| 1 | /* |
| 2 | Simple DirectMedia Layer |
| 3 | Copyright (C) 1997-2025 Sam Lantinga <slouken@libsdl.org> |
| 4 | |
| 5 | This software is provided 'as-is', without any express or implied |
| 6 | warranty. In no event will the authors be held liable for any damages |
| 7 | arising from the use of this software. |
| 8 | |
| 9 | Permission is granted to anyone to use this software for any purpose, |
| 10 | including commercial applications, and to alter it and redistribute it |
| 11 | freely, subject to the following restrictions: |
| 12 | |
| 13 | 1. The origin of this software must not be misrepresented; you must not |
| 14 | claim that you wrote the original software. If you use this software |
| 15 | in a product, an acknowledgment in the product documentation would be |
| 16 | appreciated but is not required. |
| 17 | 2. Altered source versions must be plainly marked as such, and must not be |
| 18 | misrepresented as being the original software. |
| 19 | 3. This notice may not be removed or altered from any source distribution. |
| 20 | */ |
| 21 | #include "SDL_internal.h" |
| 22 | |
| 23 | #include "SDL_surface_c.h" |
| 24 | #include "SDL_blit_slow.h" |
| 25 | #include "SDL_pixels_c.h" |
| 26 | |
| 27 | typedef enum |
| 28 | { |
| 29 | SlowBlitPixelAccess_Index8, |
| 30 | SlowBlitPixelAccess_RGB, |
| 31 | SlowBlitPixelAccess_RGBA, |
| 32 | SlowBlitPixelAccess_10Bit, |
| 33 | SlowBlitPixelAccess_Large, |
| 34 | } SlowBlitPixelAccess; |
| 35 | |
| 36 | static SlowBlitPixelAccess GetPixelAccessMethod(SDL_PixelFormat format) |
| 37 | { |
| 38 | if (SDL_BYTESPERPIXEL(format) > 4) { |
| 39 | return SlowBlitPixelAccess_Large; |
| 40 | } else if (SDL_ISPIXELFORMAT_10BIT(format)) { |
| 41 | return SlowBlitPixelAccess_10Bit; |
| 42 | } else if (format == SDL_PIXELFORMAT_INDEX8) { |
| 43 | return SlowBlitPixelAccess_Index8; |
| 44 | } else if (SDL_ISPIXELFORMAT_ALPHA(format)) { |
| 45 | return SlowBlitPixelAccess_RGBA; |
| 46 | } else { |
| 47 | return SlowBlitPixelAccess_RGB; |
| 48 | } |
| 49 | } |
| 50 | |
| 51 | /* The ONE TRUE BLITTER |
| 52 | * This puppy has to handle all the unoptimized cases - yes, it's slow. |
| 53 | */ |
| 54 | void SDL_Blit_Slow(SDL_BlitInfo *info) |
| 55 | { |
| 56 | const int flags = info->flags; |
| 57 | const Uint32 modulateR = info->r; |
| 58 | const Uint32 modulateG = info->g; |
| 59 | const Uint32 modulateB = info->b; |
| 60 | const Uint32 modulateA = info->a; |
| 61 | Uint32 srcpixel = 0; |
| 62 | Uint32 srcR = 0, srcG = 0, srcB = 0, srcA = 0; |
| 63 | Uint32 dstpixel = 0; |
| 64 | Uint32 dstR = 0, dstG = 0, dstB = 0, dstA = 0; |
| 65 | Uint64 srcy, srcx; |
| 66 | Uint64 posy, posx; |
| 67 | Uint64 incy, incx; |
| 68 | const SDL_PixelFormatDetails *src_fmt = info->src_fmt; |
| 69 | const SDL_Palette *src_pal = info->src_pal; |
| 70 | const SDL_PixelFormatDetails *dst_fmt = info->dst_fmt; |
| 71 | const SDL_Palette *dst_pal = info->dst_pal; |
| 72 | SDL_HashTable *palette_map = info->palette_map; |
| 73 | int srcbpp = src_fmt->bytes_per_pixel; |
| 74 | int dstbpp = dst_fmt->bytes_per_pixel; |
| 75 | SlowBlitPixelAccess src_access; |
| 76 | SlowBlitPixelAccess dst_access; |
| 77 | Uint32 rgbmask = ~src_fmt->Amask; |
| 78 | Uint32 ckey = info->colorkey & rgbmask; |
| 79 | Uint32 last_pixel = 0; |
| 80 | Uint8 last_index = 0; |
| 81 | |
| 82 | src_access = GetPixelAccessMethod(src_fmt->format); |
| 83 | dst_access = GetPixelAccessMethod(dst_fmt->format); |
| 84 | if (dst_access == SlowBlitPixelAccess_Index8) { |
| 85 | last_index = SDL_LookupRGBAColor(palette_map, last_pixel, dst_pal); |
| 86 | } |
| 87 | |
| 88 | incy = ((Uint64)info->src_h << 16) / info->dst_h; |
| 89 | incx = ((Uint64)info->src_w << 16) / info->dst_w; |
| 90 | posy = incy / 2; // start at the middle of pixel |
| 91 | |
| 92 | while (info->dst_h--) { |
| 93 | Uint8 *src = 0; |
| 94 | Uint8 *dst = info->dst; |
| 95 | int n = info->dst_w; |
| 96 | posx = incx / 2; // start at the middle of pixel |
| 97 | srcy = posy >> 16; |
| 98 | while (n--) { |
| 99 | srcx = posx >> 16; |
| 100 | src = (info->src + (srcy * info->src_pitch) + (srcx * srcbpp)); |
| 101 | |
| 102 | switch (src_access) { |
| 103 | case SlowBlitPixelAccess_Index8: |
| 104 | srcpixel = *src; |
| 105 | srcR = src_pal->colors[srcpixel].r; |
| 106 | srcG = src_pal->colors[srcpixel].g; |
| 107 | srcB = src_pal->colors[srcpixel].b; |
