1/*
2 *  Copyright (c) 2010 The WebM project authors. All Rights Reserved.
3 *
4 *  Use of this source code is governed by a BSD-style license
5 *  that can be found in the LICENSE file in the root of the source
6 *  tree. An additional intellectual property rights grant can be found
7 *  in the file PATENTS.  All contributing project authors may
8 *  be found in the AUTHORS file in the root of the source tree.
9 */
10
11#include "./vp8_rtcd.h"
12
13/****************************************************************************
14 * Notes:
15 *
16 * This implementation makes use of 16 bit fixed point verio of two multiply
17 * constants:
18 *         1.   sqrt(2) * cos (pi/8)
19 *         2.   sqrt(2) * sin (pi/8)
20 * Becuase the first constant is bigger than 1, to maintain the same 16 bit
21 * fixed point precision as the second one, we use a trick of
22 *         x * a = x + x*(a-1)
23 * so
24 *         x * sqrt(2) * cos (pi/8) = x + x * (sqrt(2) *cos(pi/8)-1).
25 **************************************************************************/
26static const int cospi8sqrt2minus1 = 20091;
27static const int sinpi8sqrt2      = 35468;
28
29void vp8_short_idct4x4llm_c(short *input, unsigned char *pred_ptr,
30                            int pred_stride, unsigned char *dst_ptr,
31                            int dst_stride)
32{
33    int i;
34    int r, c;
35    int a1, b1, c1, d1;
36    short output[16];
37    short *ip = input;
38    short *op = output;
39    int temp1, temp2;
40    int shortpitch = 4;
41
42    for (i = 0; i < 4; i++)
43    {
44        a1 = ip[0] + ip[8];
45        b1 = ip[0] - ip[8];
46
47        temp1 = (ip[4] * sinpi8sqrt2) >> 16;
48        temp2 = ip[12] + ((ip[12] * cospi8sqrt2minus1) >> 16);
49        c1 = temp1 - temp2;
50
51        temp1 = ip[4] + ((ip[4] * cospi8sqrt2minus1) >> 16);
52        temp2 = (ip[12] * sinpi8sqrt2) >> 16;
53        d1 = temp1 + temp2;
54
55        op[shortpitch*0] = a1 + d1;
56        op[shortpitch*3] = a1 - d1;
57
58        op[shortpitch*1] = b1 + c1;
59        op[shortpitch*2] = b1 - c1;
60
61        ip++;
62        op++;
63    }
64
65    ip = output;
66    op = output;
67
68    for (i = 0; i < 4; i++)
69    {
70        a1 = ip[0] + ip[2];
71        b1 = ip[0] - ip[2];
72
73        temp1 = (ip[1] * sinpi8sqrt2) >> 16;
74        temp2 = ip[3] + ((ip[3] * cospi8sqrt2minus1) >> 16);
75        c1 = temp1 - temp2;
76
77        temp1 = ip[1] + ((ip[1] * cospi8sqrt2minus1) >> 16);
78        temp2 = (ip[3] * sinpi8sqrt2) >> 16;
79        d1 = temp1 + temp2;
80
81
82        op[0] = (a1 + d1 + 4) >> 3;
83        op[3] = (a1 - d1 + 4) >> 3;
84
85        op[1] = (b1 + c1 + 4) >> 3;
86        op[2] = (b1 - c1 + 4) >> 3;
87
88        ip += shortpitch;
89        op += shortpitch;
90    }
91
92    ip = output;
93    for (r = 0; r < 4; r++)
94    {
95        for (c = 0; c < 4; c++)
96        {
97            int a = ip[c] + pred_ptr[c] ;
98
99            if (a < 0)
100                a = 0;
101
102            if (a > 255)
103                a = 255;
104
105            dst_ptr[c] = (unsigned char) a ;
106        }
107        ip += 4;
108        dst_ptr += dst_stride;
109        pred_ptr += pred_stride;
110    }
111}
112
113void vp8_dc_only_idct_add_c(short input_dc, unsigned char *pred_ptr,
114                            int pred_stride, unsigned char *dst_ptr,
115                            int dst_stride)
116{
117    int a1 = ((input_dc + 4) >> 3);
118    int r, c;
119
120    for (r = 0; r < 4; r++)
121    {
122        for (c = 0; c < 4; c++)
123        {
124            int a = a1 + pred_ptr[c] ;
125
126            if (a < 0)
127                a = 0;
128
129            if (a > 255)
130                a = 255;
131
132            dst_ptr[c] = (unsigned char) a ;
133        }
134
135        dst_ptr += dst_stride;
136        pred_ptr += pred_stride;
137    }
138
139}
140
141void vp8_short_inv_walsh4x4_c(short *input, short *mb_dqcoeff)
142{
143    short output[16];
144    int i;
145    int a1, b1, c1, d1;
146    int a2, b2, c2, d2;
147    short *ip = input;
148    short *op = output;
149
150    for (i = 0; i < 4; i++)
151    {
152        a1 = ip[0] + ip[12];
153        b1 = ip[4] + ip[8];
154        c1 = ip[4] - ip[8];
155        d1 = ip[0] - ip[12];
156
157        op[0] = a1 + b1;
158        op[4] = c1 + d1;
159        op[8] = a1 - b1;
160        op[12] = d1 - c1;
161        ip++;
162        op++;
163    }
164
165    ip = output;
166    op = output;
167
168    for (i = 0; i < 4; i++)
169    {
170        a1 = ip[0] + ip[3];
171        b1 = ip[1] + ip[2];
172        c1 = ip[1] - ip[2];
173        d1 = ip[0] - ip[3];
174
175        a2 = a1 + b1;
176        b2 = c1 + d1;
177        c2 = a1 - b1;
178        d2 = d1 - c1;
179
180        op[0] = (a2 + 3) >> 3;
181        op[1] = (b2 + 3) >> 3;
182        op[2] = (c2 + 3) >> 3;
183        op[3] = (d2 + 3) >> 3;
184
185        ip += 4;
186        op += 4;
187    }
188
189    for(i = 0; i < 16; i++)
190    {
191        mb_dqcoeff[i * 16] = output[i];
192    }
193}
194
195void vp8_short_inv_walsh4x4_1_c(short *input, short *mb_dqcoeff)
196{
197    int i;
198    int a1;
199
200    a1 = ((input[0] + 3) >> 3);
201    for(i = 0; i < 16; i++)
202    {
203        mb_dqcoeff[i * 16] = a1;
204    }
205}
206