yading@11: /* yading@11: * software YUV to RGB converter yading@11: * yading@11: * Copyright (C) 2009 Konstantin Shishkov yading@11: * yading@11: * 1,4,8bpp support and context / deglobalize stuff yading@11: * by Michael Niedermayer (michaelni@gmx.at) yading@11: * yading@11: * This file is part of FFmpeg. yading@11: * yading@11: * FFmpeg is free software; you can redistribute it and/or yading@11: * modify it under the terms of the GNU Lesser General Public yading@11: * License as published by the Free Software Foundation; either yading@11: * version 2.1 of the License, or (at your option) any later version. yading@11: * yading@11: * FFmpeg is distributed in the hope that it will be useful, yading@11: * but WITHOUT ANY WARRANTY; without even the implied warranty of yading@11: * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU yading@11: * Lesser General Public License for more details. yading@11: * yading@11: * You should have received a copy of the GNU Lesser General Public yading@11: * License along with FFmpeg; if not, write to the Free Software yading@11: * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA yading@11: */ yading@11: yading@11: #include yading@11: #include yading@11: #include yading@11: yading@11: #include "libavutil/cpu.h" yading@11: #include "libavutil/bswap.h" yading@11: #include "config.h" yading@11: #include "rgb2rgb.h" yading@11: #include "swscale.h" yading@11: #include "swscale_internal.h" yading@11: #include "libavutil/pixdesc.h" yading@11: yading@11: extern const uint8_t dither_2x2_4[3][8]; yading@11: extern const uint8_t dither_2x2_8[3][8]; yading@11: extern const uint8_t dither_4x4_16[5][8]; yading@11: extern const uint8_t dither_8x8_32[9][8]; yading@11: extern const uint8_t dither_8x8_73[9][8]; yading@11: extern const uint8_t dither_8x8_220[9][8]; yading@11: yading@11: const int32_t ff_yuv2rgb_coeffs[8][4] = { yading@11: { 117504, 138453, 13954, 34903 }, /* no sequence_display_extension */ yading@11: { 117504, 138453, 13954, 34903 }, /* ITU-R Rec. 709 (1990) */ yading@11: { 104597, 132201, 25675, 53279 }, /* unspecified */ yading@11: { 104597, 132201, 25675, 53279 }, /* reserved */ yading@11: { 104448, 132798, 24759, 53109 }, /* FCC */ yading@11: { 104597, 132201, 25675, 53279 }, /* ITU-R Rec. 624-4 System B, G */ yading@11: { 104597, 132201, 25675, 53279 }, /* SMPTE 170M */ yading@11: { 117579, 136230, 16907, 35559 } /* SMPTE 240M (1987) */ yading@11: }; yading@11: yading@11: const int *sws_getCoefficients(int colorspace) yading@11: { yading@11: if (colorspace > 7 || colorspace < 0) yading@11: colorspace = SWS_CS_DEFAULT; yading@11: return ff_yuv2rgb_coeffs[colorspace]; yading@11: } yading@11: yading@11: #define LOADCHROMA(i) \ yading@11: U = pu[i]; \ yading@11: V = pv[i]; \ yading@11: r = (void *)c->table_rV[V+YUVRGB_TABLE_HEADROOM]; \ yading@11: g = (void *)(c->table_gU[U+YUVRGB_TABLE_HEADROOM] + c->table_gV[V+YUVRGB_TABLE_HEADROOM]); \ yading@11: b = (void *)c->table_bU[U+YUVRGB_TABLE_HEADROOM]; yading@11: yading@11: #define PUTRGB(dst, src, i) \ yading@11: Y = src[2 * i]; \ yading@11: dst[2 * i] = r[Y] + g[Y] + b[Y]; \ yading@11: Y = src[2 * i + 1]; \ yading@11: dst[2 * i + 1] = r[Y] + g[Y] + b[Y]; yading@11: yading@11: #define PUTRGB24(dst, src, i) \ yading@11: Y = src[2 * i]; \ yading@11: dst[6 * i + 0] = r[Y]; \ yading@11: dst[6 * i + 1] = g[Y]; \ yading@11: dst[6 * i + 2] = b[Y]; \ yading@11: Y = src[2 * i + 1]; \ yading@11: dst[6 * i + 3] = r[Y]; \ yading@11: dst[6 * i + 4] = g[Y]; \ yading@11: dst[6 * i + 5] = b[Y]; yading@11: yading@11: #define PUTBGR24(dst, src, i) \ yading@11: Y = src[2 * i]; \ yading@11: dst[6 * i + 0] = b[Y]; \ yading@11: dst[6 * i + 1] = g[Y]; \ yading@11: dst[6 * i + 2] = r[Y]; \ yading@11: Y = src[2 * i + 1]; \ yading@11: dst[6 * i + 3] = b[Y]; \ yading@11: dst[6 * i + 4] = g[Y]; \ yading@11: dst[6 * i + 5] = r[Y]; yading@11: yading@11: #define PUTRGBA(dst, ysrc, asrc, i, s) \ yading@11: Y = ysrc[2 * i]; \ yading@11: dst[2 * i] = r[Y] + g[Y] + b[Y] + (asrc[2 * i] << s); \ yading@11: Y = ysrc[2 * i + 1]; \ yading@11: dst[2 * i + 1] = r[Y] + g[Y] + b[Y] + (asrc[2 * i + 1] << s); yading@11: yading@11: #define PUTRGB48(dst, src, i) \ yading@11: Y = src[ 2 * i]; \ yading@11: dst[12 * i + 0] = dst[12 * i + 1] = r[Y]; \ yading@11: dst[12 * i + 2] = dst[12 * i + 3] = g[Y]; \ yading@11: dst[12 * i + 4] = dst[12 * i + 5] = b[Y]; \ yading@11: Y = src[ 2 * i + 1]; \ yading@11: dst[12 * i + 6] = dst[12 * i + 7] = r[Y]; \ yading@11: dst[12 * i + 8] = dst[12 * i + 9] = g[Y]; \ yading@11: dst[12 * i + 10] = dst[12 * i + 11] = b[Y]; yading@11: yading@11: #define PUTBGR48(dst, src, i) \ yading@11: Y = src[2 * i]; \ yading@11: dst[12 * i + 0] = dst[12 * i + 1] = b[Y]; \ yading@11: dst[12 * i + 2] = dst[12 * i + 3] = g[Y]; \ yading@11: dst[12 * i + 4] = dst[12 * i + 5] = r[Y]; \ yading@11: Y = src[2 * i + 1]; \ yading@11: dst[12 * i + 6] = dst[12 * i + 7] = b[Y]; \ yading@11: dst[12 * i + 8] = dst[12 * i + 9] = g[Y]; \ yading@11: dst[12 * i + 10] = dst[12 * i + 11] = r[Y]; yading@11: yading@11: #define YUV2RGBFUNC(func_name, dst_type, alpha) \ yading@11: static int func_name(SwsContext *c, const uint8_t *src[], \ yading@11: int srcStride[], int srcSliceY, int srcSliceH, \ yading@11: uint8_t *dst[], int dstStride[]) \ yading@11: { \ yading@11: int y; \ yading@11: \ yading@11: if (!alpha && c->srcFormat == AV_PIX_FMT_YUV422P) { \ yading@11: srcStride[1] *= 2; \ yading@11: srcStride[2] *= 2; \ yading@11: } \ yading@11: for (y = 0; y < srcSliceH; y += 2) { \ yading@11: dst_type *dst_1 = \ yading@11: (dst_type *)(dst[0] + (y + srcSliceY) * dstStride[0]); \ yading@11: dst_type *dst_2 = \ yading@11: (dst_type *)(dst[0] + (y + srcSliceY + 1) * dstStride[0]); \ yading@11: dst_type av_unused *r, *g, *b; \ yading@11: const uint8_t *py_1 = src[0] + y * srcStride[0]; \ yading@11: const uint8_t *py_2 = py_1 + srcStride[0]; \ yading@11: const uint8_t *pu = src[1] + (y >> 1) * srcStride[1]; \ yading@11: const uint8_t *pv = src[2] + (y >> 1) * srcStride[2]; \ yading@11: const uint8_t av_unused *pa_1, *pa_2; \ yading@11: unsigned int h_size = c->dstW >> 3; \ yading@11: if (alpha) { \ yading@11: pa_1 = src[3] + y * srcStride[3]; \ yading@11: pa_2 = pa_1 + srcStride[3]; \ yading@11: } \ yading@11: while (h_size--) { \ yading@11: int av_unused U, V, Y; \ yading@11: yading@11: #define ENDYUV2RGBLINE(dst_delta, ss) \ yading@11: pu += 4 >> ss; \ yading@11: pv += 4 >> ss; \ yading@11: py_1 += 8 >> ss; \ yading@11: py_2 += 8 >> ss; \ yading@11: dst_1 += dst_delta >> ss; \ yading@11: dst_2 += dst_delta >> ss; \ yading@11: } \ yading@11: if (c->dstW & (4 >> ss)) { \ yading@11: int av_unused Y, U, V; \ yading@11: yading@11: #define ENDYUV2RGBFUNC() \ yading@11: } \ yading@11: } \ yading@11: return srcSliceH; \ yading@11: } yading@11: yading@11: #define CLOSEYUV2RGBFUNC(dst_delta) \ yading@11: ENDYUV2RGBLINE(dst_delta, 0) \ yading@11: ENDYUV2RGBFUNC() yading@11: yading@11: YUV2RGBFUNC(yuv2rgb_c_48, uint8_t, 0) yading@11: LOADCHROMA(0); yading@11: PUTRGB48(dst_1, py_1, 0); yading@11: PUTRGB48(dst_2, py_2, 0); yading@11: yading@11: LOADCHROMA(1); yading@11: PUTRGB48(dst_2, py_2, 1); yading@11: PUTRGB48(dst_1, py_1, 1); yading@11: yading@11: LOADCHROMA(2); yading@11: PUTRGB48(dst_1, py_1, 2); yading@11: PUTRGB48(dst_2, py_2, 2); yading@11: yading@11: LOADCHROMA(3); yading@11: PUTRGB48(dst_2, py_2, 3); yading@11: PUTRGB48(dst_1, py_1, 3); yading@11: ENDYUV2RGBLINE(48, 0) yading@11: LOADCHROMA(0); yading@11: PUTRGB48(dst_1, py_1, 0); yading@11: PUTRGB48(dst_2, py_2, 0); yading@11: yading@11: LOADCHROMA(1); yading@11: PUTRGB48(dst_2, py_2, 1); yading@11: PUTRGB48(dst_1, py_1, 1); yading@11: ENDYUV2RGBLINE(48, 1) yading@11: LOADCHROMA(0); yading@11: PUTRGB48(dst_1, py_1, 0); yading@11: PUTRGB48(dst_2, py_2, 0); yading@11: ENDYUV2RGBFUNC() yading@11: yading@11: YUV2RGBFUNC(yuv2rgb_c_bgr48, uint8_t, 0) yading@11: LOADCHROMA(0); yading@11: PUTBGR48(dst_1, py_1, 0); yading@11: PUTBGR48(dst_2, py_2, 0); yading@11: yading@11: LOADCHROMA(1); yading@11: PUTBGR48(dst_2, py_2, 1); yading@11: PUTBGR48(dst_1, py_1, 1); yading@11: yading@11: LOADCHROMA(2); yading@11: PUTBGR48(dst_1, py_1, 2); yading@11: PUTBGR48(dst_2, py_2, 2); yading@11: yading@11: LOADCHROMA(3); yading@11: PUTBGR48(dst_2, py_2, 3); yading@11: PUTBGR48(dst_1, py_1, 3); yading@11: ENDYUV2RGBLINE(48, 0) yading@11: LOADCHROMA(0); yading@11: PUTBGR48(dst_1, py_1, 0); yading@11: PUTBGR48(dst_2, py_2, 0); yading@11: yading@11: LOADCHROMA(1); yading@11: PUTBGR48(dst_2, py_2, 1); yading@11: PUTBGR48(dst_1, py_1, 1); yading@11: ENDYUV2RGBLINE(48, 1) yading@11: LOADCHROMA(0); yading@11: PUTBGR48(dst_1, py_1, 0); yading@11: PUTBGR48(dst_2, py_2, 0); yading@11: ENDYUV2RGBFUNC() yading@11: yading@11: YUV2RGBFUNC(yuv2rgb_c_32, uint32_t, 0) yading@11: LOADCHROMA(0); yading@11: PUTRGB(dst_1, py_1, 0); yading@11: PUTRGB(dst_2, py_2, 0); yading@11: yading@11: LOADCHROMA(1); yading@11: PUTRGB(dst_2, py_2, 1); yading@11: PUTRGB(dst_1, py_1, 1); yading@11: yading@11: LOADCHROMA(2); yading@11: PUTRGB(dst_1, py_1, 2); yading@11: PUTRGB(dst_2, py_2, 2); yading@11: yading@11: LOADCHROMA(3); yading@11: PUTRGB(dst_2, py_2, 3); yading@11: PUTRGB(dst_1, py_1, 3); yading@11: ENDYUV2RGBLINE(8, 0) yading@11: LOADCHROMA(0); yading@11: PUTRGB(dst_1, py_1, 0); yading@11: PUTRGB(dst_2, py_2, 0); yading@11: yading@11: LOADCHROMA(1); yading@11: PUTRGB(dst_2, py_2, 1); yading@11: PUTRGB(dst_1, py_1, 1); yading@11: ENDYUV2RGBLINE(8, 1) yading@11: LOADCHROMA(0); yading@11: PUTRGB(dst_1, py_1, 0); yading@11: PUTRGB(dst_2, py_2, 0); yading@11: ENDYUV2RGBFUNC() yading@11: yading@11: YUV2RGBFUNC(yuva2rgba_c, uint32_t, 1) yading@11: LOADCHROMA(0); yading@11: PUTRGBA(dst_1, py_1, pa_1, 0, 24); yading@11: PUTRGBA(dst_2, py_2, pa_2, 0, 24); yading@11: yading@11: LOADCHROMA(1); yading@11: PUTRGBA(dst_2, py_2, pa_1, 1, 24); yading@11: PUTRGBA(dst_1, py_1, pa_2, 1, 24); yading@11: yading@11: LOADCHROMA(2); yading@11: PUTRGBA(dst_1, py_1, pa_1, 2, 24); yading@11: PUTRGBA(dst_2, py_2, pa_2, 2, 24); yading@11: yading@11: LOADCHROMA(3); yading@11: PUTRGBA(dst_2, py_2, pa_1, 3, 24); yading@11: PUTRGBA(dst_1, py_1, pa_2, 3, 24); yading@11: pa_1 += 8; \ yading@11: pa_2 += 8; \ yading@11: ENDYUV2RGBLINE(8, 0) yading@11: LOADCHROMA(0); yading@11: PUTRGBA(dst_1, py_1, pa_1, 0, 24); yading@11: PUTRGBA(dst_2, py_2, pa_2, 0, 24); yading@11: yading@11: LOADCHROMA(1); yading@11: PUTRGBA(dst_2, py_2, pa_1, 1, 24); yading@11: PUTRGBA(dst_1, py_1, pa_2, 1, 24); yading@11: pa_1 += 4; \ yading@11: pa_2 += 4; \ yading@11: ENDYUV2RGBLINE(8, 1) yading@11: LOADCHROMA(0); yading@11: PUTRGBA(dst_1, py_1, pa_1, 0, 24); yading@11: PUTRGBA(dst_2, py_2, pa_2, 0, 24); yading@11: ENDYUV2RGBFUNC() yading@11: yading@11: YUV2RGBFUNC(yuva2argb_c, uint32_t, 1) yading@11: LOADCHROMA(0); yading@11: PUTRGBA(dst_1, py_1, pa_1, 0, 0); yading@11: PUTRGBA(dst_2, py_2, pa_2, 0, 0); yading@11: yading@11: LOADCHROMA(1); yading@11: PUTRGBA(dst_2, py_2, pa_2, 1, 0); yading@11: PUTRGBA(dst_1, py_1, pa_1, 1, 0); yading@11: yading@11: LOADCHROMA(2); yading@11: PUTRGBA(dst_1, py_1, pa_1, 2, 0); yading@11: PUTRGBA(dst_2, py_2, pa_2, 2, 0); yading@11: yading@11: LOADCHROMA(3); yading@11: PUTRGBA(dst_2, py_2, pa_2, 3, 0); yading@11: PUTRGBA(dst_1, py_1, pa_1, 3, 0); yading@11: pa_1 += 8; \ yading@11: pa_2 += 8; \ yading@11: ENDYUV2RGBLINE(8, 0) yading@11: LOADCHROMA(0); yading@11: PUTRGBA(dst_1, py_1, pa_1, 0, 0); yading@11: PUTRGBA(dst_2, py_2, pa_2, 0, 0); yading@11: yading@11: LOADCHROMA(1); yading@11: PUTRGBA(dst_2, py_2, pa_2, 1, 0); yading@11: PUTRGBA(dst_1, py_1, pa_1, 1, 0); yading@11: pa_1 += 4; \ yading@11: pa_2 += 4; \ yading@11: ENDYUV2RGBLINE(8, 1) yading@11: LOADCHROMA(0); yading@11: PUTRGBA(dst_1, py_1, pa_1, 0, 0); yading@11: PUTRGBA(dst_2, py_2, pa_2, 0, 0); yading@11: ENDYUV2RGBFUNC() yading@11: yading@11: YUV2RGBFUNC(yuv2rgb_c_24_rgb, uint8_t, 0) yading@11: LOADCHROMA(0); yading@11: PUTRGB24(dst_1, py_1, 0); yading@11: PUTRGB24(dst_2, py_2, 0); yading@11: yading@11: LOADCHROMA(1); yading@11: PUTRGB24(dst_2, py_2, 1); yading@11: PUTRGB24(dst_1, py_1, 1); yading@11: yading@11: LOADCHROMA(2); yading@11: PUTRGB24(dst_1, py_1, 2); yading@11: PUTRGB24(dst_2, py_2, 2); yading@11: yading@11: LOADCHROMA(3); yading@11: PUTRGB24(dst_2, py_2, 3); yading@11: PUTRGB24(dst_1, py_1, 3); yading@11: ENDYUV2RGBLINE(24, 0) yading@11: LOADCHROMA(0); yading@11: PUTRGB24(dst_1, py_1, 0); yading@11: PUTRGB24(dst_2, py_2, 0); yading@11: yading@11: LOADCHROMA(1); yading@11: PUTRGB24(dst_2, py_2, 1); yading@11: PUTRGB24(dst_1, py_1, 1); yading@11: ENDYUV2RGBLINE(24, 1) yading@11: LOADCHROMA(0); yading@11: PUTRGB24(dst_1, py_1, 0); yading@11: PUTRGB24(dst_2, py_2, 