| 108 | srcA = src_pal->colors[srcpixel].a; |
| 109 | break; |
| 110 | case SlowBlitPixelAccess_RGB: |
| 111 | DISEMBLE_RGB(src, srcbpp, src_fmt, srcpixel, srcR, srcG, srcB); |
| 112 | srcA = 0xFF; |
| 113 | break; |
| 114 | case SlowBlitPixelAccess_RGBA: |
| 115 | DISEMBLE_RGBA(src, srcbpp, src_fmt, srcpixel, srcR, srcG, srcB, srcA); |
| 116 | break; |
| 117 | case SlowBlitPixelAccess_10Bit: |
| 118 | srcpixel = *((Uint32 *)(src)); |
| 119 | switch (src_fmt->format) { |
| 120 | case SDL_PIXELFORMAT_XRGB2101010: |
| 121 | RGBA_FROM_ARGB2101010(srcpixel, srcR, srcG, srcB, srcA); |
| 122 | srcA = 0xFF; |
| 123 | break; |
| 124 | case SDL_PIXELFORMAT_XBGR2101010: |
| 125 | RGBA_FROM_ABGR2101010(srcpixel, srcR, srcG, srcB, srcA); |
| 126 | srcA = 0xFF; |
| 127 | break; |
| 128 | case SDL_PIXELFORMAT_ARGB2101010: |
| 129 | RGBA_FROM_ARGB2101010(srcpixel, srcR, srcG, srcB, srcA); |
| 130 | break; |
| 131 | case SDL_PIXELFORMAT_ABGR2101010: |
| 132 | RGBA_FROM_ABGR2101010(srcpixel, srcR, srcG, srcB, srcA); |
| 133 | break; |
| 134 | default: |
| 135 | break; |
| 136 | } |
| 137 | break; |
| 138 | case SlowBlitPixelAccess_Large: |
| 139 | // Handled in SDL_Blit_Slow_Float() |
| 140 | break; |
| 141 | } |
| 142 | |
| 143 | if (flags & SDL_COPY_COLORKEY) { |
| 144 | // srcpixel isn't set for 24 bpp |
| 145 | if (srcbpp == 3) { |
| 146 | srcpixel = (srcR << src_fmt->Rshift) | |
| 147 | (srcG << src_fmt->Gshift) | (srcB << src_fmt->Bshift); |
| 148 | } |
| 149 | if ((srcpixel & rgbmask) == ckey) { |
| 150 | posx += incx; |
| 151 | dst += dstbpp; |
| 152 | continue; |
| 153 | } |
| 154 | } |
| 155 | if (flags & SDL_COPY_BLEND_MASK) { |
| 156 | switch (dst_access) { |
| 157 | case SlowBlitPixelAccess_Index8: |
| 158 | dstpixel = *dst; |
| 159 | dstR = dst_pal->colors[dstpixel].r; |
| 160 | dstG = dst_pal->colors[dstpixel].g; |
| 161 | dstB = dst_pal->colors[dstpixel].b; |
| 162 | dstA = dst_pal->colors[dstpixel].a; |
| 163 | break; |
| 164 | case SlowBlitPixelAccess_RGB: |
| 165 | DISEMBLE_RGB(dst, dstbpp, dst_fmt, dstpixel, dstR, dstG, dstB); |
| 166 | dstA = 0xFF; |
| 167 | break; |
| 168 | case SlowBlitPixelAccess_RGBA: |
| 169 | DISEMBLE_RGBA(dst, dstbpp, dst_fmt, dstpixel, dstR, dstG, dstB, dstA); |
| 170 | break; |
| 171 | case SlowBlitPixelAccess_10Bit: |
| 172 | dstpixel = *((Uint32 *)(dst)); |
| 173 | switch (dst_fmt->format) { |
| 174 | case SDL_PIXELFORMAT_XRGB2101010: |
| 175 | RGBA_FROM_ARGB2101010(dstpixel, dstR, dstG, dstB, dstA); |
| 176 | dstA = 0xFF; |
| 177 | break; |
| 178 | case SDL_PIXELFORMAT_XBGR2101010: |
| 179 | RGBA_FROM_ABGR2101010(dstpixel, dstR, dstG, dstB, dstA); |
| 180 | dstA = 0xFF; |
| 181 | break; |
| 182 | case SDL_PIXELFORMAT_ARGB2101010: |
| 183 | RGBA_FROM_ARGB2101010(dstpixel, dstR, dstG, dstB, dstA); |
| 184 | break; |
| 185 | case SDL_PIXELFORMAT_ABGR2101010: |
| 186 | RGBA_FROM_ABGR2101010(dstpixel, dstR, dstG, dstB, dstA); |
| 187 | break; |
| 188 | default: |
| 189 | break; |
| 190 | } |
| 191 | break; |
| 192 | case SlowBlitPixelAccess_Large: |
| 193 | // Handled in SDL_Blit_Slow_Float() |
| 194 | break; |
| 195 | } |
| 196 | } else { |
| 197 | // don't care |
| 198 | } |
| 199 | |
| 200 | if (flags & SDL_COPY_MODULATE_COLOR) { |
| 201 | srcR = (srcR * modulateR) / 255; |
| 202 | srcG = (srcG * modulateG) / 255; |
| 203 | srcB = (srcB * modulateB) / 255; |
| 204 | } |
| 205 | if (flags & SDL_COPY_MODULATE_ALPHA) { |
| 206 | srcA = (srcA * modulateA) / 255; |
| 207 | } |
| 208 | if (flags & (SDL_COPY_BLEND | SDL_COPY_ADD)) { |
| 209 | if (srcA < 255) { |
| 210 | srcR = (srcR * srcA) / 255; |
| 211 | srcG = (srcG * srcA) / 255; |
| 212 | srcB = (srcB * srcA) / 255; |
| 213 | } |
| 214 | } |
| 215 | switch (flags & SDL_COPY_BLEND_MASK) { |
| 216 | case 0: |
| 217 | dstR = srcR; |
| 218 | dstG = srcG; |
| 219 | dstB = srcB; |
| 220 | dstA = srcA; |
| 221 | break; |
| 222 | case SDL_COPY_BLEND: |
| 223 | dstR = srcR + ((255 - srcA) * dstR) / 255; |
| 224 | dstG = srcG + ((255 - srcA) * dstG) / 255; |
| 225 | dstB = srcB + ((255 - srcA) * dstB) / 255; |
| 226 | dstA = srcA + ((255 - srcA) * dstA) / 255; |
| 227 | break; |
| 228 | case SDL_COPY_BLEND_PREMULTIPLIED: |
| 229 | dstR = srcR + ((255 - srcA) * dstR) / 255; |
| 230 | if (dstR > 255) { |
| 231 | dstR = 255; |
| 232 | } |
| 233 | dstG = srcG + ((255 - srcA) * dstG) / 255; |
| 234 | if (dstG > 255) { |
| 235 | dstG = 255; |
