mirror of
https://chromium.googlesource.com/libyuv/libyuv
synced 2026-10-28 18:07:12 +08:00
63e54f7374
Implement 32-bit and 64-bit Arm NEON versions (in row_neon.cc and row_neon64.cc) for the 12 remaining row functions that previously fell back to C on Arm: - AR30ToAB30Row_NEON (and AR30ToAB30Row_Any_NEON) - AR30ToABGRRow_NEON (and AR30ToABGRRow_Any_NEON) - AR30ToARGBRow_NEON (and AR30ToARGBRow_Any_NEON) - ARGBAffineRow_NEON - ARGBColorTableRow_NEON - ARGBCopyAlphaRow_NEON (and ARGBCopyAlphaRow_Any_NEON) - ARGBLumaColorTableRow_NEON - ARGBPolynomialRow_NEON (and ARGBPolynomialRow_Any_NEON) - ARGBUnattenuateRow_NEON (and ARGBUnattenuateRow_Any_NEON) - CopyRow_16_NEON (and CopyRow_16_Any_NEON) - MirrorRow_16_NEON (and MirrorRow_16_Any_NEON) - RGBColorTableRow_NEON Add HAS_*_NEON macros, prototypes, _Any wrappers, NEON dispatches in convert.cc, convert_argb.cc, and planar_functions.cc, and enable unit tests in planar_test.cc. Arm row function SIMD coverage is now 164/164 (100%). ARM64 (Apple M6) Benchmark (1280 width, 10000 iterations): ---------------------------------------------------------------- Function Before C (ms) After NEON (ms) Improvement ---------------------------------------------------------------- AR30ToAB30Row 1.20 0.70 1.72x AR30ToABGRRow 1.88 1.03 1.83x AR30ToARGBRow 1.88 0.87 2.17x ARGBAffineRow 7.96 2.31 3.44x ARGBColorTableRow 8.17 7.43 1.10x ARGBCopyAlphaRow 1.37 0.53 2.58x ARGBLumaColorTableRow 8.18 5.24 1.56x ARGBPolynomialRow 8.73 7.55 1.16x ARGBUnattenuateRow 6.00 2.05 2.93x CopyRow_16 0.18 0.18 1.00x MirrorRow_16 0.34 0.20 1.73x RGBColorTableRow 6.15 5.58 1.10x ---------------------------------------------------------------- Maximum SIMD ISA supported per C row function: ---------------------------------------------------------------- C Row Function arm32 arm64 x86 riscv ---------------------------------------------------------------- AB64ToARGBRow_C NEON NEON AVX2 RVV ABGRToAR30Row_C - NEON AVX512BW - AR30ToAB30Row_C NEON NEON AVX512BW - AR30ToABGRRow_C NEON NEON AVX512BW - AR30ToARGBRow_C NEON NEON AVX512BW - AR64ShuffleRow_C NEON NEON AVX512BW RVV AR64ToAB64Row_C NEON NEON AVX512BW RVV AR64ToARGBRow_C NEON NEON AVX2 RVV ARGB1555ToARGBRow_C NEON SVE2 AVX2 - ARGB1555ToUVMatrixRow_C NEON NEON AVX2 - ARGB1555ToUVRow_C NEON NEON - - ARGB1555ToYMatrixRow_C NEON NEON AVX2 - ARGB1555ToYRow_C NEON NEON - - ARGB4444ToARGBRow_C NEON NEON AVX2 - ARGB4444ToUVMatrixRow_C NEON NEON AVX2 - ARGB4444ToUVRow_C NEON NEON - - ARGB4444ToYMatrixRow_C NEON NEON AVX2 - ARGB4444ToYRow_C NEON NEON - - ARGBAddRow_C NEON NEON AVX2 - ARGBAffineRow_C NEON NEON SSE2 - ARGBAlphaBlendRow_C NEON SME AVX512BW RVV ARGBAttenuateRow_C NEON NEON AVX512BW RVV ARGBBlendRow_C NEON NEON AVX512BW RVV ARGBColorMatrixRow_C NEON I8MM SSSE3 - ARGBColorTableRow_C NEON NEON X86 - ARGBCopyAlphaRow_C NEON NEON AVX2 - ARGBCopyYToAlphaRow_C NEON NEON AVX512BW RVV ARGBExtractAlphaRow_C NEON NEON AVX512BW RVV ARGBGrayRow_C NEON DOTPROD AVX512VNNI - ARGBLumaColorTableRow_C NEON NEON SSSE3 - ARGBMirrorRow_C NEON NEON AVX512BW RVV ARGBMultiplyRow_C NEON SME AVX512BW RVV ARGBOrRow_C NEON SME AVX512BW RVV ARGBPolynomialRow_C NEON NEON AVX2 - ARGBQuantizeRow_C NEON NEON SSE2 - ARGBSepiaRow_C NEON DOTPROD SSSE3 - ARGBSetRow_C NEON NEON X86 RVV ARGBShadeRow_C NEON NEON SSE2 - ARGBShuffleRow_C NEON NEON AVX512BW RVV ARGBSubtractRow_C NEON NEON AVX2 - ARGBToAB64Row_C NEON NEON AVX2 RVV ARGBToABGRRow_C NEON NEON AVX512BW RVV ARGBToAR30Row_C - NEON AVX512BW - ARGBToAR64Row_C NEON NEON AVX2 RVV ARGBToARGB1555Row_C NEON NEON AVX2 - ARGBToARGB4444Row_C NEON NEON AVX2 - ARGBToBGRARow_C NEON NEON AVX512BW RVV ARGBToRAWRow_C NEON SVE2 AVX512BW RVV ARGBToRGB24Row_C NEON SVE2 AVX512BW RVV ARGBToRGB565DitherRow_C NEON SVE2 AVX2 - ARGBToRGB565Row_C NEON SVE2 AVX2 - ARGBToRGBARow_C NEON NEON AVX512BW RVV ARGBToUV444MatrixRow_C NEON I8MM AVX512VNNI RVV ARGBToUVMatrixRow_C NEON SME AVX512VNNI RVV ARGBToYMatrixRow_C NEON SME AVX512VNNI RVV ARGBUnattenuateRow_C NEON NEON AVX2 - AYUVToUVRow_C NEON SVE2 AVX512VBMI - AYUVToVURow_C NEON SVE2 AVX512VBMI - AYUVToYRow_C NEON NEON AVX512VBMI - BlendPlaneRow_C NEON NEON AVX512BW RVV ByteToFloatRow_C NEON NEON - - ComputeCumulativeSumRow_C - NEON AVX2 - Convert16To8Row_C NEON SME AVX512BW RVV Convert8To16Row_C NEON SME AVX512BW RVV Convert8To8Row_C NEON SME AVX2 - CopyRow_C NEON SME AVX512BW RVV CopyRow_16_C NEON NEON - - CumulativeSumToAverageRow_C - NEON AVX2 - DetileRow_C NEON NEON SSE2 - DetileRow_16_C NEON NEON AVX - DetileSplitUVRow_C NEON NEON SSSE3 - DivideRow_16_C NEON SVE2 AVX2 - GaussRow_C NEON NEON - - GaussRow_F32_C - NEON - - HalfFloatRow_C NEON SVE2 AVX2 - HalfMergeUVRow_C NEON SVE2 AVX512BW - HalfRow_16To8_C NEON SME AVX512BW RVV HalfWidthRow_16To8_C NEON SME AVX512BW RVV I210AlphaToARGBRow_C - SME AVX2 - I210ToAR30Row_C - SME AVX2 - I210ToARGBRow_C - SME AVX2 - I212ToAR30Row_C - SME AVX2 - I212ToARGBRow_C - SME AVX2 - I400ToARGBRow_C NEON SME AVX512BW RVV I410AlphaToARGBRow_C - SME AVX2 - I410ToAR30Row_C - SME AVX2 - I410ToARGBRow_C - SME AVX2 - I422AlphaToARGBRow_C NEON SME AVX512BW RVV I422ToAR30Row_C - SME AVX512BW RVV I422ToARGB1555Row_C NEON SME AVX2 - I422ToARGB4444Row_C NEON SME AVX2 - I422ToARGBRow_C NEON SME AVX512BW RVV I422ToRGB24Row_C NEON SME AVX512VBMI RVV I422ToRGB565Row_C NEON SME AVX2 - I422ToRGBARow_C NEON SME AVX2 RVV I422ToUYVYRow_C NEON NEON AVX2 - I422ToYUY2Row_C NEON NEON AVX2 - I444AlphaToARGBRow_C NEON SME AVX2 RVV I444ToARGBRow_C NEON SME AVX2 RVV I444ToRGB24Row_C NEON SME AVX2 RVV InterpolateRow_C NEON SME AVX512BW RVV InterpolateRow_16_C NEON SME AVX512BW - J400ToARGBRow_C NEON NEON AVX512BW RVV MergeAR64Row_C NEON NEON AVX2 - MergeARGB16To8Row_C NEON NEON AVX2 - MergeARGBRow_C NEON NEON AVX2 RVV MergeRGBRow_C NEON NEON AVX512VBMI RVV MergeUVRow_C NEON SME AVX512BW RVV MergeUVRow_16_C NEON SME AVX2 - MergeXR30Row_C NEON NEON AVX512BW - MergeXR64Row_C NEON NEON AVX2 - MergeXRGB16To8Row_C NEON NEON AVX2 - MergeXRGBRow_C NEON NEON AVX2 RVV MirrorRow_C NEON NEON AVX512BW - MirrorRow_16_C NEON NEON AVX512BW - MirrorSplitUVRow_C NEON NEON AVX512BW - MirrorUVRow_C NEON NEON AVX512BW RVV MultiplyRow_16_C NEON SME AVX512BW RVV NV12ToARGBRow_C NEON SME AVX512BW RVV NV12ToRGB24Row_C NEON SME AVX512VBMI RVV NV12ToRGB565Row_C NEON NEON AVX2 - NV21ToARGBRow_C NEON SME AVX512BW RVV NV21ToRGB24Row_C NEON SME AVX512VBMI RVV NV21ToYUV24Row_C NEON NEON AVX2 - P210ToAR30Row_C - SME AVX2 - P210ToARGBRow_C - SME AVX2 - P410ToAR30Row_C - SME AVX2 - P410ToARGBRow_C - SME AVX2 - RAWToARGBRow_C NEON SVE2 AVX512BW RVV RAWToRGB24Row_C NEON SVE2 AVX512VBMI RVV RAWToRGBARow_C NEON SVE2 AVX512VBMI RVV RGB24MirrorRow_C NEON NEON AVX512VBMI - RGB24ToARGBRow_C NEON SVE2 AVX512BW RVV RGB565ToARGBRow_C NEON NEON AVX2 - RGB565ToUVMatrixRow_C NEON NEON AVX2 - RGB565ToUVRow_C NEON NEON - - RGB565ToYMatrixRow_C NEON NEON AVX2 - RGB565ToYRow_C NEON NEON - - RGBAToARGBRow_C NEON NEON AVX512BW RVV RGBColorTableRow_C NEON NEON X86 - RGBToUV444MatrixRow_C NEON NEON AVX512BW RVV RGBToUVMatrixRow_C NEON NEON AVX512BW RVV RGBToYMatrixRow_C NEON DOTPROD AVX512BW RVV SetRow_C NEON NEON ERMS - SobelRow_C NEON NEON SSE2 - SobelToPlaneRow_C NEON NEON SSE2 - SobelXRow_C NEON NEON SSE2 - SobelXYRow_C NEON NEON SSE2 - SobelYRow_C NEON NEON SSE2 - SplitARGBRow_C NEON NEON AVX2 RVV SplitRGBRow_C NEON NEON AVX2 RVV SplitUVRow_C NEON SVE2 AVX512BW RVV SplitUVRow_16_C NEON NEON AVX2 - SplitXRGBRow_C NEON NEON AVX2 RVV SwapUVRow_C NEON NEON AVX512BW RVV UYVYToARGBRow_C NEON SME AVX2 - UYVYToUV422Row_C NEON NEON AVX2 - UYVYToUVRow_C NEON SVE2 AVX2 - UYVYToYRow_C NEON SVE2 AVX2 - YUY2ToARGBRow_C NEON SME AVX2 - YUY2ToNVUVRow_C NEON NEON AVX2 - YUY2ToUV422Row_C NEON NEON AVX2 - YUY2ToUVRow_C NEON SVE2 AVX2 - YUY2ToYRow_C NEON SVE2 AVX2 - ---------------------------------------------------------------- SIMD (164/164 100%) 146 89% 164 100% 154 94% 64 39% ---------------------------------------------------------------- Test: libyuv_unittest --gtest_filter=*AR30*:*Affine*:*ColorTable*:*CopyAlpha*:*Polynomial*:*Unattenuate*:*MirrorPlane_16* Bug: None Change-Id: Ib8b68beb77b47de6b996d41fd080397c61539ffb Reviewed-on: https://chromium-review.googlesource.com/c/libyuv/libyuv/+/8666404 Reviewed-by: richard winterton <rrwinterton@gmail.com> Reviewed-by: Dale Curtis <dalecurtis@chromium.org> Commit-Queue: Frank Barchard <fbarchard@google.com>
5285 lines
162 KiB
C++
5285 lines
162 KiB
C++
/*
|
|
* Copyright 2011 The LibYuv Project Authors. All rights reserved.
|
|
*
|
|
* Use of this source code is governed by a BSD-style license
|
|
* that can be found in the LICENSE file in the root of the source
|
|
* tree. An additional intellectual property rights grant can be found
|
|
* in the file PATENTS. All contributing project authors may
|
|
* be found in the AUTHORS file in the root of the source tree.
|
|
*/
|
|
|
|
#include "libyuv/convert.h"
|
|
|
|
#include <limits.h>
|
|
|
|
#include "libyuv/convert_from_argb.h"
|
|
#include "libyuv/cpu_id.h"
|
|
#include "libyuv/planar_functions.h"
|
|
#include "libyuv/rotate.h"
|
|
#include "libyuv/row.h"
|
|
#include "libyuv/scale.h" // For ScalePlane()
|
|
#include "libyuv/scale_row.h" // For FixedDiv
|
|
#include "libyuv/scale_uv.h" // For UVScale()
|
|
|
|
#ifdef __cplusplus
|
|
namespace libyuv {
|
|
|
|
extern const struct ArgbConstants kArgbI601Constants;
|
|
extern const struct ArgbConstants kArgbJPEGConstants;
|
|
extern "C" {
|
|
#endif
|
|
|
|
// Subsample amount uses a shift.
|
|
// v is value
|
|
// a is amount to add to round up
|
|
// s is shift to subsample down
|
|
#define SUBSAMPLE(v, a, s) (v < 0) ? (-((-v + a) >> s)) : ((v + a) >> s)
|
|
static __inline int Abs(int v) {
|
|
return v >= 0 ? v : -v;
|
|
}
|
|
|
|
// Any I4xx To I420 format
|
|
static int I4xxToI420(const uint8_t* src_y,
|
|
int src_stride_y,
|
|
const uint8_t* src_u,
|
|
int src_stride_u,
|
|
const uint8_t* src_v,
|
|
int src_stride_v,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int src_y_width,
|
|
int src_y_height,
|
|
int src_uv_width,
|
|
int src_uv_height) {
|
|
int r;
|
|
if ((!src_y && dst_y) || !src_u || !src_v || !dst_u || !dst_v ||
|
|
src_y_width <= 0 || src_y_height == 0 || src_y_height == INT_MIN ||
|
|
src_uv_width <= 0 || src_uv_height == 0) {
|
|
return -1;
|
|
}
|
|
const int dst_y_width = src_y_width;
|
|
const int dst_y_height = Abs(src_y_height);
|
|
const int dst_uv_width = SUBSAMPLE(dst_y_width, 1, 1);
|
|
const int dst_uv_height = SUBSAMPLE(dst_y_height, 1, 1);
|
|
if (dst_y) {
|
|
CopyPlane(src_y, src_stride_y, dst_y, dst_stride_y, src_y_width,
|
|
src_y_height);
|
|
}
|
|
r = ScalePlane(src_u, src_stride_u, src_uv_width, src_uv_height, dst_u,
|
|
dst_stride_u, dst_uv_width, dst_uv_height, kFilterBilinear);
|
|
if (r != 0) {
|
|
return r;
|
|
}
|
|
r = ScalePlane(src_v, src_stride_v, src_uv_width, src_uv_height, dst_v,
|
|
dst_stride_v, dst_uv_width, dst_uv_height, kFilterBilinear);
|
|
return r;
|
|
}
|
|
|
|
// Copy I420 with optional vertical flipping using negative height.
|
|
LIBYUV_API
|
|
int I420Copy(const uint8_t* src_y,
|
|
int src_stride_y,
|
|
const uint8_t* src_u,
|
|
int src_stride_u,
|
|
const uint8_t* src_v,
|
|
int src_stride_v,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
int halfwidth = (width + 1) >> 1;
|
|
int halfheight = (height + 1) >> 1;
|
|
if ((!src_y && dst_y) || !src_u || !src_v || !dst_u || !dst_v || width <= 0 ||
|
|
height == 0 || height == INT_MIN) {
|
|
return -1;
|
|
}
|
|
// Negative height means invert the image.
|
|
if (height < 0) {
|
|
height = -height;
|
|
halfheight = (height + 1) >> 1;
|
|
src_y = src_y + (ptrdiff_t)(height - 1) * src_stride_y;
|
|
src_u = src_u + (ptrdiff_t)(halfheight - 1) * src_stride_u;
|
|
src_v = src_v + (ptrdiff_t)(halfheight - 1) * src_stride_v;
|
|
src_stride_y = -src_stride_y;
|
|
src_stride_u = -src_stride_u;
|
|
src_stride_v = -src_stride_v;
|
|
}
|
|
|
|
if (dst_y) {
|
|
CopyPlane(src_y, src_stride_y, dst_y, dst_stride_y, width, height);
|
|
}
|
|
// Copy UV planes.
|
|
CopyPlane(src_u, src_stride_u, dst_u, dst_stride_u, halfwidth, halfheight);
|
|
CopyPlane(src_v, src_stride_v, dst_v, dst_stride_v, halfwidth, halfheight);
|
|
return 0;
|
|
}
|
|
|
|
// Copy I010 with optional flipping.
|
|
LIBYUV_API
|
|
int I010Copy(const uint16_t* src_y,
|
|
int src_stride_y,
|
|
const uint16_t* src_u,
|
|
int src_stride_u,
|
|
const uint16_t* src_v,
|
|
int src_stride_v,
|
|
uint16_t* dst_y,
|
|
int dst_stride_y,
|
|
uint16_t* dst_u,
|
|
int dst_stride_u,
|
|
uint16_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
int halfwidth = (width + 1) >> 1;
|
|
int halfheight = (height + 1) >> 1;
|
|
if ((!src_y && dst_y) || !src_u || !src_v || !dst_u || !dst_v || width <= 0 ||
|
|
height == 0 || height == INT_MIN) {
|
|
return -1;
|
|
}
|
|
// Negative height means invert the image.
|
|
if (height < 0) {
|
|
height = -height;
|
|
halfheight = (height + 1) >> 1;
|
|
src_y = src_y + (ptrdiff_t)(height - 1) * src_stride_y;
|
|
src_u = src_u + (ptrdiff_t)(halfheight - 1) * src_stride_u;
|
|
src_v = src_v + (ptrdiff_t)(halfheight - 1) * src_stride_v;
|
|
src_stride_y = -src_stride_y;
|
|
src_stride_u = -src_stride_u;
|
|
src_stride_v = -src_stride_v;
|
|
}
|
|
|
|
if (dst_y) {
|
|
CopyPlane_16(src_y, src_stride_y, dst_y, dst_stride_y, width, height);
|
|
}
|
|
// Copy UV planes.
|
|
CopyPlane_16(src_u, src_stride_u, dst_u, dst_stride_u, halfwidth, halfheight);
|
|
CopyPlane_16(src_v, src_stride_v, dst_v, dst_stride_v, halfwidth, halfheight);
|
|
return 0;
|
|
}
|
|
|
|
static int Planar16bitTo8bit(const uint16_t* src_y,
|
|
int src_stride_y,
|
|
const uint16_t* src_u,
|
|
int src_stride_u,
|
|
const uint16_t* src_v,
|
|
int src_stride_v,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height,
|
|
int subsample_x,
|
|
int subsample_y,
|
|
int depth) {
|
|
if ((!src_y && dst_y) || !src_u || !src_v || !dst_u || !dst_v || width <= 0 ||
|
|
height == 0 || height == INT_MIN) {
|
|
return -1;
|
|
}
|
|
int uv_width = SUBSAMPLE(width, subsample_x, subsample_x);
|
|
int uv_height = SUBSAMPLE(height, subsample_y, subsample_y);
|
|
int scale = 1 << (24 - depth);
|
|
// Negative height means invert the image.
|
|
if (height < 0) {
|
|
height = -height;
|
|
uv_height = -uv_height;
|
|
src_y = src_y + (ptrdiff_t)(height - 1) * src_stride_y;
|
|
src_u = src_u + (ptrdiff_t)(uv_height - 1) * src_stride_u;
|
|
src_v = src_v + (ptrdiff_t)(uv_height - 1) * src_stride_v;
|
|
src_stride_y = -src_stride_y;
|
|
src_stride_u = -src_stride_u;
|
|
src_stride_v = -src_stride_v;
|
|
}
|
|
|
|
// Convert Y plane.
|
|
if (dst_y) {
|
|
Convert16To8Plane(src_y, src_stride_y, dst_y, dst_stride_y, scale, width,
|
|
height);
|
|
}
|
|
// Convert UV planes.
|
|
Convert16To8Plane(src_u, src_stride_u, dst_u, dst_stride_u, scale, uv_width,
|
|
uv_height);
|
|
Convert16To8Plane(src_v, src_stride_v, dst_v, dst_stride_v, scale, uv_width,
|
|
uv_height);
|
|
return 0;
|
|
}
|
|
|
|
static int I41xToI420(const uint16_t* src_y,
|
|
int src_stride_y,
|
|
const uint16_t* src_u,
|
|
int src_stride_u,
|
|
const uint16_t* src_v,
|
|
int src_stride_v,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height,
|
|
int depth) {
|
|
int y;
|
|
const int scale = 1 << (24 - depth);
|
|
const int uv_width = SUBSAMPLE(width, 1, 1);
|
|
void (*Convert16To8Row)(const uint16_t* src_y, uint8_t* dst_y, int scale,
|
|
int width) = Convert16To8Row_C;
|
|
void (*HalfWidthRow_16To8)(const uint16_t* src_uv, ptrdiff_t src_uv_stride,
|
|
uint8_t* dst_uv, int scale, int width) =
|
|
(width & 1) ? HalfWidthRow_16To8_Odd_C : HalfWidthRow_16To8_C;
|
|
|
|
if ((!src_y && dst_y) || !src_u || !src_v || !dst_u || !dst_v || width <= 0 ||
|
|
height == 0 || height == INT_MIN) {
|
|
return -1;
|
|
}
|
|
// Negative height means invert the image.
