dany/libwebp
1
0
Fork 0
mirror of https://github.com/webmproject/libwebp.git synced 2025-07-14 21:09:55 +02:00
Code Issues Actions Packages Projects Releases Wiki Activity

move the decoder sourcetree to a sub-location src/dec

Browse Source 
to make room for future libs sources

also extract the types declaration into its own types.h file

Change-Id: I8bae8a323a479a29375cf509792228ae6af51c7a
This commit is contained in:
Pascal Massimino
2011-01-06 07:25:43 -08:00
parent a9b3eab6fd
commit 6421a7a4fb
22 changed files with 17747 additions and 23160 deletions
Download Patch File Download Diff File Expand all files Collapse all files

78
src/dec/bits.c Normal file
View File

@ -0,0 +1,78 @@
// Copyright 2010 Google Inc.
//
// This code is licensed under the same terms as WebM:
// Software License Agreement: http://www.webmproject.org/license/software/
// Additional IP Rights Grant: http://www.webmproject.org/license/additional/
// -----------------------------------------------------------------------------
//
// Boolean decoder
//
// Author: Skal (pascal.massimino@gmail.com)
#include "bits.h"
#if defined(__cplusplus) || defined(c_plusplus)
extern "C" {
#endif
//-----------------------------------------------------------------------------
// VP8BitReader
void VP8Init(VP8BitReader* const br, const uint8_t* buf, uint32_t size) {
assert(br);
assert(buf);
br->range_ = 255 - 1;
br->eof_ = 0;
br->buf_ = buf;
br->buf_end_ = buf + size;
// Need two initial bytes.
br->value_ = (VP8GetByte(br) << 8) | VP8GetByte(br);
br->left_ = -8;
}
const uint8_t kVP8Log2Range[128] = {
7, 6, 6, 5, 5, 5, 5, 4, 4, 4, 4, 4, 4, 4, 4,
3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,
2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
0
};
// range = ((range + 1) << kVP8Log2Range[range]) - 1
const uint8_t kVP8NewRange[128] = {
127, 127, 191, 127, 159, 191, 223, 127, 143, 159, 175, 191, 207, 223, 239,
127, 135, 143, 151, 159, 167, 175, 183, 191, 199, 207, 215, 223, 231, 239,
247, 127, 131, 135, 139, 143, 147, 151, 155, 159, 163, 167, 171, 175, 179,
183, 187, 191, 195, 199, 203, 207, 211, 215, 219, 223, 227, 231, 235, 239,
243, 247, 251, 127, 129, 131, 133, 135, 137, 139, 141, 143, 145, 147, 149,
151, 153, 155, 157, 159, 161, 163, 165, 167, 169, 171, 173, 175, 177, 179,
181, 183, 185, 187, 189, 191, 193, 195, 197, 199, 201, 203, 205, 207, 209,
211, 213, 215, 217, 219, 221, 223, 225, 227, 229, 231, 233, 235, 237, 239,
241, 243, 245, 247, 249, 251, 253, 127
};
//-----------------------------------------------------------------------------
// Higher-level calls
uint32_t VP8GetValue(VP8BitReader* const br, int bits) {
uint32_t v = 0;
while (bits-- > 0) {
v |= VP8GetBit(br, 0x80) << bits;
}
return v;
}
int32_t VP8GetSignedValue(VP8BitReader* const br, int bits) {
const int value = VP8GetValue(br, bits);
return VP8Get(br) ? -value : value;
}
//-----------------------------------------------------------------------------
#if defined(__cplusplus) || defined(c_plusplus)
} // extern "C"
#endif

106
src/dec/bits.h Normal file
View File

@ -0,0 +1,106 @@
// Copyright 2010 Google Inc.
//
// This code is licensed under the same terms as WebM:
// Software License Agreement: http://www.webmproject.org/license/software/
// Additional IP Rights Grant: http://www.webmproject.org/license/additional/
// -----------------------------------------------------------------------------
//
// Boolean decoder
//
// Author: Skal (pascal.massimino@gmail.com)
#ifndef WEBP_DECODE_BITS_H_
#define WEBP_DECODE_BITS_H_
#include <assert.h>
#include "webp/decode_vp8.h"
#if defined(__cplusplus) || defined(c_plusplus)
extern "C" {
#endif
//-----------------------------------------------------------------------------
// Bitreader and code-tree reader
typedef struct {
const uint8_t* buf_; // next byte to be read
const uint8_t* buf_end_; // end of read buffer
int eof_; // true if input is exhausted
// boolean decoder
uint32_t range_; // current range minus 1. In [127, 254] interval.
uint32_t value_; // current value
int left_; // how many unused bits (negated)
} VP8BitReader;
// Initialize the bit reader and the boolean decoder.
void VP8Init(VP8BitReader* const br, const uint8_t* buf, uint32_t size);
// return the next value made of 'num_bits' bits
uint32_t VP8GetValue(VP8BitReader* const br, int num_bits);
static inline uint32_t VP8Get(VP8BitReader* const br) {
return VP8GetValue(br, 1);
}
// return the next value with sign-extension.
int32_t VP8GetSignedValue(VP8BitReader* const br, int num_bits);
// Read a bit with proba 'prob'. Speed-critical function!
extern const uint8_t kVP8Log2Range[128];
extern const uint8_t kVP8NewRange[128];
static inline uint32_t VP8GetByte(VP8BitReader* const br) {
assert(br);
if (br->buf_ < br->buf_end_) {
assert(br->buf_);
return *br->buf_++;
}
br->eof_ = 1;
return 0x80;
}
static inline void VP8Shift(VP8BitReader* const br) {
// range_ is in [0..127] interval here.
const int shift = kVP8Log2Range[br->range_];
br->range_ = kVP8NewRange[br->range_];
br->value_ <<= shift;
br->left_ += shift;
if (br->left_ > 0) {
br->value_ |= VP8GetByte(br) << br->left_;
br->left_ -= 8;
}
}
static inline uint32_t VP8GetBit(VP8BitReader* const br, int prob) {
const uint32_t split = (br->range_ * prob) >> 8;
const uint32_t bit = ((br->value_ >> 8) > split);
if (bit) {
br->range_ -= split + 1;
br->value_ -= (split + 1) << 8;
} else {
br->range_ = split;
}
if (br->range_ < 0x7f) {
VP8Shift(br);
}
return bit;
}
static inline int VP8GetSigned(VP8BitReader* const br, int v) {
const uint32_t split = br->range_ >> 1;
const uint32_t bit = ((br->value_ >> 8) > split);
if (bit) {
br->range_ -= split + 1;
br->value_ -= (split + 1) << 8;
v = -v;
} else {
br->range_ = split;
}
VP8Shift(br);
return v;
}
#if defined(__cplusplus) || defined(c_plusplus)
} // extern "C"
#endif
#endif // WEBP_DECODE_BITS_H_

