1 |
/************************************************************************** |
/***************************************************************************** |
2 |
* |
* |
3 |
* XVID MPEG-4 VIDEO CODEC |
* XVID MPEG-4 VIDEO CODEC |
4 |
* colorspace conversions |
* - colorspace conversion module - |
5 |
* |
* |
6 |
* This program is free software; you can redistribute it and/or modify |
* Copyright(C) 2002 Peter Ross <pross@xvid.org> |
7 |
* it under the terms of the GNU General Public License as published by |
* 2002 Michael Militzer <isibaar@xvid.org> |
8 |
|
* |
9 |
|
* This file is part of XviD, a free MPEG-4 video encoder/decoder |
10 |
|
* |
11 |
|
* XviD is free software; you can redistribute it and/or modify it |
12 |
|
* under the terms of the GNU General Public License as published by |
13 |
* the Free Software Foundation; either version 2 of the License, or |
* the Free Software Foundation; either version 2 of the License, or |
14 |
* (at your option) any later version. |
* (at your option) any later version. |
15 |
* |
* |
20 |
* |
* |
21 |
* You should have received a copy of the GNU General Public License |
* You should have received a copy of the GNU General Public License |
22 |
* along with this program; if not, write to the Free Software |
* along with this program; if not, write to the Free Software |
23 |
* Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. |
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA |
24 |
* |
* |
25 |
*************************************************************************/ |
* Under section 8 of the GNU General Public License, the copyright |
26 |
|
* holders of XVID explicitly forbid distribution in the following |
27 |
/************************************************************************** |
* countries: |
28 |
|
* |
29 |
|
* - Japan |
30 |
|
* - United States of America |
31 |
|
* |
32 |
|
* Linking XviD statically or dynamically with other modules is making a |
33 |
|
* combined work based on XviD. Thus, the terms and conditions of the |
34 |
|
* GNU General Public License cover the whole combination. |
35 |
* |
* |
36 |
* History: |
* As a special exception, the copyright holders of XviD give you |
37 |
|
* permission to link XviD with independent modules that communicate with |
38 |
|
* XviD solely through the VFW1.1 and DShow interfaces, regardless of the |
39 |
|
* license terms of these independent modules, and to copy and distribute |
40 |
|
* the resulting combined work under terms of your choice, provided that |
41 |
|
* every copy of the combined work is accompanied by a complete copy of |
42 |
|
* the source code of XviD (the version of XviD used to produce the |
43 |
|
* combined work), being distributed under the terms of the GNU General |
44 |
|
* Public License plus this exception. An independent module is a module |
45 |
|
* which is not derived from or based on XviD. |
46 |
* |
* |
47 |
* 14.04.2002 added user_to_yuv_c() |
* Note that people who make modified versions of XviD are not obligated |
48 |
* 30.02.2002 out_yuv dst_stride2 fix |
* to grant this special exception for their modified versions; it is |
49 |
* 26.02.2002 rgb555, rgb565 |
* their choice whether to do so. The GNU General Public License gives |
50 |
* 24.11.2001 accuracy improvement to yuyv/vyuy conversion |
* permission to release a modified version without this exception; this |
51 |
* 28.10.2001 total rewrite <pross@cs.rmit.edu.au> |
* exception also makes it possible to release a modified version which |
52 |
|
* carries forward this exception. |
53 |
* |
* |
54 |
**************************************************************************/ |
* $Id$ |
55 |
|
* |
56 |
|
****************************************************************************/ |
57 |
|
|
58 |
#include <string.h> // memcpy |
#include <string.h> // memcpy |
59 |
|
|
|
#include "../global.h" |
|
|
#include "../divx4.h" // DEC_PICTURE |
|
60 |
#include "colorspace.h" |
#include "colorspace.h" |
61 |
|
#include "../divx4.h" // DEC_PICTURE |
62 |
|
|
63 |
// function pointers |
// function pointers |
64 |
|
|
65 |
/* input */ |
/* input */ |
66 |
packedFuncPtr rgb555_to_yv12; |
color_inputFuncPtr rgb555_to_yv12; |
67 |
packedFuncPtr rgb565_to_yv12; |
color_inputFuncPtr rgb565_to_yv12; |
68 |
packedFuncPtr bgr_to_yv12; |
color_inputFuncPtr rgb24_to_yv12; |
69 |
packedFuncPtr bgra_to_yv12; |
color_inputFuncPtr rgb32_to_yv12; |
70 |
packedFuncPtr abgr_to_yv12; |
color_inputFuncPtr yuv_to_yv12; |
71 |
packedFuncPtr rgba_to_yv12; |
color_inputFuncPtr yuyv_to_yv12; |
72 |
packedFuncPtr yuv_to_yv12; |
color_inputFuncPtr uyvy_to_yv12; |
|
packedFuncPtr yuyv_to_yv12; |
|
|
packedFuncPtr uyvy_to_yv12; |
|
|
|
|
|
packedFuncPtr rgb555i_to_yv12; |
|
|
packedFuncPtr rgb565i_to_yv12; |
|
|
packedFuncPtr bgri_to_yv12; |
|
|
packedFuncPtr bgrai_to_yv12; |
|
|
packedFuncPtr abgri_to_yv12; |
|
|
packedFuncPtr rgbai_to_yv12; |
|
|
packedFuncPtr yuyvi_to_yv12; |
|
|
packedFuncPtr uyvyi_to_yv12; |
|
73 |
|
|
74 |
/* output */ |
/* output */ |
75 |
packedFuncPtr yv12_to_rgb555; |
color_outputFuncPtr yv12_to_rgb555; |
76 |
packedFuncPtr yv12_to_rgb565; |
color_outputFuncPtr yv12_to_rgb565; |
77 |
packedFuncPtr yv12_to_bgr; |
color_outputFuncPtr yv12_to_rgb24; |
78 |
packedFuncPtr yv12_to_bgra; |
color_outputFuncPtr yv12_to_rgb32; |
79 |
packedFuncPtr yv12_to_abgr; |
color_outputFuncPtr yv12_to_yuv; |
80 |
packedFuncPtr yv12_to_rgba; |
color_outputFuncPtr yv12_to_yuyv; |
81 |
packedFuncPtr yv12_to_yuv; |
color_outputFuncPtr yv12_to_uyvy; |
|
packedFuncPtr yv12_to_yuyv; |
|
|
packedFuncPtr yv12_to_uyvy; |
|
|
|
|
|
packedFuncPtr yv12_to_rgb555i; |
|
|
packedFuncPtr yv12_to_rgb565i; |
|
|
packedFuncPtr yv12_to_bgri; |
|
|
packedFuncPtr yv12_to_bgrai; |
|
|
packedFuncPtr yv12_to_abgri; |
|
|
packedFuncPtr yv12_to_rgbai; |
|
|
packedFuncPtr yv12_to_yuyvi; |
|
|
packedFuncPtr yv12_to_uyvyi; |
|
|
|
|
|
planarFuncPtr yv12_to_yv12; |
|
|
|
|
|
|
|
|
int32_t RGB_Y_tab[256]; |
|
|
int32_t B_U_tab[256]; |
|
|
int32_t G_U_tab[256]; |
|
|
int32_t G_V_tab[256]; |
|
|
int32_t R_V_tab[256]; |
|
|
|
|
|
|
|
82 |
|
|
83 |
/********** generic colorspace macro **********/ |
#define MIN(A,B) ((A)<(B)?(A):(B)) |
84 |
|
#define MAX(A,B) ((A)>(B)?(A):(B)) |
|
|
|
|
#define MAKE_COLORSPACE(NAME,SIZE,PIXELS,VPIXELS,FUNC,C1,C2,C3,C4) \ |
|
|
void \ |
|
|
NAME(uint8_t * x_ptr, int x_stride, \ |
|
|
uint8_t * y_ptr, uint8_t * u_ptr, uint8_t * v_ptr, \ |
|
|
int y_stride, int uv_stride, \ |
|
|
int width, int height, int vflip) \ |
|
|
{ \ |
|
|