|
|
if (height < 0) {
|
|
height = -height;
|
|
if (src_y) {
|
|
src_y = src_y + (ptrdiff_t)(height - 1) * src_stride_y;
|
|
src_stride_y = -src_stride_y;
|
|
}
|
|
src_u = src_u + (ptrdiff_t)(height - 1) * src_stride_u;
|
|
src_v = src_v + (ptrdiff_t)(height - 1) * src_stride_v;
|
|
src_stride_u = -src_stride_u;
|
|
src_stride_v = -src_stride_v;
|
|
}
|
|
|
|
#if defined(HAS_CONVERT16TO8ROW_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
Convert16To8Row = Convert16To8Row_Any_NEON;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
Convert16To8Row = Convert16To8Row_NEON;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_CONVERT16TO8ROW_SVE2)
|
|
if (TestCpuFlag(kCpuHasSVE2)) {
|
|
Convert16To8Row = Convert16To8Row_SVE2;
|
|
}
|
|
#endif
|
|
#if defined(HAS_CONVERT16TO8ROW_SME)
|
|
if (TestCpuFlag(kCpuHasSME)) {
|
|
Convert16To8Row = Convert16To8Row_SME;
|
|
}
|
|
#endif
|
|
#if defined(HAS_CONVERT16TO8ROW_SSSE3)
|
|
if (TestCpuFlag(kCpuHasSSSE3)) {
|
|
Convert16To8Row = Convert16To8Row_Any_SSSE3;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
Convert16To8Row = Convert16To8Row_SSSE3;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_CONVERT16TO8ROW_AVX2)
|
|
if (TestCpuFlag(kCpuHasAVX2)) {
|
|
Convert16To8Row = Convert16To8Row_Any_AVX2;
|
|
if (IS_ALIGNED(width, 32)) {
|
|
Convert16To8Row = Convert16To8Row_AVX2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_CONVERT16TO8ROW_AVX512BW)
|
|
if (TestCpuFlag(kCpuHasAVX512BW)) {
|
|
Convert16To8Row = Convert16To8Row_Any_AVX512BW;
|
|
if (IS_ALIGNED(width, 64)) {
|
|
Convert16To8Row = Convert16To8Row_AVX512BW;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_CONVERT16TO8ROW_RVV)
|
|
if (TestCpuFlag(kCpuHasRVV)) {
|
|
Convert16To8Row = Convert16To8Row_RVV;
|
|
}
|
|
#endif
|
|
#if defined(HAS_CONVERT16TO8ROW_HVX)
|
|
if (TestCpuFlag(kCpuHasHVX)) {
|
|
Convert16To8Row = Convert16To8Row_HVX;
|
|
}
|
|
#endif
|
|
|
|
if (IS_ALIGNED(width, 2)) {
|
|
#if defined(HAS_HALFWIDTHROW_16TO8_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
HalfWidthRow_16To8 = HalfWidthRow_16To8_Any_NEON;
|
|
if (IS_ALIGNED(uv_width, 8)) {
|
|
HalfWidthRow_16To8 = HalfWidthRow_16To8_NEON;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_HALFWIDTHROW_16TO8_SVE2)
|
|
if (TestCpuFlag(kCpuHasSVE2)) {
|
|
HalfWidthRow_16To8 = HalfWidthRow_16To8_SVE2;
|
|
}
|
|
#endif
|
|
#if defined(HAS_HALFWIDTHROW_16TO8_SME)
|
|
if (TestCpuFlag(kCpuHasSME)) {
|
|
HalfWidthRow_16To8 = HalfWidthRow_16To8_SME;
|
|
}
|
|
#endif
|
|
#if defined(HAS_HALFWIDTHROW_16TO8_SSSE3)
|
|
if (TestCpuFlag(kCpuHasSSSE3)) {
|
|
HalfWidthRow_16To8 = HalfWidthRow_16To8_Any_SSSE3;
|
|
if (IS_ALIGNED(uv_width, 8)) {
|
|
HalfWidthRow_16To8 = HalfWidthRow_16To8_SSSE3;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_HALFWIDTHROW_16TO8_AVX2)
|
|
if (TestCpuFlag(kCpuHasAVX2)) {
|
|
HalfWidthRow_16To8 = HalfWidthRow_16To8_Any_AVX2;
|
|
if (IS_ALIGNED(uv_width, 16)) {
|
|
HalfWidthRow_16To8 = HalfWidthRow_16To8_AVX2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_HALFWIDTHROW_16TO8_AVX512BW)
|
|
if (TestCpuFlag(kCpuHasAVX512BW)) {
|
|
HalfWidthRow_16To8 = HalfWidthRow_16To8_Any_AVX512BW;
|
|
if (IS_ALIGNED(uv_width, 32)) {
|
|
HalfWidthRow_16To8 = HalfWidthRow_16To8_AVX512BW;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_HALFWIDTHROW_16TO8_RVV)
|
|
if (TestCpuFlag(kCpuHasRVV)) {
|
|
HalfWidthRow_16To8 = HalfWidthRow_16To8_RVV;
|
|
}
|
|
#endif
|
|
}
|
|
|
|
for (y = 0; y < height - 1; y += 2) {
|
|
HalfWidthRow_16To8(src_u, src_stride_u, dst_u, scale, uv_width);
|
|
HalfWidthRow_16To8(src_v, src_stride_v, dst_v, scale, uv_width);
|
|
if (dst_y) {
|
|
Convert16To8Row(src_y, dst_y, scale, width);
|
|
Convert16To8Row(src_y + src_stride_y, dst_y + dst_stride_y, scale, width);
|
|
src_y += src_stride_y * 2;
|
|
dst_y += dst_stride_y * 2;
|
|
}
|
|
src_u += src_stride_u * 2;
|
|
src_v += src_stride_v * 2;
|
|
dst_u += dst_stride_u;
|
|
dst_v += dst_stride_v;
|
|
}
|
|
if (height & 1) {
|
|
HalfWidthRow_16To8(src_u, 0, dst_u, scale, uv_width);
|
|
HalfWidthRow_16To8(src_v, 0, dst_v, scale, uv_width);
|
|
if (dst_y) {
|
|
Convert16To8Row(src_y, dst_y, scale, width);
|
|
}
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
static int I21xToI420(const uint16_t* src_y,
|
|
int src_stride_y,
|
|
const uint16_t* src_u,
|
|
int src_stride_u,
|
|
const uint16_t* src_v,
|
|
int src_stride_v,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height,
|
|
int depth) {
|
|
int y;
|
|
const int scale = 1 << (24 - depth);
|
|
const int uv_width = SUBSAMPLE(width, 1, 1);
|
|
void (*Convert16To8Row)(const uint16_t* src_y, uint8_t* dst_y, int scale,
|
|
int width) = Convert16To8Row_C;
|
|
void (*HalfRow_16To8)(const uint16_t* src_uv, ptrdiff_t src_uv_stride,
|
|
uint8_t* dst_uv, int scale, int width) =
|
|
HalfRow_16To8_C;
|
|
|
|
if ((!src_y && dst_y) || !src_u || !src_v || !dst_u || !dst_v || width <= 0 ||
|
|
height == 0 || height == INT_MIN) {
|
|
return -1;
|
|
}
|
|
// Negative height means invert the image.
|
|
if (height < 0) {
|
|
height = -height;
|
|
if (src_y) {
|
|
src_y = src_y + (ptrdiff_t)(height - 1) * src_stride_y;
|
|
src_stride_y = -src_stride_y;
|
|
}
|
|
src_u = src_u + (ptrdiff_t)(height - 1) * src_stride_u;
|
|
src_v = src_v + (ptrdiff_t)(height - 1) * src_stride_v;
|
|
src_stride_u = -src_stride_u;
|
|
src_stride_v = -src_stride_v;
|
|
}
|
|
|
|
#if defined(HAS_CONVERT16TO8ROW_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
Convert16To8Row = Convert16To8Row_Any_NEON;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
Convert16To8Row = Convert16To8Row_NEON;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_CONVERT16TO8ROW_SVE2)
|
|
if (TestCpuFlag(kCpuHasSVE2)) {
|
|
Convert16To8Row = Convert16To8Row_SVE2;
|
|
}
|
|
#endif
|
|
#if defined(HAS_CONVERT16TO8ROW_SME)
|
|
if (TestCpuFlag(kCpuHasSME)) {
|
|
Convert16To8Row = Convert16To8Row_SME;
|
|
}
|
|
#endif
|
|
#if defined(HAS_CONVERT16TO8ROW_SSSE3)
|
|
if (TestCpuFlag(kCpuHasSSSE3)) {
|
|
Convert16To8Row = Convert16To8Row_Any_SSSE3;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
Convert16To8Row = Convert16To8Row_SSSE3;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_CONVERT16TO8ROW_AVX2)
|
|
if (TestCpuFlag(kCpuHasAVX2)) {
|
|
Convert16To8Row = Convert16To8Row_Any_AVX2;
|
|
if (IS_ALIGNED(width, 32)) {
|
|
Convert16To8Row = Convert16To8Row_AVX2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_CONVERT16TO8ROW_AVX512BW)
|
|
if (TestCpuFlag(kCpuHasAVX512BW)) {
|
|
Convert16To8Row = Convert16To8Row_Any_AVX512BW;
|
|
if (IS_ALIGNED(width, 64)) {
|
|
Convert16To8Row = Convert16To8Row_AVX512BW;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_CONVERT16TO8ROW_RVV)
|
|
if (TestCpuFlag(kCpuHasRVV)) {
|
|
Convert16To8Row = Convert16To8Row_RVV;
|
|
}
|
|
#endif
|
|
#if defined(HAS_CONVERT16TO8ROW_HVX)
|
|
if (TestCpuFlag(kCpuHasHVX)) {
|
|
Convert16To8Row = Convert16To8Row_HVX;
|
|
}
|
|
#endif
|
|
|
|
#if defined(HAS_HALFROW_16TO8_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
HalfRow_16To8 = HalfRow_16To8_Any_NEON;
|
|
if (IS_ALIGNED(uv_width, 16)) {
|
|
HalfRow_16To8 = HalfRow_16To8_NEON;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_HALFROW_16TO8_SVE2)
|
|
if (TestCpuFlag(kCpuHasSVE2)) {
|
|
HalfRow_16To8 = HalfRow_16To8_SVE2;
|
|
}
|
|
#endif
|
|
#if defined(HAS_HALFROW_16TO8_SME)
|
|
if (TestCpuFlag(kCpuHasSME)) {
|
|
HalfRow_16To8 = HalfRow_16To8_SME;
|
|
}
|
|
#endif
|
|
#if defined(HAS_HALFROW_16TO8_SSSE3)
|
|
if (TestCpuFlag(kCpuHasSSSE3)) {
|
|
HalfRow_16To8 = HalfRow_16To8_Any_SSSE3;
|
|
if (IS_ALIGNED(uv_width, 16)) {
|
|
HalfRow_16To8 = HalfRow_16To8_SSSE3;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_HALFROW_16TO8_AVX2)
|
|
if (TestCpuFlag(kCpuHasAVX2)) {
|
|
HalfRow_16To8 = HalfRow_16To8_Any_AVX2;
|
|
if (IS_ALIGNED(uv_width, 32)) {
|
|
HalfRow_16To8 = HalfRow_16To8_AVX2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_HALFROW_16TO8_AVX512BW)
|
|
if (TestCpuFlag(kCpuHasAVX512BW)) {
|
|
HalfRow_16To8 = HalfRow_16To8_Any_AVX512BW;
|
|
if (IS_ALIGNED(uv_width, 64)) {
|
|
HalfRow_16To8 = HalfRow_16To8_AVX512BW;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_HALFROW_16TO8_RVV)
|
|
if (TestCpuFlag(kCpuHasRVV)) {
|
|
HalfRow_16To8 = HalfRow_16To8_RVV;
|
|
}
|
|
#endif
|
|
#if defined(HAS_HALFROW_16TO8_HVX)
|
|
if (TestCpuFlag(kCpuHasHVX)) {
|
|
HalfRow_16To8 = HalfRow_16To8_HVX;
|
|
}
|
|
#endif
|
|
|
|
for (y = 0; y < height - 1; y += 2) {
|
|
HalfRow_16To8(src_u, src_stride_u, dst_u, scale, uv_width);
|
|
HalfRow_16To8(src_v, src_stride_v, dst_v, scale, uv_width);
|
|
if (dst_y) {
|
|
Convert16To8Row(src_y, dst_y, scale, width);
|
|
Convert16To8Row(src_y + src_stride_y, dst_y + dst_stride_y, scale, width);
|
|
src_y += src_stride_y * 2;
|
|
dst_y += dst_stride_y * 2;
|
|
}
|
|
src_u += src_stride_u * 2;
|
|
src_v += src_stride_v * 2;
|
|
dst_u += dst_stride_u;
|
|
dst_v += dst_stride_v;
|
|
}
|
|
if (height & 1) {
|
|
HalfRow_16To8(src_u, 0, dst_u, scale, uv_width);
|
|
HalfRow_16To8(src_v, 0, dst_v, scale, uv_width);
|
|
if (dst_y) {
|
|
Convert16To8Row(src_y, dst_y, scale, width);
|
|
}
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
// Convert 10 bit YUV to 8 bit.
|
|
LIBYUV_API
|
|
int I010ToI420(const uint16_t* src_y,
|
|
int src_stride_y,
|
|
const uint16_t* src_u,
|
|
int src_stride_u,
|
|
const uint16_t* src_v,
|
|
int src_stride_v,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
return Planar16bitTo8bit(src_y, src_stride_y, src_u, src_stride_u, src_v,
|
|
src_stride_v, dst_y, dst_stride_y, dst_u,
|
|
dst_stride_u, dst_v, dst_stride_v, width, height, 1,
|
|
1, 10);
|
|
}
|
|
|
|
LIBYUV_API
|
|
int I210ToI420(const uint16_t* src_y,
|
|
int src_stride_y,
|
|
const uint16_t* src_u,
|
|
int src_stride_u,
|
|
const uint16_t* src_v,
|
|
int src_stride_v,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
return I21xToI420(src_y, src_stride_y, src_u, src_stride_u, src_v,
|
|
src_stride_v, dst_y, dst_stride_y, dst_u, dst_stride_u,
|
|
dst_v, dst_stride_v, width, height, 10);
|
|
}
|
|
|
|
LIBYUV_API
|
|
int I210ToI422(const uint16_t* src_y,
|
|
int src_stride_y,
|
|
const uint16_t* src_u,
|
|
int src_stride_u,
|
|
const uint16_t* src_v,
|
|
int src_stride_v,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
return Planar16bitTo8bit(src_y, src_stride_y, src_u, src_stride_u, src_v,
|
|
src_stride_v, dst_y, dst_stride_y, dst_u,
|
|
dst_stride_u, dst_v, dst_stride_v, width, height, 1,
|
|
0, 10);
|
|
}
|
|
|
|
LIBYUV_API
|
|
int I410ToI420(const uint16_t* src_y,
|
|
int src_stride_y,
|
|
const uint16_t* src_u,
|
|
int src_stride_u,
|
|
const uint16_t* src_v,
|
|
int src_stride_v,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
return I41xToI420(src_y, src_stride_y, src_u, src_stride_u, src_v,
|
|
src_stride_v, dst_y, dst_stride_y, dst_u, dst_stride_u,
|
|
dst_v, dst_stride_v, width, height, 10);
|
|
}
|
|
|
|
LIBYUV_API
|
|
int I410ToI444(const uint16_t* src_y,
|
|
int src_stride_y,
|
|
const uint16_t* src_u,
|
|
int src_stride_u,
|
|
const uint16_t* src_v,
|
|
int src_stride_v,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
return Planar16bitTo8bit(src_y, src_stride_y, src_u, src_stride_u, src_v,
|
|
src_stride_v, dst_y, dst_stride_y, dst_u,
|
|
dst_stride_u, dst_v, dst_stride_v, width, height, 0,
|
|
0, 10);
|
|
}
|
|
|
|
LIBYUV_API
|
|
int I012ToI420(const uint16_t* src_y,
|
|
int src_stride_y,
|
|
const uint16_t* src_u,
|
|
int src_stride_u,
|
|
const uint16_t* src_v,
|
|
int src_stride_v,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
return Planar16bitTo8bit(src_y, src_stride_y, src_u, src_stride_u, src_v,
|
|
src_stride_v, dst_y, dst_stride_y, dst_u,
|
|
dst_stride_u, dst_v, dst_stride_v, width, height, 1,
|
|
1, 12);
|
|
}
|
|
|
|
LIBYUV_API
|
|
int I212ToI422(const uint16_t* src_y,
|
|
int src_stride_y,
|
|
const uint16_t* src_u,
|
|
int src_stride_u,
|
|
const uint16_t* src_v,
|
|
int src_stride_v,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
return Planar16bitTo8bit(src_y, src_stride_y, src_u, src_stride_u, src_v,
|
|
src_stride_v, dst_y, dst_stride_y, dst_u,
|
|
dst_stride_u, dst_v, dst_stride_v, width, height, 1,
|
|
0, 12);
|
|
}
|
|
|
|
LIBYUV_API
|
|
int I212ToI420(const uint16_t* src_y,
|
|
int src_stride_y,
|
|
const uint16_t* src_u,
|
|
int src_stride_u,
|
|
const uint16_t* src_v,
|
|
int src_stride_v,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
return I21xToI420(src_y, src_stride_y, src_u, src_stride_u, src_v,
|
|
src_stride_v, dst_y, dst_stride_y, dst_u, dst_stride_u,
|
|
dst_v, dst_stride_v, width, height, 12);
|
|
}
|
|
|
|
LIBYUV_API
|
|
int I412ToI444(const uint16_t* src_y,
|
|
int src_stride_y,
|
|
const uint16_t* src_u,
|
|
int src_stride_u,
|
|
const uint16_t* src_v,
|
|
int src_stride_v,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
return Planar16bitTo8bit(src_y, src_stride_y, src_u, src_stride_u, src_v,
|
|
src_stride_v, dst_y, dst_stride_y, dst_u,
|
|
dst_stride_u, dst_v, dst_stride_v, width, height, 0,
|
|
0, 12);
|
|
}
|
|
|
|
LIBYUV_API
|
|
int I412ToI420(const uint16_t* src_y,
|
|
int src_stride_y,
|
|
const uint16_t* src_u,
|
|
int src_stride_u,
|
|
const uint16_t* src_v,
|
|
int src_stride_v,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
return I41xToI420(src_y, src_stride_y, src_u, src_stride_u, src_v,
|
|
src_stride_v, dst_y, dst_stride_y, dst_u, dst_stride_u,
|
|
dst_v, dst_stride_v, width, height, 12);
|
|
}
|
|
|
|
// Any Ix10 To I010 format
|
|
static int Ix10ToI010(const uint16_t* src_y,
|
|
int src_stride_y,
|
|
const uint16_t* src_u,
|
|
int src_stride_u,
|
|
const uint16_t* src_v,
|
|
int src_stride_v,
|
|
uint16_t* dst_y,
|
|
int dst_stride_y,
|
|
uint16_t* dst_u,
|
|
int dst_stride_u,
|
|
uint16_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height,
|
|
int subsample_x,
|
|
int subsample_y) {
|
|
int r;
|
|
if ((!src_y && dst_y) || !src_u || !src_v || !dst_u || !dst_v || width <= 0 ||
|
|
height == 0 || height == INT_MIN) {
|
|
return -1;
|
|
}
|
|
const int dst_y_width = width;
|
|
const int dst_y_height = Abs(height);
|
|
const int src_uv_width = SUBSAMPLE(width, subsample_x, subsample_x);
|
|
const int src_uv_height = SUBSAMPLE(height, subsample_y, subsample_y);
|
|
const int dst_uv_width = SUBSAMPLE(dst_y_width, 1, 1);
|
|
const int dst_uv_height = SUBSAMPLE(dst_y_height, 1, 1);
|
|
if (dst_y) {
|
|
CopyPlane_16(src_y, src_stride_y, dst_y, dst_stride_y, width, height);
|
|
}
|
|
r = ScalePlane_12(src_u, src_stride_u, src_uv_width, src_uv_height, dst_u,
|
|
dst_stride_u, dst_uv_width, dst_uv_height, kFilterBilinear);
|
|
if (r != 0) {
|
|
return r;
|
|
}
|
|
r = ScalePlane_12(src_v, src_stride_v, src_uv_width, src_uv_height, dst_v,
|
|
dst_stride_v, dst_uv_width, dst_uv_height, kFilterBilinear);
|
|
return r;
|
|
}
|
|
|
|
LIBYUV_API
|
|
int I410ToI010(const uint16_t* src_y,
|
|
int src_stride_y,
|
|
const uint16_t* src_u,
|
|
int src_stride_u,
|
|
const uint16_t* src_v,
|
|
int src_stride_v,
|
|
uint16_t* dst_y,
|
|
int dst_stride_y,
|
|
uint16_t* dst_u,
|
|
int dst_stride_u,
|
|
uint16_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
return Ix10ToI010(src_y, src_stride_y, src_u, src_stride_u, src_v,
|
|
src_stride_v, dst_y, dst_stride_y, dst_u, dst_stride_u,
|
|
dst_v, dst_stride_v, width, height, 0, 0);
|
|
}
|
|
|
|
LIBYUV_API
|
|
int I210ToI010(const uint16_t* src_y,
|
|
int src_stride_y,
|
|
const uint16_t* src_u,
|
|
int src_stride_u,
|
|
const uint16_t* src_v,
|
|
int src_stride_v,
|
|
uint16_t* dst_y,
|
|
int dst_stride_y,
|
|
uint16_t* dst_u,
|
|
int dst_stride_u,
|
|
uint16_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
return Ix10ToI010(src_y, src_stride_y, src_u, src_stride_u, src_v,
|
|
src_stride_v, dst_y, dst_stride_y, dst_u, dst_stride_u,
|
|
dst_v, dst_stride_v, width, height, 1, 0);
|
|
}
|
|
|
|
// Any I[420]1[02] to P[420]1[02] format
|
|
static int IxxxToPxxx(const uint16_t* src_y,
|
|
int src_stride_y,
|
|
const uint16_t* src_u,
|
|
int src_stride_u,
|
|
const uint16_t* src_v,
|
|
int src_stride_v,
|
|
uint16_t* dst_y,
|
|
int dst_stride_y,
|
|
uint16_t* dst_uv,
|
|
int dst_stride_uv,
|
|
int width,
|
|
int height,
|
|
int subsample_x,
|
|
int subsample_y,
|
|
int depth) {
|
|
if (width <= 0 || height == 0 || height == INT_MIN) {
|
|
return -1;
|
|
}
|
|
const int uv_width = SUBSAMPLE(width, subsample_x, subsample_x);
|
|
const int uv_height = SUBSAMPLE(height, subsample_y, subsample_y);
|
|
|
|
ConvertToMSBPlane_16(src_y, src_stride_y, dst_y, dst_stride_y, width, height,
|
|
depth);
|
|
MergeUVPlane_16(src_u, src_stride_u, src_v, src_stride_v, dst_uv,
|
|
dst_stride_uv, uv_width, uv_height, depth);
|
|
return 0;
|
|
}
|
|
|
|
LIBYUV_API
|
|
int I010ToP010(const uint16_t* src_y,
|
|
int src_stride_y,
|
|
const uint16_t* src_u,
|
|
int src_stride_u,
|
|
const uint16_t* src_v,
|
|
int src_stride_v,
|
|
uint16_t* dst_y,
|
|
int dst_stride_y,
|
|
uint16_t* dst_uv,
|
|
int dst_stride_uv,
|
|
int width,
|
|
int height) {
|
|
return IxxxToPxxx(src_y, src_stride_y, src_u, src_stride_u, src_v,
|
|
src_stride_v, dst_y, dst_stride_y, dst_uv, dst_stride_uv,
|
|
width, height, 1, 1, 10);
|
|
}
|
|
|
|
LIBYUV_API
|
|
int I010ToNV12(const uint16_t* src_y,
|
|
int src_stride_y,
|
|
const uint16_t* src_u,
|
|
int src_stride_u,
|
|
const uint16_t* src_v,
|
|
int src_stride_v,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_uv,
|
|
int dst_stride_uv,
|
|
int width,
|
|
int height) {
|
|
int y;
|
|
int halfwidth = (width + 1) >> 1;
|
|
int halfheight = (height + 1) >> 1;
|
|
const int scale = 16385; // 16384 for 10 bits
|
|
void (*Convert16To8Row)(const uint16_t* src_y, uint8_t* dst_y, int scale,
|
|
int width) = Convert16To8Row_C;
|
|
void (*MergeUVRow)(const uint8_t* src_u, const uint8_t* src_v,
|
|
uint8_t* dst_uv, int width) = MergeUVRow_C;
|
|
if ((!src_y && dst_y) || !src_u || !src_v || !dst_uv || width <= 0 ||
|
|
height == 0 || height == INT_MIN) {
|
|
return -1;
|
|
}
|
|
// Negative height means invert the image.
