gba-x3dtest: 62390f9cc93e src/x3d.c

gba-x3dtest

view src/x3d.c @ 19:62390f9cc93e

texture cache optimization failed

author	John Tsiombikas <nuclear@member.fsf.org>
date	Mon, 30 Jun 2014 09:07:41 +0300
parents	0a7f402892b3
children	2e903e27e35a

line source

1 #include "config.h"

2 #include <stdio.h>

3 #include <stdlib.h>

4 #include <string.h>

5 #include <math.h>

6 #include "x3d.h"

7 #include "fixed.h"

8 #include "sincos.h"

9 #include "logger.h"

10 #include "polyfill.h"

11 #include "gbasys.h"

12 #include "x3d_impl.h"

14 int dbg_fill_dump;

16 #define MAT_STACK_SIZE 4

18 struct matrix {

19 int32_t m[12];

20 };

22 static void proc_vertex(const int32_t *vin, const int32_t *cin, const int32_t *tin,

23 pvec3 *vout, pvec3 *cout, pvec2 *tout);

24 static int dump_frame(struct pixel_buffer *frame);

27 static int32_t proj_fov = M_PI_X16;

28 static int32_t proj_aspect = 65536;

29 static int32_t inv_proj_aspect = 65536;

30 static int32_t proj_near = ftox16(0.5);

31 static int32_t proj_far = 500 << 16;

32 static int32_t inv_tan_half_xfov, inv_tan_half_yfov;

34 #define ID_INIT {65536, 0, 0, 0, 0, 65536, 0, 0, 0, 0, 65536, 0}

36 static struct matrix identity = { ID_INIT };

38 static short mtop;

39 static struct matrix mstack[MAT_STACK_SIZE] = { {ID_INIT}, {ID_INIT} };

41 static const int32_t *vertex_array;

42 static unsigned short vertex_count;

43 static const int32_t *color_array;

44 static unsigned short color_count;

45 static const int32_t *texcoord_array;

46 static unsigned short texcoord_count;

48 static int32_t im_color[3];

49 static int32_t im_texcoord[2];

50 static uint8_t im_color_index;

52 #define MAX_TEXTURES 64

53 static struct texture textures[MAX_TEXTURES];

54 static short cur_tex = -1;

57 void x3d_projection(int fov, int32_t aspect, int32_t nearz, int32_t farz)

58 {

59 proj_fov = (M_PI_X16 * fov) / 180;

60 proj_aspect = aspect;

61 inv_proj_aspect = x16div(65536, proj_aspect);

62 proj_near = nearz;

63 proj_far = farz;

65 inv_tan_half_yfov = (int32_t)(65536.0 / tan(0.5 * proj_fov / 65536.0));

66 inv_tan_half_xfov = x16mul(inv_tan_half_yfov, aspect);

67 }

69 int x3d_push_matrix(void)

70 {

71 short newtop = mtop + 1;

72 if(newtop >= MAT_STACK_SIZE) {

73 return -1;

74 }

75 memcpy(mstack + newtop, mstack + mtop, sizeof *mstack);

76 mtop = newtop;

77 return 0;

78 }

80 int x3d_pop_matrix(void)

81 {

82 if(mtop <= 0) {

83 return -1;

84 }

85 --mtop;

86 return 0;

87 }

89 void x3d_load_matrix(int32_t *m)

90 {

91 memcpy(mstack[mtop].m, m, sizeof *mstack);

92 }

95 #define M(i,j) (((i) << 2) + (j))

96 void x3d_mult_matrix(int32_t *m)

97 {

98 int i, j;

99 struct matrix tmp;

100

101 memcpy(tmp.m, mstack[mtop].m, sizeof tmp);

102

103 for(i=0; i<3; i++) {

104 for(j=0; j<4; j++) {

105 mstack[mtop].m[M(i, j)] =

106 x16mul(m[M(0, j)], tmp.m[M(i, 0)]) +

107 x16mul(m[M(1, j)], tmp.m[M(i, 1)]) +

108 x16mul(m[M(2, j)], tmp.m[M(i, 2)]);

109 }

110 mstack[mtop].m[M(i, 3)] += tmp.m[M(i, 3)];

111 }

112 }

113

114 void x3d_load_identity(void)

115 {

116 memcpy(mstack[mtop].m, identity.m, sizeof identity);

117 }

118

119 void x3d_translate(int32_t x, int32_t y, int32_t z)

120 {

121 int32_t m[] = ID_INIT;

122 m[3] = x;

123 m[7] = y;

124 m[11] = z;

125

126 x3d_mult_matrix(m);

127 }

128

129 void x3d_rotate(int32_t deg, int32_t x, int32_t y, int32_t z)

130 {

131 int32_t xform[] = ID_INIT;

132

133 int32_t angle = x16mul(M_PI_X16, deg) / 180;

134 int32_t sina = sin_x16(angle);

