23 |
* |
* |
24 |
* History: |
* History: |
25 |
* |
* |
26 |
* 02.05.2002 add "interpolate8x8_c" for B-frame; MinChen <chenm002@163.com> |
* 05.10.2002 new bilinear and qpel interpolation code - Isibaar |
27 |
* 27.12.2001 modified "compensate_halfpel" |
* 27.12.2001 modified "compensate_halfpel" |
28 |
* 05.11.2001 initial version; (c)2001 peter ross <pross@cs.rmit.edu.au> |
* 05.11.2001 initial version; (c)2001 peter ross <pross@cs.rmit.edu.au> |
29 |
* |
* |
38 |
INTERPOLATE8X8_PTR interpolate8x8_halfpel_v; |
INTERPOLATE8X8_PTR interpolate8x8_halfpel_v; |
39 |
INTERPOLATE8X8_PTR interpolate8x8_halfpel_hv; |
INTERPOLATE8X8_PTR interpolate8x8_halfpel_hv; |
40 |
|
|
41 |
|
INTERPOLATE8X8_AVG2_PTR interpolate8x8_avg2; |
42 |
|
INTERPOLATE8X8_AVG4_PTR interpolate8x8_avg4; |
43 |
|
|
44 |
|
INTERPOLATE8X8_LOWPASS_PTR interpolate8x8_lowpass_h; |
45 |
|
INTERPOLATE8X8_LOWPASS_PTR interpolate8x8_lowpass_v; |
46 |
|
INTERPOLATE8X8_LOWPASS_HV_PTR interpolate8x8_lowpass_hv; |
47 |
|
|
48 |
|
INTERPOLATE8X8_6TAP_LOWPASS_PTR interpolate8x8_6tap_lowpass_h; |
49 |
|
INTERPOLATE8X8_6TAP_LOWPASS_PTR interpolate8x8_6tap_lowpass_v; |
50 |
|
|
51 |
|
void interpolate8x8_avg2_c(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, const uint32_t stride, const uint32_t rounding) |
52 |
|
{ |
53 |
|
int32_t i; |
54 |
|
const int32_t round = 1 - rounding; |
55 |
|
|
56 |
|
for(i = 0; i < 8; i++) |
57 |
|
{ |
58 |
|
dst[0] = (src1[0] + src2[0] + round) >> 1; |
59 |
|
dst[1] = (src1[1] + src2[1] + round) >> 1; |
60 |
|
dst[2] = (src1[2] + src2[2] + round) >> 1; |
61 |
|
dst[3] = (src1[3] + src2[3] + round) >> 1; |
62 |
|
dst[4] = (src1[4] + src2[4] + round) >> 1; |
63 |
|
dst[5] = (src1[5] + src2[5] + round) >> 1; |
64 |
|
dst[6] = (src1[6] + src2[6] + round) >> 1; |
65 |
|
dst[7] = (src1[7] + src2[7] + round) >> 1; |
66 |
|
|
67 |
|
dst += stride; |
68 |
|
src1 += stride; |
69 |
|
src2 += stride; |
70 |
|
} |
71 |
|
} |
72 |
|
|
73 |
|
void interpolate8x8_avg4_c(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, const uint8_t *src3, const uint8_t *src4, const uint32_t stride, const uint32_t rounding) |
74 |
|
{ |
75 |
|
int32_t i; |
76 |
|
const int32_t round = 2 - rounding; |
77 |
|
|
78 |
|
for(i = 0; i < 8; i++) |
79 |
|
{ |
80 |
|
dst[0] = (src1[0] + src2[0] + src3[0] + src4[0] + round) >> 2; |
81 |
|
dst[1] = (src1[1] + src2[1] + src3[1] + src4[1] + round) >> 2; |
82 |
|
dst[2] = (src1[2] + src2[2] + src3[2] + src4[2] + round) >> 2; |
83 |
|