694
src/dec/dsp.c Normal file
View File

@ -0,0 +1,694 @@
// Copyright 2010 Google Inc.
//
// This code is licensed under the same terms as WebM:
// Software License Agreement: http://www.webmproject.org/license/software/
// Additional IP Rights Grant: http://www.webmproject.org/license/additional/
// -----------------------------------------------------------------------------
//
// speed-critical functions.
//
// Author: Skal (pascal.massimino@gmail.com)
#include "vp8i.h"
#if defined(__SSE2__)
#include <emmintrin.h>
#endif
#if defined(__cplusplus) || defined(c_plusplus)
extern "C" {
#endif
//-----------------------------------------------------------------------------
// run-time tables (~4k)
static uint8_t abs0[255 + 255 + 1]; // abs(i)
static uint8_t abs1[255 + 255 + 1]; // abs(i)>>1
static int8_t sclip1[1020 + 1020 + 1]; // clips [-1020, 1020] to [-128, 127]
static int8_t sclip2[112 + 112 + 1]; // clips [-112, 112] to [-16, 15]
static uint8_t clip1[255 + 510 + 1]; // clips [-255,510] to [0,255]
static int tables_ok = 0;
void VP8DspInitTables() {
if (!tables_ok) {
int i;
for (i = -255; i <= 255; ++i) {
abs0[255 + i] = (i < 0) ? -i : i;
abs1[255 + i] = abs0[255 + i] >> 1;
}
for (i = -1020; i <= 1020; ++i) {
sclip1[1020 + i] = (i < -128) ? -128 : (i > 127) ? 127 : i;
}
for (i = -112; i <= 112; ++i) {
sclip2[112 + i] = (i < -16) ? -16 : (i > 15) ? 15 : i;
}
for (i = -255; i <= 255 + 255; ++i) {
clip1[255 + i] = (i < 0) ? 0 : (i > 255) ? 255 : i;
}
tables_ok = 1;
}
}
static inline uint8_t clip_8b(int v) {
return (!(v & ~0xff)) ? v : (v < 0) ? 0 : 255;
}
//-----------------------------------------------------------------------------
// Transforms (Paragraph 14.4)
#define STORE(x, y, v) \
dst[x + y * BPS] = clip_8b(dst[x + y * BPS] + ((v) >> 3))
static const int kC1 = 20091 + (1 << 16);
static const int kC2 = 35468;
#define MUL(a, b) (((a) * (b)) >> 16)
static void Transform(const int16_t* in, uint8_t* dst) {
int C[4 * 4], *tmp;
int i;
tmp = C;
for (i = 0; i < 4; ++i) { // vertical pass
const int a = in[0] + in[8]; // [-4096, 4094]
const int b = in[0] - in[8]; // [-4095, 4095]
const int c = MUL(in[4], kC2) - MUL(in[12], kC1); // [-3783, 3783]
const int d = MUL(in[4], kC1) + MUL(in[12], kC2); // [-3785, 3781]
tmp[0] = a + d; // [-7881, 7875]
tmp[1] = b + c; // [-7878, 7878]
tmp[2] = b - c; // [-7878, 7878]
tmp[3] = a - d; // [-7877, 7879]
tmp += 4;
in++;
}
// Each pass is expanding the dynamic range by ~3.85 (upper bound).
// The exact value is (2. + (kC1 + kC2) / 65536).
// After the second pass, maximum interval is [-3794, 3794], assuming
// an input in [-2048, 2047] interval. We then need to add a dst value
// in the [0, 255] range.
// In the worst case scenario, the input to clip_8b() can be as large as
// [-60713, 60968].
tmp = C;
for (i = 0; i < 4; ++i) { // horizontal pass
const int dc = tmp[0] + 4;
const int a = dc + tmp[8];
const int b = dc - tmp[8];
const int c = MUL(tmp[4], kC2) - MUL(tmp[12], kC1);
const int d = MUL(tmp[4], kC1) + MUL(tmp[12], kC2);
STORE(0, 0, a + d);
STORE(1, 0, b + c);
STORE(2, 0, b - c);
STORE(3, 0, a - d);
tmp++;
dst += BPS;
}
}
#undef MUL
static void TransformUV(const int16_t* in, uint8_t* dst) {
Transform(in + 0 * 16, dst);
Transform(in + 1 * 16, dst + 4);
Transform(in + 2 * 16, dst + 4 * BPS);
Transform(in + 3 * 16, dst + 4 * BPS + 4);
}
static void TransformDC(const int16_t *in, uint8_t* dst) {
const int DC = in[0] + 4;
int i, j;
for (j = 0; j < 4; ++j) {
for (i = 0; i < 4; ++i) {
STORE(i, j, DC);
}
}
}
static void TransformDCUV(const int16_t* in, uint8_t* dst) {
if (in[0 * 16]) TransformDC(in + 0 * 16, dst);
if (in[1 * 16]) TransformDC(in + 1 * 16, dst + 4);
if (in[2 * 16]) TransformDC(in + 2 * 16, dst + 4 * BPS);
if (in[3 * 16]) TransformDC(in + 3 * 16, dst + 4 * BPS + 4);
}
#undef STORE
// default C implementations:
VP8Idct VP8Transform = Transform;
VP8Idct VP8TransformUV = TransformUV;
VP8Idct VP8TransformDC = TransformDC;
VP8Idct VP8TransformDCUV = TransformDCUV;
//-----------------------------------------------------------------------------
// Paragraph 14.3
static void TransformWHT(const int16_t* in, int16_t* out) {
int tmp[16];
int i;
for (i = 0; i < 4; ++i) {
const int a0 = in[0 + i] + in[12 + i];
const int a1 = in[4 + i] + in[ 8 + i];
const int a2 = in[4 + i] - in[ 8 + i];
const int a3 = in[0 + i] - in[12 + i];
tmp[0 + i] = a0 + a1;
tmp[8 + i] = a0 - a1;
tmp[4 + i] = a3 + a2;
tmp[12 + i] = a3 - a2;
}
for (i = 0; i < 4; ++i) {
const int dc = tmp[0 + i * 4] + 3; // w/ rounder
const int a0 = dc + tmp[3 + i * 4];
const int a1 = tmp[1 + i * 4] + tmp[2 + i * 4];
const int a2 = tmp[1 + i * 4] - tmp[2 + i * 4];
const int a3 = dc - tmp[3 + i * 4];
out[ 0] = (a0 + a1) >> 3;
out[16] = (a3 + a2) >> 3;
out[32] = (a0 - a1) >> 3;
out[48] = (a3 - a2) >> 3;
out += 64;
}
}
void (*VP8TransformWHT)(const int16_t* in, int16_t* out) = TransformWHT;
//-----------------------------------------------------------------------------
// Intra predictions
#define OUT(x, y) dst[(x) + (y) * BPS]
static inline void TrueMotion(uint8_t *dst, int size) {
const uint8_t* top = dst - BPS;
const uint8_t* const clip0 = clip1 + 255 - top[-1];
int y;
for (y = 0; y < size; ++y) {
const uint8_t* const clip = clip0 + dst[-1];
int x;
for (x = 0; x < size; ++x) {
dst[x] = clip[top[x]];
}
dst += BPS;
}
}
static void TM4(uint8_t *dst) { TrueMotion(dst, 4); }
static void TM8uv(uint8_t *dst) { TrueMotion(dst, 8); }
static void TM16(uint8_t *dst) { TrueMotion(dst, 16); }
//-----------------------------------------------------------------------------
// 16x16
static void VE16(uint8_t *dst) { // vertical
int j;
for (j = 0; j < 16; ++j) {
memcpy(dst + j * BPS, dst - BPS, 16);
}
}
static void HE16(uint8_t *dst) { // horizontal
int j;
for (j = 16; j > 0; --j) {
memset(dst, dst[-1], 16);
dst += BPS;
}
}
static inline void Put16(int v, uint8_t* dst) {
int j;
for (j = 0; j < 16; ++j) {
memset(dst + j * BPS, v, 16);
}
}
static void DC16(uint8_t *dst) { // DC
int DC = 16;
int j;
for (j = 0; j < 16; ++j) {
DC += dst[-1 + j * BPS] + dst[j - BPS];
}
Put16(DC >> 5, dst);
}
static void DC16NoTop(uint8_t *dst) { // DC with top samples not available
int DC = 8;
int j;
for (j = 0; j < 16; ++j) {
DC += dst[-1 + j * BPS];
}
Put16(DC >> 4, dst);
}
static void DC16NoLeft(uint8_t *dst) { // DC with left samples not available
int DC = 8;
int i;
for (i = 0; i < 16; ++i) {
DC += dst[i - BPS];
}
Put16(DC >> 4, dst);
}
static void DC16NoTopLeft(uint8_t *dst) { // DC with no top and left samples
Put16(0x80, dst);
}
//-----------------------------------------------------------------------------
// 4x4
#define AVG3(a, b, c) (((a) + 2 * (b) + (c) + 2) >> 2)
#define AVG2(a, b) (((a) + (b) + 1) >> 1)
static void VE4(uint8_t *dst) { // vertical
const uint8_t* top = dst - BPS;
const uint8_t vals[4] = {
AVG3(top[-1], top[0], top[1]),
AVG3(top[ 0], top[1], top[2]),
AVG3(top[ 1], top[2], top[3]),
AVG3(top[ 2], top[3], top[4])
};
int i;
for (i = 0; i < 4; ++i) {
memcpy(dst + i * BPS, vals, sizeof(vals));
}
}
static void HE4(uint8_t *dst) { // horizontal
const int A = dst[-1 - BPS];
const int B = dst[-1];
const int C = dst[-1 + BPS];
const int D = dst[-1 + 2 * BPS];
const int E = dst[-1 + 3 * BPS];
*(uint32_t*)(dst + 0 * BPS) = 0x01010101U * AVG3(A, B, C);
*(uint32_t*)(dst + 1 * BPS) = 0x01010101U * AVG3(B, C, D);
*(uint32_t*)(dst + 2 * BPS) = 0x01010101U * AVG3(C, D, E);
*(uint32_t*)(dst + 3 * BPS) = 0x01010101U * AVG3(D, E, E);
}
static void DC4(uint8_t *dst) { // DC
uint32_t dc = 4;
int i;
for (i = 0; i < 4; ++i) dc += dst[i - BPS] + dst[-1 + i * BPS];
dc >>= 3;
for (i = 0; i < 4; ++i) memset(dst + i * BPS, dc, 4);
}
static void RD4(uint8_t *dst) { // Down-right
const int I = dst[-1 + 0 * BPS];
const int J = dst[-1 + 1 * BPS];
const int K = dst[-1 + 2 * BPS];
const int L = dst[-1 + 3 * BPS];
const int X = dst[-1 - BPS];
const int A = dst[0 - BPS];
const int B = dst[1 - BPS];
const int C = dst[2 - BPS];
const int D = dst[3 - BPS];
OUT(0, 3) = AVG3(J, K, L);
OUT(0, 2) = OUT(1, 3) = AVG3(I, J, K);
OUT(0, 1) = OUT(1, 2) = OUT(2, 3) = AVG3(X, I, J);
OUT(0, 0) = OUT(1, 1) = OUT(2, 2) = OUT(3, 3) = AVG3(A, X, I);
OUT(1, 0) = OUT(2, 1) = OUT(3, 2) = AVG3(B, A, X);
OUT(2, 0) = OUT(3, 1) = AVG3(C, B, A);
OUT(3, 0) = AVG3(D, C, B);
}
static void LD4(uint8_t *dst) { // Down-Left
const int A = dst[0 - BPS];
const int B = dst[1 - BPS];
const int C = dst[2 - BPS];
const int D = dst[3 - BPS];
const int E = dst[4 - BPS];
const int F = dst[5 - BPS];
const int G = dst[6 - BPS];
const int H = dst[7 - BPS];
OUT(0, 0) = AVG3(A, B, C);
OUT(1, 0) = OUT(0, 1) = AVG3(B, C, D);
OUT(2, 0) = OUT(1, 1) = OUT(0, 2) = AVG3(C, D, E);
OUT(3, 0) = OUT(2, 1) = OUT(1, 2) = OUT(0, 3) = AVG3(D, E, F);
OUT(3, 1) = OUT(2, 2) = OUT(1, 3) = AVG3(E, F, G);
OUT(3, 2) = OUT(2, 3) = AVG3(F, G, H);
OUT(3, 3) = AVG3(G, H, H);
}
static void VR4(uint8_t *dst) { // Vertical-Right
const int I = dst[-1 + 0 * BPS];
const int J = dst[-1 + 1 * BPS];
const int K = dst[-1 + 2 * BPS];
const int X = dst[-1 - BPS];
const int A = dst[0 - BPS];
const int B = dst[1 - BPS];
const int C = dst[2 - BPS];
const int D = dst[3 - BPS];
OUT(0, 0) = OUT(1, 2) = AVG2(X, A);
OUT(1, 0) = OUT(2, 2) = AVG2(A, B);
OUT(2, 0) = OUT(3, 2) = AVG2(B, C);
OUT(3, 0) = AVG2(C, D);
OUT(0, 3) = AVG3(K, J, I);
OUT(0, 2) = AVG3(J, I, X);
OUT(0, 1) = OUT(1, 3) = AVG3(I, X, A);
OUT(1, 1) = OUT(2, 3) = AVG3(X, A, B);
OUT(2, 1) = OUT(3, 3) = AVG3(A, B, C);
OUT(3, 1) = AVG3(B, C, D);
}
static void VL4(uint8_t *dst) { // Vertical-Left
const int A = dst[0 - BPS];
const int B = dst[1 - BPS];
const int C = dst[2 - BPS];
const int D = dst[3 - BPS];
const int E = dst[4 - BPS];
const int F = dst[5 - BPS];
const int G = dst[6 - BPS];
const int H = dst[7 - BPS];
OUT(0, 0) = AVG2(A, B);
OUT(1, 0) = OUT(0, 2) = AVG2(B, C);
OUT(2, 0) = OUT(1, 2) = AVG2(C, D);
