blend.c 28.4 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29
/* 
 * Copyright (C) 2004 Wim Taymans <wim@fluendo.com>
 * Copyright (C) 2006 Mindfruit Bv.
 *   Author: Sjoerd Simons <sjoerd@luon.net>
 *   Author: Alex Ugarte <alexugarte@gmail.com>
 * Copyright (C) 2009 Alex Ugarte <augarte@vicomtech.org>
 * Copyright (C) 2009 Sebastian Dröge <sebastian.droege@collabora.co.uk>
 *
 * This library is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Library General Public
 * License as published by the Free Software Foundation; either
 * version 2 of the License, or (at your option) any later version.
 *
 * This library is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Library General Public License for more details.
 *
 * You should have received a copy of the GNU Library General Public
 * License along with this library; if not, write to the
 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
 * Boston, MA 02111-1307, USA.
 */

#ifdef HAVE_CONFIG_H
#include "config.h"
#endif

#include "blend.h"
30
#include "blendorc.h"
31 32 33

#include <string.h>

34
#include <gst/video/video.h>
35 36 37
#ifdef HAVE_ORC
#include <orc/orc.h>
#endif
38

39 40
#define BLEND(D,S,alpha) (((D) * (256 - (alpha)) + (S) * (alpha)) >> 8)

41
#ifdef HAVE_GCC_ASM
42
#if defined(HAVE_ORC) && (defined(HAVE_CPU_I386) || defined(HAVE_CPU_X86_64))
43 44 45 46
#define BUILD_X86_ASM
#endif
#endif

47 48 49
GST_DEBUG_CATEGORY_STATIC (gst_videomixer_blend_debug);
#define GST_CAT_DEFAULT gst_videomixer_blend_debug

50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138
/* Below are the implementations of everything */

/* A32 is for AYUV, ARGB and BGRA */
#define BLEND_A32(name, LOOP) \
static void \
blend_##name (const guint8 * src, gint xpos, gint ypos, \
    gint src_width, gint src_height, gdouble src_alpha, \
    guint8 * dest, gint dest_width, gint dest_height) \
{ \
  guint s_alpha; \
  gint src_stride, dest_stride; \
  \
  src_stride = src_width * 4; \
  dest_stride = dest_width * 4; \
  \
  s_alpha = CLAMP ((gint) (src_alpha * 256), 0, 256); \
  \
  /* If it's completely transparent... we just return */ \
  if (G_UNLIKELY (s_alpha == 0)) \
    return; \
  \
  /* adjust src pointers for negative sizes */ \
  if (xpos < 0) { \
    src += -xpos * 4; \
    src_width -= -xpos; \
    xpos = 0; \
  } \
  if (ypos < 0) { \
    src += -ypos * src_stride; \
    src_height -= -ypos; \
    ypos = 0; \
  } \
  /* adjust width/height if the src is bigger than dest */ \
  if (xpos + src_width > dest_width) { \
    src_width = dest_width - xpos; \
  } \
  if (ypos + src_height > dest_height) { \
    src_height = dest_height - ypos; \
  } \
  \
  dest = dest + 4 * xpos + (ypos * dest_stride); \
  \
  LOOP (dest, src, src_height, src_width, src_stride, dest_stride, s_alpha); \
}

#define BLEND_A32_LOOP_C(name, A, C1, C2, C3) \
static inline void \
_blend_loop_##name##_c (guint8 *dest, const guint8 *src, gint src_height, gint src_width, gint src_stride, gint dest_stride, guint s_alpha) { \
  gint i, j; \
  gint alpha; \
  gint src_add = src_stride - (4 * src_width); \
  gint dest_add = dest_stride - (4 * src_width); \
  \
  for (i = 0; i < src_height; i++) { \
    for (j = 0; j < src_width; j++) { \
      alpha = (src[A] * s_alpha) >> 8; \
      dest[A] = 0xff; \
      dest[C1] = BLEND(dest[C1], src[C1], alpha); \
      dest[C2] = BLEND(dest[C2], src[C2], alpha); \
      dest[C3] = BLEND(dest[C3], src[C3], alpha); \
      \
      src += 4; \
      dest += 4; \
    } \
    src += src_add; \
    dest += dest_add; \
  } \
}

BLEND_A32_LOOP_C (argb, 0, 1, 2, 3);
BLEND_A32_LOOP_C (bgra, 3, 2, 1, 0);
BLEND_A32 (argb_c, _blend_loop_argb_c);
BLEND_A32 (bgra_c, _blend_loop_bgra_c);

#define A32_CHECKER_C(name, RGB, A, C1, C2, C3) \
static void \
fill_checker_##name##_c (guint8 * dest, gint width, gint height) \
{ \
  gint i, j; \
  gint val; \
  static const gint tab[] = { 80, 160, 80, 160 }; \
  \
  if (!RGB) { \
    for (i = 0; i < height; i++) { \
      for (j = 0; j < width; j++) { \
        dest[A] = 0xff; \
        dest[C1] = tab[((i & 0x8) >> 3) + ((j & 0x8) >> 3)]; \
        dest[C2] = 128; \
        dest[C3] = 128; \
139
        dest += 4; \
140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163
      } \
    } \
  } else { \
    for (i = 0; i < height; i++) { \
      for (j = 0; j < width; j++) { \
        val = tab[((i & 0x8) >> 3) + ((j & 0x8) >> 3)]; \
        dest[A] = 0xFF; \
        dest[C1] = val; \
        dest[C2] = val; \
        dest[C3] = val; \
        dest += 4; \
      } \
    } \
  } \
}