| 236 | } |
| 237 | dstB = srcB + ((255 - srcA) * dstB) / 255; |
| 238 | if (dstB > 255) { |
| 239 | dstB = 255; |
| 240 | } |
| 241 | dstA = srcA + ((255 - srcA) * dstA) / 255; |
| 242 | if (dstA > 255) { |
| 243 | dstA = 255; |
| 244 | } |
| 245 | break; |
| 246 | case SDL_COPY_ADD: |
| 247 | case SDL_COPY_ADD_PREMULTIPLIED: |
| 248 | dstR = srcR + dstR; |
| 249 | if (dstR > 255) { |
| 250 | dstR = 255; |
| 251 | } |
| 252 | dstG = srcG + dstG; |
| 253 | if (dstG > 255) { |
| 254 | dstG = 255; |
| 255 | } |
| 256 | dstB = srcB + dstB; |
| 257 | if (dstB > 255) { |
| 258 | dstB = 255; |
| 259 | } |
| 260 | break; |
| 261 | case SDL_COPY_MOD: |
| 262 | dstR = (srcR * dstR) / 255; |
| 263 | dstG = (srcG * dstG) / 255; |
| 264 | dstB = (srcB * dstB) / 255; |
| 265 | break; |
| 266 | case SDL_COPY_MUL: |
| 267 | dstR = ((srcR * dstR) + (dstR * (255 - srcA))) / 255; |
| 268 | if (dstR > 255) { |
| 269 | dstR = 255; |
| 270 | } |
| 271 | dstG = ((srcG * dstG) + (dstG * (255 - srcA))) / 255; |
| 272 | if (dstG > 255) { |
| 273 | dstG = 255; |
| 274 | } |
| 275 | dstB = ((srcB * dstB) + (dstB * (255 - srcA))) / 255; |
| 276 | if (dstB > 255) { |
| 277 | dstB = 255; |
| 278 | } |
| 279 | break; |
| 280 | } |
| 281 | |
| 282 | switch (dst_access) { |
| 283 | case SlowBlitPixelAccess_Index8: |
| 284 | dstpixel = ((dstR << 24) | (dstG << 16) | (dstB << 8) | dstA); |
| 285 | if (dstpixel != last_pixel) { |
| 286 | last_pixel = dstpixel; |
| 287 | last_index = SDL_LookupRGBAColor(palette_map, dstpixel, dst_pal); |
| 288 | } |
| 289 | *dst = last_index; |
| 290 | break; |
| 291 | case SlowBlitPixelAccess_RGB: |
| 292 | ASSEMBLE_RGB(dst, dstbpp, dst_fmt, dstR, dstG, dstB); |
| 293 | break; |
| 294 | case SlowBlitPixelAccess_RGBA: |
| 295 | ASSEMBLE_RGBA(dst, dstbpp, dst_fmt, dstR, dstG, dstB, dstA); |
| 296 | break; |
| 297 | case SlowBlitPixelAccess_10Bit: |
| 298 | { |
| 299 | Uint32 pixel; |
| 300 | switch (dst_fmt->format) { |
| 301 | case SDL_PIXELFORMAT_XRGB2101010: |
| 302 | dstA = 0xFF; |
| 303 | SDL_FALLTHROUGH; |
| 304 | case SDL_PIXELFORMAT_ARGB2101010: |
| 305 | ARGB2101010_FROM_RGBA(pixel, dstR, dstG, dstB, dstA); |
| 306 | break; |
| 307 | case SDL_PIXELFORMAT_XBGR2101010: |
| 308 | dstA = 0xFF; |
| 309 | SDL_FALLTHROUGH; |
| 310 | case SDL_PIXELFORMAT_ABGR2101010: |
| 311 | ABGR2101010_FROM_RGBA(pixel, dstR, dstG, dstB, dstA); |
| 312 | break; |
| 313 | default: |
| 314 | pixel = 0; |
| 315 | break; |
| 316 | } |
| 317 | *(Uint32 *)dst = pixel; |
| 318 | break; |
| 319 | } |
| 320 | case SlowBlitPixelAccess_Large: |
| 321 | // Handled in SDL_Blit_Slow_Float() |
| 322 | break; |
| 323 | } |
| 324 | |
| 325 | posx += incx; |
| 326 | dst += dstbpp; |
| 327 | } |
| 328 | posy += incy; |
| 329 | info->dst += info->dst_pitch; |
| 330 | } |
| 331 | } |
| 332 | |
| 333 | /* Convert from F16 to float |
| 334 | * Public domain implementation from https://gist.github.com/rygorous/2144712 |
| 335 | */ |
| 336 | typedef union |
| 337 | { |
| 338 | Uint32 u; |
| 339 | float f; |
| 340 | struct |
| 341 | { |
| 342 | Uint32 Mantissa : 23; |
| 343 | Uint32 Exponent : 8; |
| 344 | Uint32 Sign : 1; |
| 345 | } x; |
| 346 | } FP32; |
| 347 | |
| 348 | #ifdef _MSC_VER |
| 349 | #pragma warning(push) |
| 350 | #pragma warning(disable:4214) |
| 351 | #endif |
| 352 | |
| 353 | typedef union |
| 354 | { |
| 355 | Uint16 u; |
| 356 | struct |
| 357 | { |
| 358 | Uint16 Mantissa : 10; |
| 359 | Uint16 Exponent : 5; |
| 360 | Uint16 Sign : 1; |
| 361 | } x; |
| 362 | } FP16; |
| 363 | |
| 364 | #ifdef _MSC_VER |
| 365 | #pragma warning(pop) |
| 366 | #endif |
| 367 | |
| 368 | static float half_to_float(Uint16 unValue) |
| 369 | { |
| 370 | static const FP32 magic = { (254 - 15) << 23 }; |
| 371 | static const FP32 was_infnan = { (127 + 16) << 23 }; |
| 372 | FP16 h; |
| 373 | FP32 o; |
| 374 | |
| 375 | h.u = unValue; |
| 376 | o.u = (h.u & 0x7fff) << 13; // exponent/mantissa bits |
| 377 | o.f *= magic.f; // exponent adjust |
| 378 | if (o.f >= was_infnan.f) // make sure Inf/NaN survive |
| 379 | o.u |= 255 << 23; |
| 380 | o.u |= (h.u & 0x8000) << 16; // sign bit |
| 381 | return o.f; |
| 382 | } |
| 383 | |
| 384 | /* Convert from float to F16 |
| 385 | * Public domain implementation from https://stackoverflow.com/questions/76799117/how-to-convert-a-float-to-a-half-type-and-the-other-way-around-in-c |
| 386 | */ |