int fixed_width = (width + 1) & ~1; \ |
|
|
int x_dif = x_stride - (SIZE)*fixed_width; \ |
|
|
int y_dif = y_stride - fixed_width; \ |
|
|
int uv_dif = uv_stride - (fixed_width / 2); \ |
|
|
int x, y; \ |
|
|
if (vflip) { \ |
|
|
x_ptr += (height - 1) * x_stride; \ |
|
|
x_dif = -(SIZE)*fixed_width - x_stride; \ |
|
|
x_stride = -x_stride; \ |
|
|
} \ |
|
|
for (y = 0; y < height; y+=(VPIXELS)) { \ |
|
|
FUNC##_ROW(SIZE,C1,C2,C3,C4); \ |
|
|
for (x = 0; x < fixed_width; x+=(PIXELS)) { \ |
|
|
FUNC(SIZE,C1,C2,C3,C4); \ |
|
|
x_ptr += (PIXELS)*(SIZE); \ |
|
|
y_ptr += (PIXELS); \ |
|
|
u_ptr += (PIXELS)/2; \ |
|
|
v_ptr += (PIXELS)/2; \ |
|
|
} \ |
|
|
x_ptr += x_dif + (VPIXELS-1)*x_stride; \ |
|
|
y_ptr += y_dif + (VPIXELS-1)*y_stride; \ |
|
|
u_ptr += uv_dif + ((VPIXELS/2)-1)*uv_stride; \ |
|
|
v_ptr += uv_dif + ((VPIXELS/2)-1)*uv_stride; \ |
|
|
} \ |
|
|
} |
|
|
|
|
|
|
|
|
|
|
|
/********** colorspace input (xxx_to_yv12) functions **********/ |
|
85 |
|
|
86 |
/* rgb -> yuv def's |
/* rgb -> yuv def's |
87 |
|
|
112 |
#define FIX_IN(x) ((uint16_t) ((x) * (1L<<SCALEBITS_IN) + 0.5)) |
#define FIX_IN(x) ((uint16_t) ((x) * (1L<<SCALEBITS_IN) + 0.5)) |
113 |
|
|
114 |
|
|
115 |
/* rgb16/rgb16i input */ |
int32_t RGB_Y_tab[256]; |
116 |
|
int32_t B_U_tab[256]; |
117 |
|
int32_t G_U_tab[256]; |
118 |
|
int32_t G_V_tab[256]; |
119 |
|
int32_t R_V_tab[256]; |
120 |
|
|
121 |
|
|
122 |
|
/* rgb555 -> yuv 4:2:0 planar */ |
123 |
|
void |
124 |
|
rgb555_to_yv12_c(uint8_t * y_out, |
125 |
|
uint8_t * u_out, |
126 |
|
uint8_t * v_out, |
127 |
|
uint8_t * src, |
128 |
|
int width, |
129 |
|
int height, |
130 |
|
int y_stride) |
131 |
|
{ |
132 |
|
int32_t src_stride = width * 2; |
133 |
|
uint32_t y_dif = y_stride - width; |
134 |
|
uint32_t uv_dif = (y_stride - width) / 2; |
135 |
|
uint32_t x, y; |
136 |
|
|
137 |
|
if (height < 0) { |
138 |
|
height = -height; |
139 |
|
src += (height - 1) * src_stride; |
140 |
|
src_stride = -src_stride; |
141 |
|
} |
142 |
|
|
143 |
|
|
144 |
|
for (y = height / 2; y; y--) { |
145 |
|
// process one 2x2 block per iteration |
146 |
|
for (x = 0; x < (uint32_t) width; x += 2) { |
147 |
|
int rgb, r, g, b, r4, g4, b4; |
148 |
|
|
149 |
|
rgb = *(uint16_t *) (src + x * 2); |
150 |
|
b4 = b = (rgb << 3) & 0xf8; |
151 |
|
g4 = g = (rgb >> 2) & 0xf8; |
152 |
|
r4 = r = (rgb >> 7) & 0xf8; |
153 |
|
y_out[0] = |
154 |
|
(uint8_t) ((FIX_IN(Y_R_IN) * r + FIX_IN(Y_G_IN) * g + |
155 |
|
FIX_IN(Y_B_IN) * b) >> SCALEBITS_IN) + Y_ADD_IN; |
156 |
|
|
157 |
|
rgb = *(uint16_t *) (src + x * 2 + src_stride); |
158 |
|
b4 += b = (rgb << 3) & 0xf8; |
159 |
|
g4 += g = (rgb >> 2) & 0xf8; |
160 |
|
r4 += r = (rgb >> 7) & 0xf8; |
161 |
|
y_out[y_stride] = |
162 |
|
(uint8_t) ((FIX_IN(Y_R_IN) * r + FIX_IN(Y_G_IN) * g + |
163 |
|
FIX_IN(Y_B_IN) * b) >> SCALEBITS_IN) + Y_ADD_IN; |
164 |
|
|
165 |
|
rgb = *(uint16_t *) (src + x * 2 + 2); |
166 |
|
b4 += b = (rgb << 3) & 0xf8; |
167 |
|
g4 += g = (rgb >> 2) & 0xf8; |
168 |
|
r4 += r = (rgb >> 7) & 0xf8; |
169 |
|
y_out[1] = |
170 |
|
(uint8_t) ((FIX_IN(Y_R_IN) * r + FIX_IN(Y_G_IN) * g + |
171 |
|
FIX_IN(Y_B_IN) * b) >> SCALEBITS_IN) + Y_ADD_IN; |
172 |
|
|
173 |
|
rgb = *(uint16_t *) (src + x * 2 + src_stride + 2); |
174 |
|
b4 += b = (rgb << 3) & 0xf8; |
175 |
|
g4 += g = (rgb >> 2) & 0xf8; |
176 |
|
r4 += r = (rgb >> 7) & 0xf8; |
177 |
|
y_out[y_stride + 1] = |
178 |
|
(uint8_t) ((FIX_IN(Y_R_IN) * r + FIX_IN(Y_G_IN) * g + |
179 |
|
FIX_IN(Y_B_IN) * b) >> SCALEBITS_IN) + Y_ADD_IN; |
180 |
|
|
181 |
|
*u_out++ = |
182 |
|
(uint8_t) ((-FIX_IN(U_R_IN) * r4 - FIX_IN(U_G_IN) * g4 + |
183 |
|
FIX_IN(U_B_IN) * b4) >> (SCALEBITS_IN + 2)) + |
184 |
|
U_ADD_IN; |
185 |
|
|
186 |
|
|
187 |
|
*v_out++ = |
188 |
|
(uint8_t) ((FIX_IN(V_R_IN) * r4 - FIX_IN(V_G_IN) * g4 - |
189 |
|
FIX_IN(V_B_IN) * b4) >> (SCALEBITS_IN + 2)) + |
190 |
|
V_ADD_IN; |
191 |
|
|
192 |
|
y_out += 2; |
193 |
|
} |
194 |
|
src += src_stride * 2; |
195 |
|
y_out += y_dif + y_stride; |
196 |
|
u_out += uv_dif; |
197 |
|
v_out += uv_dif; |
198 |
|
} |
199 |
|
} |
200 |
|
|
201 |
|
|
202 |
|
|
203 |
|
/* rgb565_to_yuv_c |
204 |
|
NOTE: identical to rgb555 except for shift/mask |
205 |
|
not tested */ |
206 |
|
|
207 |
|
void |
208 |
|
rgb565_to_yv12_c(uint8_t * y_out, |
209 |
|
uint8_t * u_out, |
210 |
|
uint8_t * v_out, |
211 |
|
uint8_t * src, |
212 |
|
int width, |
213 |
|
int height, |
214 |
|
int y_stride) |
215 |
|
{ |
216 |
|
int32_t src_stride = width * 2; |
217 |
|
|
218 |
|
uint32_t y_dif = y_stride - width; |
219 |
|
uint32_t uv_dif = (y_stride - width) / 2; |
220 |
|
uint32_t x, y; |
221 |
|
|
222 |
|
if (height < 0) { |
223 |
|
height = -height; |
224 |
|
src += (height - 1) * src_stride; |
225 |
|
src_stride = -src_stride; |
226 |
|
} |
227 |
|
|
228 |
|
|
229 |
|
for (y = height / 2; y; y--) { |
230 |
|
// process one 2x2 block per iteration |
231 |
|
for (x = 0; x < (uint32_t) width; x += 2) { |
232 |
|
int rgb, r, g, b, r4, g4, b4; |
233 |
|
|
234 |
|
rgb = *(uint16_t *) (src + x * 2); |
235 |
|
b4 = b = (rgb << 3) & 0xf8; |
236 |
|
g4 = g = (rgb >> 3) & 0xfc; |
237 |
|
r4 = r = (rgb >> 8) & 0xf8; |
238 |
|
y_out[0] = |
239 |
|
(uint8_t) ((FIX_IN(Y_R_IN) * r + FIX_IN(Y_G_IN) * g + |
240 |
|
FIX_IN(Y_B_IN) * b) >> SCALEBITS_IN) + Y_ADD_IN; |
241 |
|
|
242 |
|
rgb = *(uint16_t *) (src + x * 2 + src_stride); |
243 |
|
b4 += b = (rgb << 3) & 0xf8; |
244 |
|
g4 += g = (rgb >> 3) & 0xfc; |
245 |
|
r4 += r = (rgb >> 8) & 0xf8; |
246 |
|
y_out[y_stride] = |
247 |
|
(uint8_t) ((FIX_IN(Y_R_IN) * r + FIX_IN(Y_G_IN) * g + |
248 |
|
FIX_IN(Y_B_IN) * b) >> SCALEBITS_IN) + Y_ADD_IN; |
249 |
|
|
250 |
|
rgb = *(uint16_t *) (src + x * 2 + 2); |
251 |
|
b4 += b = (rgb << 3) & 0xf8; |
252 |
|
g4 += g = (rgb >> 3) & 0xfc; |
253 |
|
r4 += r = (rgb >> 8) & 0xf8; |
254 |
|
y_out[1] = |
255 |
|
(uint8_t) ((FIX_IN(Y_R_IN) * r + FIX_IN(Y_G_IN) * g + |
256 |
|
FIX_IN(Y_B_IN) * b) >> SCALEBITS_IN) + Y_ADD_IN; |
257 |
|
|
258 |
|
rgb = *(uint16_t *) (src + x * 2 + src_stride + 2); |
259 |
|
b4 += b = (rgb << 3) & 0xf8; |
260 |
|
g4 += g = (rgb >> 3) & 0xfc; |
261 |
|
r4 += r = (rgb >> 8) & 0xf8; |
262 |
|
y_out[y_stride + 1] = |
263 |
|
(uint8_t) ((FIX_IN(Y_R_IN) * r + FIX_IN(Y_G_IN) * g + |
264 |
|
FIX_IN(Y_B_IN) * b) >> SCALEBITS_IN) + Y_ADD_IN; |
265 |
|
|
266 |
|
*u_out++ = |
267 |
|
(uint8_t) ((-FIX_IN(U_R_IN) * r4 - FIX_IN(U_G_IN) * g4 + |
268 |
|
FIX_IN(U_B_IN) * b4) >> (SCALEBITS_IN + 2)) + |
269 |
|
U_ADD_IN; |
270 |
|
|
|
#define MK_RGB555_B(RGB) ((RGB) << 3) & 0xf8 |
|
|
#define MK_RGB555_G(RGB) ((RGB) >> 2) & 0xf8 |
|
|