OUT(3, 0) = OUT(2, 2) = AVG2(D, E);
OUT(0, 1) = AVG3(A, B, C);
OUT(1, 1) = OUT(0, 3) = AVG3(B, C, D);
OUT(2, 1) = OUT(1, 3) = AVG3(C, D, E);
OUT(3, 1) = OUT(2, 3) = AVG3(D, E, F);
OUT(3, 2) = AVG3(E, F, G);
OUT(3, 3) = AVG3(F, G, H);
}
static void HU4(uint8_t *dst) { // Horizontal-Up
const int I = dst[-1 + 0 * BPS];
const int J = dst[-1 + 1 * BPS];
const int K = dst[-1 + 2 * BPS];
const int L = dst[-1 + 3 * BPS];
OUT(0, 0) = AVG2(I, J);
OUT(2, 0) = OUT(0, 1) = AVG2(J, K);
OUT(2, 1) = OUT(0, 2) = AVG2(K, L);
OUT(1, 0) = AVG3(I, J, K);
OUT(3, 0) = OUT(1, 1) = AVG3(J, K, L);
OUT(3, 1) = OUT(1, 2) = AVG3(K, L, L);
OUT(3, 2) = OUT(2, 2) =
OUT(0, 3) = OUT(1, 3) = OUT(2, 3) = OUT(3, 3) = L;
}
static void HD4(uint8_t *dst) { // Horizontal-Down
const int I = dst[-1 + 0 * BPS];
const int J = dst[-1 + 1 * BPS];
const int K = dst[-1 + 2 * BPS];
const int L = dst[-1 + 3 * BPS];
const int X = dst[-1 - BPS];
const int A = dst[0 - BPS];
const int B = dst[1 - BPS];
const int C = dst[2 - BPS];
OUT(0, 0) = OUT(2, 1) = AVG2(I, X);
OUT(0, 1) = OUT(2, 2) = AVG2(J, I);
OUT(0, 2) = OUT(2, 3) = AVG2(K, J);
OUT(0, 3) = AVG2(L, K);
OUT(3, 0) = AVG3(A, B, C);
OUT(2, 0) = AVG3(X, A, B);
OUT(1, 0) = OUT(3, 1) = AVG3(I, X, A);
OUT(1, 1) = OUT(3, 2) = AVG3(J, I, X);
OUT(1, 2) = OUT(3, 3) = AVG3(K, J, I);
OUT(1, 3) = AVG3(L, K, J);
}
#undef AVG3
#undef AVG2
//-----------------------------------------------------------------------------
// Chroma
static void VE8uv(uint8_t *dst) { // vertical
int j;
for (j = 0; j < 8; ++j) {
memcpy(dst + j * BPS, dst - BPS, 8);
}
}
static void HE8uv(uint8_t *dst) { // horizontal
int j;
for (j = 0; j < 8; ++j) {
memset(dst, dst[-1], 8);
dst += BPS;
}
}
// helper for chroma-DC predictions
static inline void Put8x8uv(uint64_t v, uint8_t* dst) {
int j;
for (j = 0; j < 8; ++j) {
*(uint64_t*)(dst + j * BPS) = v;
}
}
static void DC8uv(uint8_t *dst) { // DC
int dc0 = 8;
int i;
for (i = 0; i < 8; ++i) {
dc0 += dst[i - BPS] + dst[-1 + i * BPS];
}
Put8x8uv((uint64_t)((dc0 >> 4) * 0x0101010101010101ULL), dst);
}
static void DC8uvNoLeft(uint8_t *dst) { // DC with no left samples
int dc0 = 4;
int i;
for (i = 0; i < 8; ++i) {
dc0 += dst[i - BPS];
}
Put8x8uv((uint64_t)((dc0 >> 3) * 0x0101010101010101ULL), dst);
}
static void DC8uvNoTop(uint8_t *dst) { // DC with no top samples
int dc0 = 4;
int i;
for (i = 0; i < 8; ++i) {
dc0 += dst[-1 + i * BPS];
}
Put8x8uv((uint64_t)((dc0 >> 3) * 0x0101010101010101ULL), dst);
}
static void DC8uvNoTopLeft(uint8_t *dst) { // DC with nothing
Put8x8uv(0x8080808080808080ULL, dst);
}
//-----------------------------------------------------------------------------
// default C implementations
VP8PredFunc VP8PredLuma4[11] = {
DC4, TM4, VE4, HE4, RD4, VR4, LD4, VL4, HD4, HU4
};
VP8PredFunc VP8PredLuma16[7] = {
DC16, TM16, VE16, HE16,
DC16NoTop, DC16NoLeft, DC16NoTopLeft
};
VP8PredFunc VP8PredChroma8[7] = {
DC8uv, TM8uv, VE8uv, HE8uv,
DC8uvNoTop, DC8uvNoLeft, DC8uvNoTopLeft
};
//-----------------------------------------------------------------------------
// Edge filtering functions
// 4 pixels in, 2 pixels out
static inline void do_filter2(uint8_t* p, int step) {
const int p1 = p[-2*step], p0 = p[-step], q0 = p[0], q1 = p[step];
const int a = 3 * (q0 - p0) + sclip1[1020 + p1 - q1];
const int a1 = sclip2[112 + ((a + 4) >> 3)];
const int a2 = sclip2[112 + ((a + 3) >> 3)];
p[-step] = clip1[255 + p0 + a2];
p[ 0] = clip1[255 + q0 - a1];
}
// 4 pixels in, 4 pixels out
static inline void do_filter4(uint8_t* p, int step) {
const int p1 = p[-2*step], p0 = p[-step], q0 = p[0], q1 = p[step];
const int a = 3 * (q0 - p0);
const int a1 = sclip2[112 + ((a + 4) >> 3)];
const int a2 = sclip2[112 + ((a + 3) >> 3)];
const int a3 = (a1 + 1) >> 1;
p[-2*step] = clip1[255 + p1 + a3];
p[- step] = clip1[255 + p0 + a2];
p[ 0] = clip1[255 + q0 - a1];
p[ step] = clip1[255 + q1 - a3];
}
// 6 pixels in, 6 pixels out
static inline void do_filter6(uint8_t* p, int step) {
const int p2 = p[-3*step], p1 = p[-2*step], p0 = p[-step];
const int q0 = p[0], q1 = p[step], q2 = p[2*step];
const int a = sclip1[1020 + 3 * (q0 - p0) + sclip1[1020 + p1 - q1]];
const int a1 = (27 * a + 63) >> 7; // eq. to ((3 * a + 7) * 9) >> 7
const int a2 = (18 * a + 63) >> 7; // eq. to ((2 * a + 7) * 9) >> 7
const int a3 = (9 * a + 63) >> 7; // eq. to ((1 * a + 7) * 9) >> 7
p[-3*step] = clip1[255 + p2 + a3];
p[-2*step] = clip1[255 + p1 + a2];
p[- step] = clip1[255 + p0 + a1];
p[ 0] = clip1[255 + q0 - a1];
p[ step] = clip1[255 + q1 - a2];
p[ 2*step] = clip1[255 + q2 - a3];
}
static inline int hev(const uint8_t* p, int step, int thresh) {
const int p1 = p[-2*step], p0 = p[-step], q0 = p[0], q1 = p[step];
return (abs0[255 + p1 - p0] > thresh) || (abs0[255 + q1 - q0] > thresh);
}
static inline int needs_filter(const uint8_t* p, int step, int thresh) {
const int p1 = p[-2*step], p0 = p[-step], q0 = p[0], q1 = p[step];
return (2 * abs0[255 + p0 - q0] + abs1[255 + p1 - q1]) <= thresh;
}
static inline int needs_filter2(const uint8_t* p, int step, int t, int it) {
const int p3 = p[-4*step], p2 = p[-3*step], p1 = p[-2*step], p0 = p[-step];
const int q0 = p[0], q1 = p[step], q2 = p[2*step], q3 = p[3*step];
if ((2 * abs0[255 + p0 - q0] + abs1[255 + p1 - q1]) > t)
return 0;
return abs0[255 + p3 - p2] <= it && abs0[255 + p2 - p1] <= it &&
abs0[255 + p1 - p0] <= it && abs0[255 + q3 - q2] <= it &&
abs0[255 + q2 - q1] <= it && abs0[255 + q1 - q0] <= it;
}
//-----------------------------------------------------------------------------
// Simple In-loop filtering (Paragraph 15.2)
static void SimpleVFilter16(uint8_t* p, int stride, int thresh) {
int i;
for (i = 0; i < 16; ++i) {
if (needs_filter(p + i, stride, thresh)) {
do_filter2(p + i, stride);
}
}
}
static void SimpleHFilter16(uint8_t* p, int stride, int thresh) {
int i;
for (i = 0; i < 16; ++i) {
if (needs_filter(p + i * stride, 1, thresh)) {
do_filter2(p + i * stride, 1);
}
}
}
static void SimpleVFilter16i(uint8_t* p, int stride, int thresh) {
int k;
for (k = 3; k > 0; --k) {
p += 4 * stride;
SimpleVFilter16(p, stride, thresh);
}
}
static void SimpleHFilter16i(uint8_t* p, int stride, int thresh) {
int k;
for (k = 3; k > 0; --k) {
p += 4;
SimpleHFilter16(p, stride, thresh);
}
}
//-----------------------------------------------------------------------------
// Complex In-loop filtering (Paragraph 15.3)
static inline void FilterLoop26(uint8_t* p, int hstride, int vstride, int size,
int thresh, int ithresh, int hev_thresh) {
while (size-- > 0) {
if (needs_filter2(p, hstride, thresh, ithresh)) {
if (hev(p, hstride, hev_thresh)) {
do_filter2(p, hstride);
} else {
do_filter6(p, hstride);
}
}
p += vstride;
}
}
static inline void FilterLoop24(uint8_t* p, int hstride, int vstride, int size,
int thresh, int ithresh, int hev_thresh) {
while (size-- > 0) {
if (needs_filter2(p, hstride, thresh, ithresh)) {
if (hev(p, hstride, hev_thresh)) {
do_filter2(p, hstride);
} else {
do_filter4(p, hstride);
}
}
p += vstride;
}
}
// on macroblock edges
static void VFilter16(uint8_t* p, int stride,
int thresh, int ithresh, int hev_thresh) {
FilterLoop26(p, stride, 1, 16, thresh, ithresh, hev_thresh);
}
static void HFilter16(uint8_t* p, int stride,
int thresh, int ithresh, int hev_thresh) {
FilterLoop26(p, 1, stride, 16, thresh, ithresh, hev_thresh);
}
// on three inner edges
static void VFilter16i(uint8_t* p, int stride,
int thresh, int ithresh, int hev_thresh) {
int k;
for (k = 3; k > 0; --k) {
p += 4 * stride;
FilterLoop24(p, stride, 1, 16, thresh, ithresh, hev_thresh);
}
}
static void HFilter16i(uint8_t* p, int stride,
int thresh, int ithresh, int hev_thresh) {
int k;
for (k = 3; k > 0; --k) {
p += 4;
FilterLoop24(p, 1, stride, 16, thresh, ithresh, hev_thresh);
}
}
// 8-pixels wide variant, for chroma filtering
static void VFilter8(uint8_t* u, uint8_t* v, int stride,
int thresh, int ithresh, int hev_thresh) {
FilterLoop26(u, stride, 1, 8, thresh, ithresh, hev_thresh);
FilterLoop26(v, stride, 1, 8, thresh, ithresh, hev_thresh);
}
static void HFilter8(uint8_t* u, uint8_t* v, int stride,
int thresh, int ithresh, int hev_thresh) {
FilterLoop26(u, 1, stride, 8, thresh, ithresh, hev_thresh);
FilterLoop26(v, 1, stride, 8, thresh, ithresh, hev_thresh);
}
static void VFilter8i(uint8_t* u, uint8_t* v, int stride,
int thresh, int ithresh, int hev_thresh) {
FilterLoop24(u + 4 * stride, stride, 1, 8, thresh, ithresh, hev_thresh);
FilterLoop24(v + 4 * stride, stride, 1, 8, thresh, ithresh, hev_thresh);
}
static void HFilter8i(uint8_t* u, uint8_t* v, int stride,
int thresh, int ithresh, int hev_thresh) {
FilterLoop24(u + 4, 1, stride, 8, thresh, ithresh, hev_thresh);
FilterLoop24(v + 4, 1, stride, 8, thresh, ithresh, hev_thresh);
}
//-----------------------------------------------------------------------------
void (*VP8VFilter16)(uint8_t*, int, int, int, int) = VFilter16;
void (*VP8HFilter16)(uint8_t*, int, int, int, int) = HFilter16;
void (*VP8VFilter8)(uint8_t*, uint8_t*, int, int, int, int) = VFilter8;
void (*VP8HFilter8)(uint8_t*, uint8_t*, int, int, int, int) = HFilter8;
void (*VP8VFilter16i)(uint8_t*, int, int, int, int) = VFilter16i;
void (*VP8HFilter16i)(uint8_t*, int, int, int, int) = HFilter16i;
void (*VP8VFilter8i)(uint8_t*, uint8_t*, int, int, int, int) = VFilter8i;
void (*VP8HFilter8i)(uint8_t*, uint8_t*, int, int, int, int) = HFilter8i;
void (*VP8SimpleVFilter16)(uint8_t*, int, int) = SimpleVFilter16;
void (*VP8SimpleHFilter16)(uint8_t*, int, int) = SimpleHFilter16;
void (*VP8SimpleVFilter16i)(uint8_t*, int, int) = SimpleVFilter16i;
void (*VP8SimpleHFilter16i)(uint8_t*, int, int) = SimpleHFilter16i;
//-----------------------------------------------------------------------------
void VP8DspInit() {
// later we'll plug some SSE2 variant here
}
#if defined(__cplusplus) || defined(c_plusplus)
} // extern "C"
#endif