| 387 | static Uint16 float_to_half(float a) |
| 388 | { |
| 389 | Uint32 ia; |
| 390 | Uint16 ir; |
| 391 | |
| 392 | SDL_memcpy(&ia, &a, sizeof(ia)); |
| 393 | |
| 394 | ir = (ia >> 16) & 0x8000; |
| 395 | if ((ia & 0x7f800000) == 0x7f800000) { |
| 396 | if ((ia & 0x7fffffff) == 0x7f800000) { |
| 397 | ir |= 0x7c00; // infinity |
| 398 | } else { |
| 399 | ir |= 0x7e00 | ((ia >> (24 - 11)) & 0x1ff); // NaN, quietened |
| 400 | } |
| 401 | } else if ((ia & 0x7f800000) >= 0x33000000) { |
| 402 | int shift = (int)((ia >> 23) & 0xff) - 127; |
| 403 | if (shift > 15) { |
| 404 | ir |= 0x7c00; // infinity |
| 405 | } else { |
| 406 | ia = (ia & 0x007fffff) | 0x00800000; // extract mantissa |
| 407 | if (shift < -14) { // denormal |
| 408 | ir |= ia >> (-1 - shift); |
| 409 | ia = ia << (32 - (-1 - shift)); |
| 410 | } else { // normal |
| 411 | ir |= ia >> (24 - 11); |
| 412 | ia = ia << (32 - (24 - 11)); |
| 413 | ir = ir + ((14 + shift) << 10); |
| 414 | } |
| 415 | // IEEE-754 round to nearest of even |
| 416 | if ((ia > 0x80000000) || ((ia == 0x80000000) && (ir & 1))) { |
| 417 | ir++; |
| 418 | } |
| 419 | } |
| 420 | } |
| 421 | return ir; |
| 422 | } |
| 423 | |
| 424 | static void ReadFloatPixel(Uint8 *pixels, SlowBlitPixelAccess access, const SDL_PixelFormatDetails *fmt, const SDL_Palette *pal, SDL_Colorspace colorspace, float SDR_white_point, |
| 425 | float *outR, float *outG, float *outB, float *outA) |
| 426 | { |
| 427 | Uint32 pixel; |
| 428 | Uint32 R, G, B, A; |
| 429 | float fR = 0.0f, fG = 0.0f, fB = 0.0f, fA = 0.0f; |
| 430 | float v[4]; |
| 431 | |
| 432 | switch (access) { |
| 433 | case SlowBlitPixelAccess_Index8: |
| 434 | pixel = *pixels; |
| 435 | fR = (float)pal->colors[pixel].r / 255.0f; |
| 436 | fG = (float)pal->colors[pixel].g / 255.0f; |
| 437 | fB = (float)pal->colors[pixel].b / 255.0f; |
| 438 | fA = (float)pal->colors[pixel].a / 255.0f; |
| 439 | break; |
| 440 | case SlowBlitPixelAccess_RGB: |
| 441 | DISEMBLE_RGB(pixels, fmt->bytes_per_pixel, fmt, pixel, R, G, B); |
| 442 | fR = (float)R / 255.0f; |
| 443 | fG = (float)G / 255.0f; |
| 444 | fB = (float)B / 255.0f; |
| 445 | fA = 1.0f; |
| 446 | break; |
| 447 | case SlowBlitPixelAccess_RGBA: |
| 448 | DISEMBLE_RGBA(pixels, fmt->bytes_per_pixel, fmt, pixel, R, G, B, A); |
| 449 | fR = (float)R / 255.0f; |
| 450 | fG = (float)G / 255.0f; |
| 451 | fB = (float)B / 255.0f; |
| 452 | fA = (float)A / 255.0f; |
| 453 | break; |
| 454 | case SlowBlitPixelAccess_10Bit: |
| 455 | pixel = *((Uint32 *)pixels); |
| 456 | switch (fmt->format) { |
| 457 | case SDL_PIXELFORMAT_XRGB2101010: |
| 458 | RGBAFLOAT_FROM_ARGB2101010(pixel, fR, fG, fB, fA); |
| 459 | fA = 1.0f; |
| 460 | break; |
| 461 | case SDL_PIXELFORMAT_XBGR2101010: |
| 462 | RGBAFLOAT_FROM_ABGR2101010(pixel, fR, fG, fB, fA); |
| 463 | fA = 1.0f; |
| 464 | break; |
| 465 | case SDL_PIXELFORMAT_ARGB2101010: |
| 466 | RGBAFLOAT_FROM_ARGB2101010(pixel, fR, fG, fB, fA); |
| 467 | break; |
| 468 | case SDL_PIXELFORMAT_ABGR2101010: |
| 469 | RGBAFLOAT_FROM_ABGR2101010(pixel, fR, fG, fB, fA); |
| 470 | break; |
| 471 | default: |
| 472 | fR = fG = fB = fA = 0.0f; |
| 473 | break; |
| 474 | } |
| 475 | break; |
| 476 | case SlowBlitPixelAccess_Large: |
| 477 | switch (SDL_PIXELTYPE(fmt->format)) { |
| 478 | case SDL_PIXELTYPE_ARRAYU16: |
| 479 | v[0] = (float)(((Uint16 *)pixels)[0]) / SDL_MAX_UINT16; |
| 480 | v[1] = (float)(((Uint16 *)pixels)[1]) / SDL_MAX_UINT16; |
| 481 | v[2] = (float)(((Uint16 *)pixels)[2]) / SDL_MAX_UINT16; |
| 482 | if (fmt->bytes_per_pixel == 8) { |
| 483 | v[3] = (float)(((Uint16 *)pixels)[3]) / SDL_MAX_UINT16; |
| 484 | } else { |
| 485 | v[3] = 1.0f; |
| 486 | } |
| 487 | break; |
| 488 | case SDL_PIXELTYPE_ARRAYF16: |
| 489 | v[0] = half_to_float(((Uint16 *)pixels)[0]); |
| 490 | v[1] = half_to_float(((Uint16 *)pixels)[1]); |
| 491 | v[2] = half_to_float(((Uint16 *)pixels)[2]); |
| 492 | if (fmt->bytes_per_pixel == 8) { |
| 493 | v[3] = half_to_float(((Uint16 *)pixels)[3]); |
| 494 | } else { |
| 495 | v[3] = 1.0f; |
| 496 | } |
| 497 | break; |
| 498 | case SDL_PIXELTYPE_ARRAYF32: |
| 499 | v[0] = ((float *)pixels)[0]; |
| 500 | v[1] = ((float *)pixels)[1]; |
| 501 | v[2] = ((float *)pixels)[2]; |
| 502 | if (fmt->bytes_per_pixel == 16) { |
| 503 | v[3] = ((float *)pixels)[3]; |
| 504 | } else { |
| 505 | v[3] = 1.0f; |
| 506 | } |
| 507 | break; |
| 508 | default: |