dst[3] = (src1[3] + src2[3] + src3[3] + src4[3] + round) >> 2; |
84 |
|
dst[4] = (src1[4] + src2[4] + src3[4] + src4[4] + round) >> 2; |
85 |
|
dst[5] = (src1[5] + src2[5] + src3[5] + src4[5] + round) >> 2; |
86 |
|
dst[6] = (src1[6] + src2[6] + src3[6] + src4[6] + round) >> 2; |
87 |
|
dst[7] = (src1[7] + src2[7] + src3[7] + src4[7] + round) >> 2; |
88 |
|
|
89 |
|
dst += stride; |
90 |
|
src1 += stride; |
91 |
|
src2 += stride; |
92 |
|
src3 += stride; |
93 |
|
src4 += stride; |
94 |
|
} |
95 |
|
} |
96 |
|
|
97 |
// dst = interpolate(src) |
// dst = interpolate(src) |
98 |
|
|
158 |
} |
} |
159 |
} |
} |
160 |
|
|
|
// add by MinChen <chenm001@163.com> |
|
|
// interpolate8x8 two pred block |
|
|
void |
|
|
interpolate8x8_c(uint8_t * const dst, |
|
|
const uint8_t * const src, |
|
|
const uint32_t x, |
|
|
const uint32_t y, |
|
|
const uint32_t stride) |
|
|
{ |
|
|
uint32_t i, j; |
|
|
|
|
|
for (j = 0; j < 8; j++) { |
|
|
for (i = 0; i < 8; i++) { |
|
|
int32_t tot = |
|
|
((src[(y + j) * stride + x + i] + |
|
|
dst[(y + j) * stride + x + i] + 1) >> 1); |
|
|
dst[(y + j) * stride + x + i] = (uint8_t) tot; |
|
|
} |
|
|
} |
|
|
} |
|
|
|
|
161 |
/************************************************************* |
/************************************************************* |
162 |
* QPEL STUFF STARTS HERE * |
* QPEL STUFF STARTS HERE * |
163 |
*************************************************************/ |
*************************************************************/ |
164 |
|
|
165 |
#define CLIP(X,A,B) (X < A) ? (A) : ((X > B) ? (B) : (X)) |
#define CLIP(X,A,B) (X < A) ? (A) : ((X > B) ? (B) : (X)) |
166 |
|
|
167 |
void interpolate8x8_lowpass_h(uint8_t *dst, uint8_t *src, int32_t dst_stride, int32_t src_stride, int32_t rounding) |
void interpolate8x8_6tap_lowpass_h_c(uint8_t *dst, uint8_t *src, int32_t stride, int32_t rounding) |
168 |
{ |
{ |
169 |
int32_t i; |
int32_t i; |
170 |
|
uint8_t round_add = 16 - rounding; |
171 |
|
|
172 |
for(i = 0; i < 8; i++) |
for(i = 0; i < 8; i++) |
173 |
{ |
{ |
|
dst[0] = CLIP((((src[0] + src[1]) * 160 - (src[0] + src[2]) * 48 + (src[1] + src[3]) * 24 - (src[2] + src[4]) * 8 + (128 - rounding)) / 256), 0, 255); |
|
|
dst[1] = CLIP((((src[1] + src[2]) * 160 - (src[0] + src[3]) * 48 + (src[0] + src[4]) * 24 - (src[1] + src[5]) * 8 + (128 - rounding)) / 256), 0, 255); |
|
|
dst[2] = CLIP((((src[2] + src[3]) * 160 - (src[1] + src[4]) * 48 + (src[0] + src[5]) * 24 - (src[0] + src[6]) * 8 + (128 - rounding)) / 256), 0, 255); |