0); yading@11: ENDYUV2RGBFUNC() yading@11: yading@11: // only trivial mods from yuv2rgb_c_24_rgb yading@11: YUV2RGBFUNC(yuv2rgb_c_24_bgr, uint8_t, 0) yading@11: LOADCHROMA(0); yading@11: PUTBGR24(dst_1, py_1, 0); yading@11: PUTBGR24(dst_2, py_2, 0); yading@11: yading@11: LOADCHROMA(1); yading@11: PUTBGR24(dst_2, py_2, 1); yading@11: PUTBGR24(dst_1, py_1, 1); yading@11: yading@11: LOADCHROMA(2); yading@11: PUTBGR24(dst_1, py_1, 2); yading@11: PUTBGR24(dst_2, py_2, 2); yading@11: yading@11: LOADCHROMA(3); yading@11: PUTBGR24(dst_2, py_2, 3); yading@11: PUTBGR24(dst_1, py_1, 3); yading@11: ENDYUV2RGBLINE(24, 0) yading@11: LOADCHROMA(0); yading@11: PUTBGR24(dst_1, py_1, 0); yading@11: PUTBGR24(dst_2, py_2, 0); yading@11: yading@11: LOADCHROMA(1); yading@11: PUTBGR24(dst_2, py_2, 1); yading@11: PUTBGR24(dst_1, py_1, 1); yading@11: ENDYUV2RGBLINE(24, 1) yading@11: LOADCHROMA(0); yading@11: PUTBGR24(dst_1, py_1, 0); yading@11: PUTBGR24(dst_2, py_2, 0); yading@11: ENDYUV2RGBFUNC() yading@11: yading@11: YUV2RGBFUNC(yuv2rgb_c_16_ordered_dither, uint16_t, 0) yading@11: const uint8_t *d16 = dither_2x2_8[y & 1]; yading@11: const uint8_t *e16 = dither_2x2_4[y & 1]; yading@11: const uint8_t *f16 = dither_2x2_8[(y & 1)^1]; yading@11: yading@11: #define PUTRGB16(dst, src, i, o) \ yading@11: Y = src[2 * i]; \ yading@11: dst[2 * i] = r[Y + d16[0 + o]] + \ yading@11: g[Y + e16[0 + o]] + \ yading@11: b[Y + f16[0 + o]]; \ yading@11: Y = src[2 * i + 1]; \ yading@11: dst[2 * i + 1] = r[Y + d16[1 + o]] + \ yading@11: g[Y + e16[1 + o]] + \ yading@11: b[Y + f16[1 + o]]; yading@11: LOADCHROMA(0); yading@11: PUTRGB16(dst_1, py_1, 0, 0); yading@11: PUTRGB16(dst_2, py_2, 0, 0 + 8); yading@11: yading@11: LOADCHROMA(1); yading@11: PUTRGB16(dst_2, py_2, 1, 2 + 8); yading@11: PUTRGB16(dst_1, py_1, 1, 2); yading@11: yading@11: LOADCHROMA(2); yading@11: PUTRGB16(dst_1, py_1, 2, 4); yading@11: PUTRGB16(dst_2, py_2, 2, 4 + 8); yading@11: yading@11: LOADCHROMA(3); yading@11: PUTRGB16(dst_2, py_2, 3, 6 + 8); yading@11: PUTRGB16(dst_1, py_1, 3, 6); yading@11: CLOSEYUV2RGBFUNC(8) yading@11: yading@11: YUV2RGBFUNC(yuv2rgb_c_15_ordered_dither, uint16_t, 0) yading@11: const uint8_t *d16 = dither_2x2_8[y & 1]; yading@11: const uint8_t *e16 = dither_2x2_8[(y & 1)^1]; yading@11: yading@11: #define PUTRGB15(dst, src, i, o) \ yading@11: Y = src[2 * i]; \ yading@11: dst[2 * i] = r[Y + d16[0 + o]] + \ yading@11: g[Y + d16[1 + o]] + \ yading@11: b[Y + e16[0 + o]]; \ yading@11: Y = src[2 * i + 1]; \ yading@11: dst[2 * i + 1] = r[Y + d16[1 + o]] + \ yading@11: g[Y + d16[0 + o]] + \ yading@11: b[Y + e16[1 + o]]; yading@11: LOADCHROMA(0); yading@11: PUTRGB15(dst_1, py_1, 0, 0); yading@11: PUTRGB15(dst_2, py_2, 0, 0 + 8); yading@11: yading@11: LOADCHROMA(1); yading@11: PUTRGB15(dst_2, py_2, 1, 2 + 8); yading@11: PUTRGB15(dst_1, py_1, 1, 2); yading@11: yading@11: LOADCHROMA(2); yading@11: PUTRGB15(dst_1, py_1, 2, 4); yading@11: PUTRGB15(dst_2, py_2, 2, 4 + 8); yading@11: yading@11: LOADCHROMA(3); yading@11: PUTRGB15(dst_2, py_2, 3, 6 + 8); yading@11: PUTRGB15(dst_1, py_1, 3, 6); yading@11: CLOSEYUV2RGBFUNC(8) yading@11: yading@11: // r, g, b, dst_1, dst_2 yading@11: YUV2RGBFUNC(yuv2rgb_c_12_ordered_dither, uint16_t, 0) yading@11: const uint8_t *d16 = dither_4x4_16[y & 3]; yading@11: yading@11: #define PUTRGB12(dst, src, i, o) \ yading@11: Y = src[2 * i]; \ yading@11: dst[2 * i] = r[Y + d16[0 + o]] + \ yading@11: g[Y + d16[0 + o]] + \ yading@11: b[Y + d16[0 + o]]; \ yading@11: Y = src[2 * i + 1]; \ yading@11: dst[2 * i + 1] = r[Y + d16[1 + o]] + \ yading@11: g[Y + d16[1 + o]] + \ yading@11: b[Y + d16[1 + o]]; yading@11: yading@11: LOADCHROMA(0); yading@11: PUTRGB12(dst_1, py_1, 0, 0); yading@11: PUTRGB12(dst_2, py_2, 0, 0 + 8); yading@11: yading@11: LOADCHROMA(1); yading@11: PUTRGB12(dst_2, py_2, 1, 2 + 8); yading@11: PUTRGB12(dst_1, py_1, 1, 2); yading@11: yading@11: LOADCHROMA(2); yading@11: PUTRGB12(dst_1, py_1, 2, 4); yading@11: PUTRGB12(dst_2, py_2, 2, 4 + 8); yading@11: yading@11: LOADCHROMA(3); yading@11: PUTRGB12(dst_2, py_2, 3, 6 + 8); yading@11: PUTRGB12(dst_1, py_1, 3, 6); yading@11: CLOSEYUV2RGBFUNC(8) yading@11: yading@11: // r, g, b, dst_1, dst_2 yading@11: YUV2RGBFUNC(yuv2rgb_c_8_ordered_dither, uint8_t, 