| 509 | // Unknown array type |
| 510 | v[0] = v[1] = v[2] = v[3] = 0.0f; |
| 511 | break; |
| 512 | } |
| 513 | switch (SDL_PIXELORDER(fmt->format)) { |
| 514 | case SDL_ARRAYORDER_RGB: |
| 515 | fR = v[0]; |
| 516 | fG = v[1]; |
| 517 | fB = v[2]; |
| 518 | fA = 1.0f; |
| 519 | break; |
| 520 | case SDL_ARRAYORDER_RGBA: |
| 521 | fR = v[0]; |
| 522 | fG = v[1]; |
| 523 | fB = v[2]; |
| 524 | fA = v[3]; |
| 525 | break; |
| 526 | case SDL_ARRAYORDER_ARGB: |
| 527 | fA = v[0]; |
| 528 | fR = v[1]; |
| 529 | fG = v[2]; |
| 530 | fB = v[3]; |
| 531 | break; |
| 532 | case SDL_ARRAYORDER_BGR: |
| 533 | fB = v[0]; |
| 534 | fG = v[1]; |
| 535 | fR = v[2]; |
| 536 | fA = 1.0f; |
| 537 | break; |
| 538 | case SDL_ARRAYORDER_BGRA: |
| 539 | fB = v[0]; |
| 540 | fG = v[1]; |
| 541 | fR = v[2]; |
| 542 | fA = v[3]; |
| 543 | break; |
| 544 | case SDL_ARRAYORDER_ABGR: |
| 545 | fA = v[0]; |
| 546 | fB = v[1]; |
| 547 | fG = v[2]; |
| 548 | fR = v[3]; |
| 549 | break; |
| 550 | default: |
| 551 | // Unknown array order |
| 552 | fA = fR = fG = fB = 0.0f; |
| 553 | break; |
| 554 | } |
| 555 | break; |
| 556 | } |
| 557 | |
| 558 | // Convert to nits so src and dst are guaranteed to be linear and in the same units |
| 559 | switch (SDL_COLORSPACETRANSFER(colorspace)) { |
| 560 | case SDL_TRANSFER_CHARACTERISTICS_SRGB: |
| 561 | fR = SDL_sRGBtoLinear(fR); |
| 562 | fG = SDL_sRGBtoLinear(fG); |
| 563 | fB = SDL_sRGBtoLinear(fB); |
| 564 | break; |
| 565 | case SDL_TRANSFER_CHARACTERISTICS_PQ: |
| 566 | fR = SDL_PQtoNits(fR) / SDR_white_point; |
| 567 | fG = SDL_PQtoNits(fG) / SDR_white_point; |
| 568 | fB = SDL_PQtoNits(fB) / SDR_white_point; |
| 569 | break; |
| 570 | case SDL_TRANSFER_CHARACTERISTICS_LINEAR: |
| 571 | fR /= SDR_white_point; |
| 572 | fG /= SDR_white_point; |
| 573 | fB /= SDR_white_point; |
| 574 | break; |
| 575 | default: |
| 576 | // Unknown, leave it alone |
| 577 | break; |
| 578 | } |
| 579 | |
| 580 | *outR = fR; |
| 581 | *outG = fG; |
| 582 | *outB = fB; |
| 583 | *outA = fA; |
| 584 | } |
| 585 | |
| 586 | static void WriteFloatPixel(Uint8 *pixels, SlowBlitPixelAccess access, const SDL_PixelFormatDetails *fmt, SDL_Colorspace colorspace, float SDR_white_point, |
| 587 | float fR, float fG, float fB, float fA) |
| 588 | { |
| 589 | Uint32 R, G, B, A; |
| 590 | Uint32 pixel; |
| 591 | float v[4]; |
| 592 | |
| 593 | // We converted to nits so src and dst are guaranteed to be linear and in the same units |
| 594 | switch (SDL_COLORSPACETRANSFER(colorspace)) { |
| 595 | case SDL_TRANSFER_CHARACTERISTICS_SRGB: |
| 596 | fR = SDL_sRGBfromLinear(fR); |
| 597 | fG = SDL_sRGBfromLinear(fG); |
| 598 | fB = SDL_sRGBfromLinear(fB); |
| 599 | break; |
| 600 | case SDL_TRANSFER_CHARACTERISTICS_PQ: |
| 601 | fR = SDL_PQfromNits(fR * SDR_white_point); |
| 602 | fG = SDL_PQfromNits(fG * SDR_white_point); |
| 603 | fB = SDL_PQfromNits(fB * SDR_white_point); |
| 604 | break; |
| 605 | case SDL_TRANSFER_CHARACTERISTICS_LINEAR: |
| 606 | fR *= SDR_white_point; |
| 607 | fG *= SDR_white_point; |
| 608 | fB *= SDR_white_point; |
| 609 | break; |
| 610 | default: |
| 611 | // Unknown, leave it alone |
| 612 | break; |
| 613 | } |
| 614 | |
| 615 | switch (access) { |
| 616 | case SlowBlitPixelAccess_Index8: |
| 617 | // This should never happen, checked before this call |
| 618 | SDL_assert(0); |
| 619 | break; |
| 620 | case SlowBlitPixelAccess_RGB: |
| 621 | R = (Uint8)SDL_roundf(SDL_clamp(fR, 0.0f, 1.0f) * 255.0f); |
| 622 | G = (Uint8)SDL_roundf(SDL_clamp(fG, 0.0f, 1.0f) * 255.0f); |
| 623 | B = (Uint8)SDL_roundf(SDL_clamp(fB, 0.0f, 1.0f) * 255.0f); |
| 624 | ASSEMBLE_RGB(pixels, fmt->bytes_per_pixel, fmt, R, G, B); |
| 625 | break; |
| 626 | case SlowBlitPixelAccess_RGBA: |
| 627 | R = (Uint8)SDL_roundf(SDL_clamp(fR, 0.0f, 1.0f) * 255.0f); |
| 628 | G = (Uint8)SDL_roundf(SDL_clamp(fG, 0.0f, 1.0f) * 255.0f); |
| 629 | B = (Uint8)SDL_roundf(SDL_clamp(fB, 0.0f, 1.0f) * 255.0f); |
| 630 | A = (Uint8)SDL_roundf(SDL_clamp(fA, 0.0f, 1.0f) * 255.0f); |
| 631 | ASSEMBLE_RGBA(pixels, fmt->bytes_per_pixel, fmt, R, G, B, A); |
| 632 | break; |
| 633 | case SlowBlitPixelAccess_10Bit: |
| 634 | { |
| 635 | switch (fmt->format) { |
| 636 | case SDL_PIXELFORMAT_XRGB2101010: |
| 637 | fA = 1.0f; |
| 638 | SDL_FALLTHROUGH; |
| 639 | case SDL_PIXELFORMAT_ARGB2101010: |
| 640 | ARGB2101010_FROM_RGBAFLOAT(pixel, fR, fG, fB, fA); |