|
|
dst[3] = CLIP((((src[3] + src[4]) * 160 - (src[2] + src[5]) * 48 + (src[1] + src[6]) * 24 - (src[0] + src[7]) * 8 + (128 - rounding)) / 256), 0, 255); |
|
|
dst[4] = CLIP((((src[4] + src[5]) * 160 - (src[3] + src[6]) * 48 + (src[2] + src[7]) * 24 - (src[1] + src[8]) * 8 + (128 - rounding)) / 256), 0, 255); |
|
|
dst[5] = CLIP((((src[5] + src[6]) * 160 - (src[4] + src[7]) * 48 + (src[3] + src[8]) * 24 - (src[2] + src[8]) * 8 + (128 - rounding)) / 256), 0, 255); |
|
|
dst[6] = CLIP((((src[6] + src[7]) * 160 - (src[5] + src[8]) * 48 + (src[4] + src[8]) * 24 - (src[3] + src[7]) * 8 + (128 - rounding)) / 256), 0, 255); |
|
|
dst[7] = CLIP((((src[7] + src[8]) * 160 - (src[6] + src[8]) * 48 + (src[5] + src[7]) * 24 - (src[4] + src[6]) * 8 + (128 - rounding)) / 256), 0, 255); |
|
174 |
|
|
175 |
dst += dst_stride; |
dst[0] = CLIP((((src[-2] + src[3]) + 5 * (((src[0] + src[1])<<2) - (src[-1] + src[2])) + round_add) >> 5), 0, 255); |
176 |
src += src_stride; |
dst[1] = CLIP((((src[-1] + src[4]) + 5 * (((src[1] + src[2])<<2) - (src[0] + src[3])) + round_add) >> 5), 0, 255); |
177 |
|
dst[2] = CLIP((((src[0] + src[5]) + 5 * (((src[2] + src[3])<<2) - (src[1] + src[4])) + round_add) >> 5), 0, 255); |
178 |
|
dst[3] = CLIP((((src[1] + src[6]) + 5 * (((src[3] + src[4])<<2) - (src[2] + src[5])) + round_add) >> 5), 0, 255); |
179 |
|
dst[4] = CLIP((((src[2] + src[7]) + 5 * (((src[4] + src[5])<<2) - (src[3] + src[6])) + round_add) >> 5), 0, 255); |
180 |
|
dst[5] = CLIP((((src[3] + src[8]) + 5 * (((src[5] + src[6])<<2) - (src[4] + src[7])) + round_add) >> 5), 0, 255); |
181 |
|
dst[6] = CLIP((((src[4] + src[9]) + 5 * (((src[6] + src[7])<<2) - (src[5] + src[8])) + round_add) >> 5), 0, 255); |
182 |
|
dst[7] = CLIP((((src[5] + src[10]) + 5 * (((src[7] + src[8])<<2) - (src[6] + src[9])) + round_add) >> 5), 0, 255); |
183 |
|
|
184 |
|
dst += stride; |
185 |
|
src += stride; |
186 |
} |
} |
187 |
} |
} |
188 |
|
|
189 |
void interpolate8x8_lowpass_v(uint8_t *dst, uint8_t *src, int32_t dst_stride, int32_t src_stride, int32_t rounding) |
void interpolate8x8_lowpass_h_c(uint8_t *dst, uint8_t *src, int32_t stride, int32_t rounding) |
190 |
{ |
{ |
191 |
int32_t i; |
int32_t i; |
192 |
|
uint8_t round_add = 16 - rounding; |
193 |
|
|
194 |
for(i = 0; i < 8; i++) |
for(i = 0; i < 8; i++) |
195 |
{ |
{ |
|
int32_t src0 = src[0]; |
|
|
int32_t src1 = src[src_stride]; |
|
|
int32_t src2 = src[2 * src_stride]; |
|
|
int32_t src3 = src[3 * src_stride]; |
|
|