A32_CHECKER_C (argb, TRUE, 0, 1, 2, 3);
A32_CHECKER_C (bgra, TRUE, 3, 2, 1, 0);
A32_CHECKER_C (ayuv, FALSE, 0, 1, 2, 3);

#define YUV_TO_R(Y,U,V) (CLAMP (1.164 * (Y - 16) + 1.596 * (V - 128), 0, 255))
#define YUV_TO_G(Y,U,V) (CLAMP (1.164 * (Y - 16) - 0.813 * (V - 128) - 0.391 * (U - 128), 0, 255))
#define YUV_TO_B(Y,U,V) (CLAMP (1.164 * (Y - 16) + 2.018 * (U - 128), 0, 255))

164
#define A32_COLOR(name, RGB, A, C1, C2, C3) \
165 166 167 168
static void \
fill_color_##name (guint8 * dest, gint width, gint height, gint Y, gint U, gint V) \
{ \
  gint c1, c2, c3; \
169
  guint32 val; \
170 171 172 173 174 175 176 177 178 179
  \
  if (RGB) { \
    c1 = YUV_TO_R (Y, U, V); \
    c2 = YUV_TO_G (Y, U, V); \
    c3 = YUV_TO_B (Y, U, V); \
  } else { \
    c1 = Y; \
    c2 = U; \
    c3 = V; \
  } \
180
  val = GUINT32_FROM_BE ((0xff << A) | (c1 << C1) | (c2 << C2) | (c3 << C3)); \
181
  \
182
  orc_splat_u32 ((guint32 *) dest, val, height * width); \
183 184
}

185 186 187 188 189
A32_COLOR (argb, TRUE, 24, 16, 8, 0);
A32_COLOR (bgra, TRUE, 0, 8, 16, 24);
A32_COLOR (abgr, TRUE, 24, 0, 8, 16);
A32_COLOR (rgba, TRUE, 0, 24, 16, 8);
A32_COLOR (ayuv, FALSE, 24, 16, 8, 0);
190