|
|
if (height < 0) {
|
|
height = -height;
|
|
halfheight = (height + 1) >> 1;
|
|
src_y = src_y + (ptrdiff_t)(height - 1) * src_stride_y;
|
|
src_u = src_u + (ptrdiff_t)(halfheight - 1) * src_stride_u;
|
|
src_v = src_v + (ptrdiff_t)(halfheight - 1) * src_stride_v;
|
|
src_stride_y = -src_stride_y;
|
|
src_stride_u = -src_stride_u;
|
|
src_stride_v = -src_stride_v;
|
|
}
|
|
#if defined(HAS_CONVERT16TO8ROW_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
Convert16To8Row = Convert16To8Row_Any_NEON;
|
|
if (IS_ALIGNED(halfwidth, 16)) {
|
|
Convert16To8Row = Convert16To8Row_NEON;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_CONVERT16TO8ROW_SME)
|
|
if (TestCpuFlag(kCpuHasSME)) {
|
|
Convert16To8Row = Convert16To8Row_SME;
|
|
}
|
|
#endif
|
|
#if defined(HAS_CONVERT16TO8ROW_SSSE3)
|
|
if (TestCpuFlag(kCpuHasSSSE3)) {
|
|
Convert16To8Row = Convert16To8Row_Any_SSSE3;
|
|
if (IS_ALIGNED(halfwidth, 16)) {
|
|
Convert16To8Row = Convert16To8Row_SSSE3;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_CONVERT16TO8ROW_AVX2)
|
|
if (TestCpuFlag(kCpuHasAVX2)) {
|
|
Convert16To8Row = Convert16To8Row_Any_AVX2;
|
|
if (IS_ALIGNED(halfwidth, 32)) {
|
|
Convert16To8Row = Convert16To8Row_AVX2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_CONVERT16TO8ROW_AVX512BW)
|
|
if (TestCpuFlag(kCpuHasAVX512BW)) {
|
|
Convert16To8Row = Convert16To8Row_Any_AVX512BW;
|
|
if (IS_ALIGNED(halfwidth, 64)) {
|
|
Convert16To8Row = Convert16To8Row_AVX512BW;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_CONVERT16TO8ROW_HVX)
|
|
if (TestCpuFlag(kCpuHasHVX)) {
|
|
Convert16To8Row = Convert16To8Row_HVX;
|
|
}
|
|
#endif
|
|
|
|
#if defined(HAS_MERGEUVROW_SSE2)
|
|
if (TestCpuFlag(kCpuHasSSE2)) {
|
|
MergeUVRow = MergeUVRow_Any_SSE2;
|
|
if (IS_ALIGNED(halfwidth, 16)) {
|
|
MergeUVRow = MergeUVRow_SSE2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_MERGEUVROW_AVX2)
|
|
if (TestCpuFlag(kCpuHasAVX2)) {
|
|
MergeUVRow = MergeUVRow_Any_AVX2;
|
|
if (IS_ALIGNED(halfwidth, 32)) {
|
|
MergeUVRow = MergeUVRow_AVX2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_MERGEUVROW_AVX512BW)
|
|
if (TestCpuFlag(kCpuHasAVX512BW)) {
|
|
MergeUVRow = MergeUVRow_Any_AVX512BW;
|
|
if (IS_ALIGNED(halfwidth, 32)) {
|
|
MergeUVRow = MergeUVRow_AVX512BW;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_MERGEUVROW_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
MergeUVRow = MergeUVRow_Any_NEON;
|
|
if (IS_ALIGNED(halfwidth, 16)) {
|
|
MergeUVRow = MergeUVRow_NEON;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_MERGEUVROW_SVE2)
|
|
if (TestCpuFlag(kCpuHasSVE2)) {
|
|
MergeUVRow = MergeUVRow_SVE2;
|
|
}
|
|
#endif
|
|
#if defined(HAS_MERGEUVROW_SME)
|
|
if (TestCpuFlag(kCpuHasSME)) {
|
|
MergeUVRow = MergeUVRow_SME;
|
|
}
|
|
#endif
|
|
#if defined(HAS_MERGEUVROW_LSX)
|
|
if (TestCpuFlag(kCpuHasLSX)) {
|
|
MergeUVRow = MergeUVRow_Any_LSX;
|
|
if (IS_ALIGNED(halfwidth, 16)) {
|
|
MergeUVRow = MergeUVRow_LSX;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_MERGEUVROW_LASX)
|
|
if (TestCpuFlag(kCpuHasLASX)) {
|
|
MergeUVRow = MergeUVRow_Any_LASX;
|
|
if (IS_ALIGNED(halfwidth, 32)) {
|
|
MergeUVRow = MergeUVRow_LASX;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_MERGEUVROW_RVV)
|
|
if (TestCpuFlag(kCpuHasRVV)) {
|
|
MergeUVRow = MergeUVRow_RVV;
|
|
}
|
|
#endif
|
|
#if defined(HAS_MERGEUVROW_WASMSIMD)
|
|
if (TestCpuFlag(kCpuHasWASMSIMD)) {
|
|
MergeUVRow = MergeUVRow_Any_WASMSIMD;
|
|
if (IS_ALIGNED(halfwidth, 16)) {
|
|
MergeUVRow = MergeUVRow_WASMSIMD;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_MERGEUVROW_HVX)
|
|
if (TestCpuFlag(kCpuHasHVX)) {
|
|
MergeUVRow = MergeUVRow_HVX;
|
|
}
|
|
#endif
|
|
|
|
// Convert Y plane.
|
|
if (dst_y) {
|
|
Convert16To8Plane(src_y, src_stride_y, dst_y, dst_stride_y, scale, width,
|
|
height);
|
|
}
|
|
|
|
{
|
|
// Allocate a row of uv.
|
|
align_buffer_64(row_u, ((halfwidth + 31) & ~31) * 2);
|
|
uint8_t* row_v = row_u + ((halfwidth + 31) & ~31);
|
|
if (!row_u)
|
|
return 1;
|
|
|
|
for (y = 0; y < halfheight; ++y) {
|
|
Convert16To8Row(src_u, row_u, scale, halfwidth);
|
|
Convert16To8Row(src_v, row_v, scale, halfwidth);
|
|
MergeUVRow(row_u, row_v, dst_uv, halfwidth);
|
|
src_u += src_stride_u;
|
|
src_v += src_stride_v;
|
|
dst_uv += dst_stride_uv;
|
|
}
|
|
free_aligned_buffer_64(row_u);
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
LIBYUV_API
|
|
int I210ToP210(const uint16_t* src_y,
|
|
int src_stride_y,
|
|
const uint16_t* src_u,
|
|
int src_stride_u,
|
|
const uint16_t* src_v,
|
|
int src_stride_v,
|
|
uint16_t* dst_y,
|
|
int dst_stride_y,
|
|
uint16_t* dst_uv,
|
|
int dst_stride_uv,
|
|
int width,
|
|
int height) {
|
|
return IxxxToPxxx(src_y, src_stride_y, src_u, src_stride_u, src_v,
|
|
src_stride_v, dst_y, dst_stride_y, dst_uv, dst_stride_uv,
|
|
width, height, 1, 0, 10);
|
|
}
|
|
|
|
LIBYUV_API
|
|
int I410ToP410(const uint16_t* src_y,
|
|
int src_stride_y,
|
|
const uint16_t* src_u,
|
|
int src_stride_u,
|
|
const uint16_t* src_v,
|
|
int src_stride_v,
|
|
uint16_t* dst_y,
|
|
int dst_stride_y,
|
|
uint16_t* dst_uv,
|
|
int dst_stride_uv,
|
|
int width,
|
|
int height) {
|
|
return IxxxToPxxx(src_y, src_stride_y, src_u, src_stride_u, src_v,
|
|
src_stride_v, dst_y, dst_stride_y, dst_uv, dst_stride_uv,
|
|
width, height, 0, 0, 10);
|
|
}
|
|
|
|
LIBYUV_API
|
|
int I012ToP012(const uint16_t* src_y,
|
|
int src_stride_y,
|
|
const uint16_t* src_u,
|
|
int src_stride_u,
|
|
const uint16_t* src_v,
|
|
int src_stride_v,
|
|
uint16_t* dst_y,
|
|
int dst_stride_y,
|
|
uint16_t* dst_uv,
|
|
int dst_stride_uv,
|
|
int width,
|
|
int height) {
|
|
return IxxxToPxxx(src_y, src_stride_y, src_u, src_stride_u, src_v,
|
|
src_stride_v, dst_y, dst_stride_y, dst_uv, dst_stride_uv,
|
|
width, height, 1, 1, 12);
|
|
}
|
|
|
|
LIBYUV_API
|
|
int I212ToP212(const uint16_t* src_y,
|
|
int src_stride_y,
|
|
const uint16_t* src_u,
|
|
int src_stride_u,
|
|
const uint16_t* src_v,
|
|
int src_stride_v,
|
|
uint16_t* dst_y,
|
|
int dst_stride_y,
|
|
uint16_t* dst_uv,
|
|
int dst_stride_uv,
|
|
int width,
|
|
int height) {
|
|
return IxxxToPxxx(src_y, src_stride_y, src_u, src_stride_u, src_v,
|
|
src_stride_v, dst_y, dst_stride_y, dst_uv, dst_stride_uv,
|
|
width, height, 1, 0, 12);
|
|
}
|
|
|
|
LIBYUV_API
|
|
int I412ToP412(const uint16_t* src_y,
|
|
int src_stride_y,
|
|
const uint16_t* src_u,
|
|
int src_stride_u,
|
|
const uint16_t* src_v,
|
|
int src_stride_v,
|
|
uint16_t* dst_y,
|
|
int dst_stride_y,
|
|
uint16_t* dst_uv,
|
|
int dst_stride_uv,
|
|
int width,
|
|
int height) {
|
|
return IxxxToPxxx(src_y, src_stride_y, src_u, src_stride_u, src_v,
|
|
src_stride_v, dst_y, dst_stride_y, dst_uv, dst_stride_uv,
|
|
width, height, 0, 0, 12);
|
|
}
|
|
|
|
// 422 chroma is 1/2 width, 1x height
|
|
// 420 chroma is 1/2 width, 1/2 height
|
|
LIBYUV_API
|
|
int I422ToI420(const uint8_t* src_y,
|
|
int src_stride_y,
|
|
const uint8_t* src_u,
|
|
int src_stride_u,
|
|
const uint8_t* src_v,
|
|
int src_stride_v,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
const int src_uv_width = SUBSAMPLE(width, 1, 1);
|
|
return I4xxToI420(src_y, src_stride_y, src_u, src_stride_u, src_v,
|
|
src_stride_v, dst_y, dst_stride_y, dst_u, dst_stride_u,
|
|
dst_v, dst_stride_v, width, height, src_uv_width, height);
|
|
}
|
|
|
|
LIBYUV_API
|
|
int I422ToI210(const uint8_t* src_y,
|
|
int src_stride_y,
|
|
const uint8_t* src_u,
|
|
int src_stride_u,
|
|
const uint8_t* src_v,
|
|
int src_stride_v,
|
|
uint16_t* dst_y,
|
|
int dst_stride_y,
|
|
uint16_t* dst_u,
|
|
int dst_stride_u,
|
|
uint16_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
int halfwidth = (width + 1) >> 1;
|
|
if ((!src_y && dst_y) || !src_u || !src_v || !dst_u || !dst_v || width <= 0 ||
|
|
height == 0 || height == INT_MIN) {
|
|
return -1;
|
|
}
|
|
// Negative height means invert the image.
|
|
if (height < 0) {
|
|
height = -height;
|
|
src_y = src_y + (ptrdiff_t)(height - 1) * src_stride_y;
|
|
src_u = src_u + (ptrdiff_t)(height - 1) * src_stride_u;
|
|
src_v = src_v + (ptrdiff_t)(height - 1) * src_stride_v;
|
|
src_stride_y = -src_stride_y;
|
|
src_stride_u = -src_stride_u;
|
|
src_stride_v = -src_stride_v;
|
|
}
|
|
|
|
// Convert Y plane.
|
|
Convert8To16Plane(src_y, src_stride_y, dst_y, dst_stride_y, 1024, width,
|
|
height);
|
|
// Convert UV planes.
|
|
Convert8To16Plane(src_u, src_stride_u, dst_u, dst_stride_u, 1024, halfwidth,
|
|
height);
|
|
Convert8To16Plane(src_v, src_stride_v, dst_v, dst_stride_v, 1024, halfwidth,
|
|
height);
|
|
return 0;
|
|
}
|
|
|
|
// TODO(fbarchard): Implement row conversion.
|
|
LIBYUV_API
|
|
int I422ToNV21(const uint8_t* src_y,
|
|
int src_stride_y,
|
|
const uint8_t* src_u,
|
|
int src_stride_u,
|
|
const uint8_t* src_v,
|
|
int src_stride_v,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_vu,
|
|
int dst_stride_vu,
|
|
int width,
|
|
int height) {
|
|
int r;
|
|
if (width <= 0 || height == 0 || height == INT_MIN) {
|
|
return -1;
|
|
}
|
|
int halfwidth = (width + 1) >> 1;
|
|
int halfheight = (height + 1) >> 1;
|
|
// Negative height means invert the image.
|
|
if (height < 0) {
|
|
height = -height;
|
|
halfheight = (height + 1) >> 1;
|
|
src_y = src_y + (ptrdiff_t)(height - 1) * src_stride_y;
|
|
src_u = src_u + (ptrdiff_t)(height - 1) * src_stride_u;
|
|
src_v = src_v + (ptrdiff_t)(height - 1) * src_stride_v;
|
|
src_stride_y = -src_stride_y;
|
|
src_stride_u = -src_stride_u;
|
|
src_stride_v = -src_stride_v;
|
|
}
|
|
|
|
// Allocate u and v buffers
|
|
const uint64_t plane_size = (uint64_t)halfwidth * halfheight;
|
|
if (plane_size > SIZE_MAX / 2)
|
|
return 1;
|
|
align_buffer_64(plane_u, (size_t)plane_size * 2);
|
|
if (!plane_u)
|
|
return 1;
|
|
uint8_t* plane_v = plane_u + (size_t)plane_size;
|
|
|
|
r = I422ToI420(src_y, src_stride_y, src_u, src_stride_u, src_v, src_stride_v,
|
|
dst_y, dst_stride_y, plane_u, halfwidth, plane_v, halfwidth,
|
|
width, height);
|
|
if (r != 0) {
|
|
free_aligned_buffer_64(plane_u);
|
|
return r;
|
|
}
|
|
MergeUVPlane(plane_v, halfwidth, plane_u, halfwidth, dst_vu, dst_stride_vu,
|
|
halfwidth, halfheight);
|
|
free_aligned_buffer_64(plane_u);
|
|
return 0;
|
|
}
|
|
|
|
LIBYUV_API
|
|
int MM21ToNV12(const uint8_t* src_y,
|
|
int src_stride_y,
|
|
const uint8_t* src_uv,
|
|
int src_stride_uv,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_uv,
|
|
int dst_stride_uv,
|
|
int width,
|
|
int height) {
|
|
if (!src_uv || !dst_uv || width <= 0) {
|
|
return -1;
|
|
}
|
|
|
|
int sign = height < 0 ? -1 : 1;
|
|
|
|
if (dst_y) {
|
|
DetilePlane(src_y, src_stride_y, dst_y, dst_stride_y, width, height, 32);
|
|
}
|
|
DetilePlane(src_uv, src_stride_uv, dst_uv, dst_stride_uv, (width + 1) & ~1,
|
|
(height + sign) / 2, 16);
|
|
|
|
return 0;
|
|
}
|
|
|
|
LIBYUV_API
|
|
int MM21ToI420(const uint8_t* src_y,
|
|
int src_stride_y,
|
|
const uint8_t* src_uv,
|
|
int src_stride_uv,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
int sign = height < 0 ? -1 : 1;
|
|
|
|
if (!src_uv || !dst_u || !dst_v || width <= 0) {
|
|
return -1;
|
|
}
|
|
|
|
if (dst_y) {
|
|
DetilePlane(src_y, src_stride_y, dst_y, dst_stride_y, width, height, 32);
|
|
}
|
|
DetileSplitUVPlane(src_uv, src_stride_uv, dst_u, dst_stride_u, dst_v,
|
|
dst_stride_v, (width + 1) & ~1, (height + sign) / 2, 16);
|
|
|
|
return 0;
|
|
}
|
|
|
|
LIBYUV_API
|
|
int MM21ToYUY2(const uint8_t* src_y,
|
|
int src_stride_y,
|
|
const uint8_t* src_uv,
|
|
int src_stride_uv,
|
|
uint8_t* dst_yuy2,
|
|
int dst_stride_yuy2,
|
|
int width,
|
|
int height) {
|
|
if (!src_y || !src_uv || !dst_yuy2 || width <= 0) {
|
|
return -1;
|
|
}
|
|
|
|
DetileToYUY2(src_y, src_stride_y, src_uv, src_stride_uv, dst_yuy2,
|
|
dst_stride_yuy2, width, height, 32);
|
|
|
|
return 0;
|
|
}
|
|
|
|
// Convert MT2T into P010. See tinyurl.com/mtk-10bit-video-format for format
|
|
// documentation.
|
|
// TODO(greenjustin): Add an MT2T to I420 conversion.
|
|
LIBYUV_API
|
|
int MT2TToP010(const uint8_t* src_y,
|
|
int src_stride_y,
|
|
const uint8_t* src_uv,
|
|
int src_stride_uv,
|
|
uint16_t* dst_y,
|
|
int dst_stride_y,
|
|
uint16_t* dst_uv,
|
|
int dst_stride_uv,
|
|
int width,
|
|
int height) {
|
|
if (width <= 0 || height == 0 || height == INT_MIN || !src_uv || !dst_uv) {
|
|
return -1;
|
|
}
|
|
|
|
{
|
|
int uv_width = (width + 1) & ~1;
|
|
int uv_height = (height + 1) / 2;
|
|
int y = 0;
|
|
const int tile_width = 16;
|
|
const int y_tile_height = 32;
|
|
const int uv_tile_height = 16;
|
|
int padded_width = (width + tile_width - 1) & ~(tile_width - 1);
|
|
int y_tile_row_size = padded_width * y_tile_height * 10 / 8;
|
|
int uv_tile_row_size = padded_width * uv_tile_height * 10 / 8;
|
|
size_t row_buf_size = padded_width * y_tile_height * sizeof(uint16_t);
|
|
void (*UnpackMT2T)(const uint8_t* src, uint16_t* dst, size_t size) =
|
|
UnpackMT2T_C;
|
|
align_buffer_64(row_buf, row_buf_size);
|
|
if (!row_buf)
|
|
return 1;
|
|
|
|
#if defined(HAS_UNPACKMT2T_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
UnpackMT2T = UnpackMT2T_NEON;
|
|
}
|
|
#endif
|
|
// Negative height means invert the image.
|
|
if (height < 0) {
|
|
height = -height;
|
|
uv_height = (height + 1) / 2;
|
|
if (dst_y) {
|
|
dst_y = dst_y + (ptrdiff_t)(height - 1) * dst_stride_y;
|
|
dst_stride_y = -dst_stride_y;
|
|
}
|
|
dst_uv = dst_uv + (ptrdiff_t)(uv_height - 1) * dst_stride_uv;
|
|
dst_stride_uv = -dst_stride_uv;
|
|
}
|
|
|
|
// Unpack and detile Y in rows of tiles
|
|
if (src_y && dst_y) {
|
|
for (y = 0; y < (height & ~(y_tile_height - 1)); y += y_tile_height) {
|
|
UnpackMT2T(src_y, (uint16_t*)row_buf, y_tile_row_size);
|
|
DetilePlane_16((uint16_t*)row_buf, padded_width, dst_y, dst_stride_y,
|
|
width, y_tile_height, y_tile_height);
|
|
src_y += src_stride_y * y_tile_height;
|
|
dst_y += dst_stride_y * y_tile_height;
|
|
}
|
|
if (height & (y_tile_height - 1)) {
|
|
UnpackMT2T(src_y, (uint16_t*)row_buf, y_tile_row_size);
|
|
DetilePlane_16((uint16_t*)row_buf, padded_width, dst_y, dst_stride_y,
|
|
width, height & (y_tile_height - 1), y_tile_height);
|
|
}
|
|
}
|
|
|
|
// Unpack and detile UV plane
|
|
for (y = 0; y < (uv_height & ~(uv_tile_height - 1)); y += uv_tile_height) {
|
|
UnpackMT2T(src_uv, (uint16_t*)row_buf, uv_tile_row_size);
|
|
DetilePlane_16((uint16_t*)row_buf, padded_width, dst_uv, dst_stride_uv,
|
|
uv_width, uv_tile_height, uv_tile_height);
|
|
src_uv += src_stride_uv * uv_tile_height;
|
|
dst_uv += dst_stride_uv * uv_tile_height;
|
|
}
|
|
if (uv_height & (uv_tile_height - 1)) {
|
|
UnpackMT2T(src_uv, (uint16_t*)row_buf, uv_tile_row_size);
|
|
DetilePlane_16((uint16_t*)row_buf, padded_width, dst_uv, dst_stride_uv,
|
|
uv_width, uv_height & (uv_tile_height - 1),
|
|
uv_tile_height);
|
|
}
|
|
free_aligned_buffer_64(row_buf);
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
#ifdef I422TONV21_ROW_VERSION
|
|
// Unittest fails for this version.