int32_t src4 = src[4 * src_stride]; |
|
|
int32_t src5 = src[5 * src_stride]; |
|
|
int32_t src6 = src[6 * src_stride]; |
|
|
int32_t src7 = src[7 * src_stride]; |
|
|
int32_t src8 = src[8 * src_stride]; |
|
|
|
|
|
dst[0] = CLIP((((src0 + src1) * 160 - (src0 + src2) * 48 + (src1 + src3) * 24 - (src2 + src4) * 8 + (128 - rounding)) / 256), 0, 255); |
|
|
dst[dst_stride] = CLIP((((src1 + src2) * 160 - (src0 + src3) * 48 + (src0 + src4) * 24 - (src1 + src5) * 8 + (128 - rounding)) / 256), 0, 255); |
|
|
dst[2 * dst_stride] = CLIP((((src2 + src3) * 160 - (src1 + src4) * 48 + (src0 + src5) * 24 - (src0 + src6) * 8 + (128 - rounding)) / 256), 0, 255); |
|
|
dst[3 * dst_stride] = CLIP((((src3 + src4) * 160 - (src2 + src5) * 48 + (src1 + src6) * 24 - (src0 + src7) * 8 + (128 - rounding)) / 256), 0, 255); |
|
|
dst[4 * dst_stride] = CLIP((((src4 + src5) * 160 - (src3 + src6) * 48 + (src2 + src7) * 24 - (src1 + src8) * 8 + (128 - rounding)) / 256), 0, 255); |
|
|
dst[5 * dst_stride] = CLIP((((src5 + src6) * 160 - (src4 + src7) * 48 + (src3 + src8) * 24 - (src2 + src8) * 8 + (128 - rounding)) / 256), 0, 255); |
|
|
dst[6 * dst_stride] = CLIP((((src6 + src7) * 160 - (src5 + src8) * 48 + (src4 + src8) * 24 - (src3 + src7) * 8 + (128 - rounding)) / 256), 0, 255); |
|
|
dst[7 * dst_stride] = CLIP((((src7 + src8) * 160 - (src6 + src8) * 48 + (src5 + src7) * 24 - (src4 + src6) * 8 + (128 - rounding)) / 256), 0, 255); |
|
196 |
|
|
197 |
dst++; |
dst[0] = CLIP(((7 * ((src[0]<<1) - src[2]) + 23 * src[1] + 3 * src[3] - src[4] + round_add) >> 5), 0, 255); |
198 |
src++; |
dst[1] = CLIP(((19 * src[1] + 20 * src[2] - src[5] + 3 * (src[4] - src[0] - (src[3]<<1)) + round_add) >> 5), 0, 255); |
199 |
|
dst[2] = CLIP(((20 * (src[2] + src[3]) + (src[0]<<1) + 3 * (src[5] - ((src[1] + src[4])<<1)) - src[6] + round_add) >> 5), 0, 255); |
200 |
|
dst[3] = CLIP(((20 * (src[3] + src[4]) + 3 * ((src[6] + src[1]) - ((src[2] + src[5])<<1)) - (src[0] + src[7]) + round_add) >> 5), 0, 255); |
201 |
|
dst[4] = CLIP(((20 * (src[4] + src[5]) - 3 * (((src[3] + src[6])<<1) - (src[2] + src[7])) - (src[1] + src[8]) + round_add) >> 5), 0, 255); |
202 |
|
dst[5] = CLIP(((20 * (src[5] + src[6]) + (src[8]<<1) + 3 * (src[3] - ((src[4] + src[7]) << 1)) - src[2] + round_add) >> 5), 0, 255); |
203 |
|
dst[6] = CLIP(((19 * src[7] + 20 * src[6] + 3 * (src[4] - src[8] - (src[5] << 1)) - src[3] + round_add) >> 5), 0, 255); |
204 |
|
dst[7] = CLIP(((23 * src[7] + 7 * ((src[8]<<1) - src[6]) + 3 * src[5] - src[4] + round_add) >> 5), 0, 255); |