#define MK_RGB555_R(RGB) ((RGB) >> 7) & 0xf8 |
|
|
|
|
|
#define MK_RGB565_B(RGB) ((RGB) << 3) & 0xf8 |
|
|
#define MK_RGB565_G(RGB) ((RGB) >> 3) & 0xfc |
|
|
#define MK_RGB565_R(RGB) ((RGB) >> 8) & 0xf8 |
|
|
|
|
|
|
|
|
#define READ_RGB16_Y(ROW, UVID, C1,C2,C3,C4) \ |
|
|
rgb = *(uint16_t *) (x_ptr + ((ROW)*x_stride) + 0); \ |
|
|
b##UVID += b = C1##_B(rgb); \ |
|
|
g##UVID += g = C1##_G(rgb); \ |
|
|
r##UVID += r = C1##_R(rgb); \ |
|
|
y_ptr[(ROW)*y_stride+0] = \ |
|
|
(uint8_t) ((FIX_IN(Y_R_IN) * r + FIX_IN(Y_G_IN) * g + \ |
|
|
FIX_IN(Y_B_IN) * b) >> SCALEBITS_IN) + Y_ADD_IN; \ |
|
|
rgb = *(uint16_t *) (x_ptr + ((ROW)*x_stride) + 2); \ |
|
|
b##UVID += b = C1##_B(rgb); \ |
|
|
g##UVID += g = C1##_G(rgb); \ |
|
|
r##UVID += r = C1##_R(rgb); \ |
|
|
y_ptr[(ROW)*y_stride+1] = \ |
|
|
(uint8_t) ((FIX_IN(Y_R_IN) * r + FIX_IN(Y_G_IN) * g + \ |
|
|
FIX_IN(Y_B_IN) * b) >> SCALEBITS_IN) + Y_ADD_IN; |
|
|
|
|
|
#define READ_RGB16_UV(UV_ROW,UVID) \ |
|
|
u_ptr[(UV_ROW)*uv_stride] = \ |
|
|
(uint8_t) ((-FIX_IN(U_R_IN) * r##UVID - FIX_IN(U_G_IN) * g##UVID + \ |
|
|
FIX_IN(U_B_IN) * b##UVID) >> (SCALEBITS_IN + 2)) + U_ADD_IN; \ |
|
|
v_ptr[(UV_ROW)*uv_stride] = \ |
|
|
(uint8_t) ((FIX_IN(V_R_IN) * r##UVID - FIX_IN(V_G_IN) * g##UVID - \ |
|
|
FIX_IN(V_B_IN) * b##UVID) >> (SCALEBITS_IN + 2)) + V_ADD_IN; |
|
|
|
|
|
#define RGB16_TO_YV12_ROW(SIZE,C1,C2,C3,C4) \ |
|
|
/* nothing */ |
|
|
#define RGB16_TO_YV12(SIZE,C1,C2,C3,C4) \ |
|
|
uint32_t rgb, r, g, b, r0, g0, b0; \ |
|
|
r0 = g0 = b0 = 0; \ |
|
|
READ_RGB16_Y (0, 0, C1,C2,C3,C4) \ |
|
|
READ_RGB16_Y (1, 0, C1,C2,C3,C4) \ |
|
|
READ_RGB16_UV(0, 0) |
|
|
|
|
|
|
|
|
#define RGB16I_TO_YV12_ROW(SIZE,C1,C2,C3,C4) \ |
|
|
/* nothing */ |
|
|
#define RGB16I_TO_YV12(SIZE,C1,C2,C3,C4) \ |
|
|
uint32_t rgb, r, g, b, r0, g0, b0, r1, g1, b1; \ |
|
|
r0 = g0 = b0 = r1 = g1 = b1 = 0; \ |
|
|
READ_RGB16_Y (0, 0, C1,C2,C3,C4) \ |
|
|
READ_RGB16_Y (1, 1, C1,C2,C3,C4) \ |
|
|
READ_RGB16_Y (2, 0, C1,C2,C3,C4) \ |
|
|
READ_RGB16_Y (3, 1, C1,C2,C3,C4) \ |
|
|
READ_RGB16_UV(0, 0) \ |
|
|
READ_RGB16_UV(1, 1) |
|
|
|
|
|
|
|
|
/* rgb/rgbi input */ |
|
|
|
|
|
#define READ_RGB_Y(SIZE, ROW, UVID, C1,C2,C3,C4) \ |
|
|
r##UVID += r = x_ptr[(ROW)*x_stride+(C1)]; \ |
|
|
g##UVID += g = x_ptr[(ROW)*x_stride+(C2)]; \ |
|
|
b##UVID += b = x_ptr[(ROW)*x_stride+(C3)]; \ |
|
|
y_ptr[(ROW)*y_stride+0] = \ |
|
|
(uint8_t) ((FIX_IN(Y_R_IN) * r + FIX_IN(Y_G_IN) * g + \ |
|
|
FIX_IN(Y_B_IN) * b) >> SCALEBITS_IN) + Y_ADD_IN; \ |
|
|
r##UVID += r = x_ptr[(ROW)*x_stride+(SIZE)+(C1)]; \ |
|
|
g##UVID += g = x_ptr[(ROW)*x_stride+(SIZE)+(C2)]; \ |
|
|
b##UVID += b = x_ptr[(ROW)*x_stride+(SIZE)+(C3)]; \ |
|
|
y_ptr[(ROW)*y_stride+1] = \ |
|
|
(uint8_t) ((FIX_IN(Y_R_IN) * r + FIX_IN(Y_G_IN) * g + \ |
|
|
FIX_IN(Y_B_IN) * b) >> SCALEBITS_IN) + Y_ADD_IN; |
|
|
|
|
|
#define READ_RGB_UV(UV_ROW,UVID) \ |
|
|
u_ptr[(UV_ROW)*uv_stride] = \ |
|
|
(uint8_t) ((-FIX_IN(U_R_IN) * r##UVID - FIX_IN(U_G_IN) * g##UVID + \ |
|
|
FIX_IN(U_B_IN) * b##UVID) >> (SCALEBITS_IN + 2)) + U_ADD_IN; \ |
|
|
v_ptr[(UV_ROW)*uv_stride] = \ |
|
|
(uint8_t) ((FIX_IN(V_R_IN) * r##UVID - FIX_IN(V_G_IN) * g##UVID - \ |
|
|
FIX_IN(V_B_IN) * b##UVID) >> (SCALEBITS_IN + 2)) + V_ADD_IN; |
|
|
|
|
|
|
|
|
#define RGB_TO_YV12_ROW(SIZE,C1,C2,C3,C4) \ |
|
|
/* nothing */ |
|
|
#define RGB_TO_YV12(SIZE,C1,C2,C3,C4) \ |
|
|
uint32_t r, g, b, r0, g0, b0; \ |
|
|
r0 = g0 = b0 = 0; \ |
|
|
READ_RGB_Y(SIZE, 0, 0, C1,C2,C3,C4) \ |
|
|
READ_RGB_Y(SIZE, 1, 0, C1,C2,C3,C4) \ |
|
|
READ_RGB_UV( 0, 0) |
|
|
|
|
|
#define RGBI_TO_YV12_ROW(SIZE,C1,C2,C3,C4) \ |
|
|
/* nothing */ |
|
|
#define RGBI_TO_YV12(SIZE,C1,C2,C3,C4) \ |
|
|
uint32_t r, g, b, r0, g0, b0, r1, g1, b1; \ |
|
|
r0 = g0 = b0 = r1 = g1 = b1 = 0; \ |
|
|
READ_RGB_Y(SIZE, 0, 0, C1,C2,C3,C4) \ |
|
|
READ_RGB_Y(SIZE, 1, 1, C1,C2,C3,C4) \ |
|
|
READ_RGB_Y(SIZE, 2, 0, C1,C2,C3,C4) \ |
|
|
READ_RGB_Y(SIZE, 3, 1, C1,C2,C3,C4) \ |
|
|
READ_RGB_UV( 0, 0) \ |
|
|
READ_RGB_UV( 1, 1) |
|
|
|
|
|
|
|
|
/* yuyv/yuyvi input */ |
|
|
|
|
|
#define READ_YUYV_Y(ROW,C1,C2,C3,C4) \ |
|
|
y_ptr[(ROW)*y_stride+0] = x_ptr[(ROW)*x_stride+(C1)]; \ |
|
|
y_ptr[(ROW)*y_stride+1] = x_ptr[(ROW)*x_stride+(C3)]; |
|
|
#define READ_YUYV_UV(UV_ROW,ROW1,ROW2,C1,C2,C3,C4) \ |
|
|
u_ptr[(UV_ROW)*uv_stride] = (x_ptr[(ROW1)*x_stride+(C2)] + x_ptr[(ROW2)*x_stride+(C2)] + 1) / 2; \ |
|
|
v_ptr[(UV_ROW)*uv_stride] = (x_ptr[(ROW1)*x_stride+(C4)] + x_ptr[(ROW2)*x_stride+(C4)] + 1) / 2; |
|
|
|
|
|
#define YUYV_TO_YV12_ROW(SIZE,C1,C2,C3,C4) \ |
|
|
/* nothing */ |
|
|
#define YUYV_TO_YV12(SIZE,C1,C2,C3,C4) \ |
|
|
READ_YUYV_Y (0, C1,C2,C3,C4) \ |
|
|
READ_YUYV_Y (1, C1,C2,C3,C4) \ |
|
|
READ_YUYV_UV(0, 0,1, C1,C2,C3,C4) |
|
|
|
|
|
#define YUYVI_TO_YV12_ROW(SIZE,C1,C2,C3,C4) \ |
|
|
/* nothing */ |
|
|
#define YUYVI_TO_YV12(SIZE,C1,C2,C3,C4) \ |
|
|
READ_YUYV_Y (0, C1,C2,C3,C4) \ |
|
|
READ_YUYV_Y (1, C1,C2,C3,C4) \ |
|
|
READ_YUYV_Y (2, C1,C2,C3,C4) \ |
|
|
READ_YUYV_Y (3, C1,C2,C3,C4) \ |
|
|
READ_YUYV_UV(0, 0,2, C1,C2,C3,C4) \ |
|
|
READ_YUYV_UV(1, 1,3, C1,C2,C3,C4) |
|
|
|
|
|
|
|
|
MAKE_COLORSPACE(rgb555_to_yv12_c, 2,2,2, RGB16_TO_YV12, MK_RGB555, 0,0,0) |
|
|
MAKE_COLORSPACE(rgb565_to_yv12_c, 2,2,2, RGB16_TO_YV12, MK_RGB565, 0,0,0) |
|
|
MAKE_COLORSPACE(bgr_to_yv12_c, 3,2,2, RGB_TO_YV12, 2,1,0, 0) |
|
|
MAKE_COLORSPACE(bgra_to_yv12_c, 4,2,2, RGB_TO_YV12, 2,1,0, 0) |
|
|
MAKE_COLORSPACE(abgr_to_yv12_c, 4,2,2, RGB_TO_YV12, 3,2,1, 0) |
|
|
MAKE_COLORSPACE(rgba_to_yv12_c, 4,2,2, RGB_TO_YV12, 0,1,2, 0) |
|
|
MAKE_COLORSPACE(yuyv_to_yv12_c, 2,2,2, YUYV_TO_YV12, 0,1,2,3) |
|
|
MAKE_COLORSPACE(uyvy_to_yv12_c, 2,2,2, YUYV_TO_YV12, 1,0,3,2) |
|
|
|
|
|
MAKE_COLORSPACE(rgb555i_to_yv12_c, 2,2,4, RGB16I_TO_YV12, MK_RGB555, 0,0,0) |
|
|
MAKE_COLORSPACE(rgb565i_to_yv12_c, 2,2,4, RGB16I_TO_YV12, MK_RGB565, 0,0,0) |
|
|
MAKE_COLORSPACE(bgri_to_yv12_c, 3,2,4, RGBI_TO_YV12, 2,1,0, 0) |
|
|
MAKE_COLORSPACE(bgrai_to_yv12_c, 4,2,4, RGBI_TO_YV12, 2,1,0, 0) |
|
|
MAKE_COLORSPACE(abgri_to_yv12_c, 4,2,4, RGBI_TO_YV12, 3,2,1, 0) |
|
|
MAKE_COLORSPACE(rgbai_to_yv12_c, 4,2,4, RGBI_TO_YV12, 0,1,2, 0) |
|
|
MAKE_COLORSPACE(yuyvi_to_yv12_c, 2,2,4, YUYVI_TO_YV12, 0,1,2,3) |
|
|
MAKE_COLORSPACE(uyvyi_to_yv12_c, 2,2,4, YUYVI_TO_YV12, 1,0,3,2) |
|
271 |
|
|
272 |
|
*v_out++ = |
273 |
|
(uint8_t) ((FIX_IN(V_R_IN) * r4 - FIX_IN(V_G_IN) * g4 - |
274 |
|
FIX_IN(V_B_IN) * b4) >> (SCALEBITS_IN + 2)) + |
275 |
|
V_ADD_IN; |
276 |
|
|
277 |
|
y_out += 2; |
278 |
|
} |
279 |
|
src += src_stride * 2; |
280 |
|
y_out += y_dif + y_stride; |
281 |
|
u_out += uv_dif; |
282 |
|
v_out += uv_dif; |
283 |
|
} |
284 |
|
} |
285 |