191
/* Y444, Y42B, I420, YV12, Y41B */
192
#define PLANAR_YUV_BLEND(format_name,format_enum,x_round,y_round,MEMCPY,BLENDLOOP) \
193
inline static void \
194
_blend_##format_name (const guint8 * src, guint8 * dest, \
195
    gint src_stride, gint dest_stride, gint src_width, gint src_height, \
196
    gdouble src_alpha) \
197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219
{ \
  gint i; \
  gint b_alpha; \
  \
  /* If it's completely transparent... we just return */ \
  if (G_UNLIKELY (src_alpha == 0.0)) { \
    GST_INFO ("Fast copy (alpha == 0.0)"); \
    return; \
  } \
  \
  /* If it's completely opaque, we do a fast copy */ \
  if (G_UNLIKELY (src_alpha == 1.0)) { \
    GST_INFO ("Fast copy (alpha == 1.0)"); \
    for (i = 0; i < src_height; i++) { \
      MEMCPY (dest, src, src_width); \
      src += src_stride; \
      dest += dest_stride; \
    } \
    return; \
  } \
  \
  b_alpha = CLAMP ((gint) (src_alpha * 256), 0, 256); \
  \
220
  BLENDLOOP(dest, dest_stride, src, src_stride, b_alpha, src_width, src_height); \
221 222 223
} \
\
static void \
224
blend_##format_name (const guint8 * src, gint xpos, gint ypos, \
225 226 227 228 229 230 231 232 233
    gint src_width, gint src_height, gdouble src_alpha, \
    guint8 * dest, gint dest_width, gint dest_height) \
{ \
  const guint8 *b_src; \
  guint8 *b_dest; \
  gint b_src_width = src_width; \
  gint b_src_height = src_height; \
  gint xoffset = 0; \
  gint yoffset = 0; \
234 235 236 237 238
  gint src_comp_rowstride, dest_comp_rowstride; \
  gint src_comp_height, dest_comp_height; \
  gint src_comp_width, dest_comp_width; \
  gint comp_ypos, comp_xpos; \
  gint comp_yoffset, comp_xoffset; \
239
  \
240 241
  xpos = x_round (xpos); \
  ypos = y_round (ypos); \
242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270
  \
  /* adjust src pointers for negative sizes */ \
  if (xpos < 0) { \
    xoffset = -xpos; \
    b_src_width -= -xpos; \
    xpos = 0; \
  } \
  if (ypos < 0) { \
    yoffset += -ypos; \
    b_src_height -= -ypos; \
    ypos = 0; \
  } \
  /* If x or y offset are larger then the source it's outside of the picture */ \
  if (xoffset > src_width || yoffset > src_width) { \
    return; \
  } \
  \
  /* adjust width/height if the src is bigger than dest */ \
  if (xpos + src_width > dest_width) { \
    b_src_width = dest_width - xpos; \
  } \
  if (ypos + src_height > dest_height) { \
    b_src_height = dest_height - ypos; \
  } \
  if (b_src_width < 0 || b_src_height < 0) { \
    return; \
  } \
  \
  /* First mix Y, then U, then V */ \
271 272 273 274 275 276 277 278 279 280 281 282
  b_src = src + gst_video_format_get_component_offset (format_enum, 0, src_width, src_height); \
  b_dest = dest + gst_video_format_get_component_offset (format_enum, 0, dest_width, dest_height); \
  src_comp_rowstride = gst_video_format_get_row_stride (format_enum, 0, src_width); \
  dest_comp_rowstride = gst_video_format_get_row_stride (format_enum, 0, dest_width); \
  src_comp_height = gst_video_format_get_component_height (format_enum, 0, b_src_height); \
  dest_comp_height = gst_video_format_get_component_height (format_enum, 0, dest_height); \
  src_comp_width = gst_video_format_get_component_width (format_enum, 0, b_src_width); \
  dest_comp_width = gst_video_format_get_component_width (format_enum, 0, dest_width); \
  comp_xpos = (xpos == 0) ? 0 : gst_video_format_get_component_width (format_enum, 0, xpos); \
  comp_ypos = (ypos == 0) ? 0 : gst_video_format_get_component_height (format_enum, 0, ypos); \
  comp_xoffset = (xoffset == 0) ? 0 : gst_video_format_get_component_width (format_enum, 0, xoffset); \
  comp_yoffset = (yoffset == 0) ? 0 : gst_video_format_get_component_height (format_enum, 0, yoffset); \
283
  _blend_##format_name (b_src + comp_xoffset + comp_yoffset * src_comp_rowstride, \
284 285 286
      b_dest + comp_xpos + comp_ypos * dest_comp_rowstride, \
      src_comp_rowstride, \
      dest_comp_rowstride, src_comp_width, src_comp_height, \
287
      src_alpha); \
288 289 290 291 292 293 294 295 296 297 298 299 300
  \
  b_src = src + gst_video_format_get_component_offset (format_enum, 1, src_width, src_height); \
  b_dest = dest + gst_video_format_get_component_offset (format_enum, 1, dest_width, dest_height); \
  src_comp_rowstride = gst_video_format_get_row_stride (format_enum, 1, src_width); \
  dest_comp_rowstride = gst_video_format_get_row_stride (format_enum, 1, dest_width); \
  src_comp_height = gst_video_format_get_component_height (format_enum, 1, b_src_height); \
  dest_comp_height = gst_video_format_get_component_height (format_enum, 1, dest_height); \
  src_comp_width = gst_video_format_get_component_width (format_enum, 1, b_src_width); \
  dest_comp_width = gst_video_format_get_component_width (format_enum, 1, dest_width); \
  comp_xpos = (xpos == 0) ? 0 : gst_video_format_get_component_width (format_enum, 1, xpos); \
  comp_ypos = (ypos == 0) ? 0 : gst_video_format_get_component_height (format_enum, 1, ypos); \
  comp_xoffset = (xoffset == 0) ? 0 : gst_video_format_get_component_width (format_enum, 1, xoffset); \
  comp_yoffset = (yoffset == 0) ? 0 : gst_video_format_get_component_height (format_enum, 1, yoffset); \
301
  _blend_##format_name (b_src + comp_xoffset + comp_yoffset * src_comp_rowstride, \
302 303 304
      b_dest + comp_xpos + comp_ypos * dest_comp_rowstride, \
      src_comp_rowstride, \
      dest_comp_rowstride, src_comp_width, src_comp_height, \
305
      src_alpha); \
306 307 308 309 310 311 312 313 314 315 316 317 318
  \
  b_src = src + gst_video_format_get_component_offset (format_enum, 2, src_width, src_height); \
  b_dest = dest + gst_video_format_get_component_offset (format_enum, 2, dest_width, dest_height); \
  src_comp_rowstride = gst_video_format_get_row_stride (format_enum, 2, src_width); \
  dest_comp_rowstride = gst_video_format_get_row_stride (format_enum, 2, dest_width); \
  src_comp_height = gst_video_format_get_component_height (format_enum, 2, b_src_height); \
  dest_comp_height = gst_video_format_get_component_height (format_enum, 2, dest_height); \
  src_comp_width = gst_video_format_get_component_width (format_enum, 2, b_src_width); \
  dest_comp_width = gst_video_format_get_component_width (format_enum, 2, dest_width); \
  comp_xpos = (xpos == 0) ? 0 : gst_video_format_get_component_width (format_enum, 2, xpos); \
  comp_ypos = (ypos == 0) ? 0 : gst_video_format_get_component_height (format_enum, 2, ypos); \
  comp_xoffset = (xoffset == 0) ? 0 : gst_video_format_get_component_width (format_enum, 2, xoffset); \
  comp_yoffset = (yoffset == 0) ? 0 : gst_video_format_get_component_height (format_enum, 2, yoffset); \
319
  _blend_##format_name (b_src + comp_xoffset + comp_yoffset * src_comp_rowstride, \
320 321 322
      b_dest + comp_xpos + comp_ypos * dest_comp_rowstride, \
      src_comp_rowstride, \
      dest_comp_rowstride, src_comp_width, src_comp_height, \
323
      src_alpha); \
324 325
}