135 int32_t cosa = cos_x16(angle);

136 int32_t one_minus_cosa = 65536 - cosa;

137 int32_t nxsq = x16sq(x);

138 int32_t nysq = x16sq(y);

139 int32_t nzsq = x16sq(z);

140

141 xform[0] = nxsq + x16mul(65536 - nxsq, cosa);

142 xform[4] = x16mul(x16mul(x, y), one_minus_cosa) - x16mul(z, sina);

143 xform[8] = x16mul(x16mul(x, z), one_minus_cosa) + x16mul(y, sina);

144 xform[1] = x16mul(x16mul(x, y), one_minus_cosa) + x16mul(z, sina);

145 xform[5] = nysq + x16mul(65536 - nysq, cosa);

146 xform[9] = x16mul(x16mul(y, z), one_minus_cosa) - x16mul(x, sina);

147 xform[2] = x16mul(x16mul(x, z), one_minus_cosa) - x16mul(y, sina);

148 xform[6] = x16mul(x16mul(y, z), one_minus_cosa) + x16mul(x, sina);

149 xform[10] = nzsq + x16mul(65536 - nzsq, cosa);

150

151 x3d_mult_matrix(xform);

152 }

153

154 void x3d_scale(int32_t x, int32_t y, int32_t z)

155 {

156 int32_t m[] = ID_INIT;

157

158 m[0] = x;

159 m[5] = y;

160 m[10] = z;

161

162 x3d_mult_matrix(m);

163 }

164

165 void x3d_vertex_array(int count, const int32_t *ptr)

166 {

167 vertex_array = ptr;

168 vertex_count = count;

169 }

170

171 void x3d_color_array(int count, const int32_t *ptr)

172 {

173 color_array = ptr;

174 color_count = count;

175 }

176

177 void x3d_texcoord_array(int count, const int32_t *ptr)

178 {

179 texcoord_array = ptr;

180 texcoord_count = count;

181 }

182

183 int x3d_draw(int prim, int vnum)

184 {

185 int i, j, pverts = prim;

186 const int32_t *vptr = vertex_array;

187 const int32_t *cptr = color_array;

188 const int32_t *tptr = texcoord_array;

189 #ifndef PALMODE

190 short cr, cg, cb;

191 #endif

192 uint16_t color;

193

194 if(!vertex_array) return -1;

195

196 if(vnum > vertex_count) {

197 logmsg(LOG_DBG, "%s called with vnum=%d, but current vertex array has %d vertices\n",

198 __FUNCTION__, vnum, vertex_count);

199 vnum = vertex_count;

200 }

201 if(color_array && vnum > color_count) {

202 logmsg(LOG_DBG, "%s called with vnum=%d, but current color array has %d elements\n",

203 __FUNCTION__, vnum, color_count);

204 vnum = color_count;

205 }

206 if(texcoord_array && vnum > texcoord_count) {

207 logmsg(LOG_DBG, "%s called with vnum=%d, but current texcoord array has %d elements\n",

208 __FUNCTION__, vnum, texcoord_count);

209 vnum = texcoord_count;

210 }

211

212 for(i=0; i<vnum; i+=pverts) {

213 /* process vertices */

214 pvec3 vpos[4];

215 pvec3 col[4];

216 pvec2 tex[4];

217

218 for(j=0; j<pverts; j++) {

219 proc_vertex(vptr, cptr, tptr, vpos + j, col + j, tex + j);

220

221 if(vpos[j].z <= proj_near) {

222 goto skip_prim;

223 }

224

225 vptr += 3;

226 if(cptr) cptr += 3;

227 if(tptr) tptr += 2;

228 }

229

230 #ifdef PALMODE

231 color = im_color_index;

232 #else

233 cr = col[0].x >> 8;

234 cg = col[0].y >> 8;

235 cb = col[0].z >> 8;

236

237 if(cr > 255) cr = 255;

238 if(cg > 255) cg = 255;

239 if(cb > 255) cb = 255;

240

241 color = RGB(cr, cg, cb);

242 #endif

243

244 /* project & viewport */

245 for(j=0; j<pverts; j++) {

246 int32_t x, y;

247

248 x = x16mul(vpos[j].x, inv_tan_half_xfov);

249 x = x16div(x, vpos[j].z);

250 vpos[j].x = (x16mul(x, inv_proj_aspect) + 65536) * (WIDTH / 2);

251

252 y = x16mul(vpos[j].y, inv_tan_half_yfov);

253 y = x16div(y, vpos[j].z);

254 vpos[j].y = (65536 - y) * (HEIGHT / 2);

255 }

256

257 switch(pverts) {

258 case X3D_POINTS:

259 draw_point(vpos, color);

260 break;

261

262 case X3D_LINES:

263 break;

264

265 case X3D_TRIANGLES:

266 case X3D_QUADS:

267 draw_poly(pverts, vpos, tex, color, cur_tex >= 0 ? textures + cur_tex : 0);

268 if(dbg_fill_dump) {

269 dump_frame(back_buffer);

270 }

271 break;

272 }

273 skip_prim: ;

274 }

275

276 dbg_fill_dump = 0;

277 return 0;

278 }

279

280 static void proc_vertex(const int32_t *vin, const int32_t *cin, const int32_t *tin,

281 pvec3 *vout, pvec3 *cout, pvec2 *tout)

282 {

283 int i;

284 int32_t tvert[3];

285 int32_t *mvmat = mstack[mtop].m;

286

287 /* transform vertex with current matrix */

288 for(i=0; i<3; i++) {

289 tvert[i] = x16mul(mvmat[0], vin[0]) +

290 x16mul(mvmat[1], vin[1]) +

291 x16mul(mvmat[2], vin[2]) +

292 mvmat[3];

293 mvmat += 4;

294 }

295

296 vout->x = tvert[0];

297 vout->y = tvert[1];

298 vout->z = tvert[2];

299 /*logmsg(LOG_DBG, "%s: (%g %g %g) -> (%g %g %g)\n", __FUNCTION__,

300 x16tof(vin[0]), x16tof(vin[1]), x16tof(vin[2]),

301 x16tof(vout->x), x16tof(vout->y), x16tof(vout->z));*/

302

303 if(color_array) {

304 cout->x = cin[0];

305 cout->y = cin[1];

306 cout->z = cin[2];

307 } else {

308 cout->x = im_color[0];

309 cout->y = im_color[1];

310 cout->z = im_color[2];

311 }

312

313 if(texcoord_array) {

314 tout->x = tin[0];

315 tout->y = tin[1];

316 } else {

317 tout->x = im_texcoord[0];

318 tout->y = im_texcoord[1];

319 }

320 }

321

322 void x3d_color_index(int cidx)

323 {

324 im_color_index = cidx;

325 }

326

327 void x3d_color(int32_t r, int32_t g, int32_t b)

328 {

329 im_color[0] = r;

330 im_color[1] = g;

331 im_color[2] = b;

332 }

333

334 static int count_bits(int x)

335 {

336 int i, count = 0;

337 for(i=0; i<32; i++) {

338 if(x & 1) count++;

339 x >>= 1;

340 }

341 return count;

342 }

343

344 int x3d_create_texture_rgb(int xsz, int ysz, const uint16_t *pixels)

345 {

346 int i, j;

347

348 if(xsz == 0 || count_bits(xsz) > 1 || ysz == 0 || count_bits(ysz) > 1) {

349 logmsg(LOG_DBG, "%s: texture size (%dx%d) not power of two!\n", __func__, xsz, ysz);

350 return -1;

351 }

352

353 for(i=0; i<MAX_TEXTURES; i++) {

354 if(!textures[i].pixels) {

355 /*textures[i].pixels = pixels;*/

356 textures[i].pixels = malloc(xsz * ysz * 2);

357 memcpy(textures[i].pixels, pixels, xsz * ysz * 2);

358 textures[i].xsz = xsz;

359 textures[i].ysz = ysz;

360 textures[i].umask = xsz - 1;

361 textures[i].vmask = ysz - 1;

362

363 for(j=0; j<32; j++) {

364 if((1 << j) == xsz) {

365 textures[i].ushift = j;

366 }

367 if((1 << j) == ysz) {

368 textures[i].vshift = j;

369 }

370 }

371

372 logmsg(LOG_DBG, "create texture %dx%d: %p\n", xsz, ysz, pixels);

373

374 return i;

375 }

376 }

377 return -1;

378 }

379

380 void x3d_enable_texture(int texid)

381 {

382 cur_tex = texid;

383 }

384

385 void x3d_disable_texture(void)

386 {

387 cur_tex = 0;

388 }

389

390 int x3d_get_active_texture(void)

391 {

392 return cur_tex;

393 }

394

395 static int dump_frame(struct pixel_buffer *frame)

396 {

397 static int frameno;

398 char buf[128];

399 FILE *fp;

400 int i, npix;

401 uint16_t *ptr = frame->pixels;

402

403 sprintf(buf, "dump%03d.ppm", ++frameno);

404

405 if(!(fp = fopen(buf, "wb"))) {

406 fprintf(stderr, "failed to dump file: %s\n", buf);

407 return -1;

408 }

409

410 fprintf(fp, "P6\n%d %d\n255\n", frame->x, frame->y);

411

412 npix = frame->x * frame->y;

413 for(i=0; i<npix; i++) {

414 uint16_t pixel = *ptr++;

415 fputc(GET_R(pixel), fp);

416 fputc(GET_G(pixel), fp);

417 fputc(GET_B(pixel), fp);

418 }

419 fclose(fp);

420 return 0;

421 }