|
|
286 |
|
|
287 |
|
|
288 |
|
|
289 |
|
/* rgb24 -> yuv 4:2:0 planar |
290 |
|
|
291 |
|
NOTE: always flips. |
292 |
|
*/ |
293 |
|
|
294 |
|
void |
295 |
|
rgb24_to_yv12_c(uint8_t * y_out, |
296 |
|
uint8_t * u_out, |
297 |
|
uint8_t * v_out, |
298 |
|
uint8_t * src, |
299 |
|
int width, |
300 |
|
int height, |
301 |
|
int stride) |
302 |
|
{ |
303 |
|
uint32_t width3 = (width << 1) + width; /* width * 3 */ |
304 |
|
uint32_t src_dif = (width << 3) + width; /* width3 * 3 */ |
305 |
|
uint32_t y_dif = (stride << 1) - width; |
306 |
|
uint32_t uv_dif = (stride - width) >> 1; |
307 |
|
uint32_t x, y; |
308 |
|
|
309 |
|
src += (height - 2) * width3; |
310 |
|
|
311 |
|
|
312 |
|
for (y = height >> 1; y; y--) { |
313 |
|
for (x = width >> 1; x; x--) { |
314 |
|
uint32_t r, g, b, r4, g4, b4; |
315 |
|
|
316 |
|
b4 = b = src[0]; |
317 |
|
g4 = g = src[1]; |
318 |
|
r4 = r = src[2]; |
319 |
|
y_out[stride + 0] = |
320 |
|
(uint8_t) ((FIX_IN(Y_R_IN) * r + FIX_IN(Y_G_IN) * g + |
321 |
|
FIX_IN(Y_B_IN) * b) >> SCALEBITS_IN) + Y_ADD_IN; |
322 |
|
|
323 |
|
b4 += (b = src[3]); |
324 |
|
g4 += (g = src[4]); |
325 |
|
r4 += (r = src[5]); |
326 |
|
y_out[stride + 1] = |
327 |
|
(uint8_t) ((FIX_IN(Y_R_IN) * r + FIX_IN(Y_G_IN) * g + |
328 |
|
FIX_IN(Y_B_IN) * b) >> SCALEBITS_IN) + Y_ADD_IN; |
329 |
|
|
330 |
|
b4 += (b = src[width3 + 0]); |
331 |
|
g4 += (g = src[width3 + 1]); |
332 |
|
r4 += (r = src[width3 + 2]); |
333 |
|
y_out[0] = |
334 |
|
(uint8_t) ((FIX_IN(Y_R_IN) * r + FIX_IN(Y_G_IN) * g + |
335 |
|
FIX_IN(Y_B_IN) * b) >> SCALEBITS_IN) + Y_ADD_IN; |
336 |
|
|
337 |
|
b4 += (b = src[width3 + 3]); |
338 |
|
g4 += (g = src[width3 + 4]); |
339 |
|
r4 += (r = src[width3 + 5]); |
340 |
|
y_out[1] = |
341 |
|
(uint8_t) ((FIX_IN(Y_R_IN) * r + FIX_IN(Y_G_IN) * g + |
342 |
|
FIX_IN(Y_B_IN) * b) >> SCALEBITS_IN) + Y_ADD_IN; |
343 |
|
|
344 |
|
*u_out++ = |
345 |
|
(uint8_t) ((-FIX_IN(U_R_IN) * r4 - FIX_IN(U_G_IN) * g4 + |
346 |
|
FIX_IN(U_B_IN) * b4) >> (SCALEBITS_IN + 2)) + |
347 |
|
U_ADD_IN; |
348 |
|
|
349 |
|
|
350 |
|
*v_out++ = |
351 |
|
(uint8_t) ((FIX_IN(V_R_IN) * r4 - FIX_IN(V_G_IN) * g4 - |
352 |
|
FIX_IN(V_B_IN) * b4) >> (SCALEBITS_IN + 2)) + |
353 |
|
V_ADD_IN; |
354 |
|
|
355 |
|
|
356 |
|
src += 6; |
357 |
|
y_out += 2; |
358 |
|
} |
359 |
|
src -= src_dif; |
360 |
|
y_out += y_dif; |
361 |
|
u_out += uv_dif; |
362 |
|
v_out += uv_dif; |
363 |
|
} |
364 |
|
} |
365 |
|
|
366 |
|
|
367 |
|
/* rgb32 -> yuv 4:2:0 planar |
368 |
|
|
369 |
|
NOTE: always flips |
370 |
|
*/ |
371 |
|
|
372 |
|
void |
373 |
|
rgb32_to_yv12_c(uint8_t * y_out, |
374 |
|
uint8_t * u_out, |
375 |
|
uint8_t * v_out, |
376 |
|
uint8_t * src, |
377 |
|
int width, |
378 |
|
int height, |
379 |
|
int stride) |
380 |
|
{ |
381 |
|
uint32_t width4 = (width << 2); /* width * 4 */ |
382 |
|
uint32_t src_dif = 3 * width4; |
383 |
|
uint32_t y_dif = (stride << 1) - width; |
384 |
|
uint32_t uv_dif = (stride - width) >> 1; |
385 |
|
uint32_t x, y; |
386 |
|
|
387 |
|
src += (height - 2) * width4; |
388 |
|
|
389 |
|
for (y = height >> 1; y; y--) { |
390 |
|
for (x = width >> 1; x; x--) { |
391 |
|
uint32_t r, g, b, r4, g4, b4; |
392 |
|
|
393 |
|
b4 = b = src[0]; |
394 |
|
g4 = g = src[1]; |
395 |
|
r4 = r = src[2]; |
396 |
|
y_out[stride + 0] = |
397 |
|
(uint8_t) ((FIX_IN(Y_R_IN) * r + FIX_IN(Y_G_IN) * g + |
398 |
|
FIX_IN(Y_B_IN) * b) >> SCALEBITS_IN) + Y_ADD_IN; |
399 |
|
|
400 |
|
b4 += (b = src[4]); |
401 |
|
g4 += (g = src[5]); |
402 |
|
r4 += (r = src[6]); |
403 |
|
y_out[stride + 1] = |
404 |
|
(uint8_t) ((FIX_IN(Y_R_IN) * r + FIX_IN(Y_G_IN) * g + |
405 |
|
FIX_IN(Y_B_IN) * b) >> SCALEBITS_IN) + Y_ADD_IN; |
406 |
|
|
407 |
|
b4 += (b = src[width4 + 0]); |
408 |
|
g4 += (g = src[width4 + 1]); |
409 |
|
r4 += (r = src[width4 + 2]); |
410 |
|
|
411 |
|
y_out[0] = |
412 |
|
(uint8_t) ((FIX_IN(Y_R_IN) * r + FIX_IN(Y_G_IN) * g + |
413 |
|
FIX_IN(Y_B_IN) * b) >> SCALEBITS_IN) + Y_ADD_IN; |
414 |
|
|
415 |
|
b4 += (b = src[width4 + 4]); |
416 |
|
g4 += (g = src[width4 + 5]); |
417 |
|
r4 += (r = src[width4 + 6]); |
418 |
|
y_out[1] = |
419 |
|
(uint8_t) ((FIX_IN(Y_R_IN) * r + FIX_IN(Y_G_IN) * g + |
420 |
|
FIX_IN(Y_B_IN) * b) >> SCALEBITS_IN) + Y_ADD_IN; |
421 |
|
|
422 |
|
*u_out++ = |
423 |
|
(uint8_t) ((-FIX_IN(U_R_IN) * r4 - FIX_IN(U_G_IN) * g4 + |
424 |
|
FIX_IN(U_B_IN) * b4) >> (SCALEBITS_IN + 2)) + |
425 |
|
U_ADD_IN; |
426 |
|
|
427 |
|
*v_out++ = |
428 |
|
(uint8_t) ((FIX_IN(V_R_IN) * r4 - FIX_IN(V_G_IN) * g4 - |
429 |
|
FIX_IN(V_B_IN) * b4) >> (SCALEBITS_IN + 2)) + |
430 |
|
V_ADD_IN; |
431 |
|
|
432 |
|
src += 8; |
433 |
|
y_out += 2; |
434 |
|
} |
435 |
|
src -= src_dif; |
436 |
|
y_out += y_dif; |
437 |
|
u_out += uv_dif; |
438 |
|
v_out += uv_dif; |
439 |
|
} |
440 |
|
} |
441 |
|
|
442 |
|
/* yuv planar -> yuv 4:2:0 planar |
443 |
|
|
444 |
|
NOTE: does not flip */ |
445 |
|
|
446 |
|
void |
447 |
|
yuv_to_yv12_c(uint8_t * y_out, |
448 |
|
uint8_t * u_out, |
449 |
|
uint8_t * v_out, |
450 |
|
uint8_t * src, |
451 |
|
int width, |
452 |
|
int height, |
453 |
|
int stride) |
454 |
|
{ |
455 |
|
uint32_t stride2 = stride >> 1; |
456 |
|
uint32_t width2 = width >> 1; |
457 |
|
uint32_t y; |
458 |
|
|
459 |
|
for (y = height; y; y--) { |
460 |
|
memcpy(y_out, src, width); |
461 |
|
src += width; |
462 |
|
y_out += stride; |
463 |
|
} |
464 |
|
|
465 |
|
for (y = height >> 1; y; y--) { |
466 |
|
memcpy(u_out, src, width2); |
467 |
|
src += width2; |
468 |
|
u_out += stride2; |
469 |
|
} |
470 |
|
|
471 |
|
for (y = height >> 1; y; y--) { |
472 |
|
memcpy(v_out, src, width2); |
473 |
|
src += width2; |
474 |
|
v_out += stride2; |
475 |
|
} |
476 |
|
} |
477 |
|
|
478 |
|
|
479 |
|
|
480 |
|
/* yuyv (yuv2) packed -> yuv 4:2:0 planar |
481 |
|
|
482 |
|
NOTE: does not flip */ |
483 |
|
|
484 |
|
void |
485 |
|
yuyv_to_yv12_c(uint8_t * y_out, |
486 |
|
uint8_t * u_out, |
487 |
|
uint8_t * v_out, |
488 |
|
uint8_t * src, |
489 |
|
int width, |
490 |
|
int height, |
491 |
|
int stride) |
492 |
|
{ |
493 |
|
uint32_t width2 = width + width; |
494 |
|
uint32_t y_dif = stride - width; |
495 |
|
uint32_t uv_dif = y_dif >> 1; |
496 |
|
uint32_t x, y; |
497 |
|
|
498 |
|
for (y = height >> 1; y; y--) { |
499 |
|
|
500 |
|
for (x = width >> 1; x; x--) { |
501 |
|
*y_out++ = *src++; |
502 |
|
//*u_out++ = *src++; |
503 |
|
*u_out++ = (*(src + width2) + *src) >> 1; |
504 |
|
src++; |
505 |
|
*y_out++ = *src++; |
506 |
|
//*v_out++ = *src++; |
507 |
|
*v_out++ = (*(src + width2) + *src) >> 1; |
508 |
|
src++; |
509 |
|
|
510 |
|
} |
511 |
|
|
512 |
|
y_out += y_dif; |