410
src/dec/frame.c Normal file
View File

@ -0,0 +1,410 @@
// Copyright 2010 Google Inc.
//
// This code is licensed under the same terms as WebM:
// Software License Agreement: http://www.webmproject.org/license/software/
// Additional IP Rights Grant: http://www.webmproject.org/license/additional/
// -----------------------------------------------------------------------------
//
// Frame-reconstruction function. Memory allocation.
//
// Author: Skal (pascal.massimino@gmail.com)
#include <stdlib.h>
#include "vp8i.h"
#if defined(__cplusplus) || defined(c_plusplus)
extern "C" {
#endif
#define ALIGN_MASK (32 - 1)
//-----------------------------------------------------------------------------
// Memory setup
// how many extra luma lines are needed for caching, given a filtering level
static const uint8_t kFilterExtraRows[3] = { 0, 4, 8 };
int VP8InitFrame(VP8Decoder* const dec, VP8Io* io) {
const int mb_w = dec->mb_w_;
const int intra_pred_mode_size = 4 * mb_w * sizeof(uint8_t);
const int top_size = (16 + 8 + 8) * mb_w;
const int info_size = (mb_w + 1) * sizeof(VP8MB);
const int yuv_size = YUV_SIZE * sizeof(*dec->yuv_b_);
const int coeffs_size = 384 * sizeof(*dec->coeffs_);
const int cache_height = (16 + kFilterExtraRows[dec->filter_type_]) * 3 / 2;
const int cache_size = top_size * cache_height;
const int needed = intra_pred_mode_size
+ top_size + info_size
+ yuv_size + coeffs_size
+ cache_size + ALIGN_MASK;
uint8_t* mem;
if (needed > dec->mem_size_) {
free(dec->mem_);
dec->mem_size_ = 0;
dec->mem_ = (uint8_t*)malloc(needed);
if (dec->mem_ == NULL) {
return VP8SetError(dec, 1, "no memory during frame initialization.");
}
dec->mem_size_ = needed;
}
mem = (uint8_t*)dec->mem_;
dec->intra_t_ = (uint8_t*)mem;
mem += intra_pred_mode_size;
dec->y_t_ = (uint8_t*)mem;
mem += 16 * mb_w;
dec->u_t_ = (uint8_t*)mem;
mem += 8 * mb_w;
dec->v_t_ = (uint8_t*)mem;
mem += 8 * mb_w;
dec->mb_info_ = ((VP8MB*)mem) + 1;
mem += info_size;
mem = (uint8_t*)((uintptr_t)(mem + ALIGN_MASK) & ~ALIGN_MASK);
assert((yuv_size & ALIGN_MASK) == 0);
dec->yuv_b_ = (uint8_t*)mem;
mem += yuv_size;
dec->coeffs_ = (int16_t*)mem;
mem += coeffs_size;
dec->cache_y_stride_ = 16 * mb_w;
dec->cache_uv_stride_ = 8 * mb_w;
{
const int extra_rows = kFilterExtraRows[dec->filter_type_];
const int extra_y = extra_rows * dec->cache_y_stride_;
const int extra_uv = (extra_rows / 2) * dec->cache_uv_stride_;
dec->cache_y_ = ((uint8_t*)mem) + extra_y;
dec->cache_u_ = dec->cache_y_ + 16 * dec->cache_y_stride_ + extra_uv;
dec->cache_v_ = dec->cache_u_ + 8 * dec->cache_uv_stride_ + extra_uv;
}
mem += cache_size;
// note: left-info is initialized once for all.
memset(dec->mb_info_ - 1, 0, (mb_w + 1) * sizeof(*dec->mb_info_));
// initialize top
memset(dec->intra_t_, B_DC_PRED, intra_pred_mode_size);
// prepare 'io'
io->width = dec->pic_hdr_.width_;
io->height = dec->pic_hdr_.height_;
io->mb_y = 0;
io->y = dec->cache_y_;
io->u = dec->cache_u_;
io->v = dec->cache_v_;
io->y_stride = dec->cache_y_stride_;
io->uv_stride = dec->cache_uv_stride_;
io->fancy_upscaling = 0; // default
// Init critical function pointers and look-up tables.
VP8DspInitTables();
VP8DspInit();
return 1;
}
//-----------------------------------------------------------------------------
// Filtering
static inline int hev_thresh_from_level(int level, int keyframe) {
if (keyframe) {
return (level >= 40) ? 2 : (level >= 15) ? 1 : 0;
} else {
return (level >= 40) ? 3 : (level >= 20) ? 2 : (level >= 15) ? 1 : 0;
}
}
static void DoFilter(VP8Decoder* const dec, int mb_x, int mb_y) {
VP8MB* const mb = dec->mb_info_ + mb_x;
uint8_t* const y_dst = dec->cache_y_ + mb_x * 16;
const int y_bps = dec->cache_y_stride_;
const int level = mb->f_level_;
const int ilevel = mb->f_ilevel_;
const int limit = 2 * level + ilevel;
if (dec->filter_type_ == 1) { // simple
if (mb_x > 0) {
VP8SimpleHFilter16(y_dst, y_bps, limit + 4);
}
if (mb->f_inner_) {
VP8SimpleHFilter16i(y_dst, y_bps, limit);
}
if (mb_y > 0) {
VP8SimpleVFilter16(y_dst, y_bps, limit + 4);
}
if (mb->f_inner_) {
VP8SimpleVFilter16i(y_dst, y_bps, limit);
}
} else { // complex
uint8_t* const u_dst = dec->cache_u_ + mb_x * 8;
uint8_t* const v_dst = dec->cache_v_ + mb_x * 8;
const int uv_bps = dec->cache_uv_stride_;
const int hev_thresh =
hev_thresh_from_level(level, dec->frm_hdr_.key_frame_);
if (mb_x > 0) {
VP8HFilter16(y_dst, y_bps, limit + 4, ilevel, hev_thresh);
VP8HFilter8(u_dst, v_dst, uv_bps, limit + 4, ilevel, hev_thresh);
}
if (mb->f_inner_) {
VP8HFilter16i(y_dst, y_bps, limit, ilevel, hev_thresh);
VP8HFilter8i(u_dst, v_dst, uv_bps, limit, ilevel, hev_thresh);
}
if (mb_y > 0) {
VP8VFilter16(y_dst, y_bps, limit + 4, ilevel, hev_thresh);
VP8VFilter8(u_dst, v_dst, uv_bps, limit + 4, ilevel, hev_thresh);
}
if (mb->f_inner_) {
VP8VFilter16i(y_dst, y_bps, limit, ilevel, hev_thresh);
VP8VFilter8i(u_dst, v_dst, uv_bps, limit, ilevel, hev_thresh);
}
}
}
void VP8StoreBlock(VP8Decoder* const dec) {
if (dec->filter_type_ > 0) {
VP8MB* const info = dec->mb_info_ + dec->mb_x_;
int level = dec->filter_levels_[dec->segment_];
if (dec->filter_hdr_.use_lf_delta_) {
// TODO(skal): only CURRENT is handled for now.
level += dec->filter_hdr_.ref_lf_delta_[0];
if (dec->is_i4x4_) {
level += dec->filter_hdr_.mode_lf_delta_[0];
}
}
level = (level < 0) ? 0 : (level > 63) ? 63 : level;
info->f_level_ = level;
if (dec->filter_hdr_.sharpness_ > 0) {
if (dec->filter_hdr_.sharpness_ > 4) {
level >>= 2;
} else {
level >>= 1;
}
if (level > 9 - dec->filter_hdr_.sharpness_) {
level = 9 - dec->filter_hdr_.sharpness_;
}
}
info->f_ilevel_ = (level < 1) ? 1 : level;
info->f_inner_ = (!info->skip_ || dec->is_i4x4_);
}
{
// Transfer samples to row cache
int y;
uint8_t* const ydst = dec->cache_y_ + dec->mb_x_ * 16;
uint8_t* const udst = dec->cache_u_ + dec->mb_x_ * 8;
uint8_t* const vdst = dec->cache_v_ + dec->mb_x_ * 8;
for (y = 0; y < 16; ++y) {
memcpy(ydst + y * dec->cache_y_stride_,
dec->yuv_b_ + Y_OFF + y * BPS, 16);
}
for (y = 0; y < 8; ++y) {
memcpy(udst + y * dec->cache_uv_stride_,
dec->yuv_b_ + U_OFF + y * BPS, 8);
memcpy(vdst + y * dec->cache_uv_stride_,
dec->yuv_b_ + V_OFF + y * BPS, 8);
}
}
}
void VP8FinishRow(VP8Decoder* const dec, VP8Io* io) {
const int extra_y_rows = kFilterExtraRows[dec->filter_type_];
const int ysize = extra_y_rows * dec->cache_y_stride_;
const int uvsize = (extra_y_rows / 2) * dec->cache_uv_stride_;
const int first_row = (dec->mb_y_ == 0);
const int last_row = (dec->mb_y_ >= dec->mb_h_ - 1);
uint8_t* const ydst = dec->cache_y_ - ysize;
uint8_t* const udst = dec->cache_u_ - uvsize;
uint8_t* const vdst = dec->cache_v_ - uvsize;
if (dec->filter_type_ > 0) {
int mb_x;
for (mb_x = 0; mb_x < dec->mb_w_; ++mb_x) {
DoFilter(dec, mb_x, dec->mb_y_);
}
}
if (io->put) {
int y_start = dec->mb_y_ * 16;
int y_end = y_start + 16;
if (!first_row) {
y_start -= extra_y_rows;
io->y = ydst;
io->u = udst;
io->v = vdst;
} else {
io->y = dec->cache_y_;
io->u = dec->cache_u_;
io->v = dec->cache_v_;
}
if (!last_row) {
y_end -= extra_y_rows;
}
if (y_end > io->height) {
y_end = io->height;
}
io->mb_y = y_start;
io->mb_h = y_end - y_start;
io->put(io);
}
// rotate top samples
if (!last_row) {
memcpy(ydst, ydst + 16 * dec->cache_y_stride_, ysize);
memcpy(udst, udst + 8 * dec->cache_uv_stride_, uvsize);
memcpy(vdst, vdst + 8 * dec->cache_uv_stride_, uvsize);
}
}
//-----------------------------------------------------------------------------
// Main reconstruction function.
static const int kScan[16] = {
0 + 0 * BPS, 4 + 0 * BPS, 8 + 0 * BPS, 12 + 0 * BPS,
0 + 4 * BPS, 4 + 4 * BPS, 8 + 4 * BPS, 12 + 4 * BPS,
0 + 8 * BPS, 4 + 8 * BPS, 8 + 8 * BPS, 12 + 8 * BPS,
0 + 12 * BPS, 4 + 12 * BPS, 8 + 12 * BPS, 12 + 12 * BPS
};
static inline int CheckMode(VP8Decoder* const dec, int mode) {
if (mode == B_DC_PRED) {
if (dec->mb_x_ == 0) {
return (dec->mb_y_ == 0) ? B_DC_PRED_NOTOPLEFT : B_DC_PRED_NOLEFT;
} else {
return (dec->mb_y_ == 0) ? B_DC_PRED_NOTOP : B_DC_PRED;
}
}
return mode;
}
static inline void Copy32b(uint8_t* dst, uint8_t* src) {
*(uint32_t*)dst = *(uint32_t*)src;
}
void VP8ReconstructBlock(VP8Decoder* const dec) {
uint8_t* const y_dst = dec->yuv_b_ + Y_OFF;
uint8_t* const u_dst = dec->yuv_b_ + U_OFF;
uint8_t* const v_dst = dec->yuv_b_ + V_OFF;
// Rotate in the left samples from previously decoded block. We move four
// pixels at a time for alignment reason, and because of in-loop filter.
if (dec->mb_x_ > 0) {
int j;
for (j = -1; j < 16; ++j) {
Copy32b(&y_dst[j * BPS - 4], &y_dst[j * BPS + 12]);
}
for (j = -1; j < 8; ++j) {
Copy32b(&u_dst[j * BPS - 4], &u_dst[j * BPS + 4]);
Copy32b(&v_dst[j * BPS - 4], &v_dst[j * BPS + 4]);
}
} else {
int j;
for (j = 0; j < 16; ++j) {
y_dst[j * BPS - 1] = 129;
}
for (j = 0; j < 8; ++j) {
u_dst[j * BPS - 1] = 129;
v_dst[j * BPS - 1] = 129;
}
// Init top-left sample on left column too
if (dec->mb_y_ > 0) {
y_dst[-1 - BPS] = u_dst[-1 - BPS] = v_dst[-1 - BPS] = 129;
}
}
{
// bring top samples into the cache
uint8_t* const top_y = dec->y_t_ + dec->mb_x_ * 16;
uint8_t* const top_u = dec->u_t_ + dec->mb_x_ * 8;
uint8_t* const top_v = dec->v_t_ + dec->mb_x_ * 8;
const int16_t* coeffs = dec->coeffs_;
int n;
if (dec->mb_y_ > 0) {
memcpy(y_dst - BPS, top_y, 16);
memcpy(u_dst - BPS, top_u, 8);
memcpy(v_dst - BPS, top_v, 8);
} else if (dec->mb_x_ == 0) {
// we only need to do this init once at block (0,0).
// Afterward, it remains valid for the whole topmost row.
memset(y_dst - BPS - 1, 127, 16 + 4 + 1);
memset(u_dst - BPS - 1, 127, 8 + 1);
memset(v_dst - BPS - 1, 127, 8 + 1);
}
// predict and add residuals
if (dec->is_i4x4_) { // 4x4
uint32_t* const top_right = (uint32_t*)(y_dst - BPS + 16);
if (dec->mb_y_ > 0) {
if (dec->mb_x_ >= dec->mb_w_ - 1) { // on rightmost border
top_right[0] = top_y[15] * 0x01010101u;
} else {
memcpy(top_right, top_y + 16, sizeof(*top_right));
}
}
// replicate the top-right pixels below
top_right[BPS] = top_right[2 * BPS] = top_right[3 * BPS] = top_right[0];
// predict and add residues for all 4x4 blocks in turn.
for (n = 0; n < 16; n++) {
uint8_t* const dst = y_dst + kScan[n];
VP8PredLuma4[dec->imodes_[n]](dst);
if (dec->non_zero_ & (1 << n)) {
VP8Transform(coeffs + n * 16, dst);
} else if (dec->non_zero_ & (1 << n)) { // only DC is present
VP8TransformDC(coeffs + n * 16, dst);
}
}
} else { // 16x16
const int pred_func = CheckMode(dec, dec->imodes_[0]);
VP8PredLuma16[pred_func](y_dst);
if (dec->non_zero_) {
for (n = 0; n < 16; n++) {
uint8_t* const dst = y_dst + kScan[n];
if (dec->non_zero_ac_ & (1 << n)) {
VP8Transform(coeffs + n * 16, dst);
} else if (dec->non_zero_ & (1 << n)) { // only DC is present
VP8TransformDC(coeffs + n * 16, dst);
}
}
}
}
{
// Chroma
const int pred_func = CheckMode(dec, dec->uvmode_);
VP8PredChroma8[pred_func](u_dst);
VP8PredChroma8[pred_func](v_dst);
if (dec->non_zero_ & 0x0f0000) { // chroma-U
const int16_t* const u_coeffs = dec->coeffs_ + 16 * 16;
if (dec->non_zero_ac_ & 0x0f0000) {
VP8TransformUV(u_coeffs, u_dst);
} else {
VP8TransformDCUV(u_coeffs, u_dst);
}
}
if (dec->non_zero_ & 0xf00000) { // chroma-V
const int16_t* const v_coeffs = dec->coeffs_ + 20 * 16;
if (dec->non_zero_ac_ & 0xf00000) {
VP8TransformUV(v_coeffs, v_dst);
} else {
VP8TransformDCUV(v_coeffs, v_dst);
}
}
// stash away top samples for next block
if (dec->mb_y_ < dec->mb_h_ - 1) {
memcpy(top_y, y_dst + 15 * BPS, 16);
memcpy(top_u, u_dst + 7 * BPS, 8);
memcpy(top_v, v_dst + 7 * BPS, 8);
}
}
}
}
//-----------------------------------------------------------------------------
#if defined(__cplusplus) || defined(c_plusplus)
} // extern "C"
#endif

111
src/dec/quant.c Normal file
View File

@ -0,0 +1,111 @@
// Copyright 2010 Google Inc.
//
// This code is licensed under the same terms as WebM:
// Software License Agreement: http://www.webmproject.org/license/software/
// Additional IP Rights Grant: http://www.webmproject.org/license/additional/
// -----------------------------------------------------------------------------
//
// Quantizer initialization
//
// Author: Skal (pascal.massimino@gmail.com)
#include "vp8i.h"
#if defined(__cplusplus) || defined(c_plusplus)
extern "C" {
#endif
static inline int clip(int v, int M) {
return v < 0 ? 0 : v > M ? M : v;
}
// Paragraph 14.1
static const uint8_t kDcTable[128] = {
4, 5, 6, 7, 8, 9, 10, 10,
11, 12, 13, 14, 15, 16, 17, 17,
18, 19, 20, 20, 21, 21, 22, 22,
23, 23, 24, 25, 25, 26, 27, 28,
29, 30, 31, 32, 33, 34, 35, 36,
37, 37, 38, 39, 40, 41, 42, 43,
44, 45, 46, 46, 47, 48, 49, 50,
51, 52, 53, 54, 55, 56, 57, 58,
59, 60, 61, 62, 63, 64, 65, 66,
67, 68, 69, 70, 71, 72, 73, 74,
75, 76, 76, 77, 78, 79, 80, 81,
82, 83, 84, 85, 86, 87, 88, 89,
91, 93, 95, 96, 98, 100, 101, 102,
104, 106, 108, 110, 112, 114, 116, 118,
122, 124, 126, 128, 130, 132, 134, 136,
138, 140, 143, 145, 148, 151, 154, 157
};
static const uint16_t kAcTable[128] = {
4, 5, 6, 7, 8, 9, 10, 11,
12, 13, 14, 15, 16, 17, 18, 19,
20, 21, 22, 23, 24, 25, 26, 27,
28, 29, 30, 31, 32, 33, 34, 35,
36, 37, 38, 39, 40, 41, 42, 43,
44, 45, 46, 47, 48, 49, 50, 51,
52, 53, 54, 55, 56, 57, 58, 60,
62, 64, 66, 68, 70, 72, 74, 76,
78, 80, 82, 84, 86, 88, 90, 92,
94, 96, 98, 100, 102, 104, 106, 108,
110, 112, 114, 116, 119, 122, 125, 128,
131, 134, 137, 140, 143, 146, 149, 152,
155, 158, 161, 164, 167, 170, 173, 177,
181, 185, 189, 193, 197, 201, 205, 209,
213, 217, 221, 225, 229, 234, 239, 245,
249, 254, 259, 264, 269, 274, 279, 284
};
//-----------------------------------------------------------------------------
// Paragraph 9.6
void VP8ParseQuant(VP8Decoder* const dec) {
VP8BitReader* const br = &dec->br_;
const int base_q0 = VP8GetValue(br, 7);
const int dqy1_dc = VP8Get(br) ? VP8GetSignedValue(br, 4) : 0;
const int dqy2_dc = VP8Get(br) ? VP8GetSignedValue(br, 4) : 0;
const int dqy2_ac = VP8Get(br) ? VP8GetSignedValue(br, 4) : 0;
const int dquv_dc = VP8Get(br) ? VP8GetSignedValue(br, 4) : 0;
const int dquv_ac = VP8Get(br) ? VP8GetSignedValue(br, 4) : 0;
const VP8SegmentHeader* const hdr = &dec->segment_hdr_;
int i;
for (i = 0; i < NUM_MB_SEGMENTS; ++i) {
int q;
if (hdr->use_segment_) {
q = hdr->quantizer_[i];
if (!hdr->absolute_delta_) {
q += base_q0;
}
} else {
if (i > 0) {
dec->dqm_[i] = dec->dqm_[0];
continue;
} else {
q = base_q0;
}
}
{
VP8QuantMatrix* const m = &dec->dqm_[i];
m->y1_mat_[0] = kDcTable[clip(q + dqy1_dc, 127)];
m->y1_mat_[1] = kAcTable[clip(q + 0, 127)];
m->y2_mat_[0] = kDcTable[clip(q + dqy2_dc, 127)] * 2;
// TODO(skal): make it another table?
m->y2_mat_[1] = kAcTable[clip(q + dqy2_ac, 127)] * 155 / 100;
if (m->y2_mat_[1] < 8) m->y2_mat_[1] = 8;
m->uv_mat_[0] = kDcTable[clip(q + dquv_dc, 117)];
m->uv_mat_[1] = kAcTable[clip(q + dquv_ac, 127)];
}
}
}
//-----------------------------------------------------------------------------
#if defined(__cplusplus) || defined(c_plusplus)
} // extern "C"
#endif