205 |
|
|
206 |
|
dst += stride; |
207 |
|
src += stride; |
208 |
} |
} |
209 |
} |
} |
210 |
|
|
211 |
void interpolate8x8_lowpass_hv(uint8_t *dst1, uint8_t *dst2, uint8_t *src, int32_t dst1_stride, int32_t dst2_stride, int32_t src_stride, int32_t rounding) |
void interpolate8x8_6tap_lowpass_v_c(uint8_t *dst, uint8_t *src, int32_t stride, int32_t rounding) |
212 |
{ |
{ |
|
uint8_t data[72]; |
|
|
|
|
213 |
int32_t i; |
int32_t i; |
214 |
|
uint8_t round_add = 16 - rounding; |
|
|
|
|
for(i = 0; i < 9; i++) |
|
|
{ |
|
|
dst2[0] = data[8 * i + 0] = CLIP((((src[0] + src[1]) * 160 - (src[0] + src[2]) * 48 + (src[1] + src[3]) * 24 - (src[2] + src[4]) * 8 + (128 - rounding)) / 256), 0, 255); |
|
|
dst2[1] = data[8 * i + 1] = CLIP((((src[1] + src[2]) * 160 - (src[0] + src[3]) * 48 + (src[0] + src[4]) * 24 - (src[1] + src[5]) * 8 + (128 - rounding)) / 256), 0, 255); |
|
|
dst2[2] = data[8 * i + 2] = CLIP((((src[2] + src[3]) * 160 - (src[1] + src[4]) * 48 + (src[0] + src[5]) * 24 - (src[0] + src[6]) * 8 + (128 - rounding)) / 256), 0, 255); |
|
|
dst2[3] = data[8 * i + 3] = CLIP((((src[3] + src[4]) * 160 - (src[2] + src[5]) * 48 + (src[1] + src[6]) * 24 - (src[0] + src[7]) * 8 + (128 - rounding)) / 256), 0, 255); |
|
|
dst2[4] = data[8 * i + 4] = CLIP((((src[4] + src[5]) * 160 - (src[3] + src[6]) * 48 + (src[2] + src[7]) * 24 - (src[1] + src[8]) * 8 + (128 - rounding)) / 256), 0, 255); |
|
|
dst2[5] = data[8 * i + 5] = CLIP((((src[5] + src[6]) * 160 - (src[4] + src[7]) * 48 + (src[3] + src[8]) * 24 - (src[2] + src[8]) * 8 + (128 - rounding)) / 256), 0, 255); |
|
|
dst2[6] = data[8 * i + 6] = CLIP((((src[6] + src[7]) * 160 - (src[5] + src[8]) * 48 + (src[4] + src[8]) * 24 - (src[3] + src[7]) * 8 + (128 - rounding)) / 256), 0, 255); |
|
|
dst2[7] = data[8 * i + 7] = CLIP((((src[7] + src[8]) * 160 - (src[6] + src[8]) * 48 + (src[5] + src[7]) * 24 - (src[4] + src[6]) * 8 + (128 - rounding)) / 256), 0, 255); |
|
|
|
|
|
src += src_stride; |
|
|
dst2 += dst2_stride; |
|
|
} |
|
215 |
|
|
216 |
for(i = 0; i < 8; i++) |
for(i = 0; i < 8; i++) |
217 |
{ |
{ |
218 |
int32_t src0 = data[i]; |
int32_t src_2 = src[-2*stride]; |
219 |
int32_t src1 = data[8 + i]; |
int32_t src_1 = src[-stride]; |
220 |
int32_t src2 = data[2 * 8 + i]; |
int32_t src0 = src[0]; |
221 |
int32_t src3 = data[3 * 8 + i]; |
int32_t src1 = src[stride]; |
222 |
int32_t src4 = data[4 * 8 + i]; |
int32_t src2 = src[2 * stride]; |
223 |
int32_t src5 = data[5 * 8 + i]; |