513 |
|
u_out += uv_dif; |
514 |
|
v_out += uv_dif; |
515 |
|
|
516 |
|
for (x = width >> 1; x; x--) { |
517 |
|
*y_out++ = *src++; |
518 |
|
src++; |
519 |
|
*y_out++ = *src++; |
520 |
|
src++; |
521 |
|
} |
522 |
|
|
523 |
|
y_out += y_dif; |
524 |
|
|
525 |
|
} |
526 |
|
|
527 |
|
} |
528 |
|
|
529 |
|
|
530 |
|
|
531 |
|
/* uyvy packed -> yuv 4:2:0 planar |
532 |
|
|
533 |
|
NOTE: does not flip */ |
534 |
|
|
535 |
|
|
536 |
|
void |
537 |
|
uyvy_to_yv12_c(uint8_t * y_out, |
538 |
|
uint8_t * u_out, |
539 |
|
uint8_t * v_out, |
540 |
|
uint8_t * src, |
541 |
|
int width, |
542 |
|
int height, |
543 |
|
int stride) |
544 |
|
{ |
545 |
|
uint32_t width2 = width + width; |
546 |
|
uint32_t y_dif = stride - width; |
547 |
|
uint32_t uv_dif = y_dif >> 1; |
548 |
|
uint32_t x, y; |
549 |
|
|
550 |
|
for (y = height >> 1; y; y--) { |
551 |
|
|
552 |
|
for (x = width >> 1; x; x--) { |
553 |
|
*u_out++ = *src++; |
554 |
|
// *u_out++ = (*(src+width2) + *src++) >> 1; |
555 |
|
*y_out++ = *src++; |
556 |
|
//*v_out++ = *src++; |
557 |
|
*v_out++ = (*(src + width2) + *src) >> 1; |
558 |
|
src++; |
559 |
|
*y_out++ = *src++; |
560 |
|
} |
561 |
|
|
562 |
|
y_out += y_dif; |
563 |
|
u_out += uv_dif;; |
564 |
|
v_out += uv_dif;; |
565 |
|
|
566 |
|
for (x = width >> 1; x; x--) { |
567 |
|
src++; |
568 |
|
*y_out++ = *src++; |
569 |
|
src++; |
570 |
|
*y_out++ = *src++; |
571 |
|
} |
572 |
|
|
573 |
|
y_out += y_dif; |
574 |
|
} |
575 |
|
} |
576 |
|
|
|
/********** colorspace output (yv12_to_xxx) functions **********/ |
|
577 |
|
|
578 |
/* yuv -> rgb def's */ |
/* yuv -> rgb def's */ |
579 |
|
|
588 |
#define R_V_OUT 1.596 |
#define R_V_OUT 1.596 |
589 |
#define V_ADD_OUT 128 |
#define V_ADD_OUT 128 |
590 |
|
|
591 |
|
|
592 |
#define SCALEBITS_OUT 13 |
#define SCALEBITS_OUT 13 |
593 |
#define FIX_OUT(x) ((uint16_t) ((x) * (1L<<SCALEBITS_OUT) + 0.5)) |
#define FIX_OUT(x) ((uint16_t) ((x) * (1L<<SCALEBITS_OUT) + 0.5)) |
594 |
|
|
595 |
|
|
596 |
/* rgb16/rgb16i output */ |
/* initialize rgb lookup tables */ |
597 |
|
|
598 |
|
void |
599 |
|
colorspace_init(void) |
600 |
|
{ |
601 |
|
int32_t i; |
602 |
|
|
603 |
|
for (i = 0; i < 256; i++) { |
604 |
|
RGB_Y_tab[i] = FIX_OUT(RGB_Y_OUT) * (i - Y_ADD_OUT); |
605 |
|
B_U_tab[i] = FIX_OUT(B_U_OUT) * (i - U_ADD_OUT); |
606 |
|
G_U_tab[i] = FIX_OUT(G_U_OUT) * (i - U_ADD_OUT); |
607 |
|
G_V_tab[i] = FIX_OUT(G_V_OUT) * (i - V_ADD_OUT); |
608 |
|
R_V_tab[i] = FIX_OUT(R_V_OUT) * (i - V_ADD_OUT); |
609 |
|
} |
610 |
|
} |
611 |
|
|
612 |
|
/* yuv 4:2:0 planar -> rgb555 + very simple error diffusion |
613 |
|
*/ |
614 |
|
|
615 |
#define MK_RGB555(R,G,B) \ |
#define MK_RGB555(R,G,B) ((MAX(0,MIN(255, R)) << 7) & 0x7c00) | \ |
|
((MAX(0,MIN(255, R)) << 7) & 0x7c00) | \ |
|
616 |
((MAX(0,MIN(255, G)) << 2) & 0x03e0) | \ |
((MAX(0,MIN(255, G)) << 2) & 0x03e0) | \ |
617 |
((MAX(0,MIN(255, B)) >> 3) & 0x001f) |
((MAX(0,MIN(255, B)) >> 3) & 0x001f) |
618 |
|
|
619 |
#define MK_RGB565(R,G,B) \ |
|
620 |
((MAX(0,MIN(255, R)) << 8) & 0xf800) | \ |
void |
621 |
|
yv12_to_rgb555_c(uint8_t * dst, |
622 |
|
int dst_stride, |
623 |
|
uint8_t * y_src, |
624 |
|
uint8_t * u_src, |
625 |
|
uint8_t * v_src, |
626 |
|
int y_stride, |
627 |
|
int uv_stride, |
628 |
|
int width, |
629 |
|
int height) |
630 |
|
{ |
631 |
|
const uint32_t dst_dif = 4 * dst_stride - 2 * width; |
632 |
|
int32_t y_dif = 2 * y_stride - width; |
633 |
|
|
634 |
|
uint8_t *dst2 = dst + 2 * dst_stride; |
635 |
|
uint8_t *y_src2 = y_src + y_stride; |
636 |
|
uint32_t x, y; |
637 |
|
|
638 |
|
if (height < 0) { |
639 |
|
height = -height; |
640 |
|
y_src += (height - 1) * y_stride; |
641 |
|
y_src2 = y_src - y_stride; |
642 |
|
u_src += (height / 2 - 1) * uv_stride; |
643 |
|
v_src += (height / 2 - 1) * uv_stride; |
644 |
|
y_dif = -width - 2 * y_stride; |
645 |
|
uv_stride = -uv_stride; |
646 |
|
} |
647 |
|
|
648 |
|
for (y = height / 2; y; y--) { |
649 |
|
int r, g, b; |
650 |
|
int r2, g2, b2; |
651 |
|
|
652 |
|
r = g = b = 0; |
653 |
|
r2 = g2 = b2 = 0; |
654 |
|
|
655 |
|
// process one 2x2 block per iteration |
656 |
|
for (x = 0; x < (uint32_t) width / 2; x++) { |
657 |
|
int u, v; |
658 |
|
int b_u, g_uv, r_v, rgb_y; |
659 |
|
|
660 |
|
u = u_src[x]; |
661 |
|
v = v_src[x]; |
662 |
|
|
663 |
|
b_u = B_U_tab[u]; |
664 |
|
g_uv = G_U_tab[u] + G_V_tab[v]; |
665 |
|
r_v = R_V_tab[v]; |
666 |
|
|
667 |
|
rgb_y = RGB_Y_tab[*y_src]; |
668 |
|
b = (b & 0x7) + ((rgb_y + b_u) >> SCALEBITS_OUT); |
669 |
|
g = (g & 0x7) + ((rgb_y - g_uv) >> SCALEBITS_OUT); |
670 |
|
r = (r & 0x7) + ((rgb_y + r_v) >> SCALEBITS_OUT); |
671 |
|
*(uint16_t *) dst = MK_RGB555(r, g, b); |
672 |
|
|
673 |
|
y_src++; |
674 |
|
rgb_y = RGB_Y_tab[*y_src]; |
675 |
|
b = (b & 0x7) + ((rgb_y + b_u) >> SCALEBITS_OUT); |
676 |
|
g = (g & 0x7) + ((rgb_y - g_uv) >> SCALEBITS_OUT); |
677 |
|
r = (r & 0x7) + ((rgb_y + r_v) >> SCALEBITS_OUT); |
678 |
|
*(uint16_t *) (dst + 2) = MK_RGB555(r, g, b); |
679 |
|
y_src++; |
680 |
|
|
681 |
|
rgb_y = RGB_Y_tab[*y_src2]; |
682 |
|
b2 = (b2 & 0x7) + ((rgb_y + b_u) >> SCALEBITS_OUT); |
683 |
|
g2 = (g2 & 0x7) + ((rgb_y - g_uv) >> SCALEBITS_OUT); |
684 |
|
r2 = (r2 & 0x7) + ((rgb_y + r_v) >> SCALEBITS_OUT); |
685 |
|
*(uint16_t *) (dst2) = MK_RGB555(r2, g2, b2); |
686 |
|
y_src2++; |
687 |
|
|
688 |
|
rgb_y = RGB_Y_tab[*y_src2]; |
689 |
|
b2 = (b2 & 0x7) + ((rgb_y + b_u) >> SCALEBITS_OUT); |
690 |
|
g2 = (g2 & 0x7) + ((rgb_y - g_uv) >> SCALEBITS_OUT); |
691 |
|
r2 = (r2 & 0x7) + ((rgb_y + r_v) >> SCALEBITS_OUT); |
692 |
|
*(uint16_t *) (dst2 + 2) = MK_RGB555(r2, g2, b2); |
693 |
|
y_src2++; |
694 |
|
|
695 |
|
dst += 4; |
696 |
|
dst2 += 4; |
697 |
|
} |
698 |
|
|
699 |
|
dst += dst_dif; |
700 |
|
dst2 += dst_dif; |
701 |
|
|
702 |
|
y_src += y_dif; |
703 |
|
y_src2 += y_dif; |
704 |
|
|
705 |
|
u_src += uv_stride; |
706 |
|
v_src += uv_stride; |
707 |
|
} |
708 |
|
} |
709 |
|
|
710 |
|
|
711 |
|
/* yuv 4:2:0 planar -> rgb565 + very simple error diffusion |
712 |
|
NOTE: identical to rgb555 except for shift/mask */ |
713 |
|
|
714 |
|
|
715 |
|
#define MK_RGB565(R,G,B) ((MAX(0,MIN(255, R)) << 8) & 0xf800) | \ |
716 |
((MAX(0,MIN(255, G)) << 3) & 0x07e0) | \ |
((MAX(0,MIN(255, G)) << 3) & 0x07e0) | \ |
717 |
((MAX(0,MIN(255, B)) >> 3) & 0x001f) |
((MAX(0,MIN(255, B)) >> 3) & 0x001f) |
718 |
|
|
719 |
#define WRITE_RGB16(ROW,UV_ROW,C1) \ |
void |
720 |
rgb_y = RGB_Y_tab[ y_ptr[y_stride + 0] ]; \ |
yv12_to_rgb565_c(uint8_t * dst, |
721 |
b[ROW] = (b[ROW] & 0x7) + ((rgb_y + b_u##UV_ROW) >> SCALEBITS_OUT); \ |
int dst_stride, |
722 |