590
src/dec/tree.c Normal file
View File

@ -0,0 +1,590 @@
// Copyright 2010 Google Inc.
//
// This code is licensed under the same terms as WebM:
// Software License Agreement: http://www.webmproject.org/license/software/
// Additional IP Rights Grant: http://www.webmproject.org/license/additional/
// -----------------------------------------------------------------------------
//
// Coding trees and probas
//
// Author: Skal (pascal.massimino@gmail.com)
#include <stdio.h>
#include "vp8i.h"
#define USE_GENERIC_TREE
#if defined(__cplusplus) || defined(c_plusplus)
extern "C" {
#endif
#ifdef USE_GENERIC_TREE
static const int8_t kYModesIntra4[18] = {
-B_DC_PRED, 1,
-B_TM_PRED, 2,
-B_VE_PRED, 3,
4, 6,
-B_HE_PRED, 5,
-B_RD_PRED, -B_VR_PRED,
-B_LD_PRED, 7,
-B_VL_PRED, 8,
-B_HD_PRED, -B_HU_PRED
};
#endif
#ifndef ONLY_KEYFRAME_CODE
// inter prediction modes
enum {
LEFT4 = 0, ABOVE4 = 1, ZERO4 = 2, NEW4 = 3,
NEARESTMV, NEARMV, ZEROMV, NEWMV, SPLITMV };
static const int8_t kYModesInter[8] = {
-DC_PRED, 1,
2, 3,
-V_PRED, -H_PRED,
-TM_PRED, -B_PRED
};
static const int8_t kMBSplit[6] = {
-3, 1,
-2, 2,
-0, -1
};
static const int8_t kMVRef[8] = {
-ZEROMV, 1,
-NEARESTMV, 2,
-NEARMV, 3,
-NEWMV, -SPLITMV
};
static const int8_t kMVRef4[6] = {
-LEFT4, 1
-ABOVE4, 2
-ZERO4, -NEW4
};
#endif
//-----------------------------------------------------------------------------
// Default probabilities
// Inter
#ifndef ONLY_KEYFRAME_CODE
static const uint8_t kYModeProbaInter0[4] = { 112, 86, 140, 37 };
static const uint8_t kUVModeProbaInter0[3] = { 162, 101, 204 };
static const uint8_t kMVProba0[2][NUM_MV_PROBAS] = {
{ 162, 128, 225, 146, 172, 147, 214, 39,
156, 128, 129, 132, 75, 145, 178, 206,
239, 254, 254 },
{ 164, 128, 204, 170, 119, 235, 140, 230,
228, 128, 130, 130, 74, 148, 180, 203,
236, 254, 254 }
};
#endif
// Paragraph 13.5
static const uint8_t
CoeffsProba0[NUM_TYPES][NUM_BANDS][NUM_CTX][NUM_PROBAS] = {
// genereated using vp8_default_coef_probs() in entropy.c:129
{ { { 128, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 },
{ 128, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 },
{ 128, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 }
},
{ { 253, 136, 254, 255, 228, 219, 128, 128, 128, 128, 128 },
{ 189, 129, 242, 255, 227, 213, 255, 219, 128, 128, 128 },
{ 106, 126, 227, 252, 214, 209, 255, 255, 128, 128, 128 }
},
{ { 1, 98, 248, 255, 236, 226, 255, 255, 128, 128, 128 },
{ 181, 133, 238, 254, 221, 234, 255, 154, 128, 128, 128 },
{ 78, 134, 202, 247, 198, 180, 255, 219, 128, 128, 128 },
},
{ { 1, 185, 249, 255, 243, 255, 128, 128, 128, 128, 128 },
{ 184, 150, 247, 255, 236, 224, 128, 128, 128, 128, 128 },
{ 77, 110, 216, 255, 236, 230, 128, 128, 128, 128, 128 },
},
{ { 1, 101, 251, 255, 241, 255, 128, 128, 128, 128, 128 },
{ 170, 139, 241, 252, 236, 209, 255, 255, 128, 128, 128 },
{ 37, 116, 196, 243, 228, 255, 255, 255, 128, 128, 128 }
},
{ { 1, 204, 254, 255, 245, 255, 128, 128, 128, 128, 128 },
{ 207, 160, 250, 255, 238, 128, 128, 128, 128, 128, 128 },
{ 102, 103, 231, 255, 211, 171, 128, 128, 128, 128, 128 }
},
{ { 1, 152, 252, 255, 240, 255, 128, 128, 128, 128, 128 },
{ 177, 135, 243, 255, 234, 225, 128, 128, 128, 128, 128 },
{ 80, 129, 211, 255, 194, 224, 128, 128, 128, 128, 128 }
},
{ { 1, 1, 255, 128, 128, 128, 128, 128, 128, 128, 128 },
{ 246, 1, 255, 128, 128, 128, 128, 128, 128, 128, 128 },
{ 255, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 }
}
},
{ { { 198, 35, 237, 223, 193, 187, 162, 160, 145, 155, 62 },
{ 131, 45, 198, 221, 172, 176, 220, 157, 252, 221, 1 },
{ 68, 47, 146, 208, 149, 167, 221, 162, 255, 223, 128 }
},
{ { 1, 149, 241, 255, 221, 224, 255, 255, 128, 128, 128 },
{ 184, 141, 234, 253, 222, 220, 255, 199, 128, 128, 128 },
{ 81, 99, 181, 242, 176, 190, 249, 202, 255, 255, 128 }
},
{ { 1, 129, 232, 253, 214, 197, 242, 196, 255, 255, 128 },
{ 99, 121, 210, 250, 201, 198, 255, 202, 128, 128, 128 },
{ 23, 91, 163, 242, 170, 187, 247, 210, 255, 255, 128 }
},
{ { 1, 200, 246, 255, 234, 255, 128, 128, 128, 128, 128 },
{ 109, 178, 241, 255, 231, 245, 255, 255, 128, 128, 128 },
{ 44, 130, 201, 253, 205, 192, 255, 255, 128, 128, 128 }
},
{ { 1, 132, 239, 251, 219, 209, 255, 165, 128, 128, 128 },
{ 94, 136, 225, 251, 218, 190, 255, 255, 128, 128, 128 },
{ 22, 100, 174, 245, 186, 161, 255, 199, 128, 128, 128 }
},
{ { 1, 182, 249, 255, 232, 235, 128, 128, 128, 128, 128 },
{ 124, 143, 241, 255, 227, 234, 128, 128, 128, 128, 128 },
{ 35, 77, 181, 251, 193, 211, 255, 205, 128, 128, 128 }
},
{ { 1, 157, 247, 255, 236, 231, 255, 255, 128, 128, 128 },
{ 121, 141, 235, 255, 225, 227, 255, 255, 128, 128, 128 },
{ 45, 99, 188, 251, 195, 217, 255, 224, 128, 128, 128 }
},
{ { 1, 1, 251, 255, 213, 255, 128, 128, 128, 128, 128 },
{ 203, 1, 248, 255, 255, 128, 128, 128, 128, 128, 128 },
{ 137, 1, 177, 255, 224, 255, 128, 128, 128, 128, 128 }
}
},
{ { { 253, 9, 248, 251, 207, 208, 255, 192, 128, 128, 128 },
{ 175, 13, 224, 243, 193, 185, 249, 198, 255, 255, 128 },
{ 73, 17, 171, 221, 161, 179, 236, 167, 255, 234, 128 }
},
{ { 1, 95, 247, 253, 212, 183, 255, 255, 128, 128, 128 },
{ 239, 90, 244, 250, 211, 209, 255, 255, 128, 128, 128 },
{ 155, 77, 195, 248, 188, 195, 255, 255, 128, 128, 128 }
},
{ { 1, 24, 239, 251, 218, 219, 255, 205, 128, 128, 128 },
{ 201, 51, 219, 255, 196, 186, 128, 128, 128, 128, 128 },
{ 69, 46, 190, 239, 201, 218, 255, 228, 128, 128, 128 }
},
{ { 1, 191, 251, 255, 255, 128, 128, 128, 128, 128, 128 },
{ 223, 165, 249, 255, 213, 255, 128, 128, 128, 128, 128 },
{ 141, 124, 248, 255, 255, 128, 128, 128, 128, 128, 128 }
},
{ { 1, 16, 248, 255, 255, 128, 128, 128, 128, 128, 128 },
{ 190, 36, 230, 255, 236, 255, 128, 128, 128, 128, 128 },
{ 149, 1, 255, 128, 128, 128, 128, 128, 128, 128, 128 }
},
{ { 1, 226, 255, 128, 128, 128, 128, 128, 128, 128, 128 },
{ 247, 192, 255, 128, 128, 128, 128, 128, 128, 128, 128 },
{ 240, 128, 255, 128, 128, 128, 128, 128, 128, 128, 128 }
},
{ { 1, 134, 252, 255, 255, 128, 128, 128, 128, 128, 128 },
{ 213, 62, 250, 255, 255, 128, 128, 128, 128, 128, 128 },
{ 55, 93, 255, 128, 128, 128, 128, 128, 128, 128, 128 }
},
{ { 128, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 },
{ 128, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 },
{ 128, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 }
}
},
{ { { 202, 24, 213, 235, 186, 191, 220, 160, 240, 175, 255 },
{ 126, 38, 182, 232, 169, 184, 228, 174, 255, 187, 128 },
{ 61, 46, 138, 219, 151, 178, 240, 170, 255, 216, 128 }
},
{ { 1, 112, 230, 250, 199, 191, 247, 159, 255, 255, 128 },
{ 166, 109, 228, 252, 211, 215, 255, 174, 128, 128, 128 },
{ 39, 77, 162, 232, 172, 180, 245, 178, 255, 255, 128 }
},
{ { 1, 52, 220, 246, 198, 199, 249, 220, 255, 255, 128 },
{ 124, 74, 191, 243, 183, 193, 250, 221, 255, 255, 128 },
{ 24, 71, 130, 219, 154, 170, 243, 182, 255, 255, 128 }
},
{ { 1, 182, 225, 249, 219, 240, 255, 224, 128, 128, 128 },
{ 149, 150, 226, 252, 216, 205, 255, 171, 128, 128, 128 },
{ 28, 108, 170, 242, 183, 194, 254, 223, 255, 255, 128 }
},
{ { 1, 81, 230, 252, 204, 203, 255, 192, 128, 128, 128 },
{ 123, 102, 209, 247, 188, 196, 255, 233, 128, 128, 128 },
{ 20, 95, 153, 243, 164, 173, 255, 203, 128, 128, 128 }
},
{ { 1, 222, 248, 255, 216, 213, 128, 128, 128, 128, 128 },
{ 168, 175, 246, 252, 235, 205, 255, 255, 128, 128, 128 },
{ 47, 116, 215, 255, 211, 212, 255, 255, 128, 128, 128 }
},
{ { 1, 121, 236, 253, 212, 214, 255, 255, 128, 128, 128 },
{ 141, 84, 213, 252, 201, 202, 255, 219, 128, 128, 128 },
{ 42, 80, 160, 240, 162, 185, 255, 205, 128, 128, 128 }
},
{ { 1, 1, 255, 128, 128, 128, 128, 128, 128, 128, 128 },
{ 244, 1, 255, 128, 128, 128, 128, 128, 128, 128, 128 },
{ 238, 1, 255, 128, 128, 128, 128, 128, 128, 128, 128 }
}
}
};
// Paragraph 11.5
static const uint8_t kBModesProba[NUM_BMODES][NUM_BMODES][NUM_BMODES - 1] = {
{ { 231, 120, 48, 89, 115, 113, 120, 152, 112 },
{ 152, 179, 64, 126, 170, 118, 46, 70, 95 },
{ 175, 69, 143, 80, 85, 82, 72, 155, 103 },
{ 56, 58, 10, 171, 218, 189, 17, 13, 152 },
{ 114, 26, 17, 163, 44, 195, 21, 10, 173 },
{ 121, 24, 80, 195, 26, 62, 44, 64, 85 },
{ 144, 71, 10, 38, 171, 213, 144, 34, 26 },
{ 170, 46, 55, 19, 136, 160, 33, 206, 71 },
{ 63, 20, 8, 114, 114, 208, 12, 9, 226 },
{ 81, 40, 11, 96, 182, 84, 29, 16, 36 } },
{ { 134, 183, 89, 137, 98, 101, 106, 165, 148 },
{ 72, 187, 100, 130, 157, 111, 32, 75, 80 },
{ 66, 102, 167, 99, 74, 62, 40, 234, 128 },
{ 41, 53, 9, 178, 241, 141, 26, 8, 107 },
{ 74, 43, 26, 146, 73, 166, 49, 23, 157 },
{ 65, 38, 105, 160, 51, 52, 31, 115, 128 },
{ 104, 79, 12, 27, 217, 255, 87, 17, 7 },
{ 87, 68, 71, 44, 114, 51, 15, 186, 23 },
{ 47, 41, 14, 110, 182, 183, 21, 17, 194 },
{ 66, 45, 25, 102, 197, 189, 23, 18, 22 } },
{ { 88, 88, 147, 150, 42, 46, 45, 196, 205 },
{ 43, 97, 183, 117, 85, 38, 35, 179, 61 },
{ 39, 53, 200, 87, 26, 21, 43, 232, 171 },
{ 56, 34, 51, 104, 114, 102, 29, 93, 77 },
{ 39, 28, 85, 171, 58, 165, 90, 98, 64 },
{ 34, 22, 116, 206, 23, 34, 43, 166, 73 },
{ 107, 54, 32, 26, 51, 1, 81, 43, 31 },
{ 68, 25, 106, 22, 64, 171, 36, 225, 114 },
{ 34, 19, 21, 102, 132, 188, 16, 76, 124 },
{ 62, 18, 78, 95, 85, 57, 50, 48, 51 } },
{ { 193, 101, 35, 159, 215, 111, 89, 46, 111 },
{ 60, 148, 31, 172, 219, 228, 21, 18, 111 },
{ 112, 113, 77, 85, 179, 255, 38, 120, 114 },
{ 40, 42, 1, 196, 245, 209, 10, 25, 109 },
{ 88, 43, 29, 140, 166, 213, 37, 43, 154 },
{ 61, 63, 30, 155, 67, 45, 68, 1, 209 },
{ 100, 80, 8, 43, 154, 1, 51, 26, 71 },
{ 142, 78, 78, 16, 255, 128, 34, 197, 171 },
{ 41, 40, 5, 102, 211, 183, 4, 1, 221 },
{ 51, 50, 17, 168, 209, 192, 23, 25, 82 } },
{ { 138, 31, 36, 171, 27, 166, 38, 44, 229 },
{ 67, 87, 58, 169, 82, 115, 26, 59, 179 },
{ 63, 59, 90, 180, 59, 166, 93, 73, 154 },
{ 40, 40, 21, 116, 143, 209, 34, 39, 175 },
{ 47, 15, 16, 183, 34, 223, 49, 45, 183 },
{ 46, 17, 33, 183, 6, 98, 15, 32, 183 },
{ 57, 46, 22, 24, 128, 1, 54, 17, 37 },
{ 65, 32, 73, 115, 28, 128, 23, 128, 205 },
{ 40, 3, 9, 115, 51, 192, 18, 6, 223 },
{ 87, 37, 9, 115, 59, 77, 64, 21, 47 } },
{ { 104, 55, 44, 218, 9, 54, 53, 130, 226 },
{ 64, 90, 70, 205, 40, 41, 23, 26, 57 },
{ 54, 57, 112, 184, 5, 41, 38, 166, 213 },
{ 30, 34, 26, 133, 152, 116, 10, 32, 134 },
{ 39, 19, 53, 221, 26, 114, 32, 73, 255 },
{ 31, 9, 65, 234, 2, 15, 1, 118, 73 },
{ 75, 32, 12, 51, 192, 255, 160, 43, 51 },
{ 88, 31, 35, 67, 102, 85, 55, 186, 85 },
{ 56, 21, 23, 111, 59, 205, 45, 37, 192 },
{ 55, 38, 70, 124, 73, 102, 1, 34, 98 } },
{ { 125, 98, 42, 88, 104, 85, 117, 175, 82 },
{ 95, 84, 53, 89, 128, 100, 113, 101, 45 },
{ 75, 79, 123, 47, 51, 128, 81, 171, 1 },
{ 57, 17, 5, 71, 102, 57, 53, 41, 49 },
{ 38, 33, 13, 121, 57, 73, 26, 1, 85 },
{ 41, 10, 67, 138, 77, 110, 90, 47, 114 },
{ 115, 21, 2, 10, 102, 255, 166, 23, 6 },
{ 101, 29, 16, 10, 85, 128, 101, 196, 26 },
{ 57, 18, 10, 102, 102, 213, 34, 20, 43 },
{ 117, 20, 15, 36, 163, 128, 68, 1, 26 } },
{ { 102, 61, 71, 37, 34, 53, 31, 243, 192 },
{ 69, 60, 71, 38, 73, 119, 28, 222, 37 },
{ 68, 45, 128, 34, 1, 47, 11, 245, 171 },
{ 62, 17, 19, 70, 146, 85, 55, 62, 70 },
{ 37, 43, 37, 154, 100, 163, 85, 160, 1 },
{ 63, 9, 92, 136, 28, 64, 32, 201, 85 },
{ 75, 15, 9, 9, 64, 255, 184, 119, 16 },
{ 86, 6, 28, 5, 64, 255, 25, 248, 1 },
{ 56, 8, 17, 132, 137, 255, 55, 116, 128 },
{ 58, 15, 20, 82, 135, 57, 26, 121, 40 } },
{ { 164, 50, 31, 137, 154, 133, 25, 35, 218 },
{ 51, 103, 44, 131, 131, 123, 31, 6, 158 },
{ 86, 40, 64, 135, 148, 224, 45, 183, 128 },
{ 22, 26, 17, 131, 240, 154, 14, 1, 209 },
{ 45, 16, 21, 91, 64, 222, 7, 1, 197 },
{ 56, 21, 39, 155, 60, 138, 23, 102, 213 },
{ 83, 12, 13, 54, 192, 255, 68, 47, 28 },
{ 85, 26, 85, 85, 128, 128, 32, 146, 171 },
{ 18, 11, 7, 63, 144, 171, 4, 4, 246 },
{ 35, 27, 10, 146, 174, 171, 12, 26, 128 } },
{ { 190, 80, 35, 99, 180, 80, 126, 54, 45 },
{ 85, 126, 47, 87, 176, 51, 41, 20, 32 },
{ 101, 75, 128, 139, 118, 146, 116, 128, 85 },
{ 56, 41, 15, 176, 236, 85, 37, 9, 62 },
{ 71, 30, 17, 119, 118, 255, 17, 18, 138 },
{ 101, 38, 60, 138, 55, 70, 43, 26, 142 },
{ 146, 36, 19, 30, 171, 255, 97, 27, 20 },
{ 138, 45, 61, 62, 219, 1, 81, 188, 64 },
{ 32, 41, 20, 117, 151, 142, 20, 21, 163 },
{ 112, 19, 12, 61, 195, 128, 48, 4, 24 } }
};
void VP8ResetProba(VP8Proba* const proba) {
memset(proba->segments_, 255u, sizeof(proba->segments_));
memcpy(proba->coeffs_, CoeffsProba0, sizeof(CoeffsProba0));
#ifndef ONLY_KEYFRAME_CODE
memcpy(proba->mv_, kMVProba0, sizeof(kMVProba0));
memcpy(proba->ymode_, kYModeProbaInter0, sizeof(kYModeProbaInter0));
memcpy(proba->uvmode_, kUVModeProbaInter0, sizeof(kYModeProbaInter0));
#endif
}
void VP8ParseIntraMode(VP8BitReader* const br, VP8Decoder* const dec) {
uint8_t* const top = dec->intra_t_ + 4 * dec->mb_x_;
uint8_t* const left = dec->intra_l_;
// Hardcoded 16x16 intra-mode decision tree.
dec->is_i4x4_ = !VP8GetBit(br, 145); // decide for B_PRED first
if (!dec->is_i4x4_) {
const int ymode =
VP8GetBit(br, 156) ? (VP8GetBit(br, 128) ? TM_PRED : H_PRED)
: (VP8GetBit(br, 163) ? V_PRED : DC_PRED);
dec->imodes_[0] = ymode;
memset(top, ymode, 4 * sizeof(top[0]));
memset(left, ymode, 4 * sizeof(left[0]));
} else {
uint8_t* modes = dec->imodes_;
int y;
for (y = 0; y < 4; ++y) {
int ymode = left[y];
int x;
for (x = 0; x < 4; ++x) {
const uint8_t* const prob = kBModesProba[top[x]][ymode];
#ifdef USE_GENERIC_TREE
// Generic tree-parsing
int i = 0;
do {
i = kYModesIntra4[2 * i + VP8GetBit(br, prob[i])];
} while (i > 0);
ymode = -i;
#else
// Hardcoded tree parsing
ymode = !VP8GetBit(br, prob[0]) ? B_DC_PRED :
!VP8GetBit(br, prob[1]) ? B_TM_PRED :
!VP8GetBit(br, prob[2]) ? B_VE_PRED :
!VP8GetBit(br, prob[3]) ?
(!VP8GetBit(br, prob[4]) ? B_HE_PRED :
(!VP8GetBit(br, prob[5]) ? B_RD_PRED : B_VR_PRED)) :
(!VP8GetBit(br, prob[6]) ? B_LD_PRED :
(!VP8GetBit(br, prob[7]) ? B_VL_PRED :
(!VP8GetBit(br, prob[8]) ? B_HD_PRED : B_HU_PRED)));
#endif // USE_GENERIC_TREE
top[x] = ymode;
*modes++ = ymode;
}
left[y] = ymode;
}
}
// Hardcoded UVMode decision tree
dec->uvmode_ = !VP8GetBit(br, 142) ? DC_PRED
: !VP8GetBit(br, 114) ? V_PRED
: VP8GetBit(br, 183) ? TM_PRED : H_PRED;
}
//-----------------------------------------------------------------------------
// Paragraph 13
static const uint8_t
CoeffsUpdateProba[NUM_TYPES][NUM_BANDS][NUM_CTX][NUM_PROBAS] = {
{ { { 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 }
},
{ { 176, 246, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 223, 241, 252, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 249, 253, 253, 255, 255, 255, 255, 255, 255, 255, 255 }
},
{ { 255, 244, 252, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 234, 254, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 253, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 }
},
{ { 255, 246, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 239, 253, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 254, 255, 254, 255, 255, 255, 255, 255, 255, 255, 255 }
},
{ { 255, 248, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 251, 255, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 }
},
{ { 255, 253, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 251, 254, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 254, 255, 254, 255, 255, 255, 255, 255, 255, 255, 255 }
},
{ { 255, 254, 253, 255, 254, 255, 255, 255, 255, 255, 255 },
{ 250, 255, 254, 255, 254, 255, 255, 255, 255, 255, 255 },
{ 254, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 }
},
{ { 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 }
}
},
{ { { 217, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 225, 252, 241, 253, 255, 255, 254, 255, 255, 255, 255 },
{ 234, 250, 241, 250, 253, 255, 253, 254, 255, 255, 255 }
},
{ { 255, 254, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 223, 254, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 238, 253, 254, 254, 255, 255, 255, 255, 255, 255, 255 }
},
{ { 255, 248, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 249, 254, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 }
},
{ { 255, 253, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 247, 254, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 }
},
{ { 255, 253, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 252, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 }
},
{ { 255, 254, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 253, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 }
},
{ { 255, 254, 253, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 250, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 254, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 }
},
{ { 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 }
}
},
{ { { 186, 251, 250, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 234, 251, 244, 254, 255, 255, 255, 255, 255, 255, 255 },
{ 251, 251, 243, 253, 254, 255, 254, 255, 255, 255, 255 }
},
{ { 255, 253, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 236, 253, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 251, 253, 253, 254, 254, 255, 255, 255, 255, 255, 255 }
},
{ { 255, 254, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 254, 254, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 }
},
{ { 255, 254, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 254, 254, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 254, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 }
},
{ { 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 254, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 }
},
{ { 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 }
},
{ { 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 }
},
{ { 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 }
}
},
{ { { 248, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 250, 254, 252, 254, 255, 255, 255, 255, 255, 255, 255 },
{ 248, 254, 249, 253, 255, 255, 255, 255, 255, 255, 255 }
},
{ { 255, 253, 253, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 246, 253, 253, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 252, 254, 251, 254, 254, 255, 255, 255, 255, 255, 255 }
},
{ { 255, 254, 252, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 248, 254, 253, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 253, 255, 254, 254, 255, 255, 255, 255, 255, 255, 255 }
},
{ { 255, 251, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 245, 251, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 253, 253, 254, 255, 255, 255, 255, 255, 255, 255, 255 }
},
{ { 255, 251, 253, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 252, 253, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 254, 255, 255, 255, 255, 255, 255, 255, 255, 255 }
},
{ { 255, 252, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 249, 255, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 254, 255, 255, 255, 255, 255, 255, 255, 255 }
},
{ { 255, 255, 253, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 250, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 }
},
{ { 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 254, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 }
}
}
};
#ifndef ONLY_KEYFRAME_CODE
static const uint8_t MVUpdateProba[2][NUM_MV_PROBAS] = {
{ 237, 246, 253, 253, 254, 254, 254, 254,
254, 254, 254, 254, 254, 254, 250, 250,
252, 254, 254 },
{ 231, 243, 245, 253, 254, 254, 254, 254,
254, 254, 254, 254, 254, 254, 251, 251,
254, 254, 254 }
};
#endif
// Paragraph 9.9
void VP8ParseProba(VP8BitReader* const br, VP8Decoder* const dec) {
VP8Proba* const proba = &dec->proba_;
int t, b, c, p;
for (t = 0; t < NUM_TYPES; ++t) {
for (b = 0; b < NUM_BANDS; ++b) {
for (c = 0; c < NUM_CTX; ++c) {
for (p = 0; p < NUM_PROBAS; ++p) {
if (VP8GetBit(br, CoeffsUpdateProba[t][b][c][p])) {
proba->coeffs_[t][b][c][p] = VP8GetValue(br, 8);
}
}
}
}
}
dec->use_skip_proba_ = VP8Get(br);
if (dec->use_skip_proba_) {
dec->skip_p_ = VP8GetValue(br, 8);
}
#ifndef ONLY_KEYFRAME_CODE
if (!dec->frm_hdr_.key_frame_) {
int i;
dec->intra_p_ = VP8GetValue(br, 8);
dec->last_p_ = VP8GetValue(br, 8);
dec->golden_p_ = VP8GetValue(br, 8);
if (VP8Get(br)) { // update y-mode
for (i = 0; i < 4; ++i) {
proba->ymode_[i] = VP8GetValue(br, 8);
}
}
if (VP8Get(br)) { // update uv-mode
for (i = 0; i < 3; ++i) {
proba->uvmode_[i] = VP8GetValue(br, 8);
}
}
// update MV
for (i = 0; i < 2; ++i) {
int k;
for (k = 0; k < NUM_MV_PROBAS; ++k) {
if (VP8GetBit(br, MVUpdateProba[i][k])) {
const int v = VP8GetValue(br, 7);
proba->mv_[i][k] = v ? v << 1 : 1;
}
}
}
}
#endif
}
#if defined(__cplusplus) || defined(c_plusplus)
} // extern "C"
#endif

