1cabdff1aSopenharmony_ci/*
2cabdff1aSopenharmony_ci * RV30/40 decoder common dsp functions
3cabdff1aSopenharmony_ci * Copyright (c) 2007 Mike Melanson, Konstantin Shishkov
4cabdff1aSopenharmony_ci * Copyright (c) 2011 Janne Grunau
5cabdff1aSopenharmony_ci *
6cabdff1aSopenharmony_ci * This file is part of FFmpeg.
7cabdff1aSopenharmony_ci *
8cabdff1aSopenharmony_ci * FFmpeg is free software; you can redistribute it and/or
9cabdff1aSopenharmony_ci * modify it under the terms of the GNU Lesser General Public
10cabdff1aSopenharmony_ci * License as published by the Free Software Foundation; either
11cabdff1aSopenharmony_ci * version 2.1 of the License, or (at your option) any later version.
12cabdff1aSopenharmony_ci *
13cabdff1aSopenharmony_ci * FFmpeg is distributed in the hope that it will be useful,
14cabdff1aSopenharmony_ci * but WITHOUT ANY WARRANTY; without even the implied warranty of
15cabdff1aSopenharmony_ci * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
16cabdff1aSopenharmony_ci * Lesser General Public License for more details.
17cabdff1aSopenharmony_ci *
18cabdff1aSopenharmony_ci * You should have received a copy of the GNU Lesser General Public
19cabdff1aSopenharmony_ci * License along with FFmpeg; if not, write to the Free Software
20cabdff1aSopenharmony_ci * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
21cabdff1aSopenharmony_ci */
22cabdff1aSopenharmony_ci
23cabdff1aSopenharmony_ci/**
24cabdff1aSopenharmony_ci * @file
25cabdff1aSopenharmony_ci * RV30/40 decoder common dsp functions
26cabdff1aSopenharmony_ci */
27cabdff1aSopenharmony_ci
28cabdff1aSopenharmony_ci#include "rv34dsp.h"
29cabdff1aSopenharmony_ci#include "libavutil/common.h"
30cabdff1aSopenharmony_ci
31cabdff1aSopenharmony_ci/**
32cabdff1aSopenharmony_ci * @name RV30/40 inverse transform functions
33cabdff1aSopenharmony_ci * @{
34cabdff1aSopenharmony_ci */
35cabdff1aSopenharmony_ci
36cabdff1aSopenharmony_cistatic av_always_inline void rv34_row_transform(int temp[16], int16_t *block)
37cabdff1aSopenharmony_ci{
38cabdff1aSopenharmony_ci    int i;
39cabdff1aSopenharmony_ci
40cabdff1aSopenharmony_ci    for(i = 0; i < 4; i++){
41cabdff1aSopenharmony_ci        const int z0 = 13*(block[i+4*0] +    block[i+4*2]);
42cabdff1aSopenharmony_ci        const int z1 = 13*(block[i+4*0] -    block[i+4*2]);
43cabdff1aSopenharmony_ci        const int z2 =  7* block[i+4*1] - 17*block[i+4*3];
44cabdff1aSopenharmony_ci        const int z3 = 17* block[i+4*1] +  7*block[i+4*3];
45cabdff1aSopenharmony_ci
46cabdff1aSopenharmony_ci        temp[4*i+0] = z0 + z3;
47cabdff1aSopenharmony_ci        temp[4*i+1] = z1 + z2;
48cabdff1aSopenharmony_ci        temp[4*i+2] = z1 - z2;
49cabdff1aSopenharmony_ci        temp[4*i+3] = z0 - z3;
50cabdff1aSopenharmony_ci    }
51cabdff1aSopenharmony_ci}
52cabdff1aSopenharmony_ci
53cabdff1aSopenharmony_ci/**
54cabdff1aSopenharmony_ci * Real Video 3.0/4.0 inverse transform + sample reconstruction
55cabdff1aSopenharmony_ci * Code is almost the same as in SVQ3, only scaling is different.
56cabdff1aSopenharmony_ci */
57cabdff1aSopenharmony_cistatic void rv34_idct_add_c(uint8_t *dst, ptrdiff_t stride, int16_t *block){
58cabdff1aSopenharmony_ci    int      temp[16];
59cabdff1aSopenharmony_ci    int      i;
60cabdff1aSopenharmony_ci
61cabdff1aSopenharmony_ci    rv34_row_transform(temp, block);
62cabdff1aSopenharmony_ci    memset(block, 0, 16*sizeof(int16_t));
63cabdff1aSopenharmony_ci
64cabdff1aSopenharmony_ci    for(i = 0; i < 4; i++){
65cabdff1aSopenharmony_ci        const int z0 = 13*(temp[4*0+i] +    temp[4*2+i]) + 0x200;
66cabdff1aSopenharmony_ci        const int z1 = 13*(temp[4*0+i] -    temp[4*2+i]) + 0x200;
67cabdff1aSopenharmony_ci        const int z2 =  7* temp[4*1+i] - 17*temp[4*3+i];
68cabdff1aSopenharmony_ci        const int z3 = 17* temp[4*1+i] +  7*temp[4*3+i];
69cabdff1aSopenharmony_ci
70cabdff1aSopenharmony_ci        dst[0] = av_clip_uint8( dst[0] + ( (z0 + z3) >> 10 ) );
71cabdff1aSopenharmony_ci        dst[1] = av_clip_uint8( dst[1] + ( (z1 + z2) >> 10 ) );
72cabdff1aSopenharmony_ci        dst[2] = av_clip_uint8( dst[2] + ( (z1 - z2) >> 10 ) );
73cabdff1aSopenharmony_ci        dst[3] = av_clip_uint8( dst[3] + ( (z0 - z3) >> 10 ) );
74cabdff1aSopenharmony_ci
75cabdff1aSopenharmony_ci        dst  += stride;
76cabdff1aSopenharmony_ci    }
77cabdff1aSopenharmony_ci}
78cabdff1aSopenharmony_ci
79cabdff1aSopenharmony_ci/**
80cabdff1aSopenharmony_ci * RealVideo 3.0/4.0 inverse transform for DC block
81cabdff1aSopenharmony_ci *
82cabdff1aSopenharmony_ci * Code is almost the same as rv34_inv_transform()
83cabdff1aSopenharmony_ci * but final coefficients are multiplied by 1.5 and have no rounding.