0) yading@11: const uint8_t *d32 = dither_8x8_32[y & 7]; yading@11: const uint8_t *d64 = dither_8x8_73[y & 7]; yading@11: yading@11: #define PUTRGB8(dst, src, i, o) \ yading@11: Y = src[2 * i]; \ yading@11: dst[2 * i] = r[Y + d32[0 + o]] + \ yading@11: g[Y + d32[0 + o]] + \ yading@11: b[Y + d64[0 + o]]; \ yading@11: Y = src[2 * i + 1]; \ yading@11: dst[2 * i + 1] = r[Y + d32[1 + o]] + \ yading@11: g[Y + d32[1 + o]] + \ yading@11: b[Y + d64[1 + o]]; yading@11: yading@11: LOADCHROMA(0); yading@11: PUTRGB8(dst_1, py_1, 0, 0); yading@11: PUTRGB8(dst_2, py_2, 0, 0 + 8); yading@11: yading@11: LOADCHROMA(1); yading@11: PUTRGB8(dst_2, py_2, 1, 2 + 8); yading@11: PUTRGB8(dst_1, py_1, 1, 2); yading@11: yading@11: LOADCHROMA(2); yading@11: PUTRGB8(dst_1, py_1, 2, 4); yading@11: PUTRGB8(dst_2, py_2, 2, 4 + 8); yading@11: yading@11: LOADCHROMA(3); yading@11: PUTRGB8(dst_2, py_2, 3, 6 + 8); yading@11: PUTRGB8(dst_1, py_1, 3, 6); yading@11: CLOSEYUV2RGBFUNC(8) yading@11: yading@11: YUV2RGBFUNC(yuv2rgb_c_4_ordered_dither, uint8_t, 0) yading@11: const uint8_t * d64 = dither_8x8_73[y & 7]; yading@11: const uint8_t *d128 = dither_8x8_220[y & 7]; yading@11: int acc; yading@11: yading@11: #define PUTRGB4D(dst, src, i, o) \ yading@11: Y = src[2 * i]; \ yading@11: acc = r[Y + d128[0 + o]] + \ yading@11: g[Y + d64[0 + o]] + \ yading@11: b[Y + d128[0 + o]]; \ yading@11: Y = src[2 * i + 1]; \ yading@11: acc |= (r[Y + d128[1 + o]] + \ yading@11: g[Y + d64[1 + o]] + \ yading@11: b[Y + d128[1 + o]]) << 4; \ yading@11: dst[i] = acc; yading@11: yading@11: LOADCHROMA(0); yading@11: PUTRGB4D(dst_1, py_1, 0, 0); yading@11: PUTRGB4D(dst_2, py_2, 0, 0 + 8); yading@11: yading@11: LOADCHROMA(1); yading@11: PUTRGB4D(dst_2, py_2, 1, 2 + 8); yading@11: PUTRGB4D(dst_1, py_1, 1, 2); yading@11: yading@11: LOADCHROMA(2); yading@11: PUTRGB4D(dst_1, py_1, 2, 4); yading@11: PUTRGB4D(dst_2, py_2, 2, 4 + 8); yading@11: yading@11: LOADCHROMA(3); yading@11: PUTRGB4D(dst_2, py_2, 3, 6 + 8); yading@11: PUTRGB4D(dst_1, py_1, 3, 6); yading@11: CLOSEYUV2RGBFUNC(4) yading@11: yading@11: YUV2RGBFUNC(yuv2rgb_c_4b_ordered_dither, uint8_t, 0) yading@11: const uint8_t *d64 = dither_8x8_73[y & 7]; yading@11: const uint8_t *d128 = dither_8x8_220[y & 7]; yading@11: yading@11: #define PUTRGB4DB(dst, src, i, o) \ yading@11: Y = src[2 * i]; \ yading@11: dst[2 * i] = r[Y + d128[0 + o]] + \ yading@11: g[Y + d64[0 + o]] + \ yading@11: b[Y + d128[0 + o]]; \ yading@11: Y = src[2 * i + 1]; \ yading@11: dst[2 * i + 1] = r[Y + d128[1 + o]] + \ yading@11: g[Y + d64[1 + o]] + \ yading@11: b[Y + d128[1 + o]]; yading@11: yading@11: LOADCHROMA(0); yading@11: PUTRGB4DB(dst_1, py_1, 0, 0); yading@11: PUTRGB4DB(dst_2, py_2, 0, 0 + 8); yading@11: yading@11: LOADCHROMA(1); yading@11: PUTRGB4DB(dst_2, py_2, 1, 2 + 8); yading@11: PUTRGB4DB(dst_1, py_1, 1, 2); yading@11: yading@11: LOADCHROMA(2); yading@11: PUTRGB4DB(dst_1, py_1, 2, 4); yading@11: PUTRGB4DB(dst_2, py_2, 2, 4 + 8); yading@11: yading@11: LOADCHROMA(3); yading@11: PUTRGB4DB(dst_2, py_2, 3, 6 + 8); yading@11: PUTRGB4DB(dst_1, py_1, 3, 6); yading@11: CLOSEYUV2RGBFUNC(8) yading@11: yading@11: YUV2RGBFUNC(yuv2rgb_c_1_ordered_dither, uint8_t, 0) yading@11: const uint8_t *d128 = dither_8x8_220[y & 7]; yading@11: char out_1 = 0, out_2 = 0; yading@11: g = c->table_gU[128 + YUVRGB_TABLE_HEADROOM] + c->table_gV[128 + YUVRGB_TABLE_HEADROOM]; yading@11: yading@11: #define PUTRGB1(out, src, i, o) \ yading@11: Y = src[2 * i]; \ yading@11: out += out + g[Y + d128[0 + o]]; \ yading@11: Y = src[2 * i + 1]; \ yading@11: out += out + g[Y + d128[1 + o]]; yading@11: yading@11: PUTRGB1(out_1, py_1, 0, 0); yading@11: PUTRGB1(out_2, py_2, 0, 0 + 8); yading@11: yading@11: PUTRGB1(out_2, py_2, 1, 2 + 8); yading@11: PUTRGB1(out_1, py_1, 1, 2); yading@11: yading@11: PUTRGB1(out_1, py_1, 2, 4); yading@11: PUTRGB1(out_2, py_2, 2, 4 + 8); yading@11: yading@11: PUTRGB1(out_2, py_2, 3, 6 + 8); yading@11: PUTRGB1(out_1, py_1, 3, 6); yading@11: yading@11: dst_1[0] = out_1; yading@11: dst_2[0] = out_2; yading@11: CLOSEYUV2RGBFUNC(1) yading@11: yading@11: SwsFunc ff_yuv2rgb_get_func_ptr(SwsContext *c) yading@11: { yading@11: SwsFunc t = NULL; yading@11: yading@11: if (HAVE_MMX) yading@11: t = ff_yuv2rgb_init_mmx(c); yading@11: else if (HAVE_VIS) yading@11: t = ff_yuv2rgb_init_vis(c); yading@11: else if (HAVE_ALTIVEC) yading@11: t = ff_yuv2rgb_init_altivec(c); yading@11: else if (ARCH_BFIN) yading@11: t = ff_yuv2rgb_get_func_ptr_bfin(c); yading@11: yading@11: if (t) yading@11: return t; yading@11: yading@11: av_log(c, AV_LOG_WARNING, yading@11: "No accelerated colorspace conversion found from %s to %s.