624
src/dec/vp8.c Normal file
View File

@ -0,0 +1,624 @@
// Copyright 2010 Google Inc.
//
// This code is licensed under the same terms as WebM:
// Software License Agreement: http://www.webmproject.org/license/software/
// Additional IP Rights Grant: http://www.webmproject.org/license/additional/
// -----------------------------------------------------------------------------
//
// main entry for the decoder
//
// Author: Skal (pascal.massimino@gmail.com)
#include <stdlib.h>
#include "vp8i.h"
//-----------------------------------------------------------------------------
// VP8Decoder
static void SetOk(VP8Decoder* const dec) {
dec->status_ = 0;
dec->error_msg_ = "OK";
}
void VP8InitIo(VP8Io* const io) {
if (io) {
memset(io, 0, sizeof(*io));
}
}
VP8Decoder* VP8New() {
VP8Decoder* dec = (VP8Decoder*)calloc(1, sizeof(VP8Decoder));
if (dec) {
SetOk(dec);
dec->ready_ = 0;
}
return dec;
}
int VP8Status(VP8Decoder* const dec) {
if (!dec) return 2;
return dec->status_;
}
const char* VP8StatusMessage(VP8Decoder* const dec) {
if (!dec) return "no object";
if (!dec->error_msg_) return "OK";
return dec->error_msg_;
}
void VP8Delete(VP8Decoder* const dec) {
if (dec) {
VP8Clear(dec);
free(dec);
}
}
int VP8SetError(VP8Decoder* const dec, int error, const char *msg) {
dec->status_ = error;
dec->error_msg_ = msg;
dec->ready_ = 0;
return 0;
}
//-----------------------------------------------------------------------------
// Header parsing
static void ResetSegmentHeader(VP8SegmentHeader* const hdr) {
assert(hdr);
hdr->use_segment_ = 0;
hdr->update_map_ = 0;
hdr->absolute_delta_ = 1;
memset(hdr->quantizer_, 0, sizeof(hdr->quantizer_));
memset(hdr->filter_strength_, 0, sizeof(hdr->filter_strength_));
}
// Paragraph 9.3
static int ParseSegmentHeader(VP8BitReader* br,
VP8SegmentHeader* hdr, VP8Proba* proba) {
assert(br);
assert(hdr);
hdr->use_segment_ = VP8Get(br);
if (hdr->use_segment_) {
hdr->update_map_ = VP8Get(br);
if (VP8Get(br)) { // update data
int s;
hdr->absolute_delta_ = VP8Get(br);
for (s = 0; s < NUM_MB_SEGMENTS; ++s) {
hdr->quantizer_[s] = VP8Get(br) ? VP8GetSignedValue(br, 7) : 0;
}
for (s = 0; s < NUM_MB_SEGMENTS; ++s) {
hdr->filter_strength_[s] = VP8Get(br) ? VP8GetSignedValue(br, 6) : 0;
}
}
if (hdr->update_map_) {
int s;
for (s = 0; s < MB_FEATURE_TREE_PROBS; ++s) {
proba->segments_[s] = VP8Get(br) ? VP8GetValue(br, 8) : 255u;
}
}
} else {
hdr->update_map_ = 0;
}
return 1;
}
// Paragraph 9.5
static int ParsePartitions(VP8Decoder* const dec,
const uint8_t* buf, uint32_t size) {
VP8BitReader* const br = &dec->br_;
const uint8_t* sz = buf;
int last_part;
uint32_t offset;
int p;
dec->num_parts_ = 1 << VP8GetValue(br, 2);
last_part = dec->num_parts_ - 1;
offset = last_part * 3;
if (size <= offset) {
return 0;
}
for (p = 0; p < last_part; ++p) {
const uint32_t psize = sz[0] | (sz[1] << 8) | (sz[2] << 16);
if (offset + psize > size) {
return 0;
}
VP8Init(dec->parts_ + p, buf + offset, psize);
offset += psize;
sz += 3;
}
size -= offset;
VP8Init(dec->parts_ + last_part, buf + offset, size);
return 1;
}
// Paragraph 9.4
static int ParseFilterHeader(VP8BitReader* br, VP8Decoder* const dec) {
VP8FilterHeader* const hdr = &dec->filter_hdr_;
hdr->simple_ = VP8Get(br);
hdr->level_ = VP8GetValue(br, 6);
hdr->sharpness_ = VP8GetValue(br, 3);
hdr->use_lf_delta_ = VP8Get(br);
if (hdr->use_lf_delta_) {
if (VP8Get(br)) { // update lf-delta?
int i;
for (i = 0; i < NUM_REF_LF_DELTAS; ++i) {
if (VP8Get(br)) {
hdr->ref_lf_delta_[i] = VP8GetSignedValue(br, 6);
}
}
for (i = 0; i < NUM_MODE_LF_DELTAS; ++i) {
if (VP8Get(br)) {
hdr->mode_lf_delta_[i] = VP8GetSignedValue(br, 6);
}
}
}
}
dec->filter_type_ = (hdr->level_ == 0) ? 0 : hdr->simple_ ? 1 : 2;
if (dec->filter_type_ > 0) { // precompute filter levels per segment
if (dec->segment_hdr_.use_segment_) {
int s;
for (s = 0; s < NUM_MB_SEGMENTS; ++s) {
int strength = dec->segment_hdr_.filter_strength_[s];
if (!dec->segment_hdr_.absolute_delta_) {
strength += hdr->level_;
}
dec->filter_levels_[s] = strength;
}
} else {
dec->filter_levels_[0] = hdr->level_;
}
}
return 1;
}
static inline uint32_t get_le32(const uint8_t* const data) {
return data[0] | (data[1] << 8) | (data[2] << 16) | (data[3] << 24);
}
// Topmost call
int VP8GetHeaders(VP8Decoder* const dec, VP8Io* const io) {
uint8_t* buf;
uint32_t buf_size;
VP8FrameHeader* frm_hdr;
VP8PictureHeader* pic_hdr;
VP8BitReader* br;
if (dec == NULL) {
return 0;
}
SetOk(dec);
if (io == NULL) {
return VP8SetError(dec, 2, "null VP8Io passed to VP8GetHeaders()");
}
buf = (uint8_t *)io->data;
buf_size = io->data_size;
if (buf == NULL || buf_size <= 4) {
return VP8SetError(dec, 2, "Not enough data to parse frame header");
}
// Skip over valid RIFF headers
if (!memcmp(buf, "RIFF", 4)) {
uint32_t riff_size;
uint32_t chunk_size;
if (buf_size < 20 + 4) {
return VP8SetError(dec, 2, "RIFF: Truncated header.");
}
if (memcmp(buf + 8, "WEBP", 4)) { // wrong image file signature
return VP8SetError(dec, 2, "RIFF: WEBP signature not found.");
}
riff_size = get_le32(buf + 4);
if (memcmp(buf + 12, "VP8 ", 4)) {
return VP8SetError(dec, 2, "RIFF: Invalid compression format.");
}
chunk_size = get_le32(buf + 16);
if ((chunk_size > riff_size + 8) || (chunk_size & 1)) {
return VP8SetError(dec, 2, "RIFF: Inconsistent size information.");
}
buf += 20;
buf_size -= 20;
}
// Paragraph 9.1
{
const uint32_t bits = buf[0] | (buf[1] << 8) | (buf[2] << 16);
frm_hdr = &dec->frm_hdr_;
frm_hdr->key_frame_ = !(bits & 1);
frm_hdr->profile_ = (bits >> 1) & 7;
frm_hdr->show_ = (bits >> 4) & 1;
frm_hdr->partition_length_ = (bits >> 5);
buf += 3;
buf_size -= 3;
}
pic_hdr = &dec->pic_hdr_;
if (frm_hdr->key_frame_) {
// Paragraph 9.2
if (buf_size < 7) {
return VP8SetError(dec, 2, "cannot parse picture header");
}
if (buf[0] != 0x9d || buf[1] != 0x01 || buf[2] != 0x2a) {
return VP8SetError(dec, 2, "Bad code word");
}
pic_hdr->width_ = ((buf[4] << 8) | buf[3]) & 0x3fff;
pic_hdr->xscale_ = buf[4] >> 6; // ratio: 1, 5/4 5/3 or 2
pic_hdr->height_ = ((buf[6] << 8) | buf[5]) & 0x3fff;
pic_hdr->yscale_ = buf[6] >> 6;
buf += 7;
buf_size -= 7;
dec->mb_w_ = (pic_hdr->width_ + 15) >> 4;
dec->mb_h_ = (pic_hdr->height_ + 15) >> 4;
io->width = pic_hdr->width_;
io->height = pic_hdr->height_;
VP8ResetProba(&dec->proba_);
ResetSegmentHeader(&dec->segment_hdr_);
dec->segment_ = 0; // default for intra
}
br = &dec->br_;
VP8Init(br, buf, buf_size);
if (frm_hdr->partition_length_ > buf_size) {
return VP8SetError(dec, 2, "bad partition length");
}
buf += frm_hdr->partition_length_;
buf_size -= frm_hdr->partition_length_;
if (frm_hdr->key_frame_) {
pic_hdr->colorspace_ = VP8Get(br);
pic_hdr->clamp_type_ = VP8Get(br);
}
if (!ParseSegmentHeader(br, &dec->segment_hdr_, &dec->proba_)) {
return VP8SetError(dec, 2, "cannot parse segment header");
}
// Filter specs
if (!ParseFilterHeader(br, dec)) {
return VP8SetError(dec, 2, "cannot parse filter header");
}
if (!ParsePartitions(dec, buf, buf_size)) {
return VP8SetError(dec, 2, "cannot parse partitions");
}
// quantizer change
VP8ParseQuant(dec);
// Frame buffer marking
if (!frm_hdr->key_frame_) {
// Paragraph 9.7
#ifndef ONLY_KEYFRAME_CODE
dec->buffer_flags_ = VP8Get(br) << 0; // update golden
dec->buffer_flags_ |= VP8Get(br) << 1; // update alt ref
if (!(dec->buffer_flags_ & 1)) {
dec->buffer_flags_ |= VP8GetValue(br, 2) << 2;
}
if (!(dec->buffer_flags_ & 2)) {
dec->buffer_flags_ |= VP8GetValue(br, 2) << 4;
}
dec->buffer_flags_ |= VP8Get(br) << 6; // sign bias golden
dec->buffer_flags_ |= VP8Get(br) << 7; // sign bias alt ref
#else
return VP8SetError(dec, 2, "Not a key frame.");
#endif
} else {
dec->buffer_flags_ = 0x003 | 0x100;
}
// Paragraph 9.8
#ifndef ONLY_KEYFRAME_CODE
dec->update_proba_ = VP8Get(br);
if (!dec->update_proba_) { // save for later restore
dec->proba_saved_ = dec->proba_;
}
dec->buffer_flags_ &= 1 << 8;
dec->buffer_flags_ |=
(frm_hdr->key_frame_ || VP8Get(br)) << 8; // refresh last frame
#else
VP8Get(br); // just ignore the value of update_proba_
#endif
VP8ParseProba(br, dec);
// sanitized state
dec->ready_ = 1;
return 1;
}
//-----------------------------------------------------------------------------
// Residual decoding (Paragraph 13.2 / 13.3)
static const uint8_t kBands[16 + 1] = {
0, 1, 2, 3, 6, 4, 5, 6, 6, 6, 6, 6, 6, 6, 6, 7,
0 // extra entry as sentinel
};
static const uint8_t kCat3[] = { 173, 148, 140, 0 };
static const uint8_t kCat4[] = { 176, 155, 140, 135, 0 };
static const uint8_t kCat5[] = { 180, 157, 141, 134, 130, 0 };
static const uint8_t kCat6[] =
{ 254, 254, 243, 230, 196, 177, 153, 140, 133, 130, 129, 0 };
static const uint8_t * const kCat3456[] = { kCat3, kCat4, kCat5, kCat6 };
static const uint8_t kZigzag[16] = {
0, 1, 4, 8, 5, 2, 3, 6, 9, 12, 13, 10, 7, 11, 14, 15
};
typedef const uint8_t (*ProbaArray)[NUM_CTX][NUM_PROBAS]; // for const-casting
// Returns 1 if there's non-zero coeffs, 0 otherwise
static int GetCoeffs(VP8BitReader* const br, ProbaArray prob,
int ctx, const uint16_t dq[2], int n, int16_t* out) {
const uint8_t* p = prob[kBands[n]][ctx];
if (!VP8GetBit(br, p[0])) { // first EOB is more a 'CBP' bit.
return 0;
}
while (1) {
++n;
if (!VP8GetBit(br, p[1])) {
p = prob[kBands[n]][0];
} else { // non zero coeff
int v, j;
if (!VP8GetBit(br, p[2])) {
p = prob[kBands[n]][1];
v = 1;
} else {
if (!VP8GetBit(br, p[3])) {
if (!VP8GetBit(br, p[4])) {
v = 2;
} else {
v = 3 + VP8GetBit(br, p[5]);
}
} else {
if (!VP8GetBit(br, p[6])) {
if (!VP8GetBit(br, p[7])) {
v = 5 + VP8GetBit(br, 159);
} else {
v = 7 + 2 * VP8GetBit(br, 165) + VP8GetBit(br, 145);
}
} else {
const uint8_t* tab;
const int bit1 = VP8GetBit(br, p[8]);
const int bit0 = VP8GetBit(br, p[9 + bit1]);
const int cat = 2 * bit1 + bit0;
v = 0;
for (tab = kCat3456[cat]; *tab; ++tab) {
v += v + VP8GetBit(br, *tab);
}
v += 3 + (8 << cat);
}
}
p = prob[kBands[n]][2];
}
j = kZigzag[n - 1];
out[j] = VP8GetSigned(br, v) * dq[j > 0];
if (n == 16 || !VP8GetBit(br, p[0])) { // EOB
return 1;
}
}
if (n == 16) {
return 1;
}
}
return 0;
}
// Table to unpack four bits into four bytes
static const uint8_t kUnpackTab[16][4] = {
{0, 0, 0, 0}, {1, 0, 0, 0}, {0, 1, 0, 0}, {1, 1, 0, 0},
{0, 0, 1, 0}, {1, 0, 1, 0}, {0, 1, 1, 0}, {1, 1, 1, 0},
{0, 0, 0, 1}, {1, 0, 0, 1}, {0, 1, 0, 1}, {1, 1, 0, 1},
{0, 0, 1, 1}, {1, 0, 1, 1}, {0, 1, 1, 1}, {1, 1, 1, 1} };
// Macro to pack four LSB of four bytes into four bits.
#if defined(__PPC__) || defined(_M_PPC) || defined(_ARCH_PPC) || \
defined(__BIG_ENDIAN__)
#define PACK_CST 0x08040201U
#else
#define PACK_CST 0x01020408U
#endif
#define PACK(X, S) ((((*(uint32_t*)(X)) * PACK_CST) & 0xff000000) >> (S))
static int ParseResiduals(VP8Decoder* const dec,
VP8MB* const mb, VP8BitReader* const token_br) {
int out_t_nz, out_l_nz, first;
ProbaArray ac_prob;
const VP8QuantMatrix* q = &dec->dqm_[dec->segment_];
int16_t* dst = dec->coeffs_;
VP8MB* const left_mb = dec->mb_info_ - 1;
uint8_t nz_ac[4], nz_dc[4];
uint32_t non_zero_ac = 0;
uint32_t non_zero_dc = 0;
uint8_t tnz[4], lnz[4];
int x, y, ch;
memset(dst, 0, 384 * sizeof(*dst));
if (!dec->is_i4x4_) { // parse DC
int16_t dc[16] = { 0 };
const int ctx = mb->dc_nz_ + left_mb->dc_nz_;
mb->dc_nz_ = left_mb->dc_nz_ =
GetCoeffs(token_br, (ProbaArray)dec->proba_.coeffs_[1],
ctx, q->y2_mat_, 0, dc);
first = 1;
ac_prob = (ProbaArray)dec->proba_.coeffs_[0];
VP8TransformWHT(dc, dst);
} else {
first = 0;
ac_prob = (ProbaArray)dec->proba_.coeffs_[3];
}
memcpy(tnz, kUnpackTab[mb->nz_ & 0xf], sizeof(tnz));
memcpy(lnz, kUnpackTab[left_mb->nz_ & 0xf], sizeof(lnz));
for (y = 0; y < 4; ++y) {
int l = lnz[y];
for (x = 0; x < 4; ++x) {
const int ctx = l + tnz[x];
l = GetCoeffs(token_br, ac_prob, ctx,
q->y1_mat_, first, dst);
nz_dc[x] = (dst[0] != 0);
nz_ac[x] = tnz[x] = l;
dst += 16;
}
lnz[y] = l;
non_zero_dc |= PACK(nz_dc, 24 - y * 4);
non_zero_ac |= PACK(nz_ac, 24 - y * 4);
}
out_t_nz = PACK(tnz, 24);
out_l_nz = PACK(lnz, 24);
memcpy(tnz, kUnpackTab[mb->nz_ >> 4], sizeof(tnz));
memcpy(lnz, kUnpackTab[left_mb->nz_ >> 4], sizeof(lnz));
for (ch = 0; ch < 4; ch += 2) {
for (y = 0; y < 2; ++y) {
int l = lnz[ch + y];
for (x = 0; x < 2; ++x) {
const int ctx = l + tnz[ch + x];
l = GetCoeffs(token_br, (ProbaArray)dec->proba_.coeffs_[2],
ctx, q->uv_mat_, 0, dst);
nz_dc[y * 2 + x] = (dst[0] != 0);
nz_ac[y * 2 + x] = tnz[ch + x] = l;
dst += 16;
}
lnz[ch + y] = l;
}
non_zero_dc |= PACK(nz_dc, 8 - ch * 2);
non_zero_ac |= PACK(nz_ac, 8 - ch * 2);
}
out_t_nz |= PACK(tnz, 20);
out_l_nz |= PACK(lnz, 20);
mb->nz_ = out_t_nz;
left_mb->nz_ = out_l_nz;
dec->non_zero_ac_ = non_zero_ac;
dec->non_zero_ = non_zero_ac | non_zero_dc;
mb->skip_ = !dec->non_zero_;
return 1;
}
#undef PACK
//-----------------------------------------------------------------------------
// Main loop
static int ParseFrame(VP8Decoder* const dec, VP8Io* io) {
int ok = 1;
VP8BitReader* const br = &dec->br_;
VP8BitReader* token_br;
for (dec->mb_y_ = 0; dec->mb_y_ < dec->mb_h_; ++dec->mb_y_) {
VP8MB* const left = dec->mb_info_ - 1;
memset(dec->intra_l_, B_DC_PRED, sizeof(dec->intra_l_));
left->nz_ = 0;
left->dc_nz_ = 0;
token_br = &dec->parts_[dec->mb_y_ & (dec->num_parts_ - 1)];
for (dec->mb_x_ = 0; dec->mb_x_ < dec->mb_w_; dec->mb_x_++) {
VP8MB* const info = dec->mb_info_ + dec->mb_x_;
// Note: we don't save segment map (yet), as we don't expect
// to decode more than 1 keyframe.
if (dec->segment_hdr_.update_map_) {
// Hardcoded tree parsing
dec->segment_ = !VP8GetBit(br, dec->proba_.segments_[0]) ?
VP8GetBit(br, dec->proba_.segments_[1]) :
2 + VP8GetBit(br, dec->proba_.segments_[2]);
}
info->skip_ = dec->use_skip_proba_ ? VP8GetBit(br, dec->skip_p_) : 0;
VP8ParseIntraMode(br, dec);
if (!info->skip_) {
if (!ParseResiduals(dec, info, token_br)) {
ok = 0;
break;
}
} else {
left->nz_ = info->nz_ = 0;
if (!dec->is_i4x4_) {
left->dc_nz_ = info->dc_nz_ = 0;
}
dec->non_zero_ = 0;
dec->non_zero_ac_ = 0;
}
VP8ReconstructBlock(dec);
// Store data and save block's filtering params
VP8StoreBlock(dec);
}
if (!ok) {
break;
}
VP8FinishRow(dec, io);
if (dec->br_.eof_ || token_br->eof_) {
ok = 0;
break;
}
}
// Finish
#ifndef ONLY_KEYFRAME_CODE
if (!dec->update_proba_) {
dec->proba_ = dec->proba_saved_;
}
#endif
return ok;
}
// Main entry point
int VP8Decode(VP8Decoder* const dec, VP8Io* const io) {
if (dec == NULL) {
return 0;
}
if (io == NULL) {
return VP8SetError(dec, 2, "NULL VP8Io parameter in VP8Decode().");
}
if (!dec->ready_) {
if (!VP8GetHeaders(dec, io)) {
return 0;
}
}
assert(dec->ready_);
// will allocate memory and prepare everything.
if (!VP8InitFrame(dec, io)) {
VP8Clear(dec);
return VP8SetError(dec, 3, "Allocation failed");
}
if (io->setup && !io->setup(io)) {
VP8Clear(dec);
return VP8SetError(dec, 3, "Frame setup failed");
}
// Main decoding loop
{
const int ret = ParseFrame(dec, io);
if (io->teardown) {
io->teardown(io);
}
if (!ret) {
VP8Clear(dec);
return VP8SetError(dec, 3, "Frame decoding failed");
}
}
dec->ready_ = 0;
return 1;
}
void VP8Clear(VP8Decoder* const dec) {
if (dec == NULL) {
return;
}
if (dec->mem_) {
free(dec->mem_);
}
dec->mem_ = NULL;
dec->mem_size_ = 0;
memset(&dec->br_, 0, sizeof(dec->br_));
dec->ready_ = 0;
}