int32_t src3 = src[3 * stride]; |
224 |
int32_t src6 = data[6 * 8 + i]; |
int32_t src4 = src[4 * stride]; |
225 |
int32_t src7 = data[7 * 8 + i]; |
int32_t src5 = src[5 * stride]; |
226 |
int32_t src8 = data[8 * 8 + i]; |
int32_t src6 = src[6 * stride]; |
227 |
|
int32_t src7 = src[7 * stride]; |
228 |
dst1[0] = CLIP((((src0 + src1) * 160 - (src0 + src2) * 48 + (src1 + src3) * 24 - (src2 + src4) * 8 + (128 - rounding)) / 256), 0, 255); |
int32_t src8 = src[8 * stride]; |
229 |
dst1[dst1_stride] = CLIP((((src1 + src2) * 160 - (src0 + src3) * 48 + (src0 + src4) * 24 - (src1 + src5) * 8 + (128 - rounding)) / 256), 0, 255); |
int32_t src9 = src[9 * stride]; |
230 |
dst1[2 * dst1_stride] = CLIP((((src2 + src3) * 160 - (src1 + src4) * 48 + (src0 + src5) * 24 - (src0 + src6) * 8 + (128 - rounding)) / 256), 0, 255); |
int32_t src10 = src[10 * stride]; |
231 |
dst1[3 * dst1_stride] = CLIP((((src3 + src4) * 160 - (src2 + src5) * 48 + (src1 + src6) * 24 - (src0 + src7) * 8 + (128 - rounding)) / 256), 0, 255); |
|
232 |
dst1[4 * dst1_stride] = CLIP((((src4 + src5) * 160 - (src3 + src6) * 48 + (src2 + src7) * 24 - (src1 + src8) * 8 + (128 - rounding)) / 256), 0, 255); |
dst[0] = CLIP((((src_2 + src3) + 5 * (((src0 + src1)<<2) - (src_1 + src2)) + round_add) >> 5), 0, 255); |
233 |
dst1[5 * dst1_stride] = CLIP((((src5 + src6) * 160 - (src4 + src7) * 48 + (src3 + src8) * 24 - (src2 + src8) * 8 + (128 - rounding)) / 256), 0, 255); |
dst[stride] = CLIP((((src_1 + src4) + 5 * (((src1 + src2)<<2) - (src0 + src3)) + round_add) >> 5), 0, 255); |
234 |
dst1[6 * dst1_stride] = CLIP((((src6 + src7) * 160 - (src5 + src8) * 48 + (src4 + src8) * 24 - (src3 + src7) * 8 + (128 - rounding)) / 256), 0, 255); |
dst[2 * stride] = CLIP((((src0 + src5) + 5 * (((src2 + src3)<<2) - (src1 + src4)) + round_add) >> 5), 0, 255); |
235 |
dst1[7 * dst1_stride] = CLIP((((src7 + src8) * 160 - (src6 + src8) * 48 + (src5 + src7) * 24 - (src4 + src6) * 8 + (128 - rounding)) / 256), 0, 255); |
dst[3 * stride] = CLIP((((src1 + src6) + 5 * (((src3 + src4)<<2) - (src2 + src5)) + round_add) >> 5), 0, 255); |
236 |
|
dst[4 * stride] = CLIP((((src2 + src7) + 5 * (((src4 + src5)<<2) - (src3 + src6)) + round_add) >> 5), 0, 255); |
237 |
|
dst[5 * stride] = CLIP((((src3 + src8) + 5 * (((src5 + src6)<<2) - (src4 + src7)) + round_add) >> 5), 0, 255); |
238 |
|
dst[6 * stride] = CLIP((((src4 + src9) + 5 * (((src6 + src7)<<2) - (src5 + src8)) + round_add) >> 5), 0, 255); |
239 |
|
