| 1 | /* |
| 2 | * Glide64 - Glide video plugin for Nintendo 64 emulators. |
| 3 | * Copyright (c) 2002 Dave2001 |
| 4 | * Copyright (c) 2003-2009 Sergey 'Gonetz' Lipski |
| 5 | * |
| 6 | * This program is free software; you can redistribute it and/or modify |
| 7 | * it under the terms of the GNU General Public License as published by |
| 8 | * the Free Software Foundation; either version 2 of the License, or |
| 9 | * any later version. |
| 10 | * |
| 11 | * This program is distributed in the hope that it will be useful, |
| 12 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 13 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 14 | * GNU General Public License for more details. |
| 15 | * |
| 16 | * You should have received a copy of the GNU General Public License |
| 17 | * along with this program; if not, write to the Free Software |
| 18 | * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA |
| 19 | */ |
| 20 | |
| 21 | //**************************************************************** |
| 22 | // |
| 23 | // Glide64 - Glide Plugin for Nintendo 64 emulators |
| 24 | // Project started on December 29th, 2001 |
| 25 | // |
| 26 | // Authors: |
| 27 | // Dave2001, original author, founded the project in 2001, left it in 2002 |
| 28 | // Gugaman, joined the project in 2002, left it in 2002 |
| 29 | // Sergey 'Gonetz' Lipski, joined the project in 2002, main author since fall of 2002 |
| 30 | // Hiroshi 'KoolSmoky' Morii, joined the project in 2007 |
| 31 | // |
| 32 | //**************************************************************** |
| 33 | // |
| 34 | // To modify Glide64: |
| 35 | // * Write your name and (optional)email, commented by your work, so I know who did it, and so that you can find which parts you modified when it comes time to send it to me. |
| 36 | // * Do NOT send me the whole project or file that you modified. Take out your modified code sections, and tell me where to put them. If people sent the whole thing, I would have many different versions, but no idea how to combine them all. |
| 37 | // |
| 38 | //**************************************************************** |
| 39 | // |
| 40 | // Software rendering into N64 depth buffer |
| 41 | // Idea and N64 depth value format by Orkin |
| 42 | // Polygon rasterization algorithm is taken from FATMAP2 engine by Mats Byggmastar, mri@penti.sit.fi |
| 43 | // |
| 44 | // Created by Gonetz, Dec 2004 |
| 45 | // |
| 46 | //**************************************************************** |
| 47 | |
| 48 | #include "Gfx_1.3.h" |
| 49 | #include "rdp.h" |
| 50 | #include "DepthBufferRender.h" |
| 51 | |
| 52 | wxUint16 * zLUT = 0; |
| 53 | |
| 54 | void ZLUT_init() |
| 55 | { |
| 56 | if (zLUT) |
| 57 | return; |
| 58 | zLUT = new wxUint16[0x40000]; |
| 59 | for(int i=0; i<0x40000; i++) |
| 60 | { |
| 61 | wxUint32 exponent = 0; |
| 62 | wxUint32 testbit = 1 << 17; |
| 63 | while((i & testbit) && (exponent < 7)) |
| 64 | { |
| 65 | exponent++; |
| 66 | testbit = 1 << (17 - exponent); |
| 67 | } |
| 68 | |
| 69 | wxUint32 mantissa = (i >> (6 - (6 < exponent ? 6 : exponent))) & 0x7ff; |
| 70 | zLUT[i] = (wxUint16)(((exponent << 11) | mantissa) << 2); |
| 71 | } |
| 72 | } |