|
|
// 422 chroma is 1/2 width, 1x height
|
|
// 420 chroma is 1/2 width, 1/2 height
|
|
// Swap src_u and src_v to implement I422ToNV12
|
|
LIBYUV_API
|
|
int I422ToNV21(const uint8_t* src_y,
|
|
int src_stride_y,
|
|
const uint8_t* src_u,
|
|
int src_stride_u,
|
|
const uint8_t* src_v,
|
|
int src_stride_v,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_vu,
|
|
int dst_stride_vu,
|
|
int width,
|
|
int height) {
|
|
int y;
|
|
void (*MergeUVRow)(const uint8_t* src_u, const uint8_t* src_v,
|
|
uint8_t* dst_uv, int width) = MergeUVRow_C;
|
|
void (*InterpolateRow)(uint8_t* dst_ptr, const uint8_t* src_ptr,
|
|
ptrdiff_t src_stride, int dst_width,
|
|
int source_y_fraction) = InterpolateRow_C;
|
|
int halfwidth = (width + 1) >> 1;
|
|
int halfheight = (height + 1) >> 1;
|
|
if ((!src_y && dst_y) || !src_u || !src_v || !dst_vu || width <= 0 ||
|
|
height == 0 || height == INT_MIN) {
|
|
return -1;
|
|
}
|
|
// Negative height means invert the image.
|
|
if (height < 0) {
|
|
height = -height;
|
|
halfheight = (height + 1) >> 1;
|
|
src_y = src_y + (ptrdiff_t)(height - 1) * src_stride_y;
|
|
src_u = src_u + (ptrdiff_t)(halfheight - 1) * src_stride_u;
|
|
src_v = src_v + (ptrdiff_t)(halfheight - 1) * src_stride_v;
|
|
src_stride_y = -src_stride_y;
|
|
src_stride_u = -src_stride_u;
|
|
src_stride_v = -src_stride_v;
|
|
}
|
|
#if defined(HAS_MERGEUVROW_SSE2)
|
|
if (TestCpuFlag(kCpuHasSSE2)) {
|
|
MergeUVRow = MergeUVRow_Any_SSE2;
|
|
if (IS_ALIGNED(halfwidth, 16)) {
|
|
MergeUVRow = MergeUVRow_SSE2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_MERGEUVROW_AVX2)
|
|
if (TestCpuFlag(kCpuHasAVX2)) {
|
|
MergeUVRow = MergeUVRow_Any_AVX2;
|
|
if (IS_ALIGNED(halfwidth, 32)) {
|
|
MergeUVRow = MergeUVRow_AVX2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_MERGEUVROW_AVX512BW)
|
|
if (TestCpuFlag(kCpuHasAVX512BW)) {
|
|
MergeUVRow = MergeUVRow_Any_AVX512BW;
|
|
if (IS_ALIGNED(halfwidth, 32)) {
|
|
MergeUVRow = MergeUVRow_AVX512BW;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_MERGEUVROW_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
MergeUVRow = MergeUVRow_Any_NEON;
|
|
if (IS_ALIGNED(halfwidth, 16)) {
|
|
MergeUVRow = MergeUVRow_NEON;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_MERGEUVROW_SVE2)
|
|
if (TestCpuFlag(kCpuHasSVE2)) {
|
|
MergeUVRow = MergeUVRow_SVE2;
|
|
}
|
|
#endif
|
|
#if defined(HAS_MERGEUVROW_SME)
|
|
if (TestCpuFlag(kCpuHasSME)) {
|
|
MergeUVRow = MergeUVRow_SME;
|
|
}
|
|
#endif
|
|
#if defined(HAS_MERGEUVROW_LSX)
|
|
if (TestCpuFlag(kCpuHasLSX)) {
|
|
MergeUVRow = MergeUVRow_Any_LSX;
|
|
if (IS_ALIGNED(halfwidth, 16)) {
|
|
MergeUVRow = MergeUVRow_LSX;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_MERGEUVROW_LASX)
|
|
if (TestCpuFlag(kCpuHasLASX)) {
|
|
MergeUVRow = MergeUVRow_Any_LASX;
|
|
if (IS_ALIGNED(halfwidth, 32)) {
|
|
MergeUVRow = MergeUVRow_LASX;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_MERGEUVROW_RVV)
|
|
if (TestCpuFlag(kCpuHasRVV)) {
|
|
MergeUVRow = MergeUVRow_RVV;
|
|
}
|
|
#endif
|
|
#if defined(HAS_MERGEUVROW_WASMSIMD)
|
|
if (TestCpuFlag(kCpuHasWASMSIMD)) {
|
|
MergeUVRow = MergeUVRow_Any_WASMSIMD;
|
|
if (IS_ALIGNED(halfwidth, 16)) {
|
|
MergeUVRow = MergeUVRow_WASMSIMD;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_INTERPOLATEROW_AVX2)
|
|
if (TestCpuFlag(kCpuHasAVX2)) {
|
|
InterpolateRow = InterpolateRow_Any_AVX2;
|
|
if (IS_ALIGNED(width, 32)) {
|
|
InterpolateRow = InterpolateRow_AVX2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_INTERPOLATEROW_AVX512BW)
|
|
if (TestCpuFlag(kCpuHasAVX512BW)) {
|
|
InterpolateRow = InterpolateRow_AVX512BW;
|
|
}
|
|
#endif
|
|
#if defined(HAS_INTERPOLATEROW_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
InterpolateRow = InterpolateRow_Any_NEON;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
InterpolateRow = InterpolateRow_NEON;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_INTERPOLATEROW_SVE2)
|
|
if (TestCpuFlag(kCpuHasSVE2)) {
|
|
InterpolateRow = InterpolateRow_SVE2;
|
|
}
|
|
#endif
|
|
#if defined(HAS_INTERPOLATEROW_SME)
|
|
if (TestCpuFlag(kCpuHasSME)) {
|
|
InterpolateRow = InterpolateRow_SME;
|
|
}
|
|
#endif
|
|
#if defined(HAS_INTERPOLATEROW_LSX)
|
|
if (TestCpuFlag(kCpuHasLSX)) {
|
|
InterpolateRow = InterpolateRow_Any_LSX;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
InterpolateRow = InterpolateRow_LSX;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_INTERPOLATEROW_LASX)
|
|
if (TestCpuFlag(kCpuHasLASX)) {
|
|
InterpolateRow = InterpolateRow_Any_LASX;
|
|
if (IS_ALIGNED(width, 32)) {
|
|
InterpolateRow = InterpolateRow_LASX;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_INTERPOLATEROW_RVV)
|
|
if (TestCpuFlag(kCpuHasRVV)) {
|
|
InterpolateRow = InterpolateRow_RVV;
|
|
}
|
|
#endif
|
|
#if defined(HAS_INTERPOLATEROW_WASMSIMD)
|
|
if (TestCpuFlag(kCpuHasWASMSIMD)) {
|
|
InterpolateRow = InterpolateRow_Any_WASMSIMD;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
InterpolateRow = InterpolateRow_WASMSIMD;
|
|
}
|
|
}
|
|
#endif
|
|
|
|
if (dst_y) {
|
|
CopyPlane(src_y, src_stride_y, dst_y, dst_stride_y, halfwidth, height);
|
|
}
|
|
{
|
|
// Allocate 2 rows of vu.
|
|
int awidth = halfwidth * 2;
|
|
align_buffer_64(row_vu_0, awidth * 2);
|
|
uint8_t* row_vu_1 = row_vu_0 + awidth;
|
|
if (!row_vu_0)
|
|
return 1;
|
|
|
|
for (y = 0; y < height - 1; y += 2) {
|
|
MergeUVRow(src_v, src_u, row_vu_0, halfwidth);
|
|
MergeUVRow(src_v + src_stride_v, src_u + src_stride_u, row_vu_1,
|
|
halfwidth);
|
|
InterpolateRow(dst_vu, row_vu_0, awidth, awidth, 128);
|
|
src_u += src_stride_u * 2;
|
|
src_v += src_stride_v * 2;
|
|
dst_vu += dst_stride_vu;
|
|
}
|
|
if (height & 1) {
|
|
MergeUVRow(src_v, src_u, dst_vu, halfwidth);
|
|
}
|
|
free_aligned_buffer_64(row_vu_0);
|
|
}
|
|
return 0;
|
|
}
|
|
#endif // I422TONV21_ROW_VERSION
|
|
|
|
// 444 chroma is 1x width, 1x height
|
|
// 420 chroma is 1/2 width, 1/2 height
|
|
LIBYUV_API
|
|
int I444ToI420(const uint8_t* src_y,
|
|
int src_stride_y,
|
|
const uint8_t* src_u,
|
|
int src_stride_u,
|
|
const uint8_t* src_v,
|
|
int src_stride_v,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
return I4xxToI420(src_y, src_stride_y, src_u, src_stride_u, src_v,
|
|
src_stride_v, dst_y, dst_stride_y, dst_u, dst_stride_u,
|
|
dst_v, dst_stride_v, width, height, width, height);
|
|
}
|
|
|
|
LIBYUV_API
|
|
int I444ToNV12(const uint8_t* src_y,
|
|
int src_stride_y,
|
|
const uint8_t* src_u,
|
|
int src_stride_u,
|
|
const uint8_t* src_v,
|
|
int src_stride_v,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_uv,
|
|
int dst_stride_uv,
|
|
int width,
|
|
int height) {
|
|
if ((!src_y && dst_y) || !src_u || !src_v || !dst_uv || width <= 0 ||
|
|
height == 0 || height == INT_MIN) {
|
|
return -1;
|
|
}
|
|
// Negative height means invert the image.
|
|
if (height < 0) {
|
|
height = -height;
|
|
src_y = src_y + (ptrdiff_t)(height - 1) * src_stride_y;
|
|
src_u = src_u + (ptrdiff_t)(height - 1) * src_stride_u;
|
|
src_v = src_v + (ptrdiff_t)(height - 1) * src_stride_v;
|
|
src_stride_y = -src_stride_y;
|
|
src_stride_u = -src_stride_u;
|
|
src_stride_v = -src_stride_v;
|
|
}
|
|
if (dst_y) {
|
|
CopyPlane(src_y, src_stride_y, dst_y, dst_stride_y, width, height);
|
|
}
|
|
HalfMergeUVPlane(src_u, src_stride_u, src_v, src_stride_v, dst_uv,
|
|
dst_stride_uv, width, height);
|
|
return 0;
|
|
}
|
|
|
|
LIBYUV_API
|
|
int I444ToNV21(const uint8_t* src_y,
|
|
int src_stride_y,
|
|
const uint8_t* src_u,
|
|
int src_stride_u,
|
|
const uint8_t* src_v,
|
|
int src_stride_v,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_vu,
|
|
int dst_stride_vu,
|
|
int width,
|
|
int height) {
|
|
return I444ToNV12(src_y, src_stride_y, src_v, src_stride_v, src_u,
|
|
src_stride_u, dst_y, dst_stride_y, dst_vu, dst_stride_vu,
|
|
width, height);
|
|
}
|
|
|
|
// I400 is greyscale typically used in MJPG
|
|
LIBYUV_API
|
|
int I400ToI420(const uint8_t* src_y,
|
|
int src_stride_y,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
int halfwidth = (width + 1) >> 1;
|
|
int halfheight = (height + 1) >> 1;
|
|
if ((!src_y && dst_y) || !dst_u || !dst_v || width <= 0 || height == 0 ||
|
|
height == INT_MIN) {
|
|
return -1;
|
|
}
|
|
// Negative height means invert the image.
|
|
if (height < 0) {
|
|
height = -height;
|
|
halfheight = (height + 1) >> 1;
|
|
src_y = src_y + (ptrdiff_t)(height - 1) * src_stride_y;
|
|
src_stride_y = -src_stride_y;
|
|
}
|
|
if (dst_y) {
|
|
CopyPlane(src_y, src_stride_y, dst_y, dst_stride_y, width, height);
|
|
}
|
|
SetPlane(dst_u, dst_stride_u, halfwidth, halfheight, 128);
|
|
SetPlane(dst_v, dst_stride_v, halfwidth, halfheight, 128);
|
|
return 0;
|
|
}
|
|
|
|
// I400 is greyscale typically used in MJPG
|
|
LIBYUV_API
|
|
int I400ToNV21(const uint8_t* src_y,
|
|
int src_stride_y,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_vu,
|
|
int dst_stride_vu,
|
|
int width,
|
|
int height) {
|
|
int halfwidth = (width + 1) >> 1;
|
|
int halfheight = (height + 1) >> 1;
|
|
if ((!src_y && dst_y) || !dst_vu || width <= 0 || height == 0 ||
|
|
height == INT_MIN) {
|
|
return -1;
|
|
}
|
|
// Negative height means invert the image.
|
|
if (height < 0) {
|
|
height = -height;
|
|
halfheight = (height + 1) >> 1;
|
|
src_y = src_y + (ptrdiff_t)(height - 1) * src_stride_y;
|
|
src_stride_y = -src_stride_y;
|
|
}
|
|
if (dst_y) {
|
|
CopyPlane(src_y, src_stride_y, dst_y, dst_stride_y, width, height);
|
|
}
|
|
SetPlane(dst_vu, dst_stride_vu, halfwidth * 2, halfheight, 128);
|
|
return 0;
|
|
}
|
|
|
|
// Convert NV12 to I420.
|
|
// TODO(fbarchard): Consider inverting destination. Faster on ARM with prfm.
|
|
LIBYUV_API
|
|
int NV12ToI420(const uint8_t* src_y,
|
|
int src_stride_y,
|
|
const uint8_t* src_uv,
|
|
int src_stride_uv,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
int halfwidth = (width + 1) >> 1;
|
|
int halfheight = (height + 1) >> 1;
|
|
if ((!src_y && dst_y) || !src_uv || !dst_u || !dst_v || width <= 0 ||
|
|
height == 0 || height == INT_MIN) {
|
|
return -1;
|
|
}
|
|
// Negative height means invert the image.
|
|
if (height < 0) {
|
|
height = -height;
|
|
halfheight = (height + 1) >> 1;
|
|
src_y = src_y + (ptrdiff_t)(height - 1) * src_stride_y;
|
|
src_uv = src_uv + (ptrdiff_t)(halfheight - 1) * src_stride_uv;
|
|
src_stride_y = -src_stride_y;
|
|
src_stride_uv = -src_stride_uv;
|
|
}
|
|
// Coalesce rows.
|
|
if (src_stride_y == width && dst_stride_y == width &&
|
|
(ptrdiff_t)width * height <= INT_MAX) {
|
|
width *= height;
|
|
height = 1;
|
|
src_stride_y = dst_stride_y = 0;
|
|
}
|
|
// Coalesce rows.
|
|
if (src_stride_uv == halfwidth * 2 && dst_stride_u == halfwidth &&
|
|
dst_stride_v == halfwidth &&
|
|
(ptrdiff_t)halfwidth * halfheight <= INT_MAX) {
|
|
halfwidth *= halfheight;
|
|
halfheight = 1;
|
|
src_stride_uv = dst_stride_u = dst_stride_v = 0;
|
|
}
|
|
|
|
if (dst_y) {
|
|
CopyPlane(src_y, src_stride_y, dst_y, dst_stride_y, width, height);
|
|
}
|
|
|
|
// Split UV plane - NV12 / NV21
|
|
SplitUVPlane(src_uv, src_stride_uv, dst_u, dst_stride_u, dst_v, dst_stride_v,
|
|
halfwidth, halfheight);
|
|
|
|
return 0;
|
|
}
|
|
|
|
// Convert NV21 to I420. Same as NV12 but u and v pointers swapped.
|
|
LIBYUV_API
|
|
int NV21ToI420(const uint8_t* src_y,
|
|
int src_stride_y,
|
|
const uint8_t* src_vu,
|
|
int src_stride_vu,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
return NV12ToI420(src_y, src_stride_y, src_vu, src_stride_vu, dst_y,
|
|
dst_stride_y, dst_v, dst_stride_v, dst_u, dst_stride_u,
|
|
width, height);
|
|
}
|
|
|
|
LIBYUV_API
|
|
int NV12ToNV24(const uint8_t* src_y,
|
|
int src_stride_y,
|
|
const uint8_t* src_uv,
|
|
int src_stride_uv,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_uv,
|
|
int dst_stride_uv,
|
|
int width,
|
|
int height) {
|
|
int r;
|
|
if ((!src_y && dst_y) || !src_uv || !dst_uv || width <= 0 || height == 0 ||
|
|
height == INT_MIN) {
|
|
return -1;
|
|
}
|
|
|
|
if (dst_y) {
|
|
CopyPlane(src_y, src_stride_y, dst_y, dst_stride_y, width, height);
|
|
}
|
|
r = UVScale(src_uv, src_stride_uv, SUBSAMPLE(width, 1, 1),
|
|
SUBSAMPLE(height, 1, 1), dst_uv, dst_stride_uv, Abs(width),
|
|
Abs(height), kFilterBilinear);
|
|
return r;
|
|
}
|
|
|
|
LIBYUV_API
|
|
int NV16ToNV24(const uint8_t* src_y,
|
|
int src_stride_y,
|
|
const uint8_t* src_uv,
|
|
int src_stride_uv,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_uv,
|
|
int dst_stride_uv,
|
|
int width,
|
|
int height) {
|
|
int r;
|
|
if ((!src_y && dst_y) || !src_uv || !dst_uv || width <= 0 || height == 0 ||
|
|
height == INT_MIN) {
|
|
return -1;
|
|
}
|
|
|
|
if (dst_y) {
|
|
CopyPlane(src_y, src_stride_y, dst_y, dst_stride_y, width, height);
|
|
}
|
|
r = UVScale(src_uv, src_stride_uv, SUBSAMPLE(width, 1, 1), height, dst_uv,
|
|
dst_stride_uv, Abs(width), Abs(height), kFilterBilinear);
|
|
return r;
|
|
}
|
|
|
|
// Any P[420]1[02] to I[420]1[02] format
|
|
static int PxxxToIxxx(const uint16_t* src_y,
|
|
int src_stride_y,
|
|
const uint16_t* src_uv,
|
|
int src_stride_uv,
|
|
uint16_t* dst_y,
|
|
int dst_stride_y,
|
|
uint16_t* dst_u,
|
|
int dst_stride_u,
|
|
uint16_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height,
|
|
int subsample_x,
|
|
int subsample_y,
|
|
int depth) {
|
|
const int uv_width = SUBSAMPLE(width, subsample_x, subsample_x);
|
|
const int uv_height = SUBSAMPLE(height, subsample_y, subsample_y);
|
|
if (!src_y || !dst_y || !src_uv || !dst_u || !dst_v || width <= 0 ||
|
|
height == 0 || height == INT_MIN) {
|
|
return -1;
|
|
}
|
|
ConvertToLSBPlane_16(src_y, src_stride_y, dst_y, dst_stride_y, width, height,
|
|
depth);
|
|
SplitUVPlane_16(src_uv, src_stride_uv, dst_u, dst_stride_u, dst_v,
|
|
dst_stride_v, uv_width, uv_height, depth);
|
|
return 0;
|
|
}
|
|
|
|
LIBYUV_API
|
|
int P010ToI010(const uint16_t* src_y,
|
|
int src_stride_y,
|
|
const uint16_t* src_uv,
|
|
int src_stride_uv,
|
|
uint16_t* dst_y,
|
|
int dst_stride_y,
|
|
uint16_t* dst_u,
|
|
int dst_stride_u,
|
|
uint16_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
return PxxxToIxxx(src_y, src_stride_y, src_uv, src_stride_uv, dst_y,
|
|
dst_stride_y, dst_u, dst_stride_u, dst_v, dst_stride_v,
|
|
width, height, 1, 1, 10);
|
|
}
|
|
|
|
LIBYUV_API
|
|
int P210ToI210(const uint16_t* src_y,
|
|
int src_stride_y,
|
|
const uint16_t* src_uv,
|
|
int src_stride_uv,
|
|
uint16_t* dst_y,
|
|
int dst_stride_y,
|
|
uint16_t* dst_u,
|
|
int dst_stride_u,
|
|
uint16_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
return PxxxToIxxx(src_y, src_stride_y, src_uv, src_stride_uv, dst_y,
|
|
dst_stride_y, dst_u, dst_stride_u, dst_v, dst_stride_v,
|
|
width, height, 1, 0, 10);
|
|
}
|
|
|
|
LIBYUV_API
|
|
int P410ToI410(const uint16_t* src_y,
|
|
int src_stride_y,
|
|
const uint16_t* src_uv,
|
|
int src_stride_uv,
|
|
uint16_t* dst_y,
|
|
int dst_stride_y,
|
|
uint16_t* dst_u,
|
|
int dst_stride_u,
|
|
uint16_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
return PxxxToIxxx(src_y, src_stride_y, src_uv, src_stride_uv, dst_y,
|
|
dst_stride_y, dst_u, dst_stride_u, dst_v, dst_stride_v,
|
|
width, height, 0, 0, 10);
|
|
}
|
|
|
|
static void InvertPx10(const uint16_t** src_y,
|
|
int* src_stride_y,
|
|
const uint16_t** src_uv,
|
|
int* src_stride_uv,
|
|
int* height,
|
|
int subsample_y) {
|
|
const int abs_height = Abs(*height);
|
|
const int src_uv_height = SUBSAMPLE(abs_height, subsample_y, subsample_y);
|
|
if (*src_y) {
|
|
*src_y = *src_y + (ptrdiff_t)(abs_height - 1) * *src_stride_y;
|
|
*src_stride_y = -*src_stride_y;
|
|
}
|
|
*src_uv = *src_uv + (ptrdiff_t)(src_uv_height - 1) * *src_stride_uv;
|
|
*src_stride_uv = -*src_stride_uv;
|
|
*height = abs_height;
|
|
}
|
|
|
|
static int Px10ToI420(const uint16_t* src_y,
|
|
int src_stride_y,
|
|
const uint16_t* src_uv,
|
|
int src_stride_uv,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height,
|
|
int subsample_x,
|
|
int subsample_y) {
|
|
const int scale16 = 256; // P*10 stores 10-bit in the MSBs.