| 641 | break; |
| 642 | case SDL_PIXELFORMAT_XBGR2101010: |
| 643 | fA = 1.0f; |
| 644 | SDL_FALLTHROUGH; |
| 645 | case SDL_PIXELFORMAT_ABGR2101010: |
| 646 | ABGR2101010_FROM_RGBAFLOAT(pixel, fR, fG, fB, fA); |
| 647 | break; |
| 648 | default: |
| 649 | pixel = 0; |
| 650 | break; |
| 651 | } |
| 652 | *(Uint32 *)pixels = pixel; |
| 653 | break; |
| 654 | } |
| 655 | case SlowBlitPixelAccess_Large: |
| 656 | switch (SDL_PIXELORDER(fmt->format)) { |
| 657 | case SDL_ARRAYORDER_RGB: |
| 658 | v[0] = fR; |
| 659 | v[1] = fG; |
| 660 | v[2] = fB; |
| 661 | v[3] = 1.0f; |
| 662 | break; |
| 663 | case SDL_ARRAYORDER_RGBA: |
| 664 | v[0] = fR; |
| 665 | v[1] = fG; |
| 666 | v[2] = fB; |
| 667 | v[3] = fA; |
| 668 | break; |
| 669 | case SDL_ARRAYORDER_ARGB: |
| 670 | v[0] = fA; |
| 671 | v[1] = fR; |
| 672 | v[2] = fG; |
| 673 | v[3] = fB; |
| 674 | break; |
| 675 | case SDL_ARRAYORDER_BGR: |
| 676 | v[0] = fB; |
| 677 | v[1] = fG; |
| 678 | v[2] = fR; |
| 679 | v[3] = 1.0f; |
| 680 | break; |
| 681 | case SDL_ARRAYORDER_BGRA: |
| 682 | v[0] = fB; |
| 683 | v[1] = fG; |
| 684 | v[2] = fR; |
| 685 | v[3] = fA; |
| 686 | break; |
| 687 | case SDL_ARRAYORDER_ABGR: |
| 688 | v[0] = fA; |
| 689 | v[1] = fB; |
| 690 | v[2] = fG; |
| 691 | v[3] = fR; |
| 692 | break; |
| 693 | default: |
| 694 | // Unknown array order |
| 695 | v[0] = v[1] = v[2] = v[3] = 0.0f; |
| 696 | break; |
| 697 | } |
| 698 | switch (SDL_PIXELTYPE(fmt->format)) { |
| 699 | case SDL_PIXELTYPE_ARRAYU16: |
| 700 | ((Uint16 *)pixels)[0] = (Uint16)SDL_roundf(SDL_clamp(v[0], 0.0f, 1.0f) * SDL_MAX_UINT16); |
| 701 | ((Uint16 *)pixels)[1] = (Uint16)SDL_roundf(SDL_clamp(v[1], 0.0f, 1.0f) * SDL_MAX_UINT16); |
| 702 | ((Uint16 *)pixels)[2] = (Uint16)SDL_roundf(SDL_clamp(v[2], 0.0f, 1.0f) * SDL_MAX_UINT16); |
| 703 | if (fmt->bytes_per_pixel == 8) { |
| 704 | ((Uint16 *)pixels)[3] = (Uint16)SDL_roundf(SDL_clamp(v[3], 0.0f, 1.0f) * SDL_MAX_UINT16); |
| 705 | } |
| 706 | break; |
| 707 | case SDL_PIXELTYPE_ARRAYF16: |
| 708 | ((Uint16 *)pixels)[0] = float_to_half(v[0]); |
| 709 | ((Uint16 *)pixels)[1] = float_to_half(v[1]); |
| 710 | ((Uint16 *)pixels)[2] = float_to_half(v[2]); |
| 711 | if (fmt->bytes_per_pixel == 8) { |
| 712 | ((Uint16 *)pixels)[3] = float_to_half(v[3]); |
| 713 | } |
| 714 | break; |
| 715 | case SDL_PIXELTYPE_ARRAYF32: |
| 716 | ((float *)pixels)[0] = v[0]; |
| 717 | ((float *)pixels)[1] = v[1]; |
| 718 | ((float *)pixels)[2] = v[2]; |
| 719 | if (fmt->bytes_per_pixel == 16) { |
| 720 | ((float *)pixels)[3] = v[3]; |
| 721 | } |
| 722 | break; |
| 723 | default: |
| 724 | // Unknown array type |
| 725 | break; |
| 726 | } |
| 727 | break; |
| 728 | } |
| 729 | } |
| 730 | |
| 731 | typedef enum |
| 732 | { |
| 733 | SDL_TONEMAP_NONE, |
| 734 | SDL_TONEMAP_LINEAR, |
| 735 | SDL_TONEMAP_CHROME |
| 736 | } SDL_TonemapOperator; |
| 737 | |
| 738 | typedef struct |
| 739 | { |
| 740 | SDL_TonemapOperator op; |
| 741 | |
| 742 | union { |
| 743 | struct { |
| 744 | float scale; |
| 745 | } linear; |
| 746 | |
| 747 | struct { |
| 748 | float a; |
| 749 | float b; |
| 750 | const float *color_primaries_matrix; |
| 751 | } chrome; |
| 752 | |
| 753 | } data; |
| 754 | |
| 755 | } SDL_TonemapContext; |
| 756 | |
| 757 | static void TonemapLinear(float *r, float *g, float *b, float scale) |
| 758 | { |
| 759 | *r *= scale; |
| 760 | *g *= scale; |
| 761 | *b *= scale; |
| 762 | } |
| 763 | |
| 764 | /* This uses the same tonemapping algorithm developed by Google for Chrome: |
| 765 | * https://colab.research.google.com/drive/1hI10nq6L6ru_UFvz7-f7xQaQp0qarz_K |
| 766 | * |
| 767 | * Essentially, you use the source headroom and the destination headroom |
| 768 | * to calculate scaling factors: |
| 769 | * tonemap_a = (dst_headroom / (src_headroom * src_headroom)); |
| 770 | * tonemap_b = (1.0f / dst_headroom); |
| 771 | * |
| 772 | * Then you normalize your source color by the HDR whitepoint, |
| 773 | * and calculate a final scaling factor in BT.2020 colorspace. |
| 774 | */ |
| 775 | static void TonemapChrome(float *r, float *g, float *b, float tonemap_a, float tonemap_b) |
| 776 | { |
| 777 | float v1 = *r; |
| 778 | float v2 = *g; |
| 779 | float v3 = *b; |
| 780 | float vmax = SDL_max(v1, SDL_max(v2, v3)); |
| 781 | |
| 782 | if (vmax > 0.0f) { |
| 783 | float scale = (1.0f + tonemap_a * vmax) / (1.0f + tonemap_b * vmax); |