316
src/dec/vp8i.h Normal file
View File

@ -0,0 +1,316 @@
// Copyright 2010 Google Inc.
//
// This code is licensed under the same terms as WebM:
// Software License Agreement: http://www.webmproject.org/license/software/
// Additional IP Rights Grant: http://www.webmproject.org/license/additional/
// -----------------------------------------------------------------------------
//
// VP8 decoder: internal header.
//
// Author: Skal (pascal.massimino@gmail.com)
#ifndef WEBP_DECODE_VP8I_H_
#define WEBP_DECODE_VP8I_H_
#include <string.h> // for memcpy()
#include "bits.h"
#if defined(__cplusplus) || defined(c_plusplus)
extern "C" {
#endif
//-----------------------------------------------------------------------------
// Various defines and enums
#define ONLY_KEYFRAME_CODE // to remove any code related to P-Frames
// intra prediction modes
enum { B_DC_PRED = 0, // 4x4 modes
B_TM_PRED,
B_VE_PRED,
B_HE_PRED,
B_RD_PRED,
B_VR_PRED,
B_LD_PRED,
B_VL_PRED,
B_HD_PRED,
B_HU_PRED,
NUM_BMODES = B_HU_PRED + 1 - B_DC_PRED, // = 10
// Luma16 or UV modes
DC_PRED = B_DC_PRED, V_PRED = B_VE_PRED,
H_PRED = B_HE_PRED, TM_PRED = B_TM_PRED,
B_PRED = NUM_BMODES, // refined I4x4 mode
// special modes
B_DC_PRED_NOTOP = 4,
B_DC_PRED_NOLEFT = 5,
B_DC_PRED_NOTOPLEFT = 6 };
enum { MB_FEATURE_TREE_PROBS = 3,
NUM_MB_SEGMENTS = 4,
NUM_REF_LF_DELTAS = 4,
NUM_MODE_LF_DELTAS = 4, // I4x4, ZERO, *, SPLIT
MAX_NUM_PARTITIONS = 8,
// Probabilities
NUM_TYPES = 4,
NUM_BANDS = 8,
NUM_CTX = 3,
NUM_PROBAS = 11,
NUM_MV_PROBAS = 19 };
// YUV-cache parameters.
// Constraints are: We need to store one 16x16 block of luma samples (y),
// and two 8x8 chroma blocks (u/v). These are better be 16-bytes aligned,
// in order to be SIMD-friendly. We also need to store the top, left and
// top-left samples (from previously decoded blocks), along with four
// extra top-right samples for luma (intra4x4 prediction only).
// One possible layout is, using 32 * (17 + 9) bytes:
//
// .+------ <- only 1 pixel high
// .|yyyyt.
// .|yyyyt.
// .|yyyyt.
// .|yyyy..
// .+--.+-- <- only 1 pixel high
// .|uu.|vv
// .|uu.|vv
//
// Every character is a 4x4 block, with legend:
// '.' = unused
// 'y' = y-samples 'u' = u-samples 'v' = u-samples
// '|' = left sample, '-' = top sample, '+' = top-left sample
// 't' = extra top-right sample for 4x4 modes
// With this layout, BPS (=Bytes Per Scan-line) is one cacheline size.
#define BPS 32 // this is the common stride used by yuv[]
#define YUV_SIZE (BPS * 17 + BPS * 9)
#define Y_SIZE (BPS * 17)
#define Y_OFF (BPS * 1 + 8)
#define U_OFF (Y_OFF + BPS * 16 + BPS)
#define V_OFF (U_OFF + 16)
//-----------------------------------------------------------------------------
// Headers
typedef struct {
uint8_t key_frame_;
uint8_t profile_;
uint8_t show_;
uint32_t partition_length_;
} VP8FrameHeader;
typedef struct {
uint16_t width_;
uint16_t height_;
uint8_t xscale_;
uint8_t yscale_;
uint8_t colorspace_; // 0 = YCbCr
uint8_t clamp_type_;
} VP8PictureHeader;
// segment features
typedef struct {
int use_segment_;
int update_map_; // whether to update the segment map or not
int absolute_delta_; // absolute or delta values for quantizer and filter
int8_t quantizer_[NUM_MB_SEGMENTS]; // quantization changes
int8_t filter_strength_[NUM_MB_SEGMENTS]; // filter strength for segments
} VP8SegmentHeader;
// Struct collecting all frame-persistent probabilities.
typedef struct {
uint8_t segments_[MB_FEATURE_TREE_PROBS];
// Type: 0:Intra16-AC 1:Intra16-DC 2:Chroma 3:Intra4
uint8_t coeffs_[NUM_TYPES][NUM_BANDS][NUM_CTX][NUM_PROBAS];
#ifndef ONLY_KEYFRAME_CODE
uint8_t ymode_[4], uvmode_[3];
uint8_t mv_[2][NUM_MV_PROBAS];
#endif
} VP8Proba;
// Filter parameters
typedef struct {
int simple_; // 0=complex, 1=simple
int level_; // [0..63]
int sharpness_; // [0..7]
int use_lf_delta_;
int ref_lf_delta_[NUM_REF_LF_DELTAS];
int mode_lf_delta_[NUM_MODE_LF_DELTAS];
} VP8FilterHeader;
//-----------------------------------------------------------------------------
// Informations about the macroblocks.
typedef struct {
// block type
uint8_t skip_:1;
// filter specs
uint8_t f_level_:6; // filter strength: 0..63
uint8_t f_ilevel_:6; // inner limit: 1..63
uint8_t f_inner_:1; // do inner filtering?
// cbp
uint8_t nz_; // non-zero AC/DC coeffs
uint8_t dc_nz_; // non-zero DC coeffs
} VP8MB;
// Dequantization matrices
typedef struct {
uint16_t y1_mat_[2], y2_mat_[2], uv_mat_[2]; // [DC / AC]
} VP8QuantMatrix;
//-----------------------------------------------------------------------------
// VP8Decoder: the main opaque structure handed over to user
struct VP8Decoder {
int status_; // 0 = OK
int ready_; // true if ready to decode a picture with VP8Decode()
const char* error_msg_; // set when status_ is not OK.
// Main data source
VP8BitReader br_;
// headers
VP8FrameHeader frm_hdr_;
VP8PictureHeader pic_hdr_;
VP8FilterHeader filter_hdr_;
VP8SegmentHeader segment_hdr_;
// dimension, in macroblock units.
int mb_w_, mb_h_;
// number of partitions.
int num_parts_;
// per-partition boolean decoders.
VP8BitReader parts_[MAX_NUM_PARTITIONS];
// buffer refresh flags
// bit 0: refresh Gold, bit 1: refresh Alt
// bit 2-3: copy to Gold, bit 4-5: copy to Alt
// bit 6: Gold sign bias, bit 7: Alt sign bias
// bit 8: refresh last frame
uint32_t buffer_flags_;
// dequantization (one set of DC/AC dequant factor per segment)
VP8QuantMatrix dqm_[NUM_MB_SEGMENTS];
// probabilities
VP8Proba proba_;
int use_skip_proba_;
uint8_t skip_p_;
#ifndef ONLY_KEYFRAME_CODE
uint8_t intra_p_, last_p_, golden_p_;
VP8Proba proba_saved_;
int update_proba_;
#endif
// Boundary data cache and persistent buffers.
uint8_t* intra_t_; // top intra modes values: 4 * mb_w_
uint8_t intra_l_[4]; // left intra modes values
uint8_t *y_t_; // top luma samples: 16 * mb_w_
uint8_t *u_t_, *v_t_; // top u/v samples: 8 * mb_w_ each
VP8MB* mb_info_; // contextual macroblock infos (mb_w_ + 1)
uint8_t* yuv_b_; // main block for Y/U/V (size = YUV_SIZE)
int16_t* coeffs_; // 384 coeffs = (16+8+8) * 4*4
uint8_t* cache_y_; // macroblock row for storing unfiltered samples
uint8_t* cache_u_;
uint8_t* cache_v_;
int cache_y_stride_;
int cache_uv_stride_;
// main memory chunk for the above data. Persistent.
void* mem_;
int mem_size_;
// Per macroblock non-persistent infos.
int mb_x_, mb_y_; // current position, in macroblock units
uint8_t is_i4x4_; // true if intra4x4
uint8_t imodes_[16]; // one 16x16 mode (#0) or sixteen 4x4 modes
uint8_t uvmode_; // chroma prediction mode
uint8_t segment_; // block's segment
// bit-wise info about the content of each sub-4x4 blocks: there are 16 bits
// for luma (bits #0->#15), then 4 bits for chroma-u (#16->#19) and 4 bits for
// chroma-v (#20->#23), each corresponding to one 4x4 block in decoding order.
// If the bit is set, the 4x4 block contains some non-zero coefficients.
uint32_t non_zero_;
uint32_t non_zero_ac_;
// Filtering side-info
int filter_type_; // 0=off, 1=simple, 2=complex
uint8_t filter_levels_[NUM_MB_SEGMENTS]; // precalculated per-segment
};
//-----------------------------------------------------------------------------
// internal functions. Not public.
// in vp8.c
int VP8SetError(VP8Decoder* const dec, int error, const char *msg);
// in tree.c
void VP8ResetProba(VP8Proba* const proba);
void VP8ParseProba(VP8BitReader* const br, VP8Decoder* const dec);
void VP8ParseIntraMode(VP8BitReader* const br, VP8Decoder* const dec);
// in quant.c
void VP8ParseQuant(VP8Decoder* const dec);
// in frame.c
int VP8InitFrame(VP8Decoder* const dec, VP8Io* io);
// Predict a block and add residual
void VP8ReconstructBlock(VP8Decoder* const dec);
// Store a block, along with filtering params
void VP8StoreBlock(VP8Decoder* const dec);
// Finalize and transmit a complete row
void VP8FinishRow(VP8Decoder* const dec, VP8Io* io);
// in dsp.c
typedef void (*VP8Idct)(const int16_t* coeffs, uint8_t* dst);
extern VP8Idct VP8Transform;
extern VP8Idct VP8TransformUV;
extern VP8Idct VP8TransformDC;
extern VP8Idct VP8TransformDCUV;
extern void (*VP8TransformWHT)(const int16_t* in, int16_t* out);
// *dst is the destination block, with stride BPS. Boundary samples are
// assumed accessible when needed.
typedef void (*VP8PredFunc)(uint8_t *dst);
extern VP8PredFunc VP8PredLuma16[7];
extern VP8PredFunc VP8PredChroma8[7];
extern VP8PredFunc VP8PredLuma4[11];
void VP8DspInit(); // must be called before anything using the above
void VP8DspInitTables(); // needs to be called no matter what.
// simple filter (only for luma)
typedef void (*VP8SimpleFilterFunc)(uint8_t* p, int stride, int thresh);
extern VP8SimpleFilterFunc VP8SimpleVFilter16;
extern VP8SimpleFilterFunc VP8SimpleHFilter16;
extern VP8SimpleFilterFunc VP8SimpleVFilter16i; // filter 3 inner edges
extern VP8SimpleFilterFunc VP8SimpleHFilter16i;
// regular filter (on both macroblock edges and inner edges)
typedef void (*VP8LumaFilterFunc)(uint8_t* luma, int stride,
int thresh, int ithresh, int hev_t);
typedef void (*VP8ChromaFilterFunc)(uint8_t* u, uint8_t* v, int stride,
int thresh, int ithresh, int hev_t);
// on outter edge
extern VP8LumaFilterFunc VP8VFilter16;
extern VP8LumaFilterFunc VP8HFilter16;
extern VP8ChromaFilterFunc VP8VFilter8;
extern VP8ChromaFilterFunc VP8HFilter8;
// on inner edge
extern VP8LumaFilterFunc VP8VFilter16i; // filtering 3 inner edges altogether
extern VP8LumaFilterFunc VP8HFilter16i;
extern VP8ChromaFilterFunc VP8VFilter8i; // filtering u and v altogether
extern VP8ChromaFilterFunc VP8HFilter8i;
//-----------------------------------------------------------------------------
#if defined(__cplusplus) || defined(c_plusplus)
} // extern "C"
#endif
#endif // WEBP_DECODE_VP8I_H_