326
#define PLANAR_YUV_FILL_CHECKER(format_name, format_enum, MEMSET) \
327
static void \
328
fill_checker_##format_name (guint8 * dest, gint width, gint height) \
329 330 331
{ \
  gint i, j; \
  static const int tab[] = { 80, 160, 80, 160 }; \
332 333 334
  guint8 *p; \
  gint comp_width, comp_height; \
  gint rowstride; \
335
  \
336 337 338 339 340 341 342
  p = dest + gst_video_format_get_component_offset (format_enum, 0, width, height); \
  comp_width = gst_video_format_get_component_width (format_enum, 0, width); \
  comp_height = gst_video_format_get_component_height (format_enum, 0, height); \
  rowstride = gst_video_format_get_row_stride (format_enum, 0, width); \
  \
  for (i = 0; i < comp_height; i++) { \
    for (j = 0; j < comp_width; j++) { \
343 344
      *p++ = tab[((i & 0x8) >> 3) + ((j & 0x8) >> 3)]; \
    } \
345 346 347 348 349 350 351 352 353 354 355
    p += rowstride - comp_width; \
  } \
  \
  p = dest + gst_video_format_get_component_offset (format_enum, 1, width, height); \
  comp_width = gst_video_format_get_component_width (format_enum, 1, width); \
  comp_height = gst_video_format_get_component_height (format_enum, 1, height); \
  rowstride = gst_video_format_get_row_stride (format_enum, 1, width); \
  \
  for (i = 0; i < comp_height; i++) { \
    MEMSET (p, 0x80, comp_width); \
    p += rowstride; \
356 357
  } \
  \
358 359 360 361
  p = dest + gst_video_format_get_component_offset (format_enum, 2, width, height); \
  comp_width = gst_video_format_get_component_width (format_enum, 2, width); \
  comp_height = gst_video_format_get_component_height (format_enum, 2, height); \
  rowstride = gst_video_format_get_row_stride (format_enum, 2, width); \
362
  \
363 364 365 366
  for (i = 0; i < comp_height; i++) { \
    MEMSET (p, 0x80, comp_width); \
    p += rowstride; \
  } \
367 368
}

369
#define PLANAR_YUV_FILL_COLOR(format_name,format_enum,MEMSET) \
370
static void \
371
fill_color_##format_name (guint8 * dest, gint width, gint height, \
372 373
    gint colY, gint colU, gint colV) \
{ \
374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392
  guint8 *p; \
  gint comp_width, comp_height; \
  gint rowstride; \
  gint i; \
  \
  p = dest + gst_video_format_get_component_offset (format_enum, 0, width, height); \
  comp_width = gst_video_format_get_component_width (format_enum, 0, width); \
  comp_height = gst_video_format_get_component_height (format_enum, 0, height); \
  rowstride = gst_video_format_get_row_stride (format_enum, 0, width); \
  \
  for (i = 0; i < comp_height; i++) { \
    MEMSET (p, colY, comp_width); \
    p += rowstride; \
  } \
  \
  p = dest + gst_video_format_get_component_offset (format_enum, 1, width, height); \
  comp_width = gst_video_format_get_component_width (format_enum, 1, width); \
  comp_height = gst_video_format_get_component_height (format_enum, 1, height); \
  rowstride = gst_video_format_get_row_stride (format_enum, 1, width); \
393
  \
394 395 396 397
  for (i = 0; i < comp_height; i++) { \
    MEMSET (p, colU, comp_width); \
    p += rowstride; \
  } \
398
  \
399 400 401 402
  p = dest + gst_video_format_get_component_offset (format_enum, 2, width, height); \
  comp_width = gst_video_format_get_component_width (format_enum, 2, width); \
  comp_height = gst_video_format_get_component_height (format_enum, 2, height); \
  rowstride = gst_video_format_get_row_stride (format_enum, 2, width); \
403
  \
404 405 406 407
  for (i = 0; i < comp_height; i++) { \
    MEMSET (p, colV, comp_width); \
    p += rowstride; \
  } \
408 409
}

410 411
#define GST_ROUND_UP_1(x) (x)