dst[7 * stride] = CLIP((((src5 + src10) + 5 * (((src7 + src8)<<2) - (src6 + src9)) + round_add) >> 5), 0, 255); |
240 |
|
|
241 |
dst1++; |
dst++; |
242 |
|
src++; |
243 |
} |
} |
244 |
} |
} |
245 |
|
|
246 |
void interpolate8x8_bilinear2(uint8_t *dst, uint8_t *src1, uint8_t *src2, int32_t dst_stride, int32_t src_stride, int32_t rounding) |
void interpolate8x8_lowpass_v_c(uint8_t *dst, uint8_t *src, int32_t stride, int32_t rounding) |
247 |
{ |
{ |
248 |
int32_t i; |
int32_t i; |
249 |
|
uint8_t round_add = 16 - rounding; |
250 |
|
|
251 |
for(i = 0; i < 8; i++) |
for(i = 0; i < 8; i++) |
252 |
{ |
{ |
253 |
dst[0] = (src1[0] + src2[0] + (1 - rounding)) >> 1; |
int32_t src0 = src[0]; |
254 |
dst[1] = (src1[1] + src2[1] + (1 - rounding)) >> 1; |
int32_t src1 = src[stride]; |
255 |
dst[2] = (src1[2] + src2[2] + (1 - rounding)) >> 1; |
int32_t src2 = src[2 * stride]; |
256 |
dst[3] = (src1[3] + src2[3] + (1 - rounding)) >> 1; |
int32_t src3 = src[3 * stride]; |
257 |
dst[4] = (src1[4] + src2[4] + (1 - rounding)) >> 1; |
int32_t src4 = src[4 * stride]; |
258 |
dst[5] = (src1[5] + src2[5] + (1 - rounding)) >> 1; |
int32_t src5 = src[5 * stride]; |
259 |
dst[6] = (src1[6] + src2[6] + (1 - rounding)) >> 1; |
int32_t src6 = src[6 * stride]; |
260 |
dst[7] = (src1[7] + src2[7] + (1 - rounding)) >> 1; |
int32_t src7 = src[7 * stride]; |
261 |
|
int32_t src8 = src[8 * stride]; |
262 |
|
|
263 |
|
dst[0] = CLIP(((7 * ((src0<<1) - src2) + 23 * src1 + 3 * src3 - src4 + round_add) >> 5), 0, 255); |
264 |
|
dst[stride] = CLIP(((19 * src1 + 20 * src2 - src5 + 3 * (src4 - src0 - (src3 << 1)) + round_add) >> 5), 0, 255); |
265 |
|
dst[2 * stride] = CLIP(((20 * (src2 + src3) + (src0<<1) + 3 * (src5 - ((src1 + src4) <<1 )) - src6 + round_add) >> 5), 0, 255); |
266 |
|
dst[3 * stride] = CLIP(((20 * (src3 + src4) + 3 * ((src6 + src1) - ((src2 + src5)<<1)) - (src0 + src7) + round_add) >> 5), 0, 255); |
267 |
|
dst[4 * stride] = CLIP(((20 * (src4 + src5) + 3 * ((src2 + src7) - ((src3 + src6)<<1)) - (src1 + src8) + round_add) >> 5), 0, 255); |
268 |
|
dst[5 * stride] = CLIP(((20 * (src5 + src6) + (src8<<1) + 3 * (src3 - ((src4 + src7) << 1)) - src2 + round_add) >> 5), 0, 255); |
269 |
|
dst[6 * stride] = CLIP(((19 * src7 + 20 * src6 - src3 + 3 * (src4 - src8 - (src5 << 1)) + round_add) >> 5), 0, 255); |
270 |
|
dst[7 * stride] = CLIP(((7 * ((src8<<1) - src6) + 23 * src7 + 3 * src5 - src4 + round_add) >> 5), 0, 255); |
271 |
|
|
272 |
dst += dst_stride; |
dst++; |
273 |
src1 += src_stride; |