g[ROW] = (g[ROW] & 0x7) + ((rgb_y - g_uv##UV_ROW) >> SCALEBITS_OUT); \ |
uint8_t * y_src, |
723 |
r[ROW] = (r[ROW] & 0x7) + ((rgb_y + r_v##UV_ROW) >> SCALEBITS_OUT); \ |
uint8_t * u_src, |
724 |
*(uint16_t *) (x_ptr+((ROW)*x_stride)+0) = C1(r[ROW], g[ROW], b[ROW]); \ |
uint8_t * v_src, |
725 |
rgb_y = RGB_Y_tab[ y_ptr[y_stride + 1] ]; \ |
int y_stride, |
726 |
b[ROW] = (b[ROW] & 0x7) + ((rgb_y + b_u##UV_ROW) >> SCALEBITS_OUT); \ |
int uv_stride, |
727 |
g[ROW] = (g[ROW] & 0x7) + ((rgb_y - g_uv##UV_ROW) >> SCALEBITS_OUT); \ |
int width, |
728 |
r[ROW] = (r[ROW] & 0x7) + ((rgb_y + r_v##UV_ROW) >> SCALEBITS_OUT); \ |
int height) |
729 |
*(uint16_t *) (x_ptr+((ROW)*x_stride)+2) = C1(r[ROW], g[ROW], b[ROW]); |
{ |
730 |
|
const uint32_t dst_dif = 4 * dst_stride - 2 * width; |
731 |
#define YV12_TO_RGB16_ROW(SIZE,C1,C2,C3,C4) \ |
int32_t y_dif = 2 * y_stride - width; |
732 |
int r[2], g[2], b[2]; \ |
|
733 |
r[0] = r[1] = g[0] = g[1] = b[0] = b[1] = 0; |
uint8_t *dst2 = dst + 2 * dst_stride; |
734 |
#define YV12_TO_RGB16(SIZE,C1,C2,C3,C4) \ |
uint8_t *y_src2 = y_src + y_stride; |
735 |
int rgb_y; \ |
uint32_t x, y; |
736 |
int b_u0 = B_U_tab[ u_ptr[0] ]; \ |
|
737 |
int g_uv0 = G_U_tab[ u_ptr[0] ] + G_V_tab[ v_ptr[0] ]; \ |
if (height < 0) { // flip image? |
738 |
int r_v0 = R_V_tab[ v_ptr[0] ]; \ |
height = -height; |
739 |
WRITE_RGB16(0, 0, C1) \ |
y_src += (height - 1) * y_stride; |
740 |
WRITE_RGB16(1, 0, C1) |
y_src2 = y_src - y_stride; |
741 |
|
u_src += (height / 2 - 1) * uv_stride; |
742 |
#define YV12_TO_RGB16I_ROW(SIZE,C1,C2,C3,C4) \ |
v_src += (height / 2 - 1) * uv_stride; |
743 |
int r[4], g[4], b[4]; \ |
y_dif = -width - 2 * y_stride; |
744 |
r[0] = r[1] = r[2] = r[3] = 0; \ |
uv_stride = -uv_stride; |
745 |
g[0] = g[1] = g[2] = g[3] = 0; \ |
} |
746 |
b[0] = b[1] = b[2] = b[3] = 0; |
|
747 |
#define YV12_TO_RGB16I(SIZE,C1,C2,C3,C4) \ |
for (y = height / 2; y; y--) { |
748 |
int rgb_y; \ |
int r, g, b; |
749 |
int b_u0 = B_U_tab[ u_ptr[0] ]; \ |
int r2, g2, b2; |
750 |
int g_uv0 = G_U_tab[ u_ptr[0] ] + G_V_tab[ v_ptr[0] ]; \ |
|
751 |
int r_v0 = R_V_tab[ v_ptr[0] ]; \ |
r = g = b = 0; |
752 |
int b_u1 = B_U_tab[ u_ptr[uv_stride] ]; \ |
r2 = g2 = b2 = 0; |
753 |
int g_uv1 = G_U_tab[ u_ptr[uv_stride] ] + G_V_tab[ v_ptr[uv_stride] ]; \ |
|
754 |
int r_v1 = R_V_tab[ v_ptr[uv_stride] ]; \ |
// process one 2x2 block per iteration |
755 |
WRITE_RGB16(0, 0, C1) \ |
for (x = 0; x < (uint32_t) width / 2; x++) { |
756 |
WRITE_RGB16(1, 1, C1) \ |
int u, v; |
757 |
WRITE_RGB16(2, 0, C1) \ |
int b_u, g_uv, r_v, rgb_y; |
758 |
WRITE_RGB16(3, 1, C1) \ |
|
759 |
|
u = u_src[x]; |
760 |
|
v = v_src[x]; |
761 |
/* rgb/rgbi output */ |
|
762 |
|
b_u = B_U_tab[u]; |
763 |
#define WRITE_RGB(SIZE,ROW,UV_ROW,C1,C2,C3,C4) \ |
g_uv = G_U_tab[u] + G_V_tab[v]; |
764 |
rgb_y = RGB_Y_tab[ y_ptr[(ROW)*y_stride + 0] ]; \ |
r_v = R_V_tab[v]; |
765 |
x_ptr[(ROW)*x_stride+(C3)] = MAX(0, MIN(255, (rgb_y + b_u##UV_ROW) >> SCALEBITS_OUT)); \ |
|
766 |
x_ptr[(ROW)*x_stride+(C2)] = MAX(0, MIN(255, (rgb_y - g_uv##UV_ROW) >> SCALEBITS_OUT)); \ |
rgb_y = RGB_Y_tab[*y_src]; |
767 |
x_ptr[(ROW)*x_stride+(C1)] = MAX(0, MIN(255, (rgb_y + r_v##UV_ROW) >> SCALEBITS_OUT)); \ |
b = (b & 0x7) + ((rgb_y + b_u) >> SCALEBITS_OUT); |
768 |
if ((SIZE)>3) x_ptr[(ROW)*x_stride+(C4)] = 0; \ |
g = (g & 0x7) + ((rgb_y - g_uv) >> SCALEBITS_OUT); |
769 |
rgb_y = RGB_Y_tab[ y_ptr[(ROW)*y_stride + 1] ]; \ |
r = (r & 0x7) + ((rgb_y + r_v) >> SCALEBITS_OUT); |
770 |
x_ptr[(ROW)*x_stride+(SIZE)+(C3)] = MAX(0, MIN(255, (rgb_y + b_u##UV_ROW) >> SCALEBITS_OUT)); \ |
*(uint16_t *) dst = MK_RGB565(r, g, b); |
771 |
x_ptr[(ROW)*x_stride+(SIZE)+(C2)] = MAX(0, MIN(255, (rgb_y - g_uv##UV_ROW) >> SCALEBITS_OUT)); \ |
|
772 |
x_ptr[(ROW)*x_stride+(SIZE)+(C1)] = MAX(0, MIN(255, (rgb_y + r_v##UV_ROW) >> SCALEBITS_OUT)); \ |
y_src++; |
773 |
if ((SIZE)>3) x_ptr[(ROW)*x_stride+(SIZE)+(C4)] = 0; |
rgb_y = RGB_Y_tab[*y_src]; |
774 |
|
b = (b & 0x7) + ((rgb_y + b_u) >> SCALEBITS_OUT); |
775 |
|
g = (g & 0x7) + ((rgb_y - g_uv) >> SCALEBITS_OUT); |
776 |
#define YV12_TO_RGB_ROW(SIZE,C1,C2,C3,C4) /* nothing */ |
r = (r & 0x7) + ((rgb_y + r_v) >> SCALEBITS_OUT); |
777 |
#define YV12_TO_RGB(SIZE,C1,C2,C3,C4) \ |
*(uint16_t *) (dst + 2) = MK_RGB565(r, g, b); |
778 |
int rgb_y; \ |
y_src++; |
779 |
int b_u0 = B_U_tab[ u_ptr[0] ]; \ |
|
780 |
int g_uv0 = G_U_tab[ u_ptr[0] ] + G_V_tab[ v_ptr[0] ]; \ |
rgb_y = RGB_Y_tab[*y_src2]; |
781 |
int r_v0 = R_V_tab[ v_ptr[0] ]; \ |
b2 = (b2 & 0x7) + ((rgb_y + b_u) >> SCALEBITS_OUT); |
782 |
WRITE_RGB(SIZE, 0, 0, C1,C2,C3,C4) \ |
g2 = (g2 & 0x7) + ((rgb_y - g_uv) >> SCALEBITS_OUT); |
783 |
WRITE_RGB(SIZE, 1, 0, C1,C2,C3,C4) |
r2 = (r2 & 0x7) + ((rgb_y + r_v) >> SCALEBITS_OUT); |
784 |
|
*(uint16_t *) (dst2) = MK_RGB565(r2, g2, b2); |
785 |
#define YV12_TO_RGBI_ROW(SIZE,C1,C2,C3,C4) /* nothing */ |
y_src2++; |
786 |
#define YV12_TO_RGBI(SIZE,C1,C2,C3,C4) \ |
|
787 |
int rgb_y; \ |
rgb_y = RGB_Y_tab[*y_src2]; |
788 |
int b_u0 = B_U_tab[ u_ptr[0] ]; \ |
b2 = (b2 & 0x7) + ((rgb_y + b_u) >> SCALEBITS_OUT); |
789 |
int g_uv0 = G_U_tab[ u_ptr[0] ] + G_V_tab[ v_ptr[0] ]; \ |
g2 = (g2 & 0x7) + ((rgb_y - g_uv) >> SCALEBITS_OUT); |
790 |
int r_v0 = R_V_tab[ v_ptr[0] ]; \ |
r2 = (r2 & 0x7) + ((rgb_y + r_v) >> SCALEBITS_OUT); |
791 |
int b_u1 = B_U_tab[ u_ptr[uv_stride] ]; \ |
*(uint16_t *) (dst2 + 2) = MK_RGB565(r2, g2, b2); |
792 |
int g_uv1 = G_U_tab[ u_ptr[uv_stride] ] + G_V_tab[ v_ptr[uv_stride] ]; \ |
y_src2++; |
793 |
int r_v1 = R_V_tab[ v_ptr[uv_stride] ]; \ |
|
794 |
WRITE_RGB(SIZE, 0, 0, C1,C2,C3,C4) \ |
dst += 4; |
795 |
WRITE_RGB(SIZE, 1, 1, C1,C2,C3,C4) \ |
dst2 += 4; |
796 |
WRITE_RGB(SIZE, 2, 0, C1,C2,C3,C4) \ |
} |
797 |
WRITE_RGB(SIZE, 3, 1, C1,C2,C3,C4) |
|
798 |
|
dst += dst_dif; |
799 |
|
dst2 += dst_dif; |
800 |
/* yuyv/yuyvi output */ |
|
801 |
|
y_src += y_dif; |
802 |
#define WRITE_YUYV(ROW,UV_ROW,C1,C2,C3,C4) \ |
y_src2 += y_dif; |
803 |
x_ptr[(ROW)*x_stride+(C1)] = y_ptr[ (ROW)*y_stride +0]; \ |
|
804 |
x_ptr[(ROW)*x_stride+(C2)] = u_ptr[(UV_ROW)*uv_stride+0]; \ |
u_src += uv_stride; |
805 |
x_ptr[(ROW)*x_stride+(C3)] = y_ptr[ (ROW)*y_stride +1]; \ |
v_src += uv_stride; |
806 |
x_ptr[(ROW)*x_stride+(C4)] = v_ptr[(UV_ROW)*uv_stride+0]; \ |
} |
807 |
|
} |
808 |
#define YV12_TO_YUYV_ROW(SIZE,C1,C2,C3,C4) /* nothing */ |
|
809 |
#define YV12_TO_YUYV(SIZE,C1,C2,C3,C4) \ |
|
810 |
WRITE_YUYV(0, 0, C1,C2,C3,C4) \ |
|
811 |
WRITE_YUYV(1, 0, C1,C2,C3,C4) |
/* yuv 4:2:0 planar -> rgb24 */ |
812 |
|
|
813 |
#define YV12_TO_YUYVI_ROW(SIZE,C1,C2,C3,C4) /* nothing */ |
void |
814 |