\n", yading@11: av_get_pix_fmt_name(c->srcFormat), av_get_pix_fmt_name(c->dstFormat)); yading@11: yading@11: switch (c->dstFormat) { yading@11: case AV_PIX_FMT_BGR48BE: yading@11: case AV_PIX_FMT_BGR48LE: yading@11: return yuv2rgb_c_bgr48; yading@11: case AV_PIX_FMT_RGB48BE: yading@11: case AV_PIX_FMT_RGB48LE: yading@11: return yuv2rgb_c_48; yading@11: case AV_PIX_FMT_ARGB: yading@11: case AV_PIX_FMT_ABGR: yading@11: if (CONFIG_SWSCALE_ALPHA && isALPHA(c->srcFormat)) yading@11: return yuva2argb_c; yading@11: case AV_PIX_FMT_RGBA: yading@11: case AV_PIX_FMT_BGRA: yading@11: return (CONFIG_SWSCALE_ALPHA && isALPHA(c->srcFormat)) ? yuva2rgba_c : yuv2rgb_c_32; yading@11: case AV_PIX_FMT_RGB24: yading@11: return yuv2rgb_c_24_rgb; yading@11: case AV_PIX_FMT_BGR24: yading@11: return yuv2rgb_c_24_bgr; yading@11: case AV_PIX_FMT_RGB565: yading@11: case AV_PIX_FMT_BGR565: yading@11: return yuv2rgb_c_16_ordered_dither; yading@11: case AV_PIX_FMT_RGB555: yading@11: case AV_PIX_FMT_BGR555: yading@11: return yuv2rgb_c_15_ordered_dither; yading@11: case AV_PIX_FMT_RGB444: yading@11: case AV_PIX_FMT_BGR444: yading@11: return yuv2rgb_c_12_ordered_dither; yading@11: case AV_PIX_FMT_RGB8: yading@11: case AV_PIX_FMT_BGR8: yading@11: return yuv2rgb_c_8_ordered_dither; yading@11: case AV_PIX_FMT_RGB4: yading@11: case AV_PIX_FMT_BGR4: yading@11: return yuv2rgb_c_4_ordered_dither; yading@11: case AV_PIX_FMT_RGB4_BYTE: yading@11: case AV_PIX_FMT_BGR4_BYTE: yading@11: return yuv2rgb_c_4b_ordered_dither; yading@11: case AV_PIX_FMT_MONOBLACK: yading@11: return yuv2rgb_c_1_ordered_dither; yading@11: } yading@11: return NULL; yading@11: } yading@11: yading@11: static void fill_table(uint8_t* table[256 + 2*YUVRGB_TABLE_HEADROOM], const int elemsize, yading@11: const int64_t inc, void *y_tab) yading@11: { yading@11: int i; yading@11: uint8_t *y_table = y_tab; yading@11: yading@11: y_table -= elemsize * (inc >> 9); yading@11: yading@11: for (i = 0; i < 256 + 2*YUVRGB_TABLE_HEADROOM; i++) { yading@11: int64_t cb = av_clip(i-YUVRGB_TABLE_HEADROOM, 0, 255)*inc; yading@11: table[i] = y_table + elemsize * (cb >> 16); yading@11: } yading@11: } yading@11: yading@11: static void fill_gv_table(int table[256 + 2*YUVRGB_TABLE_HEADROOM], const int elemsize, const int64_t inc) yading@11: { yading@11: int i; yading@11: int off = -(inc >> 9); yading@11: yading@11: for (i = 0; i < 256 + 2*YUVRGB_TABLE_HEADROOM; i++) { yading@11: int64_t cb = av_clip(i-YUVRGB_TABLE_HEADROOM, 0, 255)*inc; yading@11: table[i] = elemsize * (off + (cb >> 16)); yading@11: } yading@11: } yading@11: yading@11: static uint16_t roundToInt16(int64_t f) yading@11: { yading@11: int r = (f + (1 << 15)) >> 16; yading@11: yading@11: if (r < -0x7FFF) yading@11: return 0x8000; yading@11: else if (r > 0x7FFF) yading@11: return 0x7FFF; yading@11: else yading@11: return r; yading@11: } yading@11: yading@11: av_cold int ff_yuv2rgb_c_init_tables(SwsContext *c, const int inv_table[4], yading@11: int fullRange, int brightness, yading@11: int contrast, int saturation) yading@11: { yading@11: const int isRgb = c->dstFormat == AV_PIX_FMT_RGB32 || yading@11: c->dstFormat == AV_PIX_FMT_RGB32_1 || yading@11: c->dstFormat == AV_PIX_FMT_BGR24 || yading@11: c->dstFormat == AV_PIX_FMT_RGB565BE || yading@11: c->dstFormat == AV_PIX_FMT_RGB565LE || yading@11: c->dstFormat == AV_PIX_FMT_RGB555BE || yading@11: c->dstFormat == AV_PIX_FMT_RGB555LE || yading@11: c->dstFormat == AV_PIX_FMT_RGB444BE || yading@11: c->dstFormat == AV_PIX_FMT_RGB444LE || yading@11: c->dstFormat == AV_PIX_FMT_RGB8 || yading@11: c->dstFormat == AV_PIX_FMT_RGB4 || yading@11: c->dstFormat == AV_PIX_FMT_RGB4_BYTE || yading@11: c->dstFormat == AV_PIX_FMT_MONOBLACK; yading@11: const int isNotNe = c->dstFormat == AV_PIX_FMT_NE(RGB565LE, RGB565BE) || yading@11: c->dstFormat == AV_PIX_FMT_NE(RGB555LE, RGB555BE) || yading@11: c->dstFormat == AV_PIX_FMT_NE(RGB444LE, RGB444BE) || yading@11: c->dstFormat == AV_PIX_FMT_NE(BGR565LE, BGR565BE) || yading@11: c->dstFormat == AV_PIX_FMT_NE(BGR555LE, BGR555BE) || yading@11: c->dstFormat == AV_PIX_FMT_NE(BGR444LE, BGR444BE); yading@11: const int bpp = c->dstFormatBpp; yading@11: uint8_t *y_table; yading@11: uint16_t *y_table16; yading@11: uint32_t *y_table32; yading@11: int i, base, rbase, gbase, bbase, av_uninit(abase), needAlpha; yading@11: const int yoffs = fullRange ? 