| 73 | |
| 74 | void ZLUT_release() |
| 75 | { |
| 76 | delete[] zLUT; |
| 77 | zLUT = 0; |
| 78 | } |
| 79 | |
| 80 | static vertexi * max_vtx; // Max y vertex (ending vertex) |
| 81 | static vertexi * start_vtx, * end_vtx; // First and last vertex in array |
| 82 | static vertexi * right_vtx, * left_vtx; // Current right and left vertex |
| 83 | |
| 84 | #ifdef HAVE_GLES |
| 85 | // using float should be faster than int in fact, especialy if using NEON |
| 86 | static float right_height, left_height; |
| 87 | static float right_x, right_dxdy, left_x, left_dxdy; |
| 88 | static float left_z, left_dzdy; |
| 89 | |
| 90 | __inline float iceil(int x) |
| 91 | { |
| 92 | return cellf((float)x*(1.0f/65536.0f)); |
| 93 | } |
| 94 | __inline float i2float(int x) |
| 95 | { |
| 96 | return (float)x*(1.0f/65536.0f); |
| 97 | } |
| 98 | __inline int float2i(float x) |
| 99 | { |
| 100 | return x*65536.0f; |
| 101 | } |
| 102 | |
| 103 | #else |
| 104 | |
| 105 | static int right_height, left_height; |
| 106 | static int right_x, right_dxdy, left_x, left_dxdy; |
| 107 | static int left_z, left_dzdy; |
| 108 | |
| 109 | __inline int imul16(int x, int y) // (x * y) >> 16 |
| 110 | { |
| 111 | return (((long long)x) * ((long long)y)) >> 16; |
| 112 | } |
| 113 | |
| 114 | __inline int imul14(int x, int y) // (x * y) >> 14 |
| 115 | { |
| 116 | return (((long long)x) * ((long long)y)) >> 14; |
| 117 | } |
| 118 | __inline int idiv16(int x, int y) // (x << 16) / y |
| 119 | { |
| 120 | //x = (((long long)x) << 16) / ((long long)y); |
| 121 | /* |
| 122 | eax = x; |
| 123 | ebx = y; |
| 124 | edx = x; |
| 125 | (x << 16) | () |
| 126 | */ |
| 127 | #if !defined(__GNUC__) && !defined(NO_ASM) |
| 128 | __asm { |
| 129 | mov eax, x |
| 130 | mov ebx, y |
| 131 | mov edx,eax |
| 132 | sar edx,16 |
| 133 | shl eax,16 |
| 134 | idiv ebx |
| 135 | mov x, eax |
| 136 | } |
| 137 | #elif !defined(NO_ASM) |
| 138 | int reminder; |
| 139 | asm ("idivl %[divisor]" |
| 140 | : "=a" (x), "=d" (reminder) |
| 141 | : [divisor] "g" (y), "d" (x >> 16), "a" (x << 16)); |
| 142 | #else |
| 143 | x = (((long long)x) << 16) / ((long long)y); |
| 144 | #endif |
| 145 | return x; |
| 146 | } |
| 147 | |
| 148 | __inline int iceil(int x) |
| 149 | { |
| 150 | x += 0xffff; |
| 151 | return (x >> 16); |
| 152 | } |
| 153 | #endif |
| 154 | |
| 155 | static void RightSection(void) |
| 156 | { |
| 157 | // Walk backwards trough the vertex array |
| 158 | |
| 159 | vertexi * v2, * v1 = right_vtx; |
| 160 | if(right_vtx > start_vtx) v2 = right_vtx-1; |
| 161 | else v2 = end_vtx; // Wrap to end of array |
| 162 | right_vtx = v2; |
| 163 | |
| 164 | // v1 = top vertex |
| 165 | // v2 = bottom vertex |
| 166 | |
| 167 | // Calculate number of scanlines in this section |
| 168 | |
| 169 | right_height = iceil(v2->y) - iceil(v1->y); |
| 170 | if(right_height <= 0) return; |
| 171 | |
| 172 | // Guard against possible div overflows |
| 173 | |
| 174 | #ifdef HAVE_GLES |
| 175 | right_dxdy = i2float(v2->x - v1->x)/i2float(v2->y - v1->y); |
| 176 | #else |
| 177 | if(right_height > 1) { |
| 178 | // OK, no worries, we have a section that is at least |
| 179 | // one pixel high. Calculate slope as usual. |