| 784 | TonemapLinear(r, g, b, scale); |
| 785 | } |
| 786 | } |
| 787 | |
| 788 | static void ApplyTonemap(SDL_TonemapContext *ctx, float *r, float *g, float *b) |
| 789 | { |
| 790 | switch (ctx->op) { |
| 791 | case SDL_TONEMAP_LINEAR: |
| 792 | TonemapLinear(r, g, b, ctx->data.linear.scale); |
| 793 | break; |
| 794 | case SDL_TONEMAP_CHROME: |
| 795 | if (ctx->data.chrome.color_primaries_matrix) { |
| 796 | SDL_ConvertColorPrimaries(r, g, b, ctx->data.chrome.color_primaries_matrix); |
| 797 | } |
| 798 | TonemapChrome(r, g, b, ctx->data.chrome.a, ctx->data.chrome.b); |
| 799 | break; |
| 800 | default: |
| 801 | break; |
| 802 | } |
| 803 | } |
| 804 | |
| 805 | /* The SECOND TRUE BLITTER |
| 806 | * This one is even slower than the first, but also handles large pixel formats and colorspace conversion |
| 807 | */ |
| 808 | void SDL_Blit_Slow_Float(SDL_BlitInfo *info) |
| 809 | { |
| 810 | const int flags = info->flags; |
| 811 | const Uint32 modulateR = info->r; |
| 812 | const Uint32 modulateG = info->g; |
| 813 | const Uint32 modulateB = info->b; |
| 814 | const Uint32 modulateA = info->a; |
| 815 | float srcR, srcG, srcB, srcA; |
| 816 | float dstR, dstG, dstB, dstA; |
| 817 | Uint64 srcy, srcx; |
| 818 | Uint64 posy, posx; |
| 819 | Uint64 incy, incx; |
| 820 | const SDL_PixelFormatDetails *src_fmt = info->src_fmt; |
| 821 | const SDL_Palette *src_pal = info->src_pal; |
| 822 | const SDL_PixelFormatDetails *dst_fmt = info->dst_fmt; |
| 823 | const SDL_Palette *dst_pal = info->dst_pal; |
| 824 | SDL_HashTable *palette_map = info->palette_map; |
| 825 | int srcbpp = src_fmt->bytes_per_pixel; |
| 826 | int dstbpp = dst_fmt->bytes_per_pixel; |
| 827 | SlowBlitPixelAccess src_access; |
| 828 | SlowBlitPixelAccess dst_access; |
| 829 | SDL_Colorspace src_colorspace; |
| 830 | SDL_Colorspace dst_colorspace; |
| 831 | SDL_ColorPrimaries src_primaries; |
| 832 | SDL_ColorPrimaries dst_primaries; |
| 833 | const float *color_primaries_matrix = NULL; |
| 834 | float src_white_point; |
| 835 | float dst_white_point; |
| 836 | float dst_headroom; |
| 837 | float src_headroom; |
| 838 | SDL_TonemapContext tonemap; |
| 839 | Uint32 last_pixel = 0; |
| 840 | Uint8 last_index = 0; |
| 841 | |
| 842 | src_colorspace = info->src_surface->colorspace; |
| 843 | dst_colorspace = info->dst_surface->colorspace; |
| 844 | src_primaries = SDL_COLORSPACEPRIMARIES(src_colorspace); |
| 845 | dst_primaries = SDL_COLORSPACEPRIMARIES(dst_colorspace); |
| 846 | |
| 847 | src_white_point = SDL_GetSurfaceSDRWhitePoint(info->src_surface, src_colorspace); |
| 848 | dst_white_point = SDL_GetSurfaceSDRWhitePoint(info->dst_surface, dst_colorspace); |
| 849 | src_headroom = SDL_GetSurfaceHDRHeadroom(info->src_surface, src_colorspace); |
| 850 | dst_headroom = SDL_GetSurfaceHDRHeadroom(info->dst_surface, dst_colorspace); |
| 851 | if (dst_headroom == 0.0f) { |
| 852 | // The destination will have the same headroom as the source |
| 853 | dst_headroom = src_headroom; |
| 854 | SDL_SetFloatProperty(SDL_GetSurfaceProperties(info->dst_surface), SDL_PROP_SURFACE_HDR_HEADROOM_FLOAT, dst_headroom); |
| 855 | } |
| 856 | |
| 857 | SDL_zero(tonemap); |
| 858 | |
| 859 | if (src_headroom > dst_headroom) { |
| 860 | const char *tonemap_operator = SDL_GetStringProperty(SDL_GetSurfaceProperties(info->src_surface), SDL_PROP_SURFACE_TONEMAP_OPERATOR_STRING, NULL); |
| 861 | if (tonemap_operator) { |
| 862 | if (SDL_strncmp(tonemap_operator, "*=" , 2) == 0) { |
| 863 | tonemap.op = SDL_TONEMAP_LINEAR; |
| 864 | tonemap.data.linear.scale = (float)SDL_atof(tonemap_operator + 2); |
| 865 | } else if (SDL_strcasecmp(tonemap_operator, "chrome" ) == 0) { |
| 866 | tonemap.op = SDL_TONEMAP_CHROME; |
| 867 | } else if (SDL_strcasecmp(tonemap_operator, "none" ) == 0) { |
| 868 | tonemap.op = SDL_TONEMAP_NONE; |
| 869 | } |
| 870 | } else { |
| 871 | tonemap.op = SDL_TONEMAP_CHROME; |
| 872 | } |
| 873 | if (tonemap.op == SDL_TONEMAP_CHROME) { |
| 874 | tonemap.data.chrome.a = (dst_headroom / (src_headroom * src_headroom)); |
| 875 | tonemap.data.chrome.b = (1.0f / dst_headroom); |
| 876 | |
| 877 | // We'll convert to BT.2020 primaries for the tonemap operation |
| 878 | tonemap.data.chrome.color_primaries_matrix = SDL_GetColorPrimariesConversionMatrix(src_primaries, SDL_COLOR_PRIMARIES_BT2020); |
| 879 | if (tonemap.data.chrome.color_primaries_matrix) { |