412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428
PLANAR_YUV_BLEND (i420, GST_VIDEO_FORMAT_I420, GST_ROUND_UP_2,
    GST_ROUND_UP_2, memcpy, orc_blend_u8);
PLANAR_YUV_FILL_CHECKER (i420, GST_VIDEO_FORMAT_I420, memset);
PLANAR_YUV_FILL_COLOR (i420, GST_VIDEO_FORMAT_I420, memset);
PLANAR_YUV_FILL_COLOR (yv12, GST_VIDEO_FORMAT_YV12, memset);
PLANAR_YUV_BLEND (y444, GST_VIDEO_FORMAT_Y444, GST_ROUND_UP_1,
    GST_ROUND_UP_1, memcpy, orc_blend_u8);
PLANAR_YUV_FILL_CHECKER (y444, GST_VIDEO_FORMAT_Y444, memset);
PLANAR_YUV_FILL_COLOR (y444, GST_VIDEO_FORMAT_Y444, memset);
PLANAR_YUV_BLEND (y42b, GST_VIDEO_FORMAT_Y42B, GST_ROUND_UP_2,
    GST_ROUND_UP_1, memcpy, orc_blend_u8);
PLANAR_YUV_FILL_CHECKER (y42b, GST_VIDEO_FORMAT_Y42B, memset);
PLANAR_YUV_FILL_COLOR (y42b, GST_VIDEO_FORMAT_Y42B, memset);
PLANAR_YUV_BLEND (y41b, GST_VIDEO_FORMAT_Y41B, GST_ROUND_UP_4,
    GST_ROUND_UP_1, memcpy, orc_blend_u8);
PLANAR_YUV_FILL_CHECKER (y41b, GST_VIDEO_FORMAT_Y41B, memset);
PLANAR_YUV_FILL_COLOR (y41b, GST_VIDEO_FORMAT_Y41B, memset);
429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483

/* RGB, BGR, xRGB, xBGR, RGBx, BGRx */

#define RGB_BLEND(name, bpp, MEMCPY, BLENDLOOP) \
static void \
blend_##name (const guint8 * src, gint xpos, gint ypos, \
    gint src_width, gint src_height, gdouble src_alpha, \
    guint8 * dest, gint dest_width, gint dest_height) \
{ \
  gint b_alpha; \
  gint i; \
  gint src_stride, dest_stride; \
  \
  src_stride = GST_ROUND_UP_4 (src_width * bpp); \
  dest_stride = GST_ROUND_UP_4 (dest_width * bpp); \
  \
  b_alpha = CLAMP ((gint) (src_alpha * 256), 0, 256); \
  \
  /* adjust src pointers for negative sizes */ \
  if (xpos < 0) { \
    src += -xpos * bpp; \
    src_width -= -xpos; \
    xpos = 0; \
  } \
  if (ypos < 0) { \
    src += -ypos * src_stride; \
    src_height -= -ypos; \
    ypos = 0; \
  } \
  /* adjust width/height if the src is bigger than dest */ \
  if (xpos + src_width > dest_width) { \
    src_width = dest_width - xpos; \
  } \
  if (ypos + src_height > dest_height) { \
    src_height = dest_height - ypos; \
  } \
  \
  dest = dest + bpp * xpos + (ypos * dest_stride); \
  /* If it's completely transparent... we just return */ \
  if (G_UNLIKELY (src_alpha == 0.0)) { \
    GST_INFO ("Fast copy (alpha == 0.0)"); \
    return; \
  } \
  \
  /* If it's completely opaque, we do a fast copy */ \
  if (G_UNLIKELY (src_alpha == 1.0)) { \
    GST_INFO ("Fast copy (alpha == 1.0)"); \
    for (i = 0; i < src_height; i++) { \
      MEMCPY (dest, src, bpp * src_width); \
      src += src_stride; \
      dest += dest_stride; \
    } \
    return; \
  } \
  \
484
  BLENDLOOP(dest, dest_stride, src, src_stride, b_alpha, src_width * bpp, src_height); \
485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514
}

#define RGB_FILL_CHECKER_C(name, bpp, r, g, b) \
static void \
fill_checker_##name##_c (guint8 * dest, gint width, gint height) \
{ \
  gint i, j; \
  static const int tab[] = { 80, 160, 80, 160 }; \
  gint dest_add = GST_ROUND_UP_4 (width * bpp) - width * bpp; \
  \
  for (i = 0; i < height; i++) { \
    for (j = 0; j < width; j++) { \
      dest[r] = tab[((i & 0x8) >> 3) + ((j & 0x8) >> 3)];       /* red */ \
      dest[g] = tab[((i & 0x8) >> 3) + ((j & 0x8) >> 3)];       /* green */ \
      dest[b] = tab[((i & 0x8) >> 3) + ((j & 0x8) >> 3)];       /* blue */ \
      dest += bpp; \
    } \
    dest += dest_add; \
  } \
}

#define RGB_FILL_COLOR(name, bpp, MEMSET_RGB) \
static void \
fill_color_##name (guint8 * dest, gint width, gint height, \
    gint colY, gint colU, gint colV) \
{ \
  gint red, green, blue; \
  gint i; \
  gint dest_stride = GST_ROUND_UP_4 (width * bpp); \
  \
515 516 517
  red = YUV_TO_R (colY, colU, colV); \
  green = YUV_TO_G (colY, colU, colV); \
  blue = YUV_TO_B (colY, colU, colV); \
518 519 520 521 522 523 524
  \
  for (i = 0; i < height; i++) { \
    MEMSET_RGB (dest, red, green, blue, width); \
    dest += dest_stride; \
  } \
}