| 180 | |
| 181 | int height = v2->y - v1->y; |
| 182 | right_dxdy = idiv16(v2->x - v1->x, height); |
| 183 | } |
| 184 | else { |
| 185 | // Height is less or equal to one pixel. |
| 186 | // Calculate slope = width * 1/height |
| 187 | // using 18:14 bit precision to avoid overflows. |
| 188 | |
| 189 | int inv_height = (0x10000 << 14) / (v2->y - v1->y); |
| 190 | right_dxdy = imul14(v2->x - v1->x, inv_height); |
| 191 | } |
| 192 | #endif |
| 193 | |
| 194 | // Prestep initial values |
| 195 | |
| 196 | #ifdef HAVE_GLES |
| 197 | float prestep = iceil(v1->y) - i2float(v1->y); |
| 198 | right_x = i2float(v1->x) + prestep*right_dxdy; |
| 199 | #else |
| 200 | int prestep = (iceil(v1->y) << 16) - v1->y; |
| 201 | right_x = v1->x + imul16(prestep, right_dxdy); |
| 202 | #endif |
| 203 | } |
| 204 | |
| 205 | static void LeftSection(void) |
| 206 | { |
| 207 | // Walk forward trough the vertex array |
| 208 | |
| 209 | vertexi * v2, * v1 = left_vtx; |
| 210 | if(left_vtx < end_vtx) v2 = left_vtx+1; |
| 211 | else v2 = start_vtx; // Wrap to start of array |
| 212 | left_vtx = v2; |
| 213 | |
| 214 | // v1 = top vertex |
| 215 | // v2 = bottom vertex |
| 216 | |
| 217 | // Calculate number of scanlines in this section |
| 218 | |
| 219 | left_height = iceil(v2->y) - iceil(v1->y); |
| 220 | if(left_height <= 0) return; |
| 221 | |
| 222 | // Guard against possible div overflows |
| 223 | |
| 224 | #ifdef HAVE_GLES |
| 225 | float invheight = 1.0f/i2float(v2->y - v1->y) |
| 226 | left_dxdy = i2float(v2->x - v1->x)*invheight; |
| 227 | left_dzdy = i2float(v2->z - v1->z)*invheight; |
| 228 | #else |
| 229 | if(left_height > 1) { |
| 230 | // OK, no worries, we have a section that is at least |
| 231 | // one pixel high. Calculate slope as usual. |
| 232 | |
| 233 | int height = v2->y - v1->y; |
| 234 | left_dxdy = idiv16(v2->x - v1->x, height); |
| 235 | left_dzdy = idiv16(v2->z - v1->z, height); |
| 236 | } |
| 237 | else { |
| 238 | // Height is less or equal to one pixel. |
| 239 | // Calculate slope = width * 1/height |
| 240 | // using 18:14 bit precision to avoid overflows. |
| 241 | |
| 242 | int inv_height = (0x10000 << 14) / (v2->y - v1->y); |
| 243 | left_dxdy = imul14(v2->x - v1->x, inv_height); |
| 244 | left_dzdy = imul14(v2->z - v1->z, inv_height); |
| 245 | } |
| 246 | #endif |
| 247 | |
| 248 | // Prestep initial values |
| 249 | |
| 250 | #ifdef HAVE_GLES |
| 251 | float prestep = iceil(v1->y) - i2float(v1->y); |
| 252 | left_x = i2float(v1->x) + prestep*left_dxdy; |
| 253 | left_z = i2float(v1->z) + prestep*left_dzdy; |
| 254 | #else |
| 255 | int prestep = (iceil(v1->y) << 16) - v1->y; |
| 256 | left_x = v1->x + imul16(prestep, left_dxdy); |
| 257 | left_z = v1->z + imul16(prestep, left_dzdy); |
| 258 | #endif |
| 259 | } |
| 260 | |
| 261 | |
| 262 | void Rasterize(vertexi * vtx, int vertices, int dzdx) |
| 263 | { |
| 264 | start_vtx = vtx; // First vertex in array |
| 265 | |
| 266 | // Search trough the vtx array to find min y, max y |
| 267 | // and the location of these structures. |
| 268 | |
| 269 | vertexi * min_vtx = vtx; |
| 270 | max_vtx = vtx; |
| 271 | |
| 272 | int min_y = vtx->y; |
| 273 | int max_y = vtx->y; |