|
|
int src_uv_width;
|
|
int dst_uv_width;
|
|
if ((!src_y && dst_y) || !src_uv || !dst_u || !dst_v || width <= 0 ||
|
|
height == 0 || height == INT_MIN) {
|
|
return -1;
|
|
}
|
|
if (height < 0) {
|
|
InvertPx10(&src_y, &src_stride_y, &src_uv, &src_stride_uv, &height,
|
|
subsample_y);
|
|
}
|
|
src_uv_width = SUBSAMPLE(width, subsample_x, subsample_x);
|
|
dst_uv_width = SUBSAMPLE(width, 1, 1);
|
|
|
|
if (dst_y) {
|
|
Convert16To8Plane(src_y, src_stride_y, dst_y, dst_stride_y, scale16, width,
|
|
height);
|
|
}
|
|
|
|
if (subsample_x && subsample_y) {
|
|
int y;
|
|
const int uv_packed = src_uv_width * 2;
|
|
const int uv_height = SUBSAMPLE(height, 1, 1);
|
|
void (*Convert16To8Row)(const uint16_t* src_y, uint8_t* dst_y, int scale,
|
|
int width) = Convert16To8Row_C;
|
|
void (*SplitUVRow)(const uint8_t* src_uv, uint8_t* dst_u, uint8_t* dst_v,
|
|
int width) = SplitUVRow_C;
|
|
|
|
#if defined(HAS_CONVERT16TO8ROW_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
Convert16To8Row = Convert16To8Row_Any_NEON;
|
|
if (IS_ALIGNED(uv_packed, 16)) {
|
|
Convert16To8Row = Convert16To8Row_NEON;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_CONVERT16TO8ROW_SVE2)
|
|
if (TestCpuFlag(kCpuHasSVE2)) {
|
|
Convert16To8Row = Convert16To8Row_SVE2;
|
|
}
|
|
#endif
|
|
#if defined(HAS_CONVERT16TO8ROW_SME)
|
|
if (TestCpuFlag(kCpuHasSME)) {
|
|
Convert16To8Row = Convert16To8Row_SME;
|
|
}
|
|
#endif
|
|
#if defined(HAS_CONVERT16TO8ROW_SSSE3)
|
|
if (TestCpuFlag(kCpuHasSSSE3)) {
|
|
Convert16To8Row = Convert16To8Row_Any_SSSE3;
|
|
if (IS_ALIGNED(uv_packed, 16)) {
|
|
Convert16To8Row = Convert16To8Row_SSSE3;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_CONVERT16TO8ROW_AVX2)
|
|
if (TestCpuFlag(kCpuHasAVX2)) {
|
|
Convert16To8Row = Convert16To8Row_Any_AVX2;
|
|
if (IS_ALIGNED(uv_packed, 32)) {
|
|
Convert16To8Row = Convert16To8Row_AVX2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_CONVERT16TO8ROW_AVX512BW)
|
|
if (TestCpuFlag(kCpuHasAVX512BW)) {
|
|
Convert16To8Row = Convert16To8Row_Any_AVX512BW;
|
|
if (IS_ALIGNED(uv_packed, 64)) {
|
|
Convert16To8Row = Convert16To8Row_AVX512BW;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_CONVERT16TO8ROW_RVV)
|
|
if (TestCpuFlag(kCpuHasRVV)) {
|
|
Convert16To8Row = Convert16To8Row_RVV;
|
|
}
|
|
#endif
|
|
#if defined(HAS_CONVERT16TO8ROW_HVX)
|
|
if (TestCpuFlag(kCpuHasHVX)) {
|
|
Convert16To8Row = Convert16To8Row_HVX;
|
|
}
|
|
#endif
|
|
#if defined(HAS_SPLITUVROW_SSE2)
|
|
if (TestCpuFlag(kCpuHasSSE2)) {
|
|
SplitUVRow = SplitUVRow_Any_SSE2;
|
|
if (IS_ALIGNED(src_uv_width, 16)) {
|
|
SplitUVRow = SplitUVRow_SSE2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_SPLITUVROW_AVX2)
|
|
if (TestCpuFlag(kCpuHasAVX2)) {
|
|
SplitUVRow = SplitUVRow_Any_AVX2;
|
|
if (IS_ALIGNED(src_uv_width, 32)) {
|
|
SplitUVRow = SplitUVRow_AVX2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_SPLITUVROW_AVX512BW)
|
|
if (TestCpuFlag(kCpuHasAVX512BW)) {
|
|
SplitUVRow = SplitUVRow_Any_AVX512BW;
|
|
if (IS_ALIGNED(src_uv_width, 64)) {
|
|
SplitUVRow = SplitUVRow_AVX512BW;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_SPLITUVROW_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
SplitUVRow = SplitUVRow_Any_NEON;
|
|
if (IS_ALIGNED(src_uv_width, 16)) {
|
|
SplitUVRow = SplitUVRow_NEON;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_SPLITUVROW_SVE2)
|
|
if (TestCpuFlag(kCpuHasSVE2)) {
|
|
SplitUVRow = SplitUVRow_SVE2;
|
|
}
|
|
#endif
|
|
#if defined(HAS_SPLITUVROW_LSX)
|
|
if (TestCpuFlag(kCpuHasLSX)) {
|
|
SplitUVRow = SplitUVRow_Any_LSX;
|
|
if (IS_ALIGNED(src_uv_width, 16)) {
|
|
SplitUVRow = SplitUVRow_LSX;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_SPLITUVROW_LASX)
|
|
if (TestCpuFlag(kCpuHasLASX)) {
|
|
SplitUVRow = SplitUVRow_Any_LASX;
|
|
if (IS_ALIGNED(src_uv_width, 32)) {
|
|
SplitUVRow = SplitUVRow_LASX;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_SPLITUVROW_RVV)
|
|
if (TestCpuFlag(kCpuHasRVV)) {
|
|
SplitUVRow = SplitUVRow_RVV;
|
|
}
|
|
#endif
|
|
#if defined(HAS_SPLITUVROW_HVX)
|
|
if (TestCpuFlag(kCpuHasHVX)) {
|
|
SplitUVRow = SplitUVRow_HVX;
|
|
}
|
|
#endif
|
|
|
|
align_buffer_64(row_uv, ((uv_packed + 63) & ~63));
|
|
if (!row_uv) {
|
|
return 1;
|
|
}
|
|
for (y = 0; y < uv_height; ++y) {
|
|
Convert16To8Row(src_uv, row_uv, scale16, uv_packed);
|
|
SplitUVRow(row_uv, dst_u, dst_v, src_uv_width);
|
|
src_uv += src_stride_uv;
|
|
dst_u += dst_stride_u;
|
|
dst_v += dst_stride_v;
|
|
}
|
|
free_aligned_buffer_64(row_uv);
|
|
return 0;
|
|
}
|
|
|
|
{
|
|
int y;
|
|
const int scale = 16384; // 10-bit LSB after SplitUVRow_16.
|
|
int row_stride;
|
|
void (*SplitUVRow_16)(const uint16_t* src_uv, uint16_t* dst_u,
|
|
uint16_t* dst_v, int depth, int width) =
|
|
SplitUVRow_16_C;
|
|
void (*Half16To8)(const uint16_t* src_uv, ptrdiff_t src_uv_stride,
|
|
uint8_t* dst_uv, int scale, int width) =
|
|
subsample_x
|
|
? HalfRow_16To8_C
|
|
: ((width & 1) ? HalfWidthRow_16To8_Odd_C : HalfWidthRow_16To8_C);
|
|
|
|
#if defined(HAS_SPLITUVROW_16_AVX2)
|
|
if (TestCpuFlag(kCpuHasAVX2)) {
|
|
SplitUVRow_16 = SplitUVRow_16_Any_AVX2;
|
|
if (IS_ALIGNED(src_uv_width, 16)) {
|
|
SplitUVRow_16 = SplitUVRow_16_AVX2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_SPLITUVROW_16_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
SplitUVRow_16 = SplitUVRow_16_Any_NEON;
|
|
if (IS_ALIGNED(src_uv_width, 8)) {
|
|
SplitUVRow_16 = SplitUVRow_16_NEON;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_SPLITUVROW_16_HVX)
|
|
if (TestCpuFlag(kCpuHasHVX)) {
|
|
SplitUVRow_16 = SplitUVRow_16_HVX;
|
|
}
|
|
#endif
|
|
if (subsample_x) {
|
|
#if defined(HAS_HALFROW_16TO8_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
Half16To8 = HalfRow_16To8_Any_NEON;
|
|
if (IS_ALIGNED(dst_uv_width, 16)) {
|
|
Half16To8 = HalfRow_16To8_NEON;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_HALFROW_16TO8_SVE2)
|
|
if (TestCpuFlag(kCpuHasSVE2)) {
|
|
Half16To8 = HalfRow_16To8_SVE2;
|
|
}
|
|
#endif
|
|
#if defined(HAS_HALFROW_16TO8_SME)
|
|
if (TestCpuFlag(kCpuHasSME)) {
|
|
Half16To8 = HalfRow_16To8_SME;
|
|
}
|
|
#endif
|
|
#if defined(HAS_HALFROW_16TO8_SSSE3)
|
|
if (TestCpuFlag(kCpuHasSSSE3)) {
|
|
Half16To8 = HalfRow_16To8_Any_SSSE3;
|
|
if (IS_ALIGNED(dst_uv_width, 16)) {
|
|
Half16To8 = HalfRow_16To8_SSSE3;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_HALFROW_16TO8_AVX2)
|
|
if (TestCpuFlag(kCpuHasAVX2)) {
|
|
Half16To8 = HalfRow_16To8_Any_AVX2;
|
|
if (IS_ALIGNED(dst_uv_width, 32)) {
|
|
Half16To8 = HalfRow_16To8_AVX2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_HALFROW_16TO8_AVX512BW)
|
|
if (TestCpuFlag(kCpuHasAVX512BW)) {
|
|
Half16To8 = HalfRow_16To8_Any_AVX512BW;
|
|
if (IS_ALIGNED(dst_uv_width, 64)) {
|
|
Half16To8 = HalfRow_16To8_AVX512BW;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_HALFROW_16TO8_RVV)
|
|
if (TestCpuFlag(kCpuHasRVV)) {
|
|
Half16To8 = HalfRow_16To8_RVV;
|
|
}
|
|
#endif
|
|
#if defined(HAS_HALFROW_16TO8_HVX)
|
|
if (TestCpuFlag(kCpuHasHVX)) {
|
|
Half16To8 = HalfRow_16To8_HVX;
|
|
}
|
|
#endif
|
|
} else if (IS_ALIGNED(width, 2)) {
|
|
#if defined(HAS_HALFWIDTHROW_16TO8_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
Half16To8 = HalfWidthRow_16To8_Any_NEON;
|
|
if (IS_ALIGNED(dst_uv_width, 8)) {
|
|
Half16To8 = HalfWidthRow_16To8_NEON;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_HALFWIDTHROW_16TO8_SVE2)
|
|
if (TestCpuFlag(kCpuHasSVE2)) {
|
|
Half16To8 = HalfWidthRow_16To8_SVE2;
|
|
}
|
|
#endif
|
|
#if defined(HAS_HALFWIDTHROW_16TO8_SME)
|
|
if (TestCpuFlag(kCpuHasSME)) {
|
|
Half16To8 = HalfWidthRow_16To8_SME;
|
|
}
|
|
#endif
|
|
#if defined(HAS_HALFWIDTHROW_16TO8_SSSE3)
|
|
if (TestCpuFlag(kCpuHasSSSE3)) {
|
|
Half16To8 = HalfWidthRow_16To8_Any_SSSE3;
|
|
if (IS_ALIGNED(dst_uv_width, 8)) {
|
|
Half16To8 = HalfWidthRow_16To8_SSSE3;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_HALFWIDTHROW_16TO8_AVX2)
|
|
if (TestCpuFlag(kCpuHasAVX2)) {
|
|
Half16To8 = HalfWidthRow_16To8_Any_AVX2;
|
|
if (IS_ALIGNED(dst_uv_width, 16)) {
|
|
Half16To8 = HalfWidthRow_16To8_AVX2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_HALFWIDTHROW_16TO8_AVX512BW)
|
|
if (TestCpuFlag(kCpuHasAVX512BW)) {
|
|
Half16To8 = HalfWidthRow_16To8_Any_AVX512BW;
|
|
if (IS_ALIGNED(dst_uv_width, 32)) {
|
|
Half16To8 = HalfWidthRow_16To8_AVX512BW;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_HALFWIDTHROW_16TO8_RVV)
|
|
if (TestCpuFlag(kCpuHasRVV)) {
|
|
Half16To8 = HalfWidthRow_16To8_RVV;
|
|
}
|
|
#endif
|
|
}
|
|
|
|
row_stride = (src_uv_width + 31) & ~31;
|
|
align_buffer_64(row_u, (size_t)row_stride * sizeof(uint16_t) * 4u);
|
|
if (!row_u) {
|
|
return 1;
|
|
}
|
|
{
|
|
uint16_t* row_u0 = (uint16_t*)row_u;
|
|
uint16_t* row_u1 = row_u0 + row_stride;
|
|
uint16_t* row_v0 = row_u1 + row_stride;
|
|
uint16_t* row_v1 = row_v0 + row_stride;
|
|
for (y = 0; y < height - 1; y += 2) {
|
|
SplitUVRow_16(src_uv, row_u0, row_v0, 10, src_uv_width);
|
|
SplitUVRow_16(src_uv + src_stride_uv, row_u1, row_v1, 10, src_uv_width);
|
|
Half16To8(row_u0, row_stride, dst_u, scale, dst_uv_width);
|
|
Half16To8(row_v0, row_stride, dst_v, scale, dst_uv_width);
|
|
src_uv += src_stride_uv * 2;
|
|
dst_u += dst_stride_u;
|
|
dst_v += dst_stride_v;
|
|
}
|
|
if (height & 1) {
|
|
SplitUVRow_16(src_uv, row_u0, row_v0, 10, src_uv_width);
|
|
Half16To8(row_u0, 0, dst_u, scale, dst_uv_width);
|
|
Half16To8(row_v0, 0, dst_v, scale, dst_uv_width);
|
|
}
|
|
}
|
|
free_aligned_buffer_64(row_u);
|
|
return 0;
|
|
}
|
|
}
|
|
|
|
LIBYUV_API
|
|
int P010ToI420(const uint16_t* src_y,
|
|
int src_stride_y,
|
|
const uint16_t* src_uv,
|
|
int src_stride_uv,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
return Px10ToI420(src_y, src_stride_y, src_uv, src_stride_uv, dst_y,
|
|
dst_stride_y, dst_u, dst_stride_u, dst_v, dst_stride_v,
|
|
width, height, 1, 1);
|
|
}
|
|
|
|
LIBYUV_API
|
|
int P210ToI420(const uint16_t* src_y,
|
|
int src_stride_y,
|
|
const uint16_t* src_uv,
|
|
int src_stride_uv,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
return Px10ToI420(src_y, src_stride_y, src_uv, src_stride_uv, dst_y,
|
|
dst_stride_y, dst_u, dst_stride_u, dst_v, dst_stride_v,
|
|
width, height, 1, 0);
|
|
}
|
|
|
|
LIBYUV_API
|
|
int P410ToI420(const uint16_t* src_y,
|
|
int src_stride_y,
|
|
const uint16_t* src_uv,
|
|
int src_stride_uv,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
return Px10ToI420(src_y, src_stride_y, src_uv, src_stride_uv, dst_y,
|
|
dst_stride_y, dst_u, dst_stride_u, dst_v, dst_stride_v,
|
|
width, height, 0, 0);
|
|
}
|
|
|
|
LIBYUV_API
|
|
int P012ToI012(const uint16_t* src_y,
|
|
int src_stride_y,
|
|
const uint16_t* src_uv,
|
|
int src_stride_uv,
|
|
uint16_t* dst_y,
|
|
int dst_stride_y,
|
|
uint16_t* dst_u,
|
|
int dst_stride_u,
|
|
uint16_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
return PxxxToIxxx(src_y, src_stride_y, src_uv, src_stride_uv, dst_y,
|
|
dst_stride_y, dst_u, dst_stride_u, dst_v, dst_stride_v,
|
|
width, height, 1, 1, 12);
|
|
}
|
|
|
|
LIBYUV_API
|
|
int P010ToP410(const uint16_t* src_y,
|
|
int src_stride_y,
|
|
const uint16_t* src_uv,
|
|
int src_stride_uv,
|
|
uint16_t* dst_y,
|
|
int dst_stride_y,
|
|
uint16_t* dst_uv,
|
|
int dst_stride_uv,
|
|
int width,
|
|
int height) {
|
|
int r;
|
|
if ((!src_y && dst_y) || !src_uv || !dst_uv || width <= 0 || height == 0 ||
|
|
height == INT_MIN) {
|
|
return -1;
|
|
}
|
|
|
|
if (dst_y) {
|
|
CopyPlane_16(src_y, src_stride_y, dst_y, dst_stride_y, width, height);
|
|
}
|
|
r = UVScale_16(src_uv, src_stride_uv, SUBSAMPLE(width, 1, 1),
|
|
SUBSAMPLE(height, 1, 1), dst_uv, dst_stride_uv, Abs(width),
|
|
Abs(height), kFilterBilinear);
|
|
return r;
|
|
}
|
|
|
|
LIBYUV_API
|
|
int P210ToP410(const uint16_t* src_y,
|
|
int src_stride_y,
|
|
const uint16_t* src_uv,
|
|
int src_stride_uv,
|
|
uint16_t* dst_y,
|
|
int dst_stride_y,
|
|
uint16_t* dst_uv,
|
|
int dst_stride_uv,
|
|
int width,
|
|
int height) {
|
|
int r;
|
|
if ((!src_y && dst_y) || !src_uv || !dst_uv || width <= 0 || height == 0 ||
|
|
height == INT_MIN) {
|
|
return -1;
|
|
}
|
|
|
|
if (dst_y) {
|
|
CopyPlane_16(src_y, src_stride_y, dst_y, dst_stride_y, width, height);
|
|
}
|
|
r = UVScale_16(src_uv, src_stride_uv, SUBSAMPLE(width, 1, 1), height, dst_uv,
|
|
dst_stride_uv, Abs(width), Abs(height), kFilterBilinear);
|
|
return r;
|
|
}
|
|
|
|
// Convert YUY2 to I420.
|
|
LIBYUV_API
|
|
int YUY2ToI420(const uint8_t* src_yuy2,
|
|
int src_stride_yuy2,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
int y;
|
|
void (*YUY2ToUVRow)(const uint8_t* src_yuy2, int src_stride_yuy2,
|
|
uint8_t* dst_u, uint8_t* dst_v, int width) =
|
|
YUY2ToUVRow_C;
|
|
void (*YUY2ToYRow)(const uint8_t* src_yuy2, uint8_t* dst_y, int width) =
|
|
YUY2ToYRow_C;
|
|
if (width <= 0 || height == 0 || height == INT_MIN) {
|
|
return -1;
|
|
}
|
|
// Negative height means invert the image.
|
|
if (height < 0) {
|
|
height = -height;
|
|
src_yuy2 = src_yuy2 + (ptrdiff_t)(height - 1) * src_stride_yuy2;
|
|
src_stride_yuy2 = -src_stride_yuy2;
|
|
}
|
|
#if defined(HAS_YUY2TOYROW_SSE2)
|
|
if (TestCpuFlag(kCpuHasSSE2)) {
|
|
YUY2ToUVRow = YUY2ToUVRow_Any_SSE2;
|
|
YUY2ToYRow = YUY2ToYRow_Any_SSE2;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
YUY2ToUVRow = YUY2ToUVRow_SSE2;
|
|
YUY2ToYRow = YUY2ToYRow_SSE2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_YUY2TOYROW_AVX2)
|
|
if (TestCpuFlag(kCpuHasAVX2)) {
|
|
YUY2ToUVRow = YUY2ToUVRow_Any_AVX2;
|
|
YUY2ToYRow = YUY2ToYRow_Any_AVX2;
|
|
if (IS_ALIGNED(width, 32)) {
|
|
YUY2ToUVRow = YUY2ToUVRow_AVX2;
|
|
YUY2ToYRow = YUY2ToYRow_AVX2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_YUY2TOYROW_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
YUY2ToYRow = YUY2ToYRow_Any_NEON;
|
|
YUY2ToUVRow = YUY2ToUVRow_Any_NEON;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
YUY2ToYRow = YUY2ToYRow_NEON;
|
|
YUY2ToUVRow = YUY2ToUVRow_NEON;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_YUY2TOYROW_SVE2) && defined(HAS_YUY2TOUVROW_SVE2)
|
|
if (TestCpuFlag(kCpuHasSVE2)) {
|
|
YUY2ToYRow = YUY2ToYRow_SVE2;
|
|
YUY2ToUVRow = YUY2ToUVRow_SVE2;
|
|
}
|
|
#endif
|
|
#if defined(HAS_YUY2TOYROW_LSX) && defined(HAS_YUY2TOUVROW_LSX)
|
|
if (TestCpuFlag(kCpuHasLSX)) {
|
|
YUY2ToYRow = YUY2ToYRow_Any_LSX;
|
|
YUY2ToUVRow = YUY2ToUVRow_Any_LSX;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
YUY2ToYRow = YUY2ToYRow_LSX;
|
|
YUY2ToUVRow = YUY2ToUVRow_LSX;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_YUY2TOYROW_LASX) && defined(HAS_YUY2TOUVROW_LASX)
|
|
if (TestCpuFlag(kCpuHasLASX)) {
|
|
YUY2ToYRow = YUY2ToYRow_Any_LASX;
|
|
YUY2ToUVRow = YUY2ToUVRow_Any_LASX;
|
|
if (IS_ALIGNED(width, 32)) {
|
|
YUY2ToYRow = YUY2ToYRow_LASX;
|
|
YUY2ToUVRow = YUY2ToUVRow_LASX;
|
|
}
|
|
}
|
|
#endif
|
|
|
|
for (y = 0; y < height - 1; y += 2) {
|
|
YUY2ToUVRow(src_yuy2, src_stride_yuy2, dst_u, dst_v, width);
|
|
YUY2ToYRow(src_yuy2, dst_y, width);
|
|
YUY2ToYRow(src_yuy2 + src_stride_yuy2, dst_y + dst_stride_y, width);
|
|
src_yuy2 += src_stride_yuy2 * 2;
|
|
dst_y += dst_stride_y * 2;
|
|
dst_u += dst_stride_u;
|
|
dst_v += dst_stride_v;
|
|
}
|
|
if (height & 1) {
|
|
YUY2ToUVRow(src_yuy2, 0, dst_u, dst_v, width);
|
|
YUY2ToYRow(src_yuy2, dst_y, width);
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
// Convert UYVY to I420.
|
|
LIBYUV_API
|
|
int UYVYToI420(const uint8_t* src_uyvy,
|
|
int src_stride_uyvy,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
int y;
|
|
void (*UYVYToUVRow)(const uint8_t* src_uyvy, int src_stride_uyvy,
|
|
uint8_t* dst_u, uint8_t* dst_v, int width) =
|
|
UYVYToUVRow_C;
|
|
void (*UYVYToYRow)(const uint8_t* src_uyvy, uint8_t* dst_y, int width) =
|
|
UYVYToYRow_C;
|
|
if (width <= 0 || height == 0 || height == INT_MIN) {
|
|
return -1;
|
|
}
|
|
// Negative height means invert the image.