84cabdff1aSopenharmony_ci */
85cabdff1aSopenharmony_cistatic void rv34_inv_transform_noround_c(int16_t *block){
86cabdff1aSopenharmony_ci    int temp[16];
87cabdff1aSopenharmony_ci    int i;
88cabdff1aSopenharmony_ci
89cabdff1aSopenharmony_ci    rv34_row_transform(temp, block);
90cabdff1aSopenharmony_ci
91cabdff1aSopenharmony_ci    for(i = 0; i < 4; i++){
92cabdff1aSopenharmony_ci        const int z0 = 39*(temp[4*0+i] +    temp[4*2+i]);
93cabdff1aSopenharmony_ci        const int z1 = 39*(temp[4*0+i] -    temp[4*2+i]);
94cabdff1aSopenharmony_ci        const int z2 = 21* temp[4*1+i] - 51*temp[4*3+i];
95cabdff1aSopenharmony_ci        const int z3 = 51* temp[4*1+i] + 21*temp[4*3+i];
96cabdff1aSopenharmony_ci
97cabdff1aSopenharmony_ci        block[i*4+0] = (z0 + z3) >> 11;
98cabdff1aSopenharmony_ci        block[i*4+1] = (z1 + z2) >> 11;
99cabdff1aSopenharmony_ci        block[i*4+2] = (z1 - z2) >> 11;
100cabdff1aSopenharmony_ci        block[i*4+3] = (z0 - z3) >> 11;
101cabdff1aSopenharmony_ci    }
102cabdff1aSopenharmony_ci}
103cabdff1aSopenharmony_ci
104cabdff1aSopenharmony_cistatic void rv34_idct_dc_add_c(uint8_t *dst, ptrdiff_t stride, int dc)
105cabdff1aSopenharmony_ci{
106cabdff1aSopenharmony_ci    int i, j;
107cabdff1aSopenharmony_ci
108cabdff1aSopenharmony_ci    dc = (13*13*dc + 0x200) >> 10;
109cabdff1aSopenharmony_ci    for (i = 0; i < 4; i++)
110cabdff1aSopenharmony_ci    {
111cabdff1aSopenharmony_ci        for (j = 0; j < 4; j++)
112cabdff1aSopenharmony_ci            dst[j] = av_clip_uint8( dst[j] + dc );
113cabdff1aSopenharmony_ci
114cabdff1aSopenharmony_ci        dst += stride;
115cabdff1aSopenharmony_ci    }
116cabdff1aSopenharmony_ci}
117cabdff1aSopenharmony_ci
118cabdff1aSopenharmony_cistatic void rv34_inv_transform_dc_noround_c(int16_t *block)
119cabdff1aSopenharmony_ci{
120cabdff1aSopenharmony_ci    int16_t dc = (13 * 13 * 3 * block[0]) >> 11;
121cabdff1aSopenharmony_ci    int i, j;
122cabdff1aSopenharmony_ci
123cabdff1aSopenharmony_ci    for (i = 0; i < 4; i++, block += 4)
124cabdff1aSopenharmony_ci        for (j = 0; j < 4; j++)
125cabdff1aSopenharmony_ci            block[j] = dc;
126cabdff1aSopenharmony_ci}
127cabdff1aSopenharmony_ci
128cabdff1aSopenharmony_ci/** @} */ // transform
129cabdff1aSopenharmony_ci
130cabdff1aSopenharmony_ci
131cabdff1aSopenharmony_ciav_cold void ff_rv34dsp_init(RV34DSPContext *c)
132cabdff1aSopenharmony_ci{
133cabdff1aSopenharmony_ci    c->rv34_inv_transform    = rv34_inv_transform_noround_c;
134cabdff1aSopenharmony_ci    c->rv34_inv_transform_dc = rv34_inv_transform_dc_noround_c;
135cabdff1aSopenharmony_ci
136cabdff1aSopenharmony_ci    c->rv34_idct_add    = rv34_idct_add_c;
137cabdff1aSopenharmony_ci    c->rv34_idct_dc_add = rv34_idct_dc_add_c;
138cabdff1aSopenharmony_ci
139cabdff1aSopenharmony_ci#if ARCH_ARM
140cabdff1aSopenharmony_ci    ff_rv34dsp_init_arm(c);
141cabdff1aSopenharmony_ci#elif ARCH_X86
142cabdff1aSopenharmony_ci    ff_rv34dsp_init_x86(c);
143cabdff1aSopenharmony_ci#endif
144cabdff1aSopenharmony_ci}
145