pos.w) * scan.invDeltaY[2];
530 		scan.w[1] = b->Pos.w;
531 #endif
532 
533 #ifdef IPOL_C0
534 		scan.slopeC[1] = (c->Color[0] - b->Color[0]) * scan.invDeltaY[2];
535 		scan.c[1] = b->Color[0];
536 #endif
537 
538 #ifdef IPOL_T0
539 		scan.slopeT[0][1] = (c->Tex[0] - b->Tex[0]) * scan.invDeltaY[2];
540 		scan.t[0][1] = b->Tex[0];
541 #endif
542 
543 #ifdef IPOL_T1
544 		scan.slopeT[1][1] = (c->Tex[1] - b->Tex[1]) * scan.invDeltaY[2];
545 		scan.t[1][1] = b->Tex[1];
546 #endif
547 
548 		// apply top-left fill convention, top part
549 		yStart = core::ceil32( b->Pos.y );
550 		yEnd = core::ceil32( c->Pos.y ) - 1;
551 
552 #ifdef SUBTEXEL
553 
554 		subPixel = ( (f32) yStart ) - b->Pos.y;
555 
556 		// correct to pixel center
557 		scan.x[0] += scan.slopeX[0] * subPixel;
558 		scan.x[1] += scan.slopeX[1] * subPixel;
559 
560 #ifdef IPOL_Z
561 		scan.z[0] += scan.slopeZ[0] * subPixel;
562 		scan.z[1] += scan.slopeZ[1] * subPixel;
563 #endif
564 
565 #ifdef IPOL_W
566 		scan.w[0] += scan.slopeW[0] * subPixel;
567 		scan.w[1] += scan.slopeW[1] * subPixel;
568 #endif
569 
570 #ifdef IPOL_C0
571 		scan.c[0] += scan.slopeC[0] * subPixel;
572 		scan.c[1] += scan.slopeC[1] * subPixel;
573 #endif
574 
575 #ifdef IPOL_T0
576 		scan.t[0][0] += scan.slopeT[0][0] * subPixel;
577 		scan.t[0][1] += scan.slopeT[0][1] * subPixel;
578 #endif
579 
580 #ifdef IPOL_T1
581 		scan.t[1][0] += scan.slopeT[1][0] * subPixel;
582 		scan.t[1][1] += scan.slopeT[1][1] * subPixel;
583 #endif
584 
585 #endif
586 
587 		// rasterize the edge scanlines
588 		for( line.y = yStart; line.y <= yEnd; ++line.y)
589 		{
590 			line.x[scan.left] = scan.x[0];
591 			line.x[scan.right] = scan.x[1];
592 
593 #ifdef IPOL_Z
594 			line.z[scan.left] = scan.z[0];
595 			line.z[scan.right] = scan.z[1];
596 #endif
597 
598 #ifdef IPOL_W
599 			line.w[scan.left] = scan.w[0];
600 			line.w[scan.right] = scan.w[1];
601 #endif
602 
603 #ifdef IPOL_C0
604 			line.c[scan.left] = scan.c[0];
605 			line.c[scan.right] = scan.c[1];
606 #endif
607 
608 #ifdef IPOL_T0
609 			line.t[0][scan.left] = scan.t[0][0];
610 			line.t[0][scan.right] = scan.t[0][1];
611 #endif
612 
613 #ifdef IPOL_T1
614 			line.t[1][scan.left] = scan.t[1][0];
615 			line.t[1][scan.right] = scan.t[1][1];
616 #endif
617 
618 			// render a scanline
619 			scanline_bilinear ();
620 
621 			scan.x[0] += scan.slopeX[0];
622 			scan.x[1] += scan.slopeX[1];
623 
624 #ifdef IPOL_Z
625 			scan.z[0] += scan.slopeZ[0];
626 			scan.z[1] += scan.slopeZ[1];
627 #endif
628 
629 #ifdef IPOL_W
630 			scan.w[0] += scan.slopeW[0];
631 			scan.w[1] += scan.slopeW[1];
632 #endif
633 
634 #ifdef IPOL_C0
635 			scan.c[0] += scan.slopeC[0];
636 			scan.c[1] += scan.slopeC[1];
637 #endif
638 
639 #ifdef IPOL_T0
640 			scan.t[0][0] += scan.slopeT[0][0];
641 			scan.t[0][1] += scan.slopeT[0][1];
642 #endif
643 
644 #ifdef IPOL_T1
645 			scan.t[1][0] += scan.slopeT[1][0];
646 			scan.t[1][1] += scan.slopeT[1][1];
647 #endif
648 
649 		}
650 	}
651 
652 }
653 
654 
655 } // end namespace video
656 } // end namespace irr
657 
658 #endif // _IRR_COMPILE_WITH_BURNINGSVIDEO_
659 
660 namespace irr
661 {
662 namespace video
663 {
664 
665 //! creates a flat triangle renderer
createTRTextureGouraudAdd2(CBurningVideoDriver * driver)666 IBurningShader* createTRTextureGouraudAdd2(CBurningVideoDriver* driver)
667 {
668 	#ifdef _IRR_COMPILE_WITH_BURNINGSVIDEO_
669 	return new CTRTextureGouraudAdd2(driver);
670 	#else
671 	return 0;
672 	#endif // _IRR_COMPILE_WITH_BURNINGSVIDEO_
673 }
674 
675 
676 } // end namespace video
677 } // end namespace irr
678 
679 
680 
681