| 880 | src_primaries = SDL_COLOR_PRIMARIES_BT2020; |
| 881 | } |
| 882 | } |
| 883 | } |
| 884 | |
| 885 | if (src_primaries != dst_primaries) { |
| 886 | color_primaries_matrix = SDL_GetColorPrimariesConversionMatrix(src_primaries, dst_primaries); |
| 887 | } |
| 888 | |
| 889 | src_access = GetPixelAccessMethod(src_fmt->format); |
| 890 | dst_access = GetPixelAccessMethod(dst_fmt->format); |
| 891 | if (dst_access == SlowBlitPixelAccess_Index8) { |
| 892 | last_index = SDL_LookupRGBAColor(palette_map, last_pixel, dst_pal); |
| 893 | } |
| 894 | |
| 895 | incy = ((Uint64)info->src_h << 16) / info->dst_h; |
| 896 | incx = ((Uint64)info->src_w << 16) / info->dst_w; |
| 897 | posy = incy / 2; // start at the middle of pixel |
| 898 | |
| 899 | while (info->dst_h--) { |
| 900 | Uint8 *src = 0; |
| 901 | Uint8 *dst = info->dst; |
| 902 | int n = info->dst_w; |
| 903 | posx = incx / 2; // start at the middle of pixel |
| 904 | srcy = posy >> 16; |
| 905 | while (n--) { |
| 906 | srcx = posx >> 16; |
| 907 | src = (info->src + (srcy * info->src_pitch) + (srcx * srcbpp)); |
| 908 | |
| 909 | ReadFloatPixel(src, src_access, src_fmt, src_pal, src_colorspace, src_white_point, &srcR, &srcG, &srcB, &srcA); |
| 910 | |
| 911 | if (tonemap.op) { |
| 912 | ApplyTonemap(&tonemap, &srcR, &srcG, &srcB); |
| 913 | } |
| 914 | |
| 915 | if (color_primaries_matrix) { |
| 916 | SDL_ConvertColorPrimaries(&srcR, &srcG, &srcB, color_primaries_matrix); |
| 917 | } |
| 918 | |
| 919 | if (flags & SDL_COPY_COLORKEY) { |
| 920 | // colorkey isn't supported |
| 921 | } |
| 922 | if ((flags & (SDL_COPY_BLEND | SDL_COPY_ADD | SDL_COPY_MOD | SDL_COPY_MUL))) { |
| 923 | ReadFloatPixel(dst, dst_access, dst_fmt, dst_pal, dst_colorspace, dst_white_point, &dstR, &dstG, &dstB, &dstA); |
| 924 | } else { |
| 925 | // don't care |
| 926 | dstR = dstG = dstB = dstA = 0.0f; |
| 927 | } |
| 928 | |
| 929 | if (flags & SDL_COPY_MODULATE_COLOR) { |
| 930 | srcR = (srcR * modulateR) / 255; |
| 931 | srcG = (srcG * modulateG) / 255; |
| 932 | srcB = (srcB * modulateB) / 255; |
| 933 | } |
| 934 | if (flags & SDL_COPY_MODULATE_ALPHA) { |
| 935 | srcA = (srcA * modulateA) / 255; |
| 936 | } |
| 937 | if (flags & (SDL_COPY_BLEND | SDL_COPY_ADD)) { |
| 938 | if (srcA < 1.0f) { |
| 939 | srcR = (srcR * srcA); |
| 940 | srcG = (srcG * srcA); |
| 941 | srcB = (srcB * srcA); |
| 942 | } |
| 943 | } |
| 944 | switch (flags & (SDL_COPY_BLEND | SDL_COPY_ADD | SDL_COPY_MOD | SDL_COPY_MUL)) { |
| 945 | case 0: |
| 946 | dstR = srcR; |
| 947 | dstG = srcG; |
| 948 | dstB = srcB; |
| 949 | dstA = srcA; |
| 950 | break; |
| 951 | case SDL_COPY_BLEND: |
| 952 | dstR = srcR + ((1.0f - srcA) * dstR); |
| 953 | dstG = srcG + ((1.0f - srcA) * dstG); |
| 954 | dstB = srcB + ((1.0f - srcA) * dstB); |
| 955 | dstA = srcA + ((1.0f - srcA) * dstA); |
| 956 | break; |
| 957 | case SDL_COPY_ADD: |
| 958 | dstR = srcR + dstR; |
| 959 | dstG = srcG + dstG; |
| 960 | dstB = srcB + dstB; |
| 961 | break; |
| 962 | case SDL_COPY_MOD: |
| 963 | dstR = (srcR * dstR); |
| 964 | dstG = (srcG * dstG); |
| 965 | dstB = (srcB * dstB); |
| 966 | break; |
| 967 | case SDL_COPY_MUL: |
| 968 | dstR = ((srcR * dstR) + (dstR * (1.0f - srcA))); |
| 969 | dstG = ((srcG * dstG) + (dstG * (1.0f - srcA))); |
| 970 | dstB = ((srcB * dstB) + (dstB * (1.0f - srcA))); |
| 971 | break; |
| 972 | } |
| 973 | |
| 974 | if (dst_access == SlowBlitPixelAccess_Index8) { |
| 975 | Uint32 R = (Uint8)SDL_roundf(SDL_clamp(SDL_sRGBfromLinear(dstR), 0.0f, 1.0f) * 255.0f); |
| 976 | Uint32 G = (Uint8)SDL_roundf(SDL_clamp(SDL_sRGBfromLinear(dstG), 0.0f, 1.0f) * 255.0f); |
| 977 | Uint32 B = (Uint8)SDL_roundf(SDL_clamp(SDL_sRGBfromLinear(dstB), 0.0f, 1.0f) * 255.0f); |
| 978 | Uint32 A = (Uint8)SDL_roundf(SDL_clamp(dstA, 0.0f, 1.0f) * 255.0f); |
| 979 | Uint32 dstpixel = ((R << 24) | (G << 16) | (B << 8) | A); |
| 980 | if (dstpixel != last_pixel) { |
| 981 | last_pixel = dstpixel; |
| 982 | last_index = SDL_LookupRGBAColor(palette_map, dstpixel, dst_pal); |
| 983 | } |
| 984 | *dst = last_index; |
| 985 | } else { |
| 986 | WriteFloatPixel(dst, dst_access, dst_fmt, dst_colorspace, dst_white_point, dstR, dstG, dstB, dstA); |
| 987 | } |
| 988 | |
| 989 | posx += incx; |
| 990 | dst += dstbpp; |
| 991 | } |
| 992 | posy += incy; |
| 993 | info->dst += info->dst_pitch; |
| 994 | } |
| 995 | } |
| 996 | |
| 997 | |