|
|
if (height < 0) {
|
|
height = -height;
|
|
src_uyvy = src_uyvy + (ptrdiff_t)(height - 1) * src_stride_uyvy;
|
|
src_stride_uyvy = -src_stride_uyvy;
|
|
}
|
|
#if defined(HAS_UYVYTOYROW_SSE2)
|
|
if (TestCpuFlag(kCpuHasSSE2)) {
|
|
UYVYToUVRow = UYVYToUVRow_Any_SSE2;
|
|
UYVYToYRow = UYVYToYRow_Any_SSE2;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
UYVYToUVRow = UYVYToUVRow_SSE2;
|
|
UYVYToYRow = UYVYToYRow_SSE2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_UYVYTOYROW_AVX2)
|
|
if (TestCpuFlag(kCpuHasAVX2)) {
|
|
UYVYToUVRow = UYVYToUVRow_Any_AVX2;
|
|
UYVYToYRow = UYVYToYRow_Any_AVX2;
|
|
if (IS_ALIGNED(width, 32)) {
|
|
UYVYToUVRow = UYVYToUVRow_AVX2;
|
|
UYVYToYRow = UYVYToYRow_AVX2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_UYVYTOYROW_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
UYVYToYRow = UYVYToYRow_Any_NEON;
|
|
UYVYToUVRow = UYVYToUVRow_Any_NEON;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
UYVYToYRow = UYVYToYRow_NEON;
|
|
UYVYToUVRow = UYVYToUVRow_NEON;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_UYVYTOYROW_SVE2) && defined(HAS_UYVYTOUVROW_SVE2)
|
|
if (TestCpuFlag(kCpuHasSVE2)) {
|
|
UYVYToYRow = UYVYToYRow_SVE2;
|
|
UYVYToUVRow = UYVYToUVRow_SVE2;
|
|
}
|
|
#endif
|
|
#if defined(HAS_UYVYTOYROW_LSX)
|
|
if (TestCpuFlag(kCpuHasLSX)) {
|
|
UYVYToYRow = UYVYToYRow_Any_LSX;
|
|
UYVYToUVRow = UYVYToUVRow_Any_LSX;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
UYVYToYRow = UYVYToYRow_LSX;
|
|
UYVYToUVRow = UYVYToUVRow_LSX;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_UYVYTOYROW_LASX)
|
|
if (TestCpuFlag(kCpuHasLASX)) {
|
|
UYVYToYRow = UYVYToYRow_Any_LASX;
|
|
UYVYToUVRow = UYVYToUVRow_Any_LASX;
|
|
if (IS_ALIGNED(width, 32)) {
|
|
UYVYToYRow = UYVYToYRow_LASX;
|
|
UYVYToUVRow = UYVYToUVRow_LASX;
|
|
}
|
|
}
|
|
#endif
|
|
|
|
for (y = 0; y < height - 1; y += 2) {
|
|
UYVYToUVRow(src_uyvy, src_stride_uyvy, dst_u, dst_v, width);
|
|
UYVYToYRow(src_uyvy, dst_y, width);
|
|
UYVYToYRow(src_uyvy + src_stride_uyvy, dst_y + dst_stride_y, width);
|
|
src_uyvy += src_stride_uyvy * 2;
|
|
dst_y += dst_stride_y * 2;
|
|
dst_u += dst_stride_u;
|
|
dst_v += dst_stride_v;
|
|
}
|
|
if (height & 1) {
|
|
UYVYToUVRow(src_uyvy, 0, dst_u, dst_v, width);
|
|
UYVYToYRow(src_uyvy, dst_y, width);
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
// Convert AYUV to NV12.
|
|
LIBYUV_API
|
|
int AYUVToNV12(const uint8_t* src_ayuv,
|
|
int src_stride_ayuv,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_uv,
|
|
int dst_stride_uv,
|
|
int width,
|
|
int height) {
|
|
int y;
|
|
void (*AYUVToUVRow)(const uint8_t* src_ayuv, int src_stride_ayuv,
|
|
uint8_t* dst_uv, int width) = AYUVToUVRow_C;
|
|
void (*AYUVToYRow)(const uint8_t* src_ayuv, uint8_t* dst_y, int width) =
|
|
AYUVToYRow_C;
|
|
if (width <= 0 || height == 0 || height == INT_MIN) {
|
|
return -1;
|
|
}
|
|
// Negative height means invert the image.
|
|
if (height < 0) {
|
|
height = -height;
|
|
src_ayuv = src_ayuv + (ptrdiff_t)(height - 1) * src_stride_ayuv;
|
|
src_stride_ayuv = -src_stride_ayuv;
|
|
}
|
|
// place holders for future intel code
|
|
#if defined(HAS_AYUVTOYROW_SSE2)
|
|
if (TestCpuFlag(kCpuHasSSE2)) {
|
|
AYUVToUVRow = AYUVToUVRow_Any_SSE2;
|
|
AYUVToYRow = AYUVToYRow_Any_SSE2;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
AYUVToUVRow = AYUVToUVRow_SSE2;
|
|
AYUVToYRow = AYUVToYRow_SSE2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_AYUVTOYROW_AVX2)
|
|
if (TestCpuFlag(kCpuHasAVX2)) {
|
|
AYUVToUVRow = AYUVToUVRow_Any_AVX2;
|
|
AYUVToYRow = AYUVToYRow_Any_AVX2;
|
|
if (IS_ALIGNED(width, 32)) {
|
|
AYUVToUVRow = AYUVToUVRow_AVX2;
|
|
AYUVToYRow = AYUVToYRow_AVX2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_AYUVTOYROW_AVX512VBMI)
|
|
if (TestCpuFlag(kCpuHasAVX512VBMI | kCpuHasAVX512VL) ==
|
|
(kCpuHasAVX512VBMI | kCpuHasAVX512VL)) {
|
|
AYUVToUVRow = AYUVToUVRow_AVX512VBMI;
|
|
AYUVToYRow = AYUVToYRow_AVX512VBMI;
|
|
}
|
|
#endif
|
|
|
|
#if defined(HAS_AYUVTOYROW_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
AYUVToYRow = AYUVToYRow_Any_NEON;
|
|
AYUVToUVRow = AYUVToUVRow_Any_NEON;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
AYUVToYRow = AYUVToYRow_NEON;
|
|
AYUVToUVRow = AYUVToUVRow_NEON;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_AYUVTOUVROW_SVE2)
|
|
if (TestCpuFlag(kCpuHasSVE2)) {
|
|
AYUVToUVRow = AYUVToUVRow_Any_SVE2;
|
|
if (IS_ALIGNED(width, 2)) {
|
|
AYUVToUVRow = AYUVToUVRow_SVE2;
|
|
}
|
|
}
|
|
#endif
|
|
|
|
for (y = 0; y < height - 1; y += 2) {
|
|
AYUVToUVRow(src_ayuv, src_stride_ayuv, dst_uv, width);
|
|
AYUVToYRow(src_ayuv, dst_y, width);
|
|
AYUVToYRow(src_ayuv + src_stride_ayuv, dst_y + dst_stride_y, width);
|
|
src_ayuv += src_stride_ayuv * 2;
|
|
dst_y += dst_stride_y * 2;
|
|
dst_uv += dst_stride_uv;
|
|
}
|
|
if (height & 1) {
|
|
AYUVToUVRow(src_ayuv, 0, dst_uv, width);
|
|
AYUVToYRow(src_ayuv, dst_y, width);
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
// Convert AYUV to NV21.
|
|
LIBYUV_API
|
|
int AYUVToNV21(const uint8_t* src_ayuv,
|
|
int src_stride_ayuv,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_vu,
|
|
int dst_stride_vu,
|
|
int width,
|
|
int height) {
|
|
int y;
|
|
void (*AYUVToVURow)(const uint8_t* src_ayuv, int src_stride_ayuv,
|
|
uint8_t* dst_vu, int width) = AYUVToVURow_C;
|
|
void (*AYUVToYRow)(const uint8_t* src_ayuv, uint8_t* dst_y, int width) =
|
|
AYUVToYRow_C;
|
|
if (width <= 0 || height == 0 || height == INT_MIN) {
|
|
return -1;
|
|
}
|
|
// Negative height means invert the image.
|
|
if (height < 0) {
|
|
height = -height;
|
|
src_ayuv = src_ayuv + (ptrdiff_t)(height - 1) * src_stride_ayuv;
|
|
src_stride_ayuv = -src_stride_ayuv;
|
|
}
|
|
// place holders for future intel code
|
|
#if defined(HAS_AYUVTOYROW_SSE2)
|
|
if (TestCpuFlag(kCpuHasSSE2)) {
|
|
AYUVToVURow = AYUVToVURow_Any_SSE2;
|
|
AYUVToYRow = AYUVToYRow_Any_SSE2;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
AYUVToVURow = AYUVToVURow_SSE2;
|
|
AYUVToYRow = AYUVToYRow_SSE2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_AYUVTOYROW_AVX2)
|
|
if (TestCpuFlag(kCpuHasAVX2)) {
|
|
AYUVToVURow = AYUVToVURow_Any_AVX2;
|
|
AYUVToYRow = AYUVToYRow_Any_AVX2;
|
|
if (IS_ALIGNED(width, 32)) {
|
|
AYUVToVURow = AYUVToVURow_AVX2;
|
|
AYUVToYRow = AYUVToYRow_AVX2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_AYUVTOYROW_AVX512VBMI)
|
|
if (TestCpuFlag(kCpuHasAVX512VBMI | kCpuHasAVX512VL) ==
|
|
(kCpuHasAVX512VBMI | kCpuHasAVX512VL)) {
|
|
AYUVToVURow = AYUVToVURow_AVX512VBMI;
|
|
AYUVToYRow = AYUVToYRow_AVX512VBMI;
|
|
}
|
|
#endif
|
|
|
|
#if defined(HAS_AYUVTOYROW_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
AYUVToYRow = AYUVToYRow_Any_NEON;
|
|
AYUVToVURow = AYUVToVURow_Any_NEON;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
AYUVToYRow = AYUVToYRow_NEON;
|
|
AYUVToVURow = AYUVToVURow_NEON;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_AYUVTOVUROW_SVE2)
|
|
if (TestCpuFlag(kCpuHasSVE2)) {
|
|
AYUVToVURow = AYUVToVURow_Any_SVE2;
|
|
if (IS_ALIGNED(width, 2)) {
|
|
AYUVToVURow = AYUVToVURow_SVE2;
|
|
}
|
|
}
|
|
#endif
|
|
|
|
for (y = 0; y < height - 1; y += 2) {
|
|
AYUVToVURow(src_ayuv, src_stride_ayuv, dst_vu, width);
|
|
AYUVToYRow(src_ayuv, dst_y, width);
|
|
AYUVToYRow(src_ayuv + src_stride_ayuv, dst_y + dst_stride_y, width);
|
|
src_ayuv += src_stride_ayuv * 2;
|
|
dst_y += dst_stride_y * 2;
|
|
dst_vu += dst_stride_vu;
|
|
}
|
|
if (height & 1) {
|
|
AYUVToVURow(src_ayuv, 0, dst_vu, width);
|
|
AYUVToYRow(src_ayuv, dst_y, width);
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
// Convert ARGB to I420.
|
|
LIBYUV_API
|
|
int ARGBToI420(const uint8_t* src_argb,
|
|
int src_stride_argb,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
return ARGBToI420Matrix(src_argb, src_stride_argb, dst_y, dst_stride_y, dst_u,
|
|
dst_stride_u, dst_v, dst_stride_v,
|
|
&kArgbI601Constants, width, height);
|
|
}
|
|
|
|
LIBYUV_API
|
|
int ARGBToI420Matrix(const uint8_t* src_argb,
|
|
int src_stride_argb,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
const struct ArgbConstants* argbconstants,
|
|
int width,
|
|
int height) {
|
|
int y;
|
|
void (*ARGBToYMatrixRow)(const uint8_t* src_argb, uint8_t* dst_y, int width,
|
|
const struct ArgbConstants* c) = ARGBToYMatrixRow_C;
|
|
void (*ARGBToUVMatrixRow)(const uint8_t* src_argb, int src_stride_argb,
|
|
uint8_t* dst_u, uint8_t* dst_v, int width,
|
|
const struct ArgbConstants* c) =
|
|
ARGBToUVMatrixRow_C;
|
|
|
|
#if defined(HAS_ARGBTOYMATRIXROW_SSSE3)
|
|
if (TestCpuFlag(kCpuHasSSSE3)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_Any_SSSE3;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_SSSE3;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOYMATRIXROW_AVX2)
|
|
if (TestCpuFlag(kCpuHasAVX2)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_Any_AVX2;
|
|
if (IS_ALIGNED(width, 32)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_AVX2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOYMATRIXROW_AVX512BW)
|
|
if (TestCpuFlag(kCpuHasAVX512BW)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_Any_AVX512BW;
|
|
if (IS_ALIGNED(width, 64)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_AVX512BW;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOYMATRIXROW_AVX512VNNI)
|
|
if (TestCpuFlag(kCpuHasAVX512VNNI)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_Any_AVX512VNNI;
|
|
if (IS_ALIGNED(width, 64)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_AVX512VNNI;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOYMATRIXROW_AVX512VNNIVBMI)
|
|
if (TestCpuFlag(kCpuHasAVX512VNNI) && TestCpuFlag(kCpuHasAVX512VBMI)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_Any_AVX512VNNIVBMI;
|
|
if (IS_ALIGNED(width, 64)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_AVX512VNNIVBMI;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOYMATRIXROW_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_Any_NEON;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_NEON;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOYMATRIXROW_NEON_DOTPROD)
|
|
if (TestCpuFlag(kCpuHasNeonDotProd)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_Any_NEON_DotProd;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_NEON_DotProd;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOYMATRIXROW_SVE2)
|
|
if (TestCpuFlag(kCpuHasSVE2)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_SVE2;
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOYMATRIXROW_SME)
|
|
if (TestCpuFlag(kCpuHasSME)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_SME;
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOYMATRIXROW_LSX)
|
|
if (TestCpuFlag(kCpuHasLSX)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_Any_LSX;
|
|
if (IS_ALIGNED(width, 8)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_LSX;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOYMATRIXROW_LASX)
|
|
if (TestCpuFlag(kCpuHasLASX)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_Any_LASX;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_LASX;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOYMATRIXROW_RVV)
|
|
if (TestCpuFlag(kCpuHasRVV)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_RVV;
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOYMATRIXROW_WASMSIMD)
|
|
if (TestCpuFlag(kCpuHasWASMSIMD)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_Any_WASMSIMD;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_WASMSIMD;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOYMATRIXROW_HVX)
|
|
if (TestCpuFlag(kCpuHasHVX)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_HVX;
|
|
}
|
|
#endif
|
|
|
|
#if defined(HAS_ARGBTOUVMATRIXROW_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_NEON;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_NEON;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOUVMATRIXROW_NEON_I8MM)
|
|
if (TestCpuFlag(kCpuHasNeonI8MM)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_NEON_I8MM;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_NEON_I8MM;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOUVMATRIXROW_SVE2)
|
|
if (TestCpuFlag(kCpuHasSVE2)) {
|
|
if (IS_ALIGNED(width, 2)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_SVE2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOUVMATRIXROW_SME)
|
|
if (TestCpuFlag(kCpuHasSME)) {
|
|
if (IS_ALIGNED(width, 2)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_SME;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOUVMATRIXROW_SSSE3)
|
|
if (TestCpuFlag(kCpuHasSSSE3)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_SSSE3;
|
|
if (IS_ALIGNED(width, 8)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_SSSE3;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOUVMATRIXROW_AVX2)
|
|
if (TestCpuFlag(kCpuHasAVX2)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_AVX2;
|
|
if (IS_ALIGNED(width, 32)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_AVX2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOUVMATRIXROW_AVX512BW)
|
|
if (TestCpuFlag(kCpuHasAVX512BW)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_AVX512BW;
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOUVMATRIXROW_AVX512VNNI)
|
|
if (TestCpuFlag(kCpuHasAVX512VNNI)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_AVX512VNNI;
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOUVMATRIXROW_AVX512VNNIVBMI)
|
|
if (TestCpuFlag(kCpuHasAVX512VNNI) && TestCpuFlag(kCpuHasAVX512VBMI)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_AVX512VNNIVBMI;
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOUVMATRIXROW_RVV)
|
|
if (TestCpuFlag(kCpuHasRVV)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_RVV;
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOUVMATRIXROW_WASMSIMD)
|
|
if (TestCpuFlag(kCpuHasWASMSIMD)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_WASMSIMD;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_WASMSIMD;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOUVMATRIXROW_HVX)
|
|
if (TestCpuFlag(kCpuHasHVX)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_HVX;
|
|
}
|
|
#endif
|
|
if (!src_argb || !dst_y || !dst_u || !dst_v || !argbconstants || width <= 0 ||
|
|
height == 0 || height == INT_MIN) {
|
|
return -1;
|
|
}
|
|
// Negative height means invert the image.
|
|
if (height < 0) {
|
|
height = -height;
|
|
src_argb = src_argb + (ptrdiff_t)(height - 1) * src_stride_argb;
|
|
src_stride_argb = -src_stride_argb;
|
|
}
|
|
|
|
for (y = 0; y < height - 1; y += 2) {
|
|
ARGBToUVMatrixRow(src_argb, src_stride_argb, dst_u, dst_v, width,
|
|
argbconstants);
|
|
ARGBToYMatrixRow(src_argb, dst_y, width, argbconstants);
|
|
ARGBToYMatrixRow(src_argb + src_stride_argb, dst_y + dst_stride_y, width,
|
|
argbconstants);
|
|
src_argb += src_stride_argb * 2;
|
|
dst_y += dst_stride_y * 2;
|
|
dst_u += dst_stride_u;
|
|
dst_v += dst_stride_v;
|
|
}
|
|
if (height & 1) {
|
|
ARGBToUVMatrixRow(src_argb, 0, dst_u, dst_v, width, argbconstants);
|
|
ARGBToYMatrixRow(src_argb, dst_y, width, argbconstants);
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
#ifdef USE_EXTRACTALPHA
|
|
// Convert ARGB to I420 with Alpha
|
|
// The following version calls ARGBExtractAlpha on the full image.
|
|
LIBYUV_API
|
|
int ARGBToI420Alpha(const uint8_t* src_argb,
|
|
int src_stride_argb,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
uint8_t* dst_a,
|
|
int dst_stride_a,
|
|
int width,
|
|
int height) {
|
|
int r = ARGBToI420(src_argb, src_stride_argb, dst_y, dst_stride_y, dst_u,
|
|
dst_stride_u, dst_v, dst_stride_v, width, height);
|
|
if (r == 0) {
|
|
r = ARGBExtractAlpha(src_argb, src_stride_argb, dst_a, dst_stride_a, width,
|
|
height);
|
|
}
|
|
return r;
|
|
}
|
|
#else // USE_EXTRACTALPHA
|
|
// Convert ARGB to I420 with Alpha
|
|
LIBYUV_API
|
|
int ARGBToI420Alpha(const uint8_t* src_argb,
|
|
int src_stride_argb,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
uint8_t* dst_a,
|
|
int dst_stride_a,
|
|
int width,
|
|
int height) {
|
|
int y;
|
|
void (*ARGBToYMatrixRow)(const uint8_t* src_argb, uint8_t* dst_y, int width,
|
|
const struct ArgbConstants* c) = ARGBToYMatrixRow_C;
|
|
void (*ARGBToUVMatrixRow)(const uint8_t* src_argb, int src_stride_argb,
|
|
uint8_t* dst_u, uint8_t* dst_v, int width,
|
|
const struct ArgbConstants* c) =
|
|
ARGBToUVMatrixRow_C;
|
|
void (*ARGBExtractAlphaRow)(const uint8_t* src_argb, uint8_t* dst_a,
|
|
int width) = ARGBExtractAlphaRow_C;
|
|
if (!src_argb || !dst_y || !dst_u || !dst_v || !dst_a || width <= 0 ||
|
|
height == 0 || height == INT_MIN) {
|
|
return -1;
|
|
}
|
|
// Negative height means invert the image.
|
|
if (height < 0) {
|
|
height = -height;
|
|
src_argb = src_argb + (ptrdiff_t)(height - 1) * src_stride_argb;
|
|
src_stride_argb = -src_stride_argb;
|
|
}
|
|
|
|
#if defined(HAS_ARGBTOYMATRIXROW_SSSE3)
|
|
if (TestCpuFlag(kCpuHasSSSE3)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_Any_SSSE3;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_SSSE3;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOYMATRIXROW_AVX2)
|
|
if (TestCpuFlag(kCpuHasAVX2)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_Any_AVX2;
|
|
if (IS_ALIGNED(width, 32)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_AVX2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOYMATRIXROW_AVX512BW)
|
|
if (TestCpuFlag(kCpuHasAVX512BW)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_Any_AVX512BW;
|
|
if (IS_ALIGNED(width, 64)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_AVX512BW;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOYMATRIXROW_AVX512VNNI)
|
|
if (TestCpuFlag(kCpuHasAVX512VNNI)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_Any_AVX512VNNI;
|
|
if (IS_ALIGNED(width, 64)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_AVX512VNNI;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOYMATRIXROW_AVX512VNNIVBMI)
|
|
if (TestCpuFlag(kCpuHasAVX512VNNI) && TestCpuFlag(kCpuHasAVX512VBMI)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_Any_AVX512VNNIVBMI;
|
|
if (IS_ALIGNED(width, 64)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_AVX512VNNIVBMI;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOYMATRIXROW_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_Any_NEON;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_NEON;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOYMATRIXROW_NEON_DOTPROD)
|
|
if (TestCpuFlag(kCpuHasNeonDotProd)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_Any_NEON_DotProd;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_NEON_DotProd;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOYMATRIXROW_SVE2)
|
|
if (TestCpuFlag(kCpuHasSVE2)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_SVE2;
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOYMATRIXROW_SME)
|
|
if (TestCpuFlag(kCpuHasSME)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_SME;
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOYMATRIXROW_LSX)
|
|
if (TestCpuFlag(kCpuHasLSX)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_Any_LSX;
|
|
if (IS_ALIGNED(width, 8)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_LSX;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOYMATRIXROW_LASX)
|
|
if (TestCpuFlag(kCpuHasLASX)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_Any_LASX;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_LASX;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOYMATRIXROW_RVV)
|
|
if (TestCpuFlag(kCpuHasRVV)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_RVV;
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOYMATRIXROW_WASMSIMD)
|
|
if (TestCpuFlag(kCpuHasWASMSIMD)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_Any_WASMSIMD;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_WASMSIMD;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOYMATRIXROW_HVX)
|
|
if (TestCpuFlag(kCpuHasHVX)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_HVX;
|
|
}
|
|
#endif
|
|
|
|
#if defined(HAS_ARGBTOUVMATRIXROW_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_NEON;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_NEON;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOUVMATRIXROW_NEON_I8MM)
|
|
if (TestCpuFlag(kCpuHasNeonI8MM)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_NEON_I8MM;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_NEON_I8MM;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOUVMATRIXROW_SVE2)
|
|
if (TestCpuFlag(kCpuHasSVE2)) {
|
|
if (IS_ALIGNED(width, 2)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_SVE2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOUVMATRIXROW_SME)
|
|
if (TestCpuFlag(kCpuHasSME)) {
|
|
if (IS_ALIGNED(width, 2)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_SME;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOUVMATRIXROW_SSSE3)
|
|
if (TestCpuFlag(kCpuHasSSSE3)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_SSSE3;
|
|
if (IS_ALIGNED(width, 8)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_SSSE3;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOUVMATRIXROW_AVX2)
|
|
if (TestCpuFlag(kCpuHasAVX2)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_AVX2;
|
|
if (IS_ALIGNED(width, 32)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_AVX2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOUVMATRIXROW_AVX512BW)
|
|
if (TestCpuFlag(kCpuHasAVX512BW)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_AVX512BW;
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOUVMATRIXROW_AVX512VNNI)
|
|
if (TestCpuFlag(kCpuHasAVX512VNNI)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_AVX512VNNI;
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOUVMATRIXROW_AVX512VNNIVBMI)
|
|
if (TestCpuFlag(kCpuHasAVX512VNNI) && TestCpuFlag(kCpuHasAVX512VBMI)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_AVX512VNNIVBMI;
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOUVMATRIXROW_RVV)
|
|
if (TestCpuFlag(kCpuHasRVV)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_RVV;
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOUVMATRIXROW_WASMSIMD)
|
|
if (TestCpuFlag(kCpuHasWASMSIMD)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_WASMSIMD;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_WASMSIMD;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOUVMATRIXROW_HVX)
|
|
if (TestCpuFlag(kCpuHasHVX)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_HVX;
|
|
}
|
|
#endif
|
|
|
|
#if defined(HAS_ARGBEXTRACTALPHAROW_SSE2)
|
|
if (TestCpuFlag(kCpuHasSSE2)) {
|
|
ARGBExtractAlphaRow = IS_ALIGNED(width, 8) ? ARGBExtractAlphaRow_SSE2
|
|
: ARGBExtractAlphaRow_Any_SSE2;
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBEXTRACTALPHAROW_AVX2)
|
|
if (TestCpuFlag(kCpuHasAVX2)) {
|
|
ARGBExtractAlphaRow = IS_ALIGNED(width, 32) ? ARGBExtractAlphaRow_AVX2
|
|
: ARGBExtractAlphaRow_Any_AVX2;
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBEXTRACTALPHAROW_AVX512BW)
|
|
if (TestCpuFlag(kCpuHasAVX512BW)) {
|
|
ARGBExtractAlphaRow = ARGBExtractAlphaRow_AVX512BW;
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBEXTRACTALPHAROW_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
ARGBExtractAlphaRow = IS_ALIGNED(width, 16) ? ARGBExtractAlphaRow_NEON
|
|
: ARGBExtractAlphaRow_Any_NEON;
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBEXTRACTALPHAROW_LSX)
|
|
if (TestCpuFlag(kCpuHasLSX)) {
|
|
ARGBExtractAlphaRow = IS_ALIGNED(width, 8) ? ARGBExtractAlphaRow_LSX
|
|
: ARGBExtractAlphaRow_Any_LSX;
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBEXTRACTALPHAROW_LASX)
|
|
if (TestCpuFlag(kCpuHasLASX)) {
|
|
ARGBExtractAlphaRow = IS_ALIGNED(width, 16) ? ARGBExtractAlphaRow_LASX
|
|
: ARGBExtractAlphaRow_Any_LASX;
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBEXTRACTALPHAROW_WASMSIMD)
|
|
if (TestCpuFlag(kCpuHasWASMSIMD)) {
|
|
ARGBExtractAlphaRow = IS_ALIGNED(width, 16)
|
|
? ARGBExtractAlphaRow_WASMSIMD
|
|
: ARGBExtractAlphaRow_Any_WASMSIMD;
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBEXTRACTALPHAROW_RVV)
|
|
if (TestCpuFlag(kCpuHasRVV)) {
|
|
ARGBExtractAlphaRow = ARGBExtractAlphaRow_RVV;
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBEXTRACTALPHAROW_HVX)
|
|
if (TestCpuFlag(kCpuHasHVX)) {
|
|
ARGBExtractAlphaRow = ARGBExtractAlphaRow_HVX;
|
|
}
|
|
#endif
|
|
|
|
for (y = 0; y < height - 1; y += 2) {
|
|
ARGBToUVMatrixRow(src_argb, src_stride_argb, dst_u, dst_v, width,
|
|
&kArgbI601Constants);
|
|
ARGBToYMatrixRow(src_argb, dst_y, width, &kArgbI601Constants);
|
|
ARGBToYMatrixRow(src_argb + src_stride_argb, dst_y + dst_stride_y, width,
|
|
&kArgbI601Constants);
|
|
ARGBExtractAlphaRow(src_argb, dst_a, width);
|
|
ARGBExtractAlphaRow(src_argb + src_stride_argb, dst_a + dst_stride_a,
|
|
width);
|
|
src_argb += src_stride_argb * 2;
|
|
dst_y += dst_stride_y * 2;
|
|
dst_u += dst_stride_u;
|
|
dst_v += dst_stride_v;
|
|
dst_a += dst_stride_a * 2;
|
|
}
|
|
if (height & 1) {
|
|
ARGBToUVMatrixRow(src_argb, 0, dst_u, dst_v, width, &kArgbI601Constants);
|
|
ARGBToYMatrixRow(src_argb, dst_y, width, &kArgbI601Constants);
|
|
ARGBExtractAlphaRow(src_argb, dst_a, width);
|
|
}
|
|
return 0;
|
|
}
|
|
#endif // USE_EXTRACTALPHA
|
|
|
|
// Convert BGRA to I420.