525
#define MEMSET_RGB_C(name, r, g, b) \
526 527 528 529 530 531 532 533
static inline void \
_memset_##name##_c (guint8* dest, gint red, gint green, gint blue, gint width) { \
  gint j; \
  \
  for (j = 0; j < width; j++) { \
    dest[r] = red; \
    dest[g] = green; \
    dest[b] = blue; \
534
    dest += 3; \
535 536 537
  } \
}

538 539 540 541 542 543 544 545 546 547 548 549
#define MEMSET_XRGB(name, r, g, b) \
static inline void \
_memset_##name (guint8* dest, gint red, gint green, gint blue, gint width) { \
  guint32 val; \
  \
  val = GUINT32_FROM_BE ((red << r) | (green << g) | (blue << b)); \
  orc_splat_u32 ((guint32 *) dest, val, width); \
}

#define _orc_memcpy_u32(dest,src,len) orc_memcpy_u32((guint32 *) dest, (const guint32 *) src, len/4)

RGB_BLEND (rgb, 3, memcpy, orc_blend_u8);
550
RGB_FILL_CHECKER_C (rgb, 3, 0, 1, 2);
551
MEMSET_RGB_C (rgb, 0, 1, 2);
552 553
RGB_FILL_COLOR (rgb_c, 3, _memset_rgb_c);

554
MEMSET_RGB_C (bgr, 2, 1, 0);
555 556
RGB_FILL_COLOR (bgr_c, 3, _memset_bgr_c);

557
RGB_BLEND (xrgb, 4, _orc_memcpy_u32, orc_blend_u8);
558
RGB_FILL_CHECKER_C (xrgb, 4, 1, 2, 3);
559 560
MEMSET_XRGB (xrgb, 24, 16, 0);
RGB_FILL_COLOR (xrgb, 4, _memset_xrgb);
561

562 563
MEMSET_XRGB (xbgr, 0, 16, 24);
RGB_FILL_COLOR (xbgr, 4, _memset_xbgr);
564

565 566
MEMSET_XRGB (rgbx, 24, 16, 8);
RGB_FILL_COLOR (rgbx, 4, _memset_rgbx);
567

568 569
MEMSET_XRGB (bgrx, 8, 16, 24);
RGB_FILL_COLOR (bgrx, 4, _memset_bgrx);
570

571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627
/* YUY2, YVYU, UYVY */

#define PACKED_422_BLEND(name, MEMCPY, BLENDLOOP) \
static void \
blend_##name (const guint8 * src, gint xpos, gint ypos, \
    gint src_width, gint src_height, gdouble src_alpha, \
    guint8 * dest, gint dest_width, gint dest_height) \
{ \
  gint b_alpha; \
  gint i; \
  gint src_stride, dest_stride; \
  \
  src_stride = GST_ROUND_UP_4 (src_width * 2); \
  dest_stride = GST_ROUND_UP_4 (dest_width * 2); \
  \
  b_alpha = CLAMP ((gint) (src_alpha * 256), 0, 256); \
  \
  xpos = GST_ROUND_UP_2 (xpos); \
  \
  /* adjust src pointers for negative sizes */ \
  if (xpos < 0) { \
    src += -xpos * 2; \
    src_width -= -xpos; \
    xpos = 0; \
  } \
  if (ypos < 0) { \
    src += -ypos * src_stride; \
    src_height -= -ypos; \
    ypos = 0; \
  } \
  \
  /* adjust width/height if the src is bigger than dest */ \
  if (xpos + src_width > dest_width) { \
    src_width = dest_width - xpos; \
  } \
  if (ypos + src_height > dest_height) { \
    src_height = dest_height - ypos; \
  } \
  \
  dest = dest + 2 * xpos + (ypos * dest_stride); \
  /* If it's completely transparent... we just return */ \
  if (G_UNLIKELY (src_alpha == 0.0)) { \
    GST_INFO ("Fast copy (alpha == 0.0)"); \
    return; \
  } \
  \
  /* If it's completely opaque, we do a fast copy */ \
  if (G_UNLIKELY (src_alpha == 1.0)) { \
    GST_INFO ("Fast copy (alpha == 1.0)"); \
    for (i = 0; i < src_height; i++) { \
      MEMCPY (dest, src, 2 * src_width); \
      src += src_stride; \
      dest += dest_stride; \
    } \
    return; \
  } \
  \
628
  BLENDLOOP(dest, dest_stride, src, src_stride, b_alpha, 2 * src_width, src_height); \
629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659
}

#define PACKED_422_FILL_CHECKER_C(name, Y1, U, Y2, V) \
static void \
fill_checker_##name##_c (guint8 * dest, gint width, gint height) \
{ \
  gint i, j; \
  static const int tab[] = { 80, 160, 80, 160 }; \
  gint dest_add; \
  \
  width = GST_ROUND_UP_2 (width); \
  dest_add = GST_ROUND_UP_4 (width * 2) - width * 2; \
  width /= 2; \
  \
  for (i = 0; i < height; i++) { \
    for (j = 0; j < width; j++) { \
      dest[Y1] = tab[((i & 0x8) >> 3) + ((j & 0x8) >> 3)]; \
      dest[Y2] = tab[((i & 0x8) >> 3) + ((j & 0x8) >> 3)]; \
      dest[U] = 128; \
      dest[V] = 128; \
      dest += 4; \
    } \
    dest += dest_add; \
  } \
}