579
src/dec/webp.c Normal file
View File

@ -0,0 +1,579 @@
// Copyright 2010 Google Inc.
//
// This code is licensed under the same terms as WebM:
// Software License Agreement: http://www.webmproject.org/license/software/
// Additional IP Rights Grant: http://www.webmproject.org/license/additional/
// -----------------------------------------------------------------------------
//
// Main decoding functions for WEBP images.
//
// Author: Skal (pascal.massimino@gmail.com)
#include <stdlib.h>
#include "vp8i.h"
#include "yuv.h"
#if defined(__cplusplus) || defined(c_plusplus)
extern "C" {
#endif
#define FANCY_UPSCALING // undefined to remove fancy upscaling support
//-----------------------------------------------------------------------------
// RIFF layout is:
// 0ffset tag
// 0...3 "RIFF" 4-byte tag
// 4...7 size of image data (including metadata) starting at offset 8
// 8...11 "WEBP" our form-type signature
// 12..15 "VP8 ": 4-bytes tags, describing the raw video format used
// 16..19 size of the raw VP8 image data, starting at offset 20
// 20.... the VP8 bytes
// There can be extra chunks after the "VP8 " chunk (ICMT, ICOP, ...)
// All 32-bits sizes are in little-endian order.
// Note: chunk data must be padded to multiple of 2 in size
static inline uint32_t get_le32(const uint8_t* const data) {
return data[0] | (data[1] << 8) | (data[2] << 16) | (data[3] << 24);
}
// If a RIFF container is detected, validate it and skip over it.
static uint32_t CheckRIFFHeader(const uint8_t** data_ptr,
uint32_t *data_size_ptr) {
uint32_t chunk_size = 0xffffffffu;
if (*data_size_ptr >= 10 + 20 && !memcmp(*data_ptr, "RIFF", 4)) {
if (memcmp(*data_ptr + 8, "WEBP", 4)) {
return 0; // wrong image file signature
} else {
const uint32_t riff_size = get_le32(*data_ptr + 4);
if (memcmp(*data_ptr + 12, "VP8 ", 4)) {
return 0; // invalid compression format
}
chunk_size = get_le32(*data_ptr + 16);
if ((chunk_size > riff_size + 8) || (chunk_size & 1)) {
return 0; // inconsistent size information.
}
// We have a IFF container. Skip it.
*data_ptr += 20;
*data_size_ptr -= 20;
}
return chunk_size;
}
return *data_size_ptr;
}
//-----------------------------------------------------------------------------
// Fancy upscaling
typedef enum { MODE_RGB = 0, MODE_RGBA = 1,
MODE_BGR = 2, MODE_BGRA = 3,
MODE_YUV = 4 } CSP_MODE;
#ifdef FANCY_UPSCALING
// Given samples laid out in a square as:
// [a b]
// [c d]
// we interpolate u/v as:
// ([9*a + 3*b + 3*c + d 3*a + 9*b + 3*c + d] + [8 8]) / 16
// ([3*a + b + 9*c + 3*d a + 3*b + 3*c + 9*d] [8 8]) / 16
// We process u and v together stashed into 32bit (16bit each).
#define LOAD_UV(u,v) ((u) | ((v) << 16))
#define UPSCALE_FUNC(FUNC_NAME, FUNC, XSTEP) \
static inline void FUNC_NAME(const uint8_t* top_y, const uint8_t* bottom_y, \
const uint8_t* top_u, const uint8_t* top_v, \
const uint8_t* cur_u, const uint8_t* cur_v, \
uint8_t* top_dst, uint8_t* bottom_dst, int len) { \
int x; \
const int last_pixel_pair = (len - 1) >> 1; \
uint32_t tl_uv = LOAD_UV(top_u[0], top_v[0]); /* top-left sample */ \
uint32_t l_uv = LOAD_UV(cur_u[0], cur_v[0]); /* left-sample */ \
if (top_y) { \
const uint32_t uv0 = (3 * tl_uv + l_uv + 0x00020002u) >> 2; \
FUNC(top_y[0], uv0 & 0xff, (uv0 >> 16), top_dst); \
} \
if (bottom_y) { \
const uint32_t uv0 = (3 * l_uv + tl_uv + 0x00020002u) >> 2; \
FUNC(bottom_y[0], uv0 & 0xff, (uv0 >> 16), bottom_dst); \
} \
for (x = 1; x <= last_pixel_pair; ++x) { \
const uint32_t t_uv = LOAD_UV(top_u[x], top_v[x]); /* top sample */ \
const uint32_t uv = LOAD_UV(cur_u[x], cur_v[x]); /* sample */ \
/* precompute invariant values associated with first and second diagonals*/\
const uint32_t avg = tl_uv + t_uv + l_uv + uv + 0x00080008u; \
const uint32_t diag_12 = (avg + 2 * (t_uv + l_uv)) >> 3; \
const uint32_t diag_03 = (avg + 2 * (tl_uv + uv)) >> 3; \
if (top_y) { \
const uint32_t uv0 = (diag_12 + tl_uv) >> 1; \
const uint32_t uv1 = (diag_03 + t_uv) >> 1; \
FUNC(top_y[2 * x - 1], uv0 & 0xff, (uv0 >> 16), \
top_dst + (2 * x - 1) * XSTEP); \
FUNC(top_y[2 * x - 0], uv1 & 0xff, (uv1 >> 16), \
top_dst + (2 * x - 0) * XSTEP); \
} \
if (bottom_y) { \
const uint32_t uv0 = (diag_03 + l_uv) >> 1; \
const uint32_t uv1 = (diag_12 + uv) >> 1; \
FUNC(bottom_y[2 * x - 1], uv0 & 0xff, (uv0 >> 16), \
bottom_dst + (2 * x - 1) * XSTEP); \
FUNC(bottom_y[2 * x + 0], uv1 & 0xff, (uv1 >> 16), \
bottom_dst + (2 * x + 0) * XSTEP); \
} \
tl_uv = t_uv; \
l_uv = uv; \
} \
if (!(len & 1)) { \
if (top_y) { \
const uint32_t uv0 = (3 * tl_uv + l_uv + 0x00020002u) >> 2; \
FUNC(top_y[len - 1], uv0 & 0xff, (uv0 >> 16), \
top_dst + (len - 1) * XSTEP); \
} \
if (bottom_y) { \
const uint32_t uv0 = (3 * l_uv + tl_uv + 0x00020002u) >> 2; \
FUNC(bottom_y[len - 1], uv0 & 0xff, (uv0 >> 16), \
bottom_dst + (len - 1) * XSTEP); \
} \
} \
}
// All variants implemented.
UPSCALE_FUNC(UpscaleRgbLinePair, VP8YuvToRgb, 3)
UPSCALE_FUNC(UpscaleBgrLinePair, VP8YuvToBgr, 3)
UPSCALE_FUNC(UpscaleRgbaLinePair, VP8YuvToRgba, 4)
UPSCALE_FUNC(UpscaleBgraLinePair, VP8YuvToBgra, 4)
// Main driver function.
static inline
void UpscaleLinePair(const uint8_t* top_y, const uint8_t* bottom_y,
const uint8_t* top_u, const uint8_t* top_v,
const uint8_t* cur_u, const uint8_t* cur_v,
uint8_t* top_dst, uint8_t* bottom_dst, int len,
CSP_MODE mode) {
if (mode == MODE_RGB) {
UpscaleRgbLinePair(top_y, bottom_y, top_u, top_v, cur_u, cur_v,
top_dst, bottom_dst, len);
} else if (mode == MODE_BGR) {
UpscaleBgrLinePair(top_y, bottom_y, top_u, top_v, cur_u, cur_v,
top_dst, bottom_dst, len);
} else if (mode == MODE_RGBA) {
UpscaleRgbaLinePair(top_y, bottom_y, top_u, top_v, cur_u, cur_v,
top_dst, bottom_dst, len);
} else {
assert(mode == MODE_BGRA);
UpscaleBgraLinePair(top_y, bottom_y, top_u, top_v, cur_u, cur_v,
top_dst, bottom_dst, len);
}
}
#undef LOAD_UV
#undef UPSCALE_FUNC
#endif // FANCY_UPSCALING
//-----------------------------------------------------------------------------
// Main conversion driver.
typedef struct {
uint8_t* output; // rgb(a) or luma
uint8_t *u, *v;
uint8_t *top_y, *top_u, *top_v;
int stride; // rgb(a) stride or luma stride
int u_stride;
int v_stride;
CSP_MODE mode;
} Params;
static void CustomPut(const VP8Io* io) {
Params *p = (Params*)io->opaque;
const int w = io->width;
const int mb_h = io->mb_h;
const int uv_w = (w + 1) / 2;
assert(!(io->mb_y & 1));
if (w <= 0 || mb_h <= 0) return;
if (p->mode == MODE_YUV) {
uint8_t* const y_dst = p->output + io->mb_y * p->stride;
uint8_t* const u_dst = p->u + (io->mb_y >> 1) * p->u_stride;
uint8_t* const v_dst = p->v + (io->mb_y >> 1) * p->v_stride;
int j;
for (j = 0; j < mb_h; ++j) {
memcpy(y_dst + j * p->stride, io->y + j * io->y_stride, w);
}
for (j = 0; j < (mb_h + 1) / 2; ++j) {
memcpy(u_dst + j * p->u_stride, io->u + j * io->uv_stride, uv_w);
memcpy(v_dst + j * p->v_stride, io->v + j * io->uv_stride, uv_w);
}
} else {
uint8_t* dst = p->output + io->mb_y * p->stride;
if (io->fancy_upscaling) {
#ifdef FANCY_UPSCALING
const uint8_t* cur_y = io->y;
const uint8_t* cur_u = io->u;
const uint8_t* cur_v = io->v;
const uint8_t* top_u = p->top_u;
const uint8_t* top_v = p->top_v;
int y = io->mb_y;
int y_end = io->mb_y + io->mb_h;
if (y == 0) {
// First line is special cased. We mirror the u/v samples at boundary.
UpscaleLinePair(NULL, cur_y, cur_u, cur_v, cur_u, cur_v,
NULL, dst, w, p->mode);
} else {
// We can finish the left-over line from previous call
UpscaleLinePair(p->top_y, cur_y, top_u, top_v, cur_u, cur_v,
dst - p->stride, dst, w, p->mode);
}
// Loop over each output pairs of row.
for (; y + 2 < y_end; y += 2) {
top_u = cur_u;
top_v = cur_v;
cur_u += io->uv_stride;
cur_v += io->uv_stride;
dst += 2 * p->stride;
cur_y += 2 * io->y_stride;
UpscaleLinePair(cur_y - io->y_stride, cur_y,
top_u, top_v, cur_u, cur_v,
dst - p->stride, dst, w, p->mode);
}
// move to last row
cur_y += io->y_stride;
if (y_end != io->height) {
// Save the unfinished samples for next call (as we're not done yet).
memcpy(p->top_y, cur_y, w * sizeof(*p->top_y));
memcpy(p->top_u, cur_u, uv_w * sizeof(*p->top_u));
memcpy(p->top_v, cur_v, uv_w * sizeof(*p->top_v));
} else {
// Process the very last row of even-sized picture
if (!(y_end & 1)) {
UpscaleLinePair(cur_y, NULL, cur_u, cur_v, cur_u, cur_v,
dst + p->stride, NULL, w, p->mode);
}
}
#else
assert(0); // shouldn't happen.
#endif
} else {
// Point-sampling U/V upscaler.
int j;
for (j = 0; j < mb_h; ++j) {
const uint8_t* y_src = io->y + j * io->y_stride;
int i;
for (i = 0; i < w; ++i) {
const int y = y_src[i];
const int u = io->u[(j / 2) * io->uv_stride + (i / 2)];
const int v = io->v[(j / 2) * io->uv_stride + (i / 2)];
if (p->mode == MODE_RGB) {
VP8YuvToRgb(y, u, v, dst + i * 3);
} else if (p->mode == MODE_BGR) {
VP8YuvToBgr(y, u, v, dst + i * 3);
} else if (p->mode == MODE_RGBA) {
VP8YuvToRgba(y, u, v, dst + i * 4);
} else {
VP8YuvToBgra(y, u, v, dst + i * 4);
}
}
dst += p->stride;
}
}
}
}
//-----------------------------------------------------------------------------
static int CustomSetup(VP8Io* io) {
#ifdef FANCY_UPSCALING
Params *p = (Params*)io->opaque;
p->top_y = p->top_u = p->top_v = NULL;
if (p->mode != MODE_YUV) {
const int uv_width = (io->width + 1) >> 1;
p->top_y = (uint8_t*)malloc(io->width + 2 * uv_width);
if (p->top_y == NULL) {
return 0; // memory error.
}
p->top_u = p->top_y + io->width;
p->top_v = p->top_u + uv_width;
io->fancy_upscaling = 1; // activate fancy upscaling
}
#endif
return 1;
}
static void CustomTeardown(const VP8Io* io) {
#ifdef FANCY_UPSCALING
Params *p = (Params*)io->opaque;
if (p->top_y) {
free(p->top_y);
p->top_y = p->top_u = p->top_v = NULL;
}
#endif
}
//-----------------------------------------------------------------------------
// "Into" variants
static uint8_t* DecodeInto(CSP_MODE mode,
const uint8_t* data, uint32_t data_size,
Params* params, int output_size,
int output_u_size, int output_v_size) {
VP8Decoder* dec = VP8New();
VP8Io io;
int ok = 1;
if (dec == NULL) {
return NULL;
}
VP8InitIo(&io);
io.data = data;
io.data_size = data_size;
params->mode = mode;
io.opaque = params;
io.put = CustomPut;
io.setup = CustomSetup;
io.teardown = CustomTeardown;
if (!VP8GetHeaders(dec, &io)) {
VP8Delete(dec);
return NULL;
}
// check output buffers
ok &= (params->stride * io.height <= output_size);
if (mode == MODE_RGB || mode == MODE_BGR) {
ok &= (params->stride >= io.width * 3);
} else if (mode == MODE_RGBA || mode == MODE_BGRA) {
ok &= (params->stride >= io.width * 4);
} else {
// some extra checks for U/V
const int u_size = params->u_stride * ((io.height + 1) / 2);
const int v_size = params->v_stride * ((io.height + 1) / 2);
ok &= (params->stride >= io.width);
ok &= (params->u_stride >= (io.width + 1) / 2) &&
(params->v_stride >= (io.width + 1) / 2);
ok &= (u_size <= output_u_size && v_size <= output_v_size);
}
if (!ok) {
VP8Delete(dec);
return NULL;
}
if (mode != MODE_YUV) {
VP8YUVInit();
}
ok = VP8Decode(dec, &io);
VP8Delete(dec);
return ok ? params->output : NULL;
}
uint8_t* WebPDecodeRGBInto(const uint8_t* data, uint32_t data_size,
uint8_t* output, int output_size,
int output_stride) {
Params params;
if (output == NULL) {
return NULL;
}
params.output = output;
params.stride = output_stride;
return DecodeInto(MODE_RGB, data, data_size, &params, output_size, 0, 0);
}
uint8_t* WebPDecodeRGBAInto(const uint8_t* data, uint32_t data_size,
uint8_t* output, int output_size,
int output_stride) {
Params params;
if (output == NULL) {
return NULL;
}
params.output = output;
params.stride = output_stride;
return DecodeInto(MODE_RGBA, data, data_size, &params, output_size, 0, 0);
}
uint8_t* WebPDecodeBGRInto(const uint8_t* data, uint32_t data_size,
uint8_t* output, int output_size,
int output_stride) {
Params params;
if (output == NULL) {
return NULL;
}
params.output = output;
params.stride = output_stride;
return DecodeInto(MODE_BGR, data, data_size, &params, output_size, 0, 0);
}
uint8_t* WebPDecodeBGRAInto(const uint8_t* data, uint32_t data_size,
uint8_t* output, int output_size,
int output_stride) {
Params params;
if (output == NULL) {
return NULL;
}
params.output = output;
params.stride = output_stride;
return DecodeInto(MODE_BGRA, data, data_size, &params, output_size, 0, 0);
}
uint8_t* WebPDecodeYUVInto(const uint8_t* data, uint32_t data_size,
uint8_t* luma, int luma_size, int luma_stride,
uint8_t* u, int u_size, int u_stride,
uint8_t* v, int v_size, int v_stride) {
Params params;
if (luma == NULL) {
return NULL;
}
params.output = luma;
params.stride = luma_stride;
params.u = u;
params.u_stride = u_stride;
params.v = v;
params.v_stride = v_stride;
return DecodeInto(MODE_YUV, data, data_size, &params,
luma_size, u_size, v_size);
}
//-----------------------------------------------------------------------------
static uint8_t* Decode(CSP_MODE mode, const uint8_t* data, uint32_t data_size,
int* width, int* height, Params* params_out) {
int w, h, stride;
int uv_size = 0;
int uv_stride = 0;
int size;
uint8_t* output;
Params params = { 0 };
if (!WebPGetInfo(data, data_size, &w, &h)) {
return NULL;
}
if (width) *width = w;
if (height) *height = h;
// initialize output buffer, now that dimensions are known.
stride = (mode == MODE_RGB || mode == MODE_BGR) ? 3 * w
: (mode == MODE_RGBA || mode == MODE_BGRA) ? 4 * w
: w;
size = stride * h;
if (mode == MODE_YUV) {
uv_stride = (w + 1) / 2;
uv_size = uv_stride * ((h + 1) / 2);
}
output = (uint8_t*)malloc(size + 2 * uv_size);
if (!output) {
return NULL;
}
params.output = output;
params.stride = stride;
if (mode == MODE_YUV) {
params.u = output + size;
params.u_stride = uv_stride;
params.v = output + size + uv_size;
params.v_stride = uv_stride;
}
if (params_out) *params_out = params;
return DecodeInto(mode, data, data_size, &params, size, uv_size, uv_size);
}
uint8_t* WebPDecodeRGB(const uint8_t* data, uint32_t data_size,
int *width, int *height) {
return Decode(MODE_RGB, data, data_size, width, height, NULL);
}
uint8_t* WebPDecodeRGBA(const uint8_t* data, uint32_t data_size,
int *width, int *height) {
return Decode(MODE_RGBA, data, data_size, width, height, NULL);
}
uint8_t* WebPDecodeBGR(const uint8_t* data, uint32_t data_size,
int *width, int *height) {
return Decode(MODE_BGR, data, data_size, width, height, NULL);
}
uint8_t* WebPDecodeBGRA(const uint8_t* data, uint32_t data_size,
int *width, int *height) {
return Decode(MODE_BGRA, data, data_size, width, height, NULL);
}
uint8_t* WebPDecodeYUV(const uint8_t* data, uint32_t data_size,
int *width, int *height, uint8_t** u, uint8_t** v,
int *stride, int* uv_stride) {
Params params;
uint8_t* const out = Decode(MODE_YUV, data, data_size,
width, height, &params);
if (out) {
*u = params.u;
*v = params.v;
*stride = params.stride;
*uv_stride = params.u_stride;
assert(params.u_stride == params.v_stride);
}
return out;
}
//-----------------------------------------------------------------------------
// WebPGetInfo()
int WebPGetInfo(const uint8_t* data, uint32_t data_size,
int *width, int *height) {
const uint32_t chunk_size = CheckRIFFHeader(&data, &data_size);
if (!chunk_size) {
return 0; // unsupported RIFF header
}
// Validate raw video data
if (data_size < 10) {
return 0; // not enough data
}
// check signature
if (data[3] != 0x9d || data[4] != 0x01 || data[5] != 0x2a) {
return 0; // Wrong signature.
} else {
const uint32_t bits = data[0] | (data[1] << 8) | (data[2] << 16);
const int key_frame = !(bits & 1);
const int w = ((data[7] << 8) | data[6]) & 0x3fff;
const int h = ((data[9] << 8) | data[8]) & 0x3fff;
if (!key_frame) { // Not a keyframe.
return 0;
}
if (((bits >> 1) & 7) > 3) {
return 0; // unknown profile
}
if (!((bits >> 4) & 1)) {
return 0; // first frame is invisible!
}
if (((bits >> 5)) >= chunk_size) { // partition_length
return 0; // inconsistent size information.
}
if (width) {
*width = w;
}
if (height) {
*height = h;
}
return 1;
}
}
#if defined(__cplusplus) || defined(c_plusplus)
} // extern "C"
#endif

46
src/dec/yuv.c Normal file
View File

@ -0,0 +1,46 @@
// Copyright 2010 Google Inc.
//
// This code is licensed under the same terms as WebM:
// Software License Agreement: http://www.webmproject.org/license/software/
// Additional IP Rights Grant: http://www.webmproject.org/license/additional/
// -----------------------------------------------------------------------------
//
// YUV->RGB conversion function
//
// Author: Skal (pascal.massimino@gmail.com)
#include "yuv.h"
#if defined(__cplusplus) || defined(c_plusplus)
extern "C" {
#endif
enum { YUV_HALF = 1 << (YUV_FIX - 1) };
int16_t VP8kVToR[256], VP8kUToB[256];
int32_t VP8kVToG[256], VP8kUToG[256];
uint8_t VP8kClip[YUV_RANGE_MAX - YUV_RANGE_MIN];
static int done = 0;
void VP8YUVInit() {
int i;
if (done) {
return;
}
for (i = 0; i < 256; ++i) {
VP8kVToR[i] = (89858 * (i - 128) + YUV_HALF) >> YUV_FIX;
VP8kUToG[i] = -22014 * (i - 128) + YUV_HALF;
VP8kVToG[i] = -45773 * (i - 128);
VP8kUToB[i] = (113618 * (i - 128) + YUV_HALF) >> YUV_FIX;
}
for (i = YUV_RANGE_MIN; i < YUV_RANGE_MAX; ++i) {
const int k = ((i - 16) * 76283 + YUV_HALF) >> YUV_FIX;
VP8kClip[i - YUV_RANGE_MIN] = (k < 0) ? 0 : (k > 255) ? 255 : k;
}
done = 1;
}
#if defined(__cplusplus) || defined(c_plusplus)
} // extern "C"
#endif

66
src/dec/yuv.h Normal file
View File

@ -0,0 +1,66 @@
// Copyright 2010 Google Inc.
//
// This code is licensed under the same terms as WebM:
// Software License Agreement: http://www.webmproject.org/license/software/
// Additional IP Rights Grant: http://www.webmproject.org/license/additional/
// -----------------------------------------------------------------------------
//
// inline YUV->RGB conversion function
//
// Author: Skal (pascal.massimino@gmail.com)
#ifndef WEBP_DECODE_YUV_H_
#define WEBP_DECODE_YUV_H_
#include "webp/decode_vp8.h"
#if defined(__cplusplus) || defined(c_plusplus)
extern "C" {
#endif
enum { YUV_FIX = 16, // fixed-point precision
YUV_RANGE_MIN = -227, // min value of r/g/b output
YUV_RANGE_MAX = 256 + 226 // max value of r/g/b output
};
extern int16_t VP8kVToR[256], VP8kUToB[256];
extern int32_t VP8kVToG[256], VP8kUToG[256];
extern uint8_t VP8kClip[YUV_RANGE_MAX - YUV_RANGE_MIN];
inline static void VP8YuvToRgb(uint8_t y, uint8_t u, uint8_t v,
uint8_t* const rgb) {
const int r_off = VP8kVToR[v];
const int g_off = (VP8kVToG[v] + VP8kUToG[u]) >> YUV_FIX;
const int b_off = VP8kUToB[u];
rgb[0] = VP8kClip[y + r_off - YUV_RANGE_MIN];
rgb[1] = VP8kClip[y + g_off - YUV_RANGE_MIN];
rgb[2] = VP8kClip[y + b_off - YUV_RANGE_MIN];
}
inline static void VP8YuvToRgba(int y, int u, int v, uint8_t* const rgba) {
VP8YuvToRgb(y, u, v, rgba);
rgba[3] = 0xff;
}
inline static void VP8YuvToBgr(uint8_t y, uint8_t u, uint8_t v,
uint8_t* const bgr) {
const int r_off = VP8kVToR[v];
const int g_off = (VP8kVToG[v] + VP8kUToG[u]) >> YUV_FIX;
const int b_off = VP8kUToB[u];
bgr[0] = VP8kClip[y + b_off - YUV_RANGE_MIN];
bgr[1] = VP8kClip[y + g_off - YUV_RANGE_MIN];
bgr[2] = VP8kClip[y + r_off - YUV_RANGE_MIN];
}
inline static void VP8YuvToBgra(int y, int u, int v, uint8_t* const bgra) {
VP8YuvToBgr(y, u, v, bgra);
bgra[3] = 0xff;
}
// Must be called before everything, to initialize the tables.
void VP8YUVInit();
#if defined(__cplusplus) || defined(c_plusplus)
} // extern "C"
#endif
#endif // WEBP_DECODE_YUV_H_