src++; |
|
src2 += 8; |
|
274 |
} |
} |
275 |
} |
} |
276 |
|
|
277 |
void interpolate8x8_bilinear4(uint8_t *dst, uint8_t *src1, uint8_t *src2, uint8_t *src3, uint8_t *src4, int32_t stride, int32_t rounding) |
void interpolate8x8_lowpass_hv_c(uint8_t *dst1, uint8_t *dst2, uint8_t *src, int32_t stride, int32_t rounding) |
278 |
{ |
{ |
279 |
int32_t i; |
int32_t i; |
280 |
|
uint8_t round_add = 16 - rounding; |
281 |
|
uint8_t *h_ptr = dst2; |
282 |
|
|
283 |
for(i = 0; i < 8; i++) |
for(i = 0; i < 9; i++) |
284 |
{ |
{ |
|
dst[0] = (src1[0] + src2[0] + src3[0] + src4[0] + (2 - rounding)) >> 2; |
|
|
dst[1] = (src1[1] + src2[1] + src3[1] + src4[1] + (2 - rounding)) >> 2; |
|
|
dst[2] = (src1[2] + src2[2] + src3[2] + src4[2] + (2 - rounding)) >> 2; |
|
|
dst[3] = (src1[3] + src2[3] + src3[3] + src4[3] + (2 - rounding)) >> 2; |
|
|
dst[4] = (src1[4] + src2[4] + src3[4] + src4[4] + (2 - rounding)) >> 2; |
|
|
dst[5] = (src1[5] + src2[5] + src3[5] + src4[5] + (2 - rounding)) >> 2; |
|
|
dst[6] = (src1[6] + src2[6] + src3[6] + src4[6] + (2 - rounding)) >> 2; |
|
|
dst[7] = (src1[7] + src2[7] + src3[7] + src4[7] + (2 - rounding)) >> 2; |
|
285 |
|
|
286 |
dst += stride; |
h_ptr[0] = CLIP(((7 * ((src[0]<<1) - src[2]) + 23 * src[1] + 3 * src[3] - src[4] + round_add) >> 5), 0, 255); |
287 |
src1 += stride; |
h_ptr[1] = CLIP(((19 * src[1] + 20 * src[2] - src[5] + 3 * (src[4] - src[0] - (src[3]<<1)) + round_add) >> 5), 0, 255); |
288 |
src2 += 8; |
h_ptr[2] = CLIP(((20 * (src[2] + src[3]) + (src[0]<<1) + 3 * (src[5] - ((src[1] + src[4])<<1)) - src[6] + round_add) >> 5), 0, 255); |
289 |
src3 += 8; |
h_ptr[3] = CLIP(((20 * (src[3] + src[4]) + 3 * ((src[6] + src[1]) - ((src[2] + src[5])<<1)) - (src[0] + src[7]) + round_add) >> 5), 0, 255); |
290 |
src4 += 8; |
h_ptr[4] = CLIP(((20 * (src[4] + src[5]) - 3 * (((src[3] + src[6])<<1) - (src[2] + src[7])) - (src[1] + src[8]) + round_add) >> 5), 0, 255); |
291 |
|
h_ptr[5] = CLIP(((20 * (src[5] + src[6]) + (src[8]<<1) + 3 * (src[3] - ((src[4] + src[7]) << 1)) - src[2] + round_add) >> 5), 0, 255); |
292 |
|
h_ptr[6] = CLIP(((19 * src[7] + 20 * src[6] + 3 * (src[4] - src[8] - (src[5] << 1)) - src[3] + round_add) >> 5), 0, 255); |
293 |
|
h_ptr[7] = CLIP(((23 * src[7] + 7 * ((src[8]<<1) - src[6]) + 3 * src[5] - src[4] + round_add) >> 5), 0, 255); |
294 |
|
|
295 |
|
h_ptr += stride; |
296 |
|
src += stride; |
297 |
} |
} |
298 |
|
|
299 |
|
interpolate8x8_lowpass_v_c(dst1, dst2, stride, rounding); |
300 |
|
|
301 |
} |
} |