|
|
LIBYUV_API
|
|
int BGRAToI420(const uint8_t* src_bgra,
|
|
int src_stride_bgra,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
return ARGBToI420Matrix(src_bgra, src_stride_bgra, dst_y, dst_stride_y, dst_u,
|
|
dst_stride_u, dst_v, dst_stride_v,
|
|
&kBgraI601Constants, width, height);
|
|
}
|
|
|
|
// Convert BGRA to I422.
|
|
LIBYUV_API
|
|
int BGRAToI422(const uint8_t* src_bgra,
|
|
int src_stride_bgra,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
return ARGBToI422Matrix(src_bgra, src_stride_bgra, dst_y, dst_stride_y, dst_u,
|
|
dst_stride_u, dst_v, dst_stride_v,
|
|
&kBgraI601Constants, width, height);
|
|
}
|
|
|
|
// Convert ABGR to I422.
|
|
LIBYUV_API
|
|
int ABGRToI422(const uint8_t* src_abgr,
|
|
int src_stride_abgr,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
return ARGBToI422Matrix(src_abgr, src_stride_abgr, dst_y, dst_stride_y, dst_u,
|
|
dst_stride_u, dst_v, dst_stride_v,
|
|
&kAbgrI601Constants, width, height);
|
|
}
|
|
|
|
// Convert RGBA to I422.
|
|
LIBYUV_API
|
|
int RGBAToI422(const uint8_t* src_rgba,
|
|
int src_stride_rgba,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
return ARGBToI422Matrix(src_rgba, src_stride_rgba, dst_y, dst_stride_y, dst_u,
|
|
dst_stride_u, dst_v, dst_stride_v,
|
|
&kRgbaI601Constants, width, height);
|
|
}
|
|
|
|
// Convert ABGR to I420.
|
|
LIBYUV_API
|
|
int ABGRToI420(const uint8_t* src_abgr,
|
|
int src_stride_abgr,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
return ARGBToI420Matrix(src_abgr, src_stride_abgr, dst_y, dst_stride_y, dst_u,
|
|
dst_stride_u, dst_v, dst_stride_v,
|
|
&kAbgrI601Constants, width, height);
|
|
}
|
|
|
|
// Convert RGBA to I420.
|
|
LIBYUV_API
|
|
int RGBAToI420(const uint8_t* src_rgba,
|
|
int src_stride_rgba,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
return ARGBToI420Matrix(src_rgba, src_stride_rgba, dst_y, dst_stride_y, dst_u,
|
|
dst_stride_u, dst_v, dst_stride_v,
|
|
&kRgbaI601Constants, width, height);
|
|
}
|
|
|
|
static int AR30ToI420Matrix(const uint8_t* src_ar30,
|
|
int src_stride_ar30,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
const struct ArgbConstants* argbconstants,
|
|
int width,
|
|
int height) {
|
|
int y;
|
|
void (*AR30ToARGBRow)(const uint8_t* src_ar30, uint8_t* dst_argb, int width) =
|
|
AR30ToARGBRow_C;
|
|
void (*ARGBToYMatrixRow)(const uint8_t* src_argb, uint8_t* dst_y, int width,
|
|
const struct ArgbConstants* c) = ARGBToYMatrixRow_C;
|
|
void (*ARGBToUVMatrixRow)(const uint8_t* src_argb, int src_stride_argb,
|
|
uint8_t* dst_u, uint8_t* dst_v, int width,
|
|
const struct ArgbConstants* c) =
|
|
ARGBToUVMatrixRow_C;
|
|
if (!src_ar30 || !dst_y || !dst_u || !dst_v || !argbconstants || width <= 0 ||
|
|
height == 0 || height == INT_MIN) {
|
|
return -1;
|
|
}
|
|
// Negative height means invert the image.
|
|
if (height < 0) {
|
|
height = -height;
|
|
src_ar30 = src_ar30 + (ptrdiff_t)(height - 1) * src_stride_ar30;
|
|
src_stride_ar30 = -src_stride_ar30;
|
|
}
|
|
#if defined(HAS_AR30TOARGBROW_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
AR30ToARGBRow = AR30ToARGBRow_Any_NEON;
|
|
if (IS_ALIGNED(width, 8)) {
|
|
AR30ToARGBRow = AR30ToARGBRow_NEON;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_AR30TOARGBROW_AVX512BW)
|
|
if (TestCpuFlag(kCpuHasAVX512BW)) {
|
|
AR30ToARGBRow = AR30ToARGBRow_AVX512BW;
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOYMATRIXROW_SSSE3)
|
|
if (TestCpuFlag(kCpuHasSSSE3)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_Any_SSSE3;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_SSSE3;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOYMATRIXROW_AVX2)
|
|
if (TestCpuFlag(kCpuHasAVX2)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_Any_AVX2;
|
|
if (IS_ALIGNED(width, 32)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_AVX2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOYMATRIXROW_AVX512BW)
|
|
if (TestCpuFlag(kCpuHasAVX512BW)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_Any_AVX512BW;
|
|
if (IS_ALIGNED(width, 64)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_AVX512BW;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOYMATRIXROW_AVX512VNNI)
|
|
if (TestCpuFlag(kCpuHasAVX512VNNI)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_Any_AVX512VNNI;
|
|
if (IS_ALIGNED(width, 64)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_AVX512VNNI;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOYMATRIXROW_AVX512VNNIVBMI)
|
|
if (TestCpuFlag(kCpuHasAVX512VNNI) && TestCpuFlag(kCpuHasAVX512VBMI)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_Any_AVX512VNNIVBMI;
|
|
if (IS_ALIGNED(width, 64)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_AVX512VNNIVBMI;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOYMATRIXROW_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_Any_NEON;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_NEON;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOYMATRIXROW_NEON_DOTPROD)
|
|
if (TestCpuFlag(kCpuHasNeonDotProd)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_Any_NEON_DotProd;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_NEON_DotProd;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOYMATRIXROW_SVE2)
|
|
if (TestCpuFlag(kCpuHasSVE2)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_SVE2;
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOYMATRIXROW_SME)
|
|
if (TestCpuFlag(kCpuHasSME)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_SME;
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOYMATRIXROW_LSX)
|
|
if (TestCpuFlag(kCpuHasLSX)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_Any_LSX;
|
|
if (IS_ALIGNED(width, 8)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_LSX;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOYMATRIXROW_LASX)
|
|
if (TestCpuFlag(kCpuHasLASX)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_Any_LASX;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_LASX;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOYMATRIXROW_RVV)
|
|
if (TestCpuFlag(kCpuHasRVV)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_RVV;
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOYMATRIXROW_WASMSIMD)
|
|
if (TestCpuFlag(kCpuHasWASMSIMD)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_Any_WASMSIMD;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_WASMSIMD;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOYMATRIXROW_HVX)
|
|
if (TestCpuFlag(kCpuHasHVX)) {
|
|
ARGBToYMatrixRow = ARGBToYMatrixRow_HVX;
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOUVMATRIXROW_SSSE3)
|
|
if (TestCpuFlag(kCpuHasSSSE3)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_SSSE3;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_SSSE3;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOUVMATRIXROW_AVX2)
|
|
if (TestCpuFlag(kCpuHasAVX2)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_AVX2;
|
|
if (IS_ALIGNED(width, 32)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_AVX2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOUVMATRIXROW_AVX512BW)
|
|
if (TestCpuFlag(kCpuHasAVX512BW)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_AVX512BW;
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOUVMATRIXROW_AVX512VNNI)
|
|
if (TestCpuFlag(kCpuHasAVX512VNNI)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_AVX512VNNI;
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOUVMATRIXROW_AVX512VNNIVBMI)
|
|
if (TestCpuFlag(kCpuHasAVX512VNNI) && TestCpuFlag(kCpuHasAVX512VBMI)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_AVX512VNNIVBMI;
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOUVMATRIXROW_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_NEON;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_NEON;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOUVMATRIXROW_NEON_I8MM)
|
|
if (TestCpuFlag(kCpuHasNeonI8MM)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_NEON_I8MM;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_NEON_I8MM;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOUVMATRIXROW_SVE2)
|
|
if (TestCpuFlag(kCpuHasSVE2)) {
|
|
if (IS_ALIGNED(width, 2)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_SVE2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOUVMATRIXROW_SME)
|
|
if (TestCpuFlag(kCpuHasSME)) {
|
|
if (IS_ALIGNED(width, 2)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_SME;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOUVMATRIXROW_RVV)
|
|
if (TestCpuFlag(kCpuHasRVV)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_RVV;
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOUVMATRIXROW_WASMSIMD)
|
|
if (TestCpuFlag(kCpuHasWASMSIMD)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_WASMSIMD;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_WASMSIMD;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGBTOUVMATRIXROW_HVX)
|
|
if (TestCpuFlag(kCpuHasHVX)) {
|
|
ARGBToUVMatrixRow = ARGBToUVMatrixRow_HVX;
|
|
}
|
|
#endif
|
|
|
|
int row_stride = width * 4;
|
|
align_buffer_64(row, row_stride * 2);
|
|
if (!row) {
|
|
return -1;
|
|
}
|
|
for (y = 0; y < height - 1; y += 2) {
|
|
AR30ToARGBRow(src_ar30, row, width);
|
|
AR30ToARGBRow(src_ar30 + src_stride_ar30, row + row_stride, width);
|
|
ARGBToUVMatrixRow(row, row_stride, dst_u, dst_v, width, argbconstants);
|
|
ARGBToYMatrixRow(row, dst_y, width, argbconstants);
|
|
ARGBToYMatrixRow(row + row_stride, dst_y + dst_stride_y, width,
|
|
argbconstants);
|
|
src_ar30 += src_stride_ar30 * 2;
|
|
dst_y += dst_stride_y * 2;
|
|
dst_u += dst_stride_u;
|
|
dst_v += dst_stride_v;
|
|
}
|
|
if (height & 1) {
|
|
AR30ToARGBRow(src_ar30, row, width);
|
|
ARGBToUVMatrixRow(row, 0, dst_u, dst_v, width, argbconstants);
|
|
ARGBToYMatrixRow(row, dst_y, width, argbconstants);
|
|
}
|
|
free_aligned_buffer_64(row);
|
|
return 0;
|
|
}
|
|
|
|
// Convert AR30 to I420.
|
|
LIBYUV_API
|
|
int AR30ToI420(const uint8_t* src_ar30,
|
|
int src_stride_ar30,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
return AR30ToI420Matrix(src_ar30, src_stride_ar30, dst_y, dst_stride_y, dst_u,
|
|
dst_stride_u, dst_v, dst_stride_v,
|
|
&kArgbI601Constants, width, height);
|
|
}
|
|
|
|
// Convert AB30 to I420.
|
|
LIBYUV_API
|
|
int AB30ToI420(const uint8_t* src_ab30,
|
|
int src_stride_ab30,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
return AR30ToI420Matrix(src_ab30, src_stride_ab30, dst_y, dst_stride_y, dst_u,
|
|
dst_stride_u, dst_v, dst_stride_v,
|
|
&kAbgrI601Constants, width, height);
|
|
}
|
|
|
|
// Enabled if 1 pass is available
|
|
#if (defined(HAS_RGB24TOYROW_NEON) || defined(HAS_RGB24TOYROW_LSX) || \
|
|
defined(HAS_RGB24TOYROW_RVV))
|
|
#define HAS_RGB24TOYROW
|
|
#endif
|
|
|
|
// Convert RGB24 to I420.
|
|
LIBYUV_API
|
|
int RGB24ToI420(const uint8_t* src_rgb24,
|
|
int src_stride_rgb24,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
int y;
|
|
void (*RGBToUVMatrixRow)(const uint8_t* src_rgb, int src_stride_rgb,
|
|
uint8_t* dst_u, uint8_t* dst_v, int width,
|
|
const struct ArgbConstants* c) = RGBToUVMatrixRow_C;
|
|
void (*RGBToYMatrixRow)(const uint8_t* src_rgb, uint8_t* dst_y, int width,
|
|
const struct ArgbConstants* c) = RGBToYMatrixRow_C;
|
|
|
|
#if defined(HAS_RGBTOYMATRIXROW_SSSE3)
|
|
if (TestCpuFlag(kCpuHasSSSE3)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_SSSE3;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_SSSE3;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_AVX2)
|
|
if (TestCpuFlag(kCpuHasAVX2)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_AVX2;
|
|
if (IS_ALIGNED(width, 32)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_AVX2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_AVX512BW)
|
|
if (TestCpuFlag(kCpuHasAVX512BW)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_AVX512BW;
|
|
if (IS_ALIGNED(width, 64)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_AVX512BW;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOUVMATRIXROW_SSSE3)
|
|
if (TestCpuFlag(kCpuHasSSSE3)) {
|
|
RGBToUVMatrixRow = RGBToUVMatrixRow_Any_SSSE3;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToUVMatrixRow = RGBToUVMatrixRow_SSSE3;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOUVMATRIXROW_AVX2)
|
|
if (TestCpuFlag(kCpuHasAVX2)) {
|
|
RGBToUVMatrixRow = RGBToUVMatrixRow_Any_AVX2;
|
|
if (IS_ALIGNED(width, 32)) {
|
|
RGBToUVMatrixRow = RGBToUVMatrixRow_AVX2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOUVMATRIXROW_AVX512BW)
|
|
if (TestCpuFlag(kCpuHasAVX512BW)) {
|
|
RGBToUVMatrixRow = RGBToUVMatrixRow_AVX512BW;
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOUVMATRIXROW_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
RGBToUVMatrixRow = RGBToUVMatrixRow_Any_NEON;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToUVMatrixRow = RGBToUVMatrixRow_NEON;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_NEON;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_NEON;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_NEON_DOTPROD)
|
|
if (TestCpuFlag(kCpuHasNeonDotProd)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_NEON_DotProd;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_NEON_DotProd;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_LSX)
|
|
if (TestCpuFlag(kCpuHasLSX)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_LSX;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_LSX;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_LASX)
|
|
if (TestCpuFlag(kCpuHasLASX)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_LASX;
|
|
if (IS_ALIGNED(width, 32)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_LASX;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOUVMATRIXROW_RVV)
|
|
if (TestCpuFlag(kCpuHasRVV)) {
|
|
RGBToUVMatrixRow = RGBToUVMatrixRow_RVV;
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_RVV)
|
|
if (TestCpuFlag(kCpuHasRVV)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_RVV;
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOUVMATRIXROW_WASMSIMD)
|
|
if (TestCpuFlag(kCpuHasWASMSIMD)) {
|
|
RGBToUVMatrixRow = RGBToUVMatrixRow_Any_WASMSIMD;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToUVMatrixRow = RGBToUVMatrixRow_WASMSIMD;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_WASMSIMD)
|
|
if (TestCpuFlag(kCpuHasWASMSIMD)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_WASMSIMD;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_WASMSIMD;
|
|
}
|
|
}
|
|
#endif
|
|
|
|
if (!src_rgb24 || !dst_y || !dst_u || !dst_v || width <= 0 || height == 0 ||
|
|
height == INT_MIN) {
|
|
return -1;
|
|
}
|
|
// Negative height means invert the image.
|
|
if (height < 0) {
|
|
height = -height;
|
|
src_rgb24 = src_rgb24 + (ptrdiff_t)(height - 1) * src_stride_rgb24;
|
|
src_stride_rgb24 = -src_stride_rgb24;
|
|
}
|
|
|
|
for (y = 0; y < height - 1; y += 2) {
|
|
RGBToUVMatrixRow(src_rgb24, src_stride_rgb24, dst_u, dst_v, width,
|
|
&kArgbI601Constants);
|
|
RGBToYMatrixRow(src_rgb24, dst_y, width, &kArgbI601Constants);
|
|
RGBToYMatrixRow(src_rgb24 + src_stride_rgb24, dst_y + dst_stride_y, width,
|
|
&kArgbI601Constants);
|
|
src_rgb24 += src_stride_rgb24 * 2;
|
|
dst_y += dst_stride_y * 2;
|
|
dst_u += dst_stride_u;
|
|
dst_v += dst_stride_v;
|
|
}
|
|
if (height & 1) {
|
|
RGBToYMatrixRow(src_rgb24, dst_y, width, &kArgbI601Constants);
|
|
RGBToUVMatrixRow(src_rgb24, 0, dst_u, dst_v, width, &kArgbI601Constants);
|
|
}
|
|
return 0;
|
|
}
|
|
#undef HAS_RGB24TOYROW
|
|
|
|
/// Convert RGB24 to J420.
|
|
LIBYUV_API
|
|
int RGB24ToJ420(const uint8_t* src_rgb24,
|
|
int src_stride_rgb24,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
int y;
|
|
void (*RGBToUVMatrixRow)(const uint8_t* src_rgb, int src_stride_rgb,
|
|
uint8_t* dst_u, uint8_t* dst_v, int width,
|
|
const struct ArgbConstants* c) = RGBToUVMatrixRow_C;
|
|
void (*RGBToYMatrixRow)(const uint8_t* src_rgb, uint8_t* dst_y, int width,
|
|
const struct ArgbConstants* c) = RGBToYMatrixRow_C;
|
|
|
|
#if defined(HAS_RGBTOYMATRIXROW_SSSE3)
|
|
if (TestCpuFlag(kCpuHasSSSE3)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_SSSE3;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_SSSE3;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_AVX2)
|
|
if (TestCpuFlag(kCpuHasAVX2)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_AVX2;
|
|
if (IS_ALIGNED(width, 32)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_AVX2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_AVX512BW)
|
|
if (TestCpuFlag(kCpuHasAVX512BW)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_AVX512BW;
|
|
if (IS_ALIGNED(width, 64)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_AVX512BW;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOUVMATRIXROW_SSSE3)
|
|
if (TestCpuFlag(kCpuHasSSSE3)) {
|
|
RGBToUVMatrixRow = RGBToUVMatrixRow_Any_SSSE3;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToUVMatrixRow = RGBToUVMatrixRow_SSSE3;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOUVMATRIXROW_AVX2)
|
|
if (TestCpuFlag(kCpuHasAVX2)) {
|
|
RGBToUVMatrixRow = RGBToUVMatrixRow_Any_AVX2;
|
|
if (IS_ALIGNED(width, 32)) {
|
|
RGBToUVMatrixRow = RGBToUVMatrixRow_AVX2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOUVMATRIXROW_AVX512BW)
|
|
if (TestCpuFlag(kCpuHasAVX512BW)) {
|
|
RGBToUVMatrixRow = RGBToUVMatrixRow_AVX512BW;
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOUVMATRIXROW_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
RGBToUVMatrixRow = RGBToUVMatrixRow_Any_NEON;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToUVMatrixRow = RGBToUVMatrixRow_NEON;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_NEON;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_NEON;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_NEON_DOTPROD)
|
|
if (TestCpuFlag(kCpuHasNeonDotProd)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_NEON_DotProd;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_NEON_DotProd;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_LSX)
|
|
if (TestCpuFlag(kCpuHasLSX)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_LSX;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_LSX;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_LASX)
|
|
if (TestCpuFlag(kCpuHasLASX)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_LASX;
|
|
if (IS_ALIGNED(width, 32)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_LASX;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOUVMATRIXROW_RVV)
|
|
if (TestCpuFlag(kCpuHasRVV)) {
|
|
RGBToUVMatrixRow = RGBToUVMatrixRow_RVV;
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_RVV)
|
|
if (TestCpuFlag(kCpuHasRVV)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_RVV;
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOUVMATRIXROW_WASMSIMD)
|
|
if (TestCpuFlag(kCpuHasWASMSIMD)) {
|
|
RGBToUVMatrixRow = RGBToUVMatrixRow_Any_WASMSIMD;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToUVMatrixRow = RGBToUVMatrixRow_WASMSIMD;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_WASMSIMD)
|
|
if (TestCpuFlag(kCpuHasWASMSIMD)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_WASMSIMD;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_WASMSIMD;
|
|
}
|
|
}
|
|
#endif
|
|
|
|
if (!src_rgb24 || !dst_y || !dst_u || !dst_v || width <= 0 || height == 0 ||
|
|
height == INT_MIN) {
|
|
return -1;
|
|
}
|
|
// Negative height means invert the image.