#define YV12_TO_YUYVI(SIZE,C1,C2,C3,C4) \ |
yv12_to_rgb24_c(uint8_t * dst, |
815 |
WRITE_YUYV(0, 0, C1,C2,C3,C4) \ |
int dst_stride, |
816 |
WRITE_YUYV(1, 1, C1,C2,C3,C4) \ |
uint8_t * y_src, |
817 |
WRITE_YUYV(2, 0, C1,C2,C3,C4) \ |
uint8_t * u_src, |
818 |
WRITE_YUYV(3, 1, C1,C2,C3,C4) |
uint8_t * v_src, |
819 |
|
int y_stride, |
820 |
|
int uv_stride, |
821 |
MAKE_COLORSPACE(yv12_to_rgb555_c, 2,2,2, YV12_TO_RGB16, MK_RGB555, 0,0,0) |
int width, |
822 |
MAKE_COLORSPACE(yv12_to_rgb565_c, 2,2,2, YV12_TO_RGB16, MK_RGB565, 0,0,0) |
int height) |
823 |
MAKE_COLORSPACE(yv12_to_bgr_c, 3,2,2, YV12_TO_RGB, 2,1,0, 0) |
{ |
824 |
MAKE_COLORSPACE(yv12_to_bgra_c, 4,2,2, YV12_TO_RGB, 2,1,0,3) |
const uint32_t dst_dif = 6 * dst_stride - 3 * width; |
825 |
MAKE_COLORSPACE(yv12_to_abgr_c, 4,2,2, YV12_TO_RGB, 3,2,1,0) |
int32_t y_dif = 2 * y_stride - width; |
826 |
MAKE_COLORSPACE(yv12_to_rgba_c, 4,2,2, YV12_TO_RGB, 0,1,2,3) |
|
827 |
MAKE_COLORSPACE(yv12_to_yuyv_c, 2,2,2, YV12_TO_YUYV, 0,1,2,3) |
uint8_t *dst2 = dst + 3 * dst_stride; |
828 |
MAKE_COLORSPACE(yv12_to_uyvy_c, 2,2,2, YV12_TO_YUYV, 1,0,3,2) |
uint8_t *y_src2 = y_src + y_stride; |
829 |
|
uint32_t x, y; |
830 |
MAKE_COLORSPACE(yv12_to_rgb555i_c, 2,2,4, YV12_TO_RGB16I, MK_RGB555, 0,0,0) |
|
831 |
MAKE_COLORSPACE(yv12_to_rgb565i_c, 2,2,4, YV12_TO_RGB16I, MK_RGB565, 0,0,0) |
if (height < 0) { // flip image? |
832 |
MAKE_COLORSPACE(yv12_to_bgri_c, 3,2,4, YV12_TO_RGBI, 2,1,0, 0) |
height = -height; |
833 |
MAKE_COLORSPACE(yv12_to_bgrai_c, 4,2,4, YV12_TO_RGBI, 2,1,0,3) |
y_src += (height - 1) * y_stride; |
834 |
MAKE_COLORSPACE(yv12_to_abgri_c, 4,2,4, YV12_TO_RGBI, 3,2,1,0) |
y_src2 = y_src - y_stride; |
835 |
MAKE_COLORSPACE(yv12_to_rgbai_c, 4,2,4, YV12_TO_RGBI, 0,1,2,3) |
u_src += (height / 2 - 1) * uv_stride; |
836 |
MAKE_COLORSPACE(yv12_to_yuyvi_c, 2,2,4, YV12_TO_YUYVI, 0,1,2,3) |
v_src += (height / 2 - 1) * uv_stride; |
837 |
MAKE_COLORSPACE(yv12_to_uyvyi_c, 2,2,4, YV12_TO_YUYVI, 1,0,3,2) |
y_dif = -width - 2 * y_stride; |
838 |
|
uv_stride = -uv_stride; |
839 |
|
} |
840 |
|
|
841 |
/* yv12 to yv12 copy function */ |
for (y = height / 2; y; y--) { |
842 |
|
// process one 2x2 block per iteration |
843 |
void |
for (x = 0; x < (uint32_t) width / 2; x++) { |
844 |
yv12_to_yv12_c(uint8_t * y_dst, uint8_t * u_dst, uint8_t * v_dst, |
int u, v; |
845 |
int y_dst_stride, int uv_dst_stride, |
int b_u, g_uv, r_v, rgb_y; |
846 |
uint8_t * y_src, uint8_t * u_src, uint8_t * v_src, |
int r, g, b; |
847 |
int y_src_stride, int uv_src_stride, |
|
848 |
int width, int height, int vflip) |
u = u_src[x]; |
849 |
{ |
v = v_src[x]; |
850 |
int width2 = width / 2; |
|
851 |
int height2 = height / 2; |
b_u = B_U_tab[u]; |
852 |
int y; |
g_uv = G_U_tab[u] + G_V_tab[v]; |
853 |
|
r_v = R_V_tab[v]; |
854 |
if (vflip) { |
|
855 |
y_src += (height - 1) * y_src_stride; |
rgb_y = RGB_Y_tab[*y_src]; |
856 |
u_src += (height2 - 1) * uv_src_stride; |
b = (rgb_y + b_u) >> SCALEBITS_OUT; |
857 |
v_src += (height2 - 1) * uv_src_stride; |
g = (rgb_y - g_uv) >> SCALEBITS_OUT; |
858 |
y_src_stride = -y_src_stride; |
r = (rgb_y + r_v) >> SCALEBITS_OUT; |
859 |
uv_src_stride = -uv_src_stride; |
dst[0] = MAX(0, MIN(255, b)); |
860 |
|
dst[1] = MAX(0, MIN(255, g)); |
861 |
|
dst[2] = MAX(0, MIN(255, r)); |
862 |
|
|
863 |
|
y_src++; |
864 |
|
rgb_y = RGB_Y_tab[*y_src]; |
865 |
|
b = (rgb_y + b_u) >> SCALEBITS_OUT; |
866 |
|
g = (rgb_y - g_uv) >> SCALEBITS_OUT; |
867 |
|
r = (rgb_y + r_v) >> SCALEBITS_OUT; |
868 |
|
dst[3] = MAX(0, MIN(255, b)); |
869 |
|
dst[4] = MAX(0, MIN(255, g)); |
870 |
|
dst[5] = MAX(0, MIN(255, r)); |
871 |
|
y_src++; |
872 |
|
|
873 |
|
rgb_y = RGB_Y_tab[*y_src2]; |
874 |
|
b = (rgb_y + b_u) >> SCALEBITS_OUT; |
875 |
|
g = (rgb_y - g_uv) >> SCALEBITS_OUT; |
876 |
|
r = (rgb_y + r_v) >> SCALEBITS_OUT; |
877 |
|
dst2[0] = MAX(0, MIN(255, b)); |
878 |
|
dst2[1] = MAX(0, MIN(255, g)); |
879 |
|
dst2[2] = MAX(0, MIN(255, r)); |
880 |
|
y_src2++; |
881 |
|
|
882 |
|
rgb_y = RGB_Y_tab[*y_src2]; |
883 |
|
b = (rgb_y + b_u) >> SCALEBITS_OUT; |
884 |
|
g = (rgb_y - g_uv) >> SCALEBITS_OUT; |
885 |
|
r = (rgb_y + r_v) >> SCALEBITS_OUT; |
886 |
|
dst2[3] = MAX(0, MIN(255, b)); |
887 |
|
dst2[4] = MAX(0, MIN(255, g)); |
888 |
|
dst2[5] = MAX(0, MIN(255, r)); |
889 |
|
y_src2++; |
890 |
|
|
891 |
|
dst += 6; |
892 |
|
dst2 += 6; |
893 |
|
} |
894 |
|
|
895 |
|
dst += dst_dif; |
896 |
|
dst2 += dst_dif; |
897 |
|
|
898 |
|
y_src += y_dif; |
899 |
|
y_src2 += y_dif; |
900 |
|
|
901 |
|
u_src += uv_stride; |
902 |
|
v_src += uv_stride; |
903 |
|
} |
904 |
|
} |
905 |
|
|
906 |
|
|
907 |
|
|
908 |
|
/* yuv 4:2:0 planar -> rgb32 */ |
909 |
|
|
910 |
|
void |
911 |
|
yv12_to_rgb32_c(uint8_t * dst, |
912 |
|
int dst_stride, |
913 |
|
uint8_t * y_src, |
914 |
|
uint8_t * v_src, |
915 |
|
uint8_t * u_src, |
916 |
|
int y_stride, |
917 |
|
int uv_stride, |
918 |
|
int width, |
919 |
|
int height) |
920 |
|
{ |
921 |
|
const uint32_t dst_dif = 8 * dst_stride - 4 * width; |
922 |
|
int32_t y_dif = 2 * y_stride - width; |
923 |
|
|
924 |
|
uint8_t *dst2 = dst + 4 * dst_stride; |
925 |
|
uint8_t *y_src2 = y_src + y_stride; |
926 |
|
uint32_t x, y; |
927 |
|
|
928 |
|
if (height < 0) { // flip image? |
929 |
|
height = -height; |
930 |
|
y_src += (height - 1) * y_stride; |
931 |
|
y_src2 = y_src - y_stride; |
932 |
|
u_src += (height / 2 - 1) * uv_stride; |
933 |
|
v_src += (height / 2 - 1) * uv_stride; |
934 |
|
y_dif = -width - 2 * y_stride; |
935 |
|
uv_stride = -uv_stride; |
936 |
|
} |
937 |
|
|
938 |
|
for (y = height / 2; y; y--) { |
939 |
|
// process one 2x2 block per iteration |
940 |
|
for (x = 0; x < (uint32_t) width / 2; x++) { |
941 |
|
int u, v; |
942 |
|
int b_u, g_uv, r_v, rgb_y; |
943 |
|
int r, g, b; |
944 |
|
|
945 |
|
u = u_src[x]; |
946 |
|
v = v_src[x]; |
947 |
|
|
948 |
|
b_u = B_U_tab[u]; |
949 |
|
g_uv = G_U_tab[u] + G_V_tab[v]; |
950 |
|
r_v = R_V_tab[v]; |
951 |
|
|
952 |
|
rgb_y = RGB_Y_tab[*y_src]; |
953 |
|
b = (rgb_y + b_u) >> SCALEBITS_OUT; |
954 |
|
g = (rgb_y - g_uv) >> SCALEBITS_OUT; |
955 |
|
r = (rgb_y + r_v) >> SCALEBITS_OUT; |
956 |
|
dst[0] = MAX(0, MIN(255, r)); |
957 |
|
dst[1] = MAX(0, MIN(255, g)); |
958 |
|
dst[2] = MAX(0, MIN(255, b)); |
959 |
|
dst[3] = 0; |
960 |
|
|
961 |
|
y_src++; |
962 |
|
rgb_y = RGB_Y_tab[*y_src]; |
963 |
|
b = (rgb_y + b_u) >> SCALEBITS_OUT; |
964 |
|
g = (rgb_y - g_uv) >> SCALEBITS_OUT; |
965 |
|
r = (rgb_y + r_v) >> SCALEBITS_OUT; |
966 |
|
dst[4] = MAX(0, MIN(255, r)); |
967 |
|
dst[5] = MAX(0, MIN(255, g)); |
968 |
|
dst[6] = MAX(0, MIN(255, b)); |
969 |
|