#define PACKED_422_FILL_COLOR(name, Y1, U, Y2, V) \
static void \
fill_color_##name (guint8 * dest, gint width, gint height, \
    gint colY, gint colU, gint colV) \
{ \
660 661 662
  gint i; \
  gint dest_stride; \
  guint32 val; \
663 664
  \
  width = GST_ROUND_UP_2 (width); \
665
  dest_stride = GST_ROUND_UP_4 (width * 2); \
666 667
  width /= 2; \
  \
668 669
  val = GUINT32_FROM_BE ((colY << Y1) | (colY << Y2) | (colU << U) | (colV << V)); \
  \
670
  for (i = 0; i < height; i++) { \
671 672
    orc_splat_u32 ((guint32 *) dest, val, width); \
    dest += dest_stride; \
673 674 675
  } \
}

676
PACKED_422_BLEND (yuy2, memcpy, orc_blend_u8);
677 678
PACKED_422_FILL_CHECKER_C (yuy2, 0, 1, 2, 3);
PACKED_422_FILL_CHECKER_C (uyvy, 1, 0, 3, 2);
679 680 681
PACKED_422_FILL_COLOR (yuy2, 24, 16, 8, 0);
PACKED_422_FILL_COLOR (yvyu, 24, 0, 8, 16);
PACKED_422_FILL_COLOR (uyvy, 16, 24, 0, 8);
682

683 684
/* MMX Implementations */
#ifdef BUILD_X86_ASM
685

686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703
#define A32
#define NAME_BLEND _blend_loop_argb_mmx
#define A_OFF 0
#include "blend_mmx.h"
#undef NAME_BLEND
#undef A_OFF

#define NAME_BLEND _blend_loop_bgra_mmx
#define A_OFF 24
#include "blend_mmx.h"
#undef NAME_BLEND
#undef A_OFF
#undef A32

BLEND_A32 (argb_mmx, _blend_loop_argb_mmx);
BLEND_A32 (bgra_mmx, _blend_loop_bgra_mmx);
#endif

704 705 706 707 708 709 710 711 712 713 714 715
static void
_blend_loop_argb_orc (guint8 * dest, const guint8 * src, gint src_height,
    gint src_width, gint src_stride, gint dest_stride, guint s_alpha)
{
  s_alpha = MIN (255, s_alpha);
  gst_videomixer_orc_blend_ayuv (dest, dest_stride, src, src_stride,
      s_alpha, src_width, src_height);
}

BLEND_A32 (argb_orc, _blend_loop_argb_orc);


716 717 718
/* Init function */
BlendFunction gst_video_mixer_blend_argb;
BlendFunction gst_video_mixer_blend_bgra;
719
/* AYUV/ABGR is equal to ARGB, RGBA is equal to BGRA */
720 721
BlendFunction gst_video_mixer_blend_y444;
BlendFunction gst_video_mixer_blend_y42b;
722
BlendFunction gst_video_mixer_blend_i420;
723 724
/* I420 is equal to YV12 */
BlendFunction gst_video_mixer_blend_y41b;
725 726 727 728
BlendFunction gst_video_mixer_blend_rgb;
/* BGR is equal to RGB */
BlendFunction gst_video_mixer_blend_rgbx;
/* BGRx, xRGB, xBGR are equal to RGBx */
729 730
BlendFunction gst_video_mixer_blend_yuy2;
/* YVYU and UYVY are equal to YUY2 */
731 732 733

FillCheckerFunction gst_video_mixer_fill_checker_argb;
FillCheckerFunction gst_video_mixer_fill_checker_bgra;
734
/* ABGR is equal to ARGB, RGBA is equal to BGRA */
735
FillCheckerFunction gst_video_mixer_fill_checker_ayuv;
736 737
FillCheckerFunction gst_video_mixer_fill_checker_y444;
FillCheckerFunction gst_video_mixer_fill_checker_y42b;
738
FillCheckerFunction gst_video_mixer_fill_checker_i420;
739 740
/* I420 is equal to YV12 */
FillCheckerFunction gst_video_mixer_fill_checker_y41b;
741 742 743 744
FillCheckerFunction gst_video_mixer_fill_checker_rgb;
/* BGR is equal to RGB */
FillCheckerFunction gst_video_mixer_fill_checker_xrgb;
/* BGRx, xRGB, xBGR are equal to RGBx */
745 746 747
FillCheckerFunction gst_video_mixer_fill_checker_yuy2;
/* YVYU is equal to YUY2 */
FillCheckerFunction gst_video_mixer_fill_checker_uyvy;
748 749 750