|
|
if (height < 0) {
|
|
height = -height;
|
|
src_rgb24 = src_rgb24 + (ptrdiff_t)(height - 1) * src_stride_rgb24;
|
|
src_stride_rgb24 = -src_stride_rgb24;
|
|
}
|
|
|
|
for (y = 0; y < height - 1; y += 2) {
|
|
RGBToUVMatrixRow(src_rgb24, src_stride_rgb24, dst_u, dst_v, width,
|
|
&kArgbJPEGConstants);
|
|
RGBToYMatrixRow(src_rgb24, dst_y, width, &kArgbJPEGConstants);
|
|
RGBToYMatrixRow(src_rgb24 + src_stride_rgb24, dst_y + dst_stride_y, width,
|
|
&kArgbJPEGConstants);
|
|
src_rgb24 += src_stride_rgb24 * 2;
|
|
dst_y += dst_stride_y * 2;
|
|
dst_u += dst_stride_u;
|
|
dst_v += dst_stride_v;
|
|
}
|
|
if (height & 1) {
|
|
RGBToYMatrixRow(src_rgb24, dst_y, width, &kArgbJPEGConstants);
|
|
RGBToUVMatrixRow(src_rgb24, 0, dst_u, dst_v, width, &kArgbJPEGConstants);
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
// Enabled if 1 pass is available
|
|
#if (defined(HAS_RAWTOYROW_NEON) || defined(HAS_RAWTOYROW_LSX) || \
|
|
defined(HAS_RAWTOYROW_RVV))
|
|
#define HAS_RAWTOYROW
|
|
#endif
|
|
|
|
// Convert RAW to I420.
|
|
LIBYUV_API
|
|
int RAWToI420(const uint8_t* src_raw,
|
|
int src_stride_raw,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
int y;
|
|
void (*RGBToUVMatrixRow)(const uint8_t* src_rgb, int src_stride_rgb,
|
|
uint8_t* dst_u, uint8_t* dst_v, int width,
|
|
const struct ArgbConstants* c) = RGBToUVMatrixRow_C;
|
|
void (*RGBToYMatrixRow)(const uint8_t* src_rgb, uint8_t* dst_y, int width,
|
|
const struct ArgbConstants* c) = RGBToYMatrixRow_C;
|
|
|
|
#if defined(HAS_RGBTOYMATRIXROW_SSSE3)
|
|
if (TestCpuFlag(kCpuHasSSSE3)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_SSSE3;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_SSSE3;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_AVX2)
|
|
if (TestCpuFlag(kCpuHasAVX2)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_AVX2;
|
|
if (IS_ALIGNED(width, 32)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_AVX2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_AVX512BW)
|
|
if (TestCpuFlag(kCpuHasAVX512BW)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_AVX512BW;
|
|
if (IS_ALIGNED(width, 64)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_AVX512BW;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOUVMATRIXROW_SSSE3)
|
|
if (TestCpuFlag(kCpuHasSSSE3)) {
|
|
RGBToUVMatrixRow = RGBToUVMatrixRow_Any_SSSE3;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToUVMatrixRow = RGBToUVMatrixRow_SSSE3;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOUVMATRIXROW_AVX2)
|
|
if (TestCpuFlag(kCpuHasAVX2)) {
|
|
RGBToUVMatrixRow = RGBToUVMatrixRow_Any_AVX2;
|
|
if (IS_ALIGNED(width, 32)) {
|
|
RGBToUVMatrixRow = RGBToUVMatrixRow_AVX2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOUVMATRIXROW_AVX512BW)
|
|
if (TestCpuFlag(kCpuHasAVX512BW)) {
|
|
RGBToUVMatrixRow = RGBToUVMatrixRow_AVX512BW;
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOUVMATRIXROW_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
RGBToUVMatrixRow = RGBToUVMatrixRow_Any_NEON;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToUVMatrixRow = RGBToUVMatrixRow_NEON;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_NEON;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_NEON;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_NEON_DOTPROD)
|
|
if (TestCpuFlag(kCpuHasNeonDotProd)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_NEON_DotProd;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_NEON_DotProd;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_LSX)
|
|
if (TestCpuFlag(kCpuHasLSX)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_LSX;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_LSX;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_LASX)
|
|
if (TestCpuFlag(kCpuHasLASX)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_LASX;
|
|
if (IS_ALIGNED(width, 32)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_LASX;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOUVMATRIXROW_WASMSIMD)
|
|
if (TestCpuFlag(kCpuHasWASMSIMD)) {
|
|
RGBToUVMatrixRow = RGBToUVMatrixRow_Any_WASMSIMD;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToUVMatrixRow = RGBToUVMatrixRow_WASMSIMD;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_WASMSIMD)
|
|
if (TestCpuFlag(kCpuHasWASMSIMD)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_WASMSIMD;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_WASMSIMD;
|
|
}
|
|
}
|
|
#endif
|
|
|
|
if (!src_raw || !dst_y || !dst_u || !dst_v || width <= 0 || height == 0 ||
|
|
height == INT_MIN) {
|
|
return -1;
|
|
}
|
|
// Negative height means invert the image.
|
|
if (height < 0) {
|
|
height = -height;
|
|
src_raw = src_raw + (ptrdiff_t)(height - 1) * src_stride_raw;
|
|
src_stride_raw = -src_stride_raw;
|
|
}
|
|
|
|
for (y = 0; y < height - 1; y += 2) {
|
|
RGBToUVMatrixRow(src_raw, src_stride_raw, dst_u, dst_v, width,
|
|
&kAbgrI601Constants);
|
|
RGBToYMatrixRow(src_raw, dst_y, width, &kAbgrI601Constants);
|
|
RGBToYMatrixRow(src_raw + src_stride_raw, dst_y + dst_stride_y, width,
|
|
&kAbgrI601Constants);
|
|
src_raw += src_stride_raw * 2;
|
|
dst_y += dst_stride_y * 2;
|
|
dst_u += dst_stride_u;
|
|
dst_v += dst_stride_v;
|
|
}
|
|
if (height & 1) {
|
|
RGBToYMatrixRow(src_raw, dst_y, width, &kAbgrI601Constants);
|
|
RGBToUVMatrixRow(src_raw, 0, dst_u, dst_v, width, &kAbgrI601Constants);
|
|
}
|
|
return 0;
|
|
}
|
|
#undef HAS_RAWTOYROW
|
|
|
|
// Convert RAW to J420.
|
|
LIBYUV_API
|
|
int RAWToJ420(const uint8_t* src_raw,
|
|
int src_stride_raw,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
int y;
|
|
void (*RGBToUVMatrixRow)(const uint8_t* src_rgb, int src_stride_rgb,
|
|
uint8_t* dst_u, uint8_t* dst_v, int width,
|
|
const struct ArgbConstants* c) = RGBToUVMatrixRow_C;
|
|
void (*RGBToYMatrixRow)(const uint8_t* src_rgb, uint8_t* dst_y, int width,
|
|
const struct ArgbConstants* c) = RGBToYMatrixRow_C;
|
|
|
|
#if defined(HAS_RGBTOYMATRIXROW_SSSE3)
|
|
if (TestCpuFlag(kCpuHasSSSE3)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_SSSE3;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_SSSE3;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_AVX2)
|
|
if (TestCpuFlag(kCpuHasAVX2)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_AVX2;
|
|
if (IS_ALIGNED(width, 32)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_AVX2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_AVX512BW)
|
|
if (TestCpuFlag(kCpuHasAVX512BW)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_AVX512BW;
|
|
if (IS_ALIGNED(width, 64)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_AVX512BW;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOUVMATRIXROW_SSSE3)
|
|
if (TestCpuFlag(kCpuHasSSSE3)) {
|
|
RGBToUVMatrixRow = RGBToUVMatrixRow_Any_SSSE3;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToUVMatrixRow = RGBToUVMatrixRow_SSSE3;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOUVMATRIXROW_AVX2)
|
|
if (TestCpuFlag(kCpuHasAVX2)) {
|
|
RGBToUVMatrixRow = RGBToUVMatrixRow_Any_AVX2;
|
|
if (IS_ALIGNED(width, 32)) {
|
|
RGBToUVMatrixRow = RGBToUVMatrixRow_AVX2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOUVMATRIXROW_AVX512BW)
|
|
if (TestCpuFlag(kCpuHasAVX512BW)) {
|
|
RGBToUVMatrixRow = RGBToUVMatrixRow_AVX512BW;
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOUVMATRIXROW_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
RGBToUVMatrixRow = RGBToUVMatrixRow_Any_NEON;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToUVMatrixRow = RGBToUVMatrixRow_NEON;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_NEON;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_NEON;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_NEON_DOTPROD)
|
|
if (TestCpuFlag(kCpuHasNeonDotProd)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_NEON_DotProd;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_NEON_DotProd;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_LSX)
|
|
if (TestCpuFlag(kCpuHasLSX)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_LSX;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_LSX;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_LASX)
|
|
if (TestCpuFlag(kCpuHasLASX)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_LASX;
|
|
if (IS_ALIGNED(width, 32)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_LASX;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOUVMATRIXROW_WASMSIMD)
|
|
if (TestCpuFlag(kCpuHasWASMSIMD)) {
|
|
RGBToUVMatrixRow = RGBToUVMatrixRow_Any_WASMSIMD;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToUVMatrixRow = RGBToUVMatrixRow_WASMSIMD;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_WASMSIMD)
|
|
if (TestCpuFlag(kCpuHasWASMSIMD)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_WASMSIMD;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_WASMSIMD;
|
|
}
|
|
}
|
|
#endif
|
|
|
|
if (!src_raw || !dst_y || !dst_u || !dst_v || width <= 0 || height == 0 ||
|
|
height == INT_MIN) {
|
|
return -1;
|
|
}
|
|
// Negative height means invert the image.
|
|
if (height < 0) {
|
|
height = -height;
|
|
src_raw = src_raw + (ptrdiff_t)(height - 1) * src_stride_raw;
|
|
src_stride_raw = -src_stride_raw;
|
|
}
|
|
|
|
for (y = 0; y < height - 1; y += 2) {
|
|
RGBToUVMatrixRow(src_raw, src_stride_raw, dst_u, dst_v, width,
|
|
&kAbgrJPEGConstants);
|
|
RGBToYMatrixRow(src_raw, dst_y, width, &kAbgrJPEGConstants);
|
|
RGBToYMatrixRow(src_raw + src_stride_raw, dst_y + dst_stride_y, width,
|
|
&kAbgrJPEGConstants);
|
|
src_raw += src_stride_raw * 2;
|
|
dst_y += dst_stride_y * 2;
|
|
dst_u += dst_stride_u;
|
|
dst_v += dst_stride_v;
|
|
}
|
|
if (height & 1) {
|
|
RGBToYMatrixRow(src_raw, dst_y, width, &kAbgrJPEGConstants);
|
|
RGBToUVMatrixRow(src_raw, 0, dst_u, dst_v, width, &kAbgrJPEGConstants);
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
// Convert RAW to I444 with matrix.
|
|
static int RAWToI444Matrix(const uint8_t* src_raw,
|
|
int src_stride_raw,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
const struct ArgbConstants* argbconstants,
|
|
int width,
|
|
int height) {
|
|
int y;
|
|
void (*RGBToUV444MatrixRow)(
|
|
const uint8_t* src_rgb, uint8_t* dst_u, uint8_t* dst_v, int width,
|
|
const struct ArgbConstants* c) = RGBToUV444MatrixRow_C;
|
|
void (*RGBToYMatrixRow)(const uint8_t* src_rgb, uint8_t* dst_y, int width,
|
|
const struct ArgbConstants* c) = RGBToYMatrixRow_C;
|
|
|
|
#if defined(HAS_RGBTOYMATRIXROW_SSSE3)
|
|
if (TestCpuFlag(kCpuHasSSSE3)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_SSSE3;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_SSSE3;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_AVX2)
|
|
if (TestCpuFlag(kCpuHasAVX2)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_AVX2;
|
|
if (IS_ALIGNED(width, 32)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_AVX2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_AVX512BW)
|
|
if (TestCpuFlag(kCpuHasAVX512BW)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_AVX512BW;
|
|
if (IS_ALIGNED(width, 64)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_AVX512BW;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_NEON;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_NEON;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_NEON_DOTPROD)
|
|
if (TestCpuFlag(kCpuHasNeonDotProd)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_NEON_DotProd;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_NEON_DotProd;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_LSX)
|
|
if (TestCpuFlag(kCpuHasLSX)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_LSX;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_LSX;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_LASX)
|
|
if (TestCpuFlag(kCpuHasLASX)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_LASX;
|
|
if (IS_ALIGNED(width, 32)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_LASX;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_RVV)
|
|
if (TestCpuFlag(kCpuHasRVV)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_RVV;
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_WASMSIMD)
|
|
if (TestCpuFlag(kCpuHasWASMSIMD)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_WASMSIMD;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_WASMSIMD;
|
|
}
|
|
}
|
|
#endif
|
|
|
|
#if defined(HAS_RGBTOUV444MATRIXROW_SSSE3)
|
|
if (TestCpuFlag(kCpuHasSSSE3)) {
|
|
RGBToUV444MatrixRow = RGBToUV444MatrixRow_Any_SSSE3;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToUV444MatrixRow = RGBToUV444MatrixRow_SSSE3;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOUV444MATRIXROW_AVX2)
|
|
if (TestCpuFlag(kCpuHasAVX2)) {
|
|
RGBToUV444MatrixRow = RGBToUV444MatrixRow_Any_AVX2;
|
|
if (IS_ALIGNED(width, 32)) {
|
|
RGBToUV444MatrixRow = RGBToUV444MatrixRow_AVX2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOUV444MATRIXROW_AVX512BW)
|
|
if (TestCpuFlag(kCpuHasAVX512BW)) {
|
|
RGBToUV444MatrixRow = RGBToUV444MatrixRow_Any_AVX512BW;
|
|
if (IS_ALIGNED(width, 64)) {
|
|
RGBToUV444MatrixRow = RGBToUV444MatrixRow_AVX512BW;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOUV444MATRIXROW_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
RGBToUV444MatrixRow = RGBToUV444MatrixRow_Any_NEON;
|
|
if (IS_ALIGNED(width, 8)) {
|
|
RGBToUV444MatrixRow = RGBToUV444MatrixRow_NEON;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOUV444MATRIXROW_RVV)
|
|
if (TestCpuFlag(kCpuHasRVV)) {
|
|
RGBToUV444MatrixRow = RGBToUV444MatrixRow_RVV;
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOUV444MATRIXROW_WASMSIMD)
|
|
if (TestCpuFlag(kCpuHasWASMSIMD)) {
|
|
RGBToUV444MatrixRow = RGBToUV444MatrixRow_Any_WASMSIMD;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToUV444MatrixRow = RGBToUV444MatrixRow_WASMSIMD;
|
|
}
|
|
}
|
|
#endif
|
|
|
|
if (!src_raw || !dst_y || !dst_u || !dst_v || !argbconstants || width <= 0 ||
|
|
height == 0 || height == INT_MIN) {
|
|
return -1;
|
|
}
|
|
if (height < 0) {
|
|
height = -height;
|
|
src_raw = src_raw + (ptrdiff_t)(height - 1) * src_stride_raw;
|
|
src_stride_raw = -src_stride_raw;
|
|
}
|
|
|
|
for (y = 0; y < height; ++y) {
|
|
RGBToYMatrixRow(src_raw, dst_y, width, argbconstants);
|
|
RGBToUV444MatrixRow(src_raw, dst_u, dst_v, width, argbconstants);
|
|
src_raw += src_stride_raw;
|
|
dst_y += dst_stride_y;
|
|
dst_u += dst_stride_u;
|
|
dst_v += dst_stride_v;
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
// Convert RAW to I444.
|
|
LIBYUV_API
|
|
int RAWToI444(const uint8_t* src_raw,
|
|
int src_stride_raw,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
return RAWToI444Matrix(src_raw, src_stride_raw, dst_y, dst_stride_y, dst_u,
|
|
dst_stride_u, dst_v, dst_stride_v, &kAbgrI601Constants,
|
|
width, height);
|
|
}
|
|
|
|
// Convert RAW to J444.
|
|
LIBYUV_API
|
|
int RAWToJ444(const uint8_t* src_raw,
|
|
int src_stride_raw,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
return RAWToI444Matrix(src_raw, src_stride_raw, dst_y, dst_stride_y, dst_u,
|
|
dst_stride_u, dst_v, dst_stride_v, &kAbgrJPEGConstants,
|
|
width, height);
|
|
}
|
|
|
|
// Convert RGB565 to I420.
|
|
LIBYUV_API
|
|
int RGB565ToI420(const uint8_t* src_rgb565,
|
|
int src_stride_rgb565,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
int y;
|
|
void (*RGB565ToUVMatrixRow)(const uint8_t* src_rgb565, int src_stride_rgb565,
|
|
uint8_t* dst_u, uint8_t* dst_v, int width,
|
|
const struct ArgbConstants* c) =
|
|
RGB565ToUVMatrixRow_C;
|
|
void (*RGB565ToYMatrixRow)(const uint8_t* src_rgb565, uint8_t* dst_y,
|
|
int width, const struct ArgbConstants* c) =
|
|
RGB565ToYMatrixRow_C;
|
|
|
|
#if defined(HAS_RGB565TOYMATRIXROW_AVX2)
|
|
if (TestCpuFlag(kCpuHasAVX2)) {
|
|
RGB565ToYMatrixRow = RGB565ToYMatrixRow_Any_AVX2;
|
|
if (IS_ALIGNED(width, 32)) {
|
|
RGB565ToYMatrixRow = RGB565ToYMatrixRow_AVX2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGB565TOUVMATRIXROW_AVX2)
|
|
if (TestCpuFlag(kCpuHasAVX2)) {
|
|
RGB565ToUVMatrixRow = RGB565ToUVMatrixRow_Any_AVX2;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGB565ToUVMatrixRow = RGB565ToUVMatrixRow_AVX2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGB565TOUVMATRIXROW_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
RGB565ToUVMatrixRow = RGB565ToUVMatrixRow_Any_NEON;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGB565ToUVMatrixRow = RGB565ToUVMatrixRow_NEON;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGB565TOYMATRIXROW_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
RGB565ToYMatrixRow = RGB565ToYMatrixRow_Any_NEON;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGB565ToYMatrixRow = RGB565ToYMatrixRow_NEON;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGB565TOUVMATRIXROW_WASMSIMD)
|
|
if (TestCpuFlag(kCpuHasWASMSIMD)) {
|
|
RGB565ToUVMatrixRow = RGB565ToUVMatrixRow_Any_WASMSIMD;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGB565ToUVMatrixRow = RGB565ToUVMatrixRow_WASMSIMD;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGB565TOYMATRIXROW_WASMSIMD)
|
|
if (TestCpuFlag(kCpuHasWASMSIMD)) {
|
|
RGB565ToYMatrixRow = RGB565ToYMatrixRow_Any_WASMSIMD;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGB565ToYMatrixRow = RGB565ToYMatrixRow_WASMSIMD;
|
|
}
|
|
}
|
|
#endif
|
|
|
|
if (!src_rgb565 || !dst_y || !dst_u || !dst_v || width <= 0 || height == 0 ||
|
|
height == INT_MIN) {
|
|
return -1;
|
|
}
|
|
// Negative height means invert the image.
|
|
if (height < 0) {
|
|
height = -height;
|
|
src_rgb565 = src_rgb565 + (ptrdiff_t)(height - 1) * src_stride_rgb565;
|
|
src_stride_rgb565 = -src_stride_rgb565;
|
|
}
|
|
|
|
for (y = 0; y < height - 1; y += 2) {
|
|
RGB565ToUVMatrixRow(src_rgb565, src_stride_rgb565, dst_u, dst_v, width,
|
|
&kArgbI601Constants);
|
|
RGB565ToYMatrixRow(src_rgb565, dst_y, width, &kArgbI601Constants);
|
|
RGB565ToYMatrixRow(src_rgb565 + src_stride_rgb565, dst_y + dst_stride_y,
|
|
width, &kArgbI601Constants);
|
|
src_rgb565 += src_stride_rgb565 * 2;
|
|
dst_y += dst_stride_y * 2;
|
|
dst_u += dst_stride_u;
|
|
dst_v += dst_stride_v;
|
|
}
|
|
if (height & 1) {
|
|
RGB565ToYMatrixRow(src_rgb565, dst_y, width, &kArgbI601Constants);
|
|
RGB565ToUVMatrixRow(src_rgb565, 0, dst_u, dst_v, width,
|
|
&kArgbI601Constants);
|
|
}
|
|
return 0;
|
|
}
|
|
// Convert ARGB1555 to I420.
|
|
LIBYUV_API
|
|
int ARGB1555ToI420(const uint8_t* src_argb1555,
|
|
int src_stride_argb1555,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
int y;
|
|
void (*ARGB1555ToUVMatrixRow)(
|
|
const uint8_t* src_argb1555, int src_stride_argb1555, uint8_t* dst_u,
|
|
uint8_t* dst_v, int width, const struct ArgbConstants* c) =
|
|
ARGB1555ToUVMatrixRow_C;
|
|
void (*ARGB1555ToYMatrixRow)(const uint8_t* src_argb1555, uint8_t* dst_y,
|
|
int width, const struct ArgbConstants* c) =
|
|
ARGB1555ToYMatrixRow_C;
|
|
|
|
#if defined(HAS_ARGB1555TOYMATRIXROW_AVX2)
|
|
if (TestCpuFlag(kCpuHasAVX2)) {
|
|
ARGB1555ToYMatrixRow = ARGB1555ToYMatrixRow_Any_AVX2;
|
|
if (IS_ALIGNED(width, 32)) {
|
|
ARGB1555ToYMatrixRow = ARGB1555ToYMatrixRow_AVX2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGB1555TOUVMATRIXROW_AVX2)
|
|
if (TestCpuFlag(kCpuHasAVX2)) {
|
|
ARGB1555ToUVMatrixRow = ARGB1555ToUVMatrixRow_Any_AVX2;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
ARGB1555ToUVMatrixRow = ARGB1555ToUVMatrixRow_AVX2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGB1555TOUVMATRIXROW_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
ARGB1555ToUVMatrixRow = ARGB1555ToUVMatrixRow_Any_NEON;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
ARGB1555ToUVMatrixRow = ARGB1555ToUVMatrixRow_NEON;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGB1555TOYMATRIXROW_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
ARGB1555ToYMatrixRow = ARGB1555ToYMatrixRow_