dst[7] = 0; |
970 |
|
y_src++; |
971 |
|
|
972 |
|
rgb_y = RGB_Y_tab[*y_src2]; |
973 |
|
b = (rgb_y + b_u) >> SCALEBITS_OUT; |
974 |
|
g = (rgb_y - g_uv) >> SCALEBITS_OUT; |
975 |
|
r = (rgb_y + r_v) >> SCALEBITS_OUT; |
976 |
|
dst2[0] = MAX(0, MIN(255, r)); |
977 |
|
dst2[1] = MAX(0, MIN(255, g)); |
978 |
|
dst2[2] = MAX(0, MIN(255, b)); |
979 |
|
dst2[3] = 0; |
980 |
|
y_src2++; |
981 |
|
|
982 |
|
rgb_y = RGB_Y_tab[*y_src2]; |
983 |
|
b = (rgb_y + b_u) >> SCALEBITS_OUT; |
984 |
|
g = (rgb_y - g_uv) >> SCALEBITS_OUT; |
985 |
|
r = (rgb_y + r_v) >> SCALEBITS_OUT; |
986 |
|
dst2[4] = MAX(0, MIN(255, r)); |
987 |
|
dst2[5] = MAX(0, MIN(255, g)); |
988 |
|
dst2[6] = MAX(0, MIN(255, b)); |
989 |
|
dst2[7] = 0; |
990 |
|
y_src2++; |
991 |
|
|
992 |
|
dst += 8; |
993 |
|
dst2 += 8; |
994 |
|
} |
995 |
|
|
996 |
|
dst += dst_dif; |
997 |
|
dst2 += dst_dif; |
998 |
|
|
999 |
|
y_src += y_dif; |
1000 |
|
y_src2 += y_dif; |
1001 |
|
|
1002 |
|
u_src += uv_stride; |
1003 |
|
v_src += uv_stride; |
1004 |
|
} |
1005 |
|
} |
1006 |
|
|
1007 |
|
|
1008 |
|
|
1009 |
|
/* yuv 4:2:0 planar -> yuv planar */ |
1010 |
|
|
1011 |
|
void |
1012 |
|
yv12_to_yuv_c(uint8_t * dst, |
1013 |
|
int dst_stride, |
1014 |
|
uint8_t * y_src, |
1015 |
|
uint8_t * u_src, |
1016 |
|
uint8_t * v_src, |
1017 |
|
int y_stride, |
1018 |
|
int uv_stride, |
1019 |
|
int width, |
1020 |
|
int height) |
1021 |
|
{ |
1022 |
|
uint32_t dst_stride2 = dst_stride >> 1; |
1023 |
|
uint32_t width2 = width >> 1; |
1024 |
|
uint32_t y; |
1025 |
|
|
1026 |
|
if (height < 0) { |
1027 |
|
height = -height; |
1028 |
|
y_src += (height - 1) * y_stride; |
1029 |
|
u_src += (height / 2 - 1) * uv_stride; |
1030 |
|
v_src += (height / 2 - 1) * uv_stride; |
1031 |
|
y_stride = -y_stride; |
1032 |
|
uv_stride = -uv_stride; |
1033 |
} |
} |
1034 |
|
|
1035 |
for (y = height; y; y--) { |
for (y = height; y; y--) { |
1036 |
memcpy(y_dst, y_src, width); |
memcpy(dst, y_src, width); |
1037 |
y_src += y_src_stride; |
dst += dst_stride; |
1038 |
y_dst += y_dst_stride; |
y_src += y_stride; |
1039 |
} |
} |
1040 |
|
|
1041 |
for (y = height2; y; y--) { |
for (y = height >> 1; y; y--) { |
1042 |
memcpy(u_dst, u_src, width2); |
memcpy(dst, u_src, width2); |
1043 |
u_src += uv_src_stride; |
dst += dst_stride2; |
1044 |
u_dst += uv_dst_stride; |
u_src += uv_stride; |
1045 |
} |
} |
1046 |
|
|
1047 |
for (y = height2; y; y--) { |
for (y = height >> 1; y; y--) { |
1048 |
memcpy(v_dst, v_src, width2); |
memcpy(dst, v_src, width2); |
1049 |
v_src += uv_src_stride; |
dst += dst_stride2; |
1050 |
v_dst += uv_dst_stride; |
v_src += uv_stride; |
1051 |
} |
} |
1052 |
} |
} |
1053 |
|
|
1054 |
|
|
1055 |
|
|
1056 |
/* initialize rgb lookup tables */ |
|
1057 |
|
/* yuv 4:2:0 planar -> yuyv (yuv2) packed */ |
1058 |
|
|
1059 |
void |
void |
1060 |
colorspace_init(void) |
yv12_to_yuyv_c(uint8_t * dst, |
1061 |
|
int dst_stride, |
1062 |
|
uint8_t * y_src, |
1063 |
|
uint8_t * u_src, |
1064 |
|
uint8_t * v_src, |
1065 |
|
int y_stride, |
1066 |
|
int uv_stride, |
1067 |
|
int width, |
1068 |
|
int height) |
1069 |
{ |
{ |
1070 |
int32_t i; |
const uint32_t dst_dif = 2 * (dst_stride - width); |
1071 |
|
uint32_t x, y; |
1072 |
|
|
1073 |
for (i = 0; i < 256; i++) { |
if (height < 0) { |
1074 |
RGB_Y_tab[i] = FIX_OUT(RGB_Y_OUT) * (i - Y_ADD_OUT); |
height = -height; |
1075 |
B_U_tab[i] = FIX_OUT(B_U_OUT) * (i - U_ADD_OUT); |
y_src += (height - 1) * y_stride; |
1076 |
G_U_tab[i] = FIX_OUT(G_U_OUT) * (i - U_ADD_OUT); |
u_src += (height / 2 - 1) * uv_stride; |
1077 |
G_V_tab[i] = FIX_OUT(G_V_OUT) * (i - V_ADD_OUT); |
v_src += (height / 2 - 1) * uv_stride; |
1078 |
R_V_tab[i] = FIX_OUT(R_V_OUT) * (i - V_ADD_OUT); |
y_stride = -y_stride; |
1079 |
|
uv_stride = -uv_stride; |
1080 |
|
} |
1081 |
|
|
1082 |
|
for (y = 0; y < (uint32_t) height; y++) { |
1083 |
|
for (x = 0; x < (uint32_t) width / 2; x++) { |
1084 |
|
dst[0] = y_src[2 * x]; |
1085 |
|
dst[1] = u_src[x]; |
1086 |
|
dst[2] = y_src[2 * x + 1]; |
1087 |
|
dst[3] = v_src[x]; |
1088 |
|
dst += 4; |
1089 |
|
} |
1090 |
|
dst += dst_dif; |
1091 |
|
y_src += y_stride; |
1092 |
|
if (y & 1) { |
1093 |
|
u_src += uv_stride; |
1094 |
|
v_src += uv_stride; |
1095 |
|
} |
1096 |
|
} |
1097 |
|
} |
1098 |
|
|
1099 |
|
|
1100 |
|
|
1101 |
|
/* yuv 4:2:0 planar -> uyvy packed */ |
1102 |
|
|
1103 |
|
void |
1104 |
|
yv12_to_uyvy_c(uint8_t * dst, |
1105 |
|
int dst_stride, |
1106 |
|
uint8_t * y_src, |
1107 |
|
uint8_t * u_src, |
1108 |
|
uint8_t * v_src, |
1109 |
|
int y_stride, |
1110 |
|
int uv_stride, |
1111 |
|
int width, |
1112 |
|
int height) |
1113 |
|
{ |
1114 |
|
const uint32_t dst_dif = 2 * (dst_stride - width); |
1115 |
|
uint32_t x, y; |
1116 |
|
|
1117 |
|
if (height < 0) { |
1118 |
|
height = -height; |
1119 |
|
y_src += (height - 1) * y_stride; |
1120 |
|
u_src += (height / 2 - 1) * uv_stride; |
1121 |
|
v_src += (height / 2 - 1) * uv_stride; |
1122 |
|
y_stride = -y_stride; |
1123 |
|
uv_stride = -uv_stride; |
1124 |
|
} |
1125 |
|
|
1126 |
|
for (y = 0; y < (uint32_t) height; y++) { |
1127 |
|
for (x = 0; x < (uint32_t) width / 2; x++) { |
1128 |
|
dst[0] = u_src[x]; |
1129 |
|
dst[1] = y_src[2 * x]; |
1130 |
|
dst[2] = v_src[x]; |
1131 |
|
dst[3] = y_src[2 * x + 1]; |
1132 |
|
dst += 4; |
1133 |
|
} |
1134 |
|
dst += dst_dif; |
1135 |
|
y_src += y_stride; |
1136 |
|
if (y & 1) { |
1137 |
|
u_src += uv_stride; |
1138 |
|
v_src += uv_stride; |
1139 |
|
} |
1140 |
|
} |
1141 |
|
} |
1142 |
|
|
1143 |
|
|
1144 |
|
/* user yuv planar -> yuv 4:2:0 planar |
1145 |
|
|
1146 |
|
NOTE: does not flip */ |
1147 |
|
|
1148 |
|
void |
1149 |
|
user_to_yuv_c(uint8_t * y_out, |
1150 |
|
uint8_t * u_out, |
1151 |
|
uint8_t * v_out, |
1152 |
|
int stride, |
1153 |
|
DEC_PICTURE * picture, |
1154 |
|
int width, |
1155 |
|
int height) |
1156 |
|
{ |
1157 |
|
uint32_t stride2 = stride >> 1; |
1158 |
|
uint32_t width2 = width >> 1; |
1159 |
|
uint32_t y; |
1160 |
|
uint8_t *src; |
1161 |
|
|
1162 |
|
src = picture->y; |
1163 |
|
for (y = height; y; y--) { |
1164 |
|
memcpy(y_out, src, width); |
1165 |
|
src += picture->stride_y; |
1166 |
|
y_out += stride; |
1167 |
|
} |
1168 |
|
|
1169 |
|
src = picture->u; |
1170 |
|
for (y = height >> 1; y; y--) { |
1171 |
|
memcpy(u_out, src, width2); |
1172 |
|
src += picture->stride_uv; |
1173 |
|
u_out += stride2; |
1174 |
|
} |
1175 |
|
|
1176 |
|
src = picture->v; |
1177 |
|
for (y = height >> 1; y; y--) { |
1178 |
|
memcpy(v_out, src, width2); |
1179 |
|
src += picture->stride_uv; |
1180 |
|
v_out += stride2; |
1181 |
} |
} |
1182 |
} |
} |