384 : 326; yading@11: yading@11: int64_t crv = inv_table[0]; yading@11: int64_t cbu = inv_table[1]; yading@11: int64_t cgu = -inv_table[2]; yading@11: int64_t cgv = -inv_table[3]; yading@11: int64_t cy = 1 << 16; yading@11: int64_t oy = 0; yading@11: int64_t yb = 0; yading@11: yading@11: if (!fullRange) { yading@11: cy = (cy * 255) / 219; yading@11: oy = 16 << 16; yading@11: } else { yading@11: crv = (crv * 224) / 255; yading@11: cbu = (cbu * 224) / 255; yading@11: cgu = (cgu * 224) / 255; yading@11: cgv = (cgv * 224) / 255; yading@11: } yading@11: yading@11: cy = (cy * contrast) >> 16; yading@11: crv = (crv * contrast * saturation) >> 32; yading@11: cbu = (cbu * contrast * saturation) >> 32; yading@11: cgu = (cgu * contrast * saturation) >> 32; yading@11: cgv = (cgv * contrast * saturation) >> 32; yading@11: oy -= 256 * brightness; yading@11: yading@11: c->uOffset = 0x0400040004000400LL; yading@11: c->vOffset = 0x0400040004000400LL; yading@11: c->yCoeff = roundToInt16(cy * 8192) * 0x0001000100010001ULL; yading@11: c->vrCoeff = roundToInt16(crv * 8192) * 0x0001000100010001ULL; yading@11: c->ubCoeff = roundToInt16(cbu * 8192) * 0x0001000100010001ULL; yading@11: c->vgCoeff = roundToInt16(cgv * 8192) * 0x0001000100010001ULL; yading@11: c->ugCoeff = roundToInt16(cgu * 8192) * 0x0001000100010001ULL; yading@11: c->yOffset = roundToInt16(oy * 8) * 0x0001000100010001ULL; yading@11: yading@11: c->yuv2rgb_y_coeff = (int16_t)roundToInt16(cy << 13); yading@11: c->yuv2rgb_y_offset = (int16_t)roundToInt16(oy << 9); yading@11: c->yuv2rgb_v2r_coeff = (int16_t)roundToInt16(crv << 13); yading@11: c->yuv2rgb_v2g_coeff = (int16_t)roundToInt16(cgv << 13); yading@11: c->yuv2rgb_u2g_coeff = (int16_t)roundToInt16(cgu << 13); yading@11: c->yuv2rgb_u2b_coeff = (int16_t)roundToInt16(cbu << 13); yading@11: yading@11: //scale coefficients by cy yading@11: crv = ((crv << 16) + 0x8000) / cy; yading@11: cbu = ((cbu << 16) + 0x8000) / cy; yading@11: cgu = ((cgu << 16) + 0x8000) / cy; yading@11: cgv = ((cgv << 16) + 0x8000) / cy; yading@11: yading@11: av_free(c->yuvTable); yading@11: yading@11: switch (bpp) { yading@11: case 1: yading@11: c->yuvTable = av_malloc(1024); yading@11: y_table = c->yuvTable; yading@11: yb = -(384 << 16) - oy; yading@11: for (i = 0; i < 1024 - 110; i++) { yading@11: y_table[i + 110] = av_clip_uint8((yb + 0x8000) >> 16) >> 7; yading@11: yb += cy; yading@11: } yading@11: fill_table(c->table_gU, 1, cgu, y_table + yoffs); yading@11: fill_gv_table(c->table_gV, 1, cgv); yading@11: break; yading@11: case 4: yading@11: case 4 | 128: yading@11: rbase = isRgb ? 3 : 0; yading@11: gbase = 1; yading@11: bbase = isRgb ? 0 : 3; yading@11: c->yuvTable = av_malloc(1024 * 3); yading@11: y_table = c->yuvTable; yading@11: yb = -(384 << 16) - oy; yading@11: for (i = 0; i < 1024 - 110; i++) { yading@11: int yval = av_clip_uint8((yb + 0x8000) >> 16); yading@11: y_table[i + 110] = (yval >> 7) << rbase; yading@11: y_table[i + 37 + 1024] = ((yval + 43) / 85) << gbase; yading@11: y_table[i + 110 + 2048] = (yval >> 7) << bbase; yading@11: yb += cy; yading@11: } yading@11: fill_table(c->table_rV, 1, crv, y_table + yoffs); yading@11: fill_table(c->table_gU, 1, cgu, y_table + yoffs + 1024); yading@11: fill_table(c->table_bU, 1, cbu, y_table + yoffs + 2048); yading@11: fill_gv_table(c->table_gV, 1, cgv); yading@11: break; yading@11: case 8: yading@11: rbase = isRgb ? 5 : 0; yading@11: gbase = isRgb ? 2 : 3; yading@11: bbase = isRgb ? 