| 274 | |
| 275 | vtx++; |
| 276 | |
| 277 | for(int n=1; n<vertices; n++) { |
| 278 | if(vtx->y < min_y) { |
| 279 | min_y = vtx->y; |
| 280 | min_vtx = vtx; |
| 281 | } |
| 282 | else |
| 283 | if(vtx->y > max_y) { |
| 284 | max_y = vtx->y; |
| 285 | max_vtx = vtx; |
| 286 | } |
| 287 | vtx++; |
| 288 | } |
| 289 | |
| 290 | // OK, now we know where in the array we should start and |
| 291 | // where to end while scanning the edges of the polygon |
| 292 | |
| 293 | left_vtx = min_vtx; // Left side starting vertex |
| 294 | right_vtx = min_vtx; // Right side starting vertex |
| 295 | end_vtx = vtx-1; // Last vertex in array |
| 296 | |
| 297 | // Search for the first usable right section |
| 298 | |
| 299 | do { |
| 300 | if(right_vtx == max_vtx) return; |
| 301 | RightSection(); |
| 302 | } while(right_height <= 0); |
| 303 | |
| 304 | // Search for the first usable left section |
| 305 | |
| 306 | do { |
| 307 | if(left_vtx == max_vtx) return; |
| 308 | LeftSection(); |
| 309 | } while(left_height <= 0); |
| 310 | |
| 311 | wxUint16 * destptr = (wxUint16*)(gfx.RDRAM+rdp.zimg); |
| 312 | int y1 = iceil(min_y); |
| 313 | if (y1 >= (int)rdp.scissor_o.lr_y) return; |
| 314 | int shift; |
| 315 | |
| 316 | for(;;) |
| 317 | { |
| 318 | int x1 = iceil(left_x); |
| 319 | if (x1 < (int)rdp.scissor_o.ul_x) |
| 320 | x1 = rdp.scissor_o.ul_x; |
| 321 | int width = iceil(right_x) - x1; |
| 322 | if (x1+width >= (int)rdp.scissor_o.lr_x) |
| 323 | width = rdp.scissor_o.lr_x - x1 - 1; |
| 324 | |
| 325 | if(width > 0 && y1 >= (int)rdp.scissor_o.ul_y) { |
| 326 | |
| 327 | |
| 328 | // Prestep initial z |
| 329 | |
| 330 | #ifdef HAVE_GLES |
| 331 | float z = left_z + (x1 - left_x)*dzdx; |
| 332 | #else |
| 333 | int prestep = (x1 << 16) - left_x; |
| 334 | int z = left_z + imul16(prestep, dzdx); |
| 335 | #endif |
| 336 | |
| 337 | shift = x1 + y1*rdp.zi_width; |
| 338 | //draw to depth buffer |
| 339 | int trueZ; |
| 340 | int idx; |
| 341 | wxUint16 encodedZ; |
| 342 | for (int x = 0; x < width; x++) |
| 343 | { |
| 344 | #ifdef HAVE_GLES |
| 345 | trueZ = (int)(z*8.0f); |
| 346 | #else |
| 347 | trueZ = z/8192; |
| 348 | #endif |
| 349 | if (trueZ < 0) trueZ = 0; |
| 350 | else if (trueZ > 0x3FFFF) trueZ = 0x3FFFF; |
| 351 | encodedZ = zLUT[trueZ]; |
| 352 | idx = (shift+x)^1; |
| 353 | if(encodedZ < destptr[idx]) |
| 354 | destptr[idx] = encodedZ; |
| 355 | z += dzdx; |
| 356 | } |
| 357 | } |
| 358 | |
| 359 | //destptr += rdp.zi_width; |
| 360 | y1++; |
| 361 | if (y1 >= (int)rdp.scissor_o.lr_y) return; |
| 362 | |
| 363 | // Scan the right side |
| 364 | |
| 365 | if(--right_height <= 0) { // End of this section? |
| 366 | do { |
| 367 | if(right_vtx == max_vtx) return; |
| 368 | RightSection(); |
| 369 | } while(right_height <= 0); |
| 370 | } |
| 371 | else |
| 372 | right_x += right_dxdy; |
| 373 | |
| 374 | // Scan the left side |
| 375 | |
| 376 | if(--left_height <= 0) { // End of this section? |
| 377 | do { |
| 378 | if(left_vtx == max_vtx) return; |
| 379 | LeftSection(); |
| 380 | } while(left_height <= 0); |
| 381 | } |
| 382 | else { |
| 383 | left_x += left_dxdy; |
| 384 | left_z += left_dzdy; |
| 385 | } |
| 386 | } |
| 387 | } |