FillColorFunction gst_video_mixer_fill_color_argb;
FillColorFunction gst_video_mixer_fill_color_bgra;
751 752
FillColorFunction gst_video_mixer_fill_color_abgr;
FillColorFunction gst_video_mixer_fill_color_rgba;
753
FillColorFunction gst_video_mixer_fill_color_ayuv;
754 755
FillColorFunction gst_video_mixer_fill_color_y444;
FillColorFunction gst_video_mixer_fill_color_y42b;
756
FillColorFunction gst_video_mixer_fill_color_i420;
757 758
FillColorFunction gst_video_mixer_fill_color_yv12;
FillColorFunction gst_video_mixer_fill_color_y41b;
759 760 761 762 763 764
FillColorFunction gst_video_mixer_fill_color_rgb;
FillColorFunction gst_video_mixer_fill_color_bgr;
FillColorFunction gst_video_mixer_fill_color_xrgb;
FillColorFunction gst_video_mixer_fill_color_xbgr;
FillColorFunction gst_video_mixer_fill_color_rgbx;
FillColorFunction gst_video_mixer_fill_color_bgrx;
765 766 767
FillColorFunction gst_video_mixer_fill_color_yuy2;
FillColorFunction gst_video_mixer_fill_color_yvyu;
FillColorFunction gst_video_mixer_fill_color_uyvy;
768 769 770 771

void
gst_video_mixer_init_blend (void)
{
772
#ifdef BUILD_X86_ASM
773 774
  guint cpu_flags;

775 776 777
  orc_init ();
  cpu_flags = orc_target_get_default_flags (orc_target_get_by_name ("mmx"));
#endif
778

779 780 781
  GST_DEBUG_CATEGORY_INIT (gst_videomixer_blend_debug, "videomixer_blend", 0,
      "video mixer blending functions");

782 783
  gst_video_mixer_blend_argb = blend_argb_c;
  gst_video_mixer_blend_bgra = blend_bgra_c;
784 785 786 787 788 789 790
  gst_video_mixer_blend_i420 = blend_i420;
  gst_video_mixer_blend_y444 = blend_y444;
  gst_video_mixer_blend_y42b = blend_y42b;
  gst_video_mixer_blend_y41b = blend_y41b;
  gst_video_mixer_blend_rgb = blend_rgb;
  gst_video_mixer_blend_xrgb = blend_xrgb;
  gst_video_mixer_blend_yuy2 = blend_yuy2;
791 792 793 794

  gst_video_mixer_fill_checker_argb = fill_checker_argb_c;
  gst_video_mixer_fill_checker_bgra = fill_checker_bgra_c;
  gst_video_mixer_fill_checker_ayuv = fill_checker_ayuv_c;
795 796 797 798
  gst_video_mixer_fill_checker_i420 = fill_checker_i420;
  gst_video_mixer_fill_checker_y444 = fill_checker_y444;
  gst_video_mixer_fill_checker_y42b = fill_checker_y42b;
  gst_video_mixer_fill_checker_y41b = fill_checker_y41b;
799 800
  gst_video_mixer_fill_checker_rgb = fill_checker_rgb_c;
  gst_video_mixer_fill_checker_xrgb = fill_checker_xrgb_c;
801 802
  gst_video_mixer_fill_checker_yuy2 = fill_checker_yuy2_c;
  gst_video_mixer_fill_checker_uyvy = fill_checker_uyvy_c;
803

804 805 806 807 808 809 810 811 812 813
  gst_video_mixer_fill_color_argb = fill_color_argb;
  gst_video_mixer_fill_color_bgra = fill_color_bgra;
  gst_video_mixer_fill_color_abgr = fill_color_abgr;
  gst_video_mixer_fill_color_rgba = fill_color_rgba;
  gst_video_mixer_fill_color_ayuv = fill_color_ayuv;
  gst_video_mixer_fill_color_i420 = fill_color_i420;
  gst_video_mixer_fill_color_yv12 = fill_color_yv12;
  gst_video_mixer_fill_color_y444 = fill_color_y444;
  gst_video_mixer_fill_color_y42b = fill_color_y42b;
  gst_video_mixer_fill_color_y41b = fill_color_y41b;
814 815
  gst_video_mixer_fill_color_rgb = fill_color_rgb_c;
  gst_video_mixer_fill_color_bgr = fill_color_bgr_c;
816 817 818 819 820 821 822
  gst_video_mixer_fill_color_xrgb = fill_color_xrgb;
  gst_video_mixer_fill_color_xbgr = fill_color_xbgr;
  gst_video_mixer_fill_color_rgbx = fill_color_rgbx;
  gst_video_mixer_fill_color_bgrx = fill_color_bgrx;
  gst_video_mixer_fill_color_yuy2 = fill_color_yuy2;
  gst_video_mixer_fill_color_yvyu = fill_color_yvyu;
  gst_video_mixer_fill_color_uyvy = fill_color_uyvy;
823 824

#ifdef BUILD_X86_ASM
825
  if (cpu_flags & ORC_TARGET_MMX_MMX) {
826 827 828 829
    gst_video_mixer_blend_argb = blend_argb_mmx;
    gst_video_mixer_blend_bgra = blend_bgra_mmx;
  }
#endif
830 831

  gst_video_mixer_blend_argb = blend_argb_orc;
832
}