0 : 6; yading@11: c->yuvTable = av_malloc(1024 * 3); yading@11: y_table = c->yuvTable; yading@11: yb = -(384 << 16) - oy; yading@11: for (i = 0; i < 1024 - 38; i++) { yading@11: int yval = av_clip_uint8((yb + 0x8000) >> 16); yading@11: y_table[i + 16] = ((yval + 18) / 36) << rbase; yading@11: y_table[i + 16 + 1024] = ((yval + 18) / 36) << gbase; yading@11: y_table[i + 37 + 2048] = ((yval + 43) / 85) << bbase; yading@11: yb += cy; yading@11: } yading@11: fill_table(c->table_rV, 1, crv, y_table + yoffs); yading@11: fill_table(c->table_gU, 1, cgu, y_table + yoffs + 1024); yading@11: fill_table(c->table_bU, 1, cbu, y_table + yoffs + 2048); yading@11: fill_gv_table(c->table_gV, 1, cgv); yading@11: break; yading@11: case 12: yading@11: rbase = isRgb ? 8 : 0; yading@11: gbase = 4; yading@11: bbase = isRgb ? 0 : 8; yading@11: c->yuvTable = av_malloc(1024 * 3 * 2); yading@11: y_table16 = c->yuvTable; yading@11: yb = -(384 << 16) - oy; yading@11: for (i = 0; i < 1024; i++) { yading@11: uint8_t yval = av_clip_uint8((yb + 0x8000) >> 16); yading@11: y_table16[i] = (yval >> 4) << rbase; yading@11: y_table16[i + 1024] = (yval >> 4) << gbase; yading@11: y_table16[i + 2048] = (yval >> 4) << bbase; yading@11: yb += cy; yading@11: } yading@11: if (isNotNe) yading@11: for (i = 0; i < 1024 * 3; i++) yading@11: y_table16[i] = av_bswap16(y_table16[i]); yading@11: fill_table(c->table_rV, 2, crv, y_table16 + yoffs); yading@11: fill_table(c->table_gU, 2, cgu, y_table16 + yoffs + 1024); yading@11: fill_table(c->table_bU, 2, cbu, y_table16 + yoffs + 2048); yading@11: fill_gv_table(c->table_gV, 2, cgv); yading@11: break; yading@11: case 15: yading@11: case 16: yading@11: rbase = isRgb ? bpp - 5 : 0; yading@11: gbase = 5; yading@11: bbase = isRgb ? 0 : (bpp - 5); yading@11: c->yuvTable = av_malloc(1024 * 3 * 2); yading@11: y_table16 = c->yuvTable; yading@11: yb = -(384 << 16) - oy; yading@11: for (i = 0; i < 1024; i++) { yading@11: uint8_t yval = av_clip_uint8((yb + 0x8000) >> 16); yading@11: y_table16[i] = (yval >> 3) << rbase; yading@11: y_table16[i + 1024] = (yval >> (18 - bpp)) << gbase; yading@11: y_table16[i + 2048] = (yval >> 3) << bbase; yading@11: yb += cy; yading@11: } yading@11: if (isNotNe) yading@11: for (i = 0; i < 1024 * 3; i++) yading@11: y_table16[i] = av_bswap16(y_table16[i]); yading@11: fill_table(c->table_rV, 2, crv, y_table16 + yoffs); yading@11: fill_table(c->table_gU, 2, cgu, y_table16 + yoffs + 1024); yading@11: fill_table(c->table_bU, 2, cbu, y_table16 + yoffs + 2048); yading@11: fill_gv_table(c->table_gV, 2, cgv); yading@11: break; yading@11: case 24: yading@11: case 48: yading@11: c->yuvTable = av_malloc(1024); yading@11: y_table = c->yuvTable; yading@11: yb = -(384 << 16) - oy; yading@11: for (i = 0; i < 1024; i++) { yading@11: y_table[i] = av_clip_uint8((yb + 0x8000) >> 16); yading@11: yb += cy; yading@11: } yading@11: fill_table(c->table_rV, 1, crv, y_table + yoffs); yading@11: fill_table(c->table_gU, 1, cgu, y_table + yoffs); yading@11: fill_table(c->table_bU, 1, cbu, y_table + yoffs); yading@11: fill_gv_table(c->table_gV, 1, cgv); yading@11: break; yading@11: case 32: yading@11: base = (c->dstFormat == AV_PIX_FMT_RGB32_1 || yading@11: c->dstFormat == AV_PIX_FMT_BGR32_1) ? 8 : 0; yading@11: rbase = base + (isRgb ? 16 : 0); yading@11: gbase = base + 8; yading@11: bbase = base + (isRgb ? 0 : 16); yading@11: needAlpha = CONFIG_SWSCALE_ALPHA && isALPHA(c->srcFormat); yading@11: if (!needAlpha) yading@11: abase = (base + 24) & 31; yading@11: c->yuvTable = av_malloc(1024 * 3 * 4); yading@11: y_table32 = c->yuvTable; yading@11: yb = -(384 << 16) - oy; yading@11: for (i = 0; i < 1024; i++) { yading@11: unsigned yval = av_clip_uint8((yb + 0x8000) >> 16); yading@11: y_table32[i] = (yval << rbase) + yading@11: (needAlpha ? 0 : (255u << abase)); yading@11: y_table32[i + 1024] = yval << gbase; yading@11: y_table32[i + 2048] = yval << bbase; yading@11: yb += cy; yading@11: } yading@11: fill_table(c->table_rV, 4, crv, y_table32 + yoffs); yading@11: fill_table(c->table_gU, 4, cgu, y_table32 + yoffs + 1024); yading@11: fill_table(c->table_bU, 4, cbu, y_table32 + yoffs + 2048); yading@11: fill_gv_table(c->table_gV, 4, cgv); yading@11: break; yading@11: default: yading@11: c->yuvTable = NULL; yading@11: if(!isPlanar(c->dstFormat) || bpp <= 24) yading@11: av_log(c, AV_LOG_ERROR, "%ibpp not supported by yuv2rgb\n", bpp); yading@11: return -1; yading@11: } yading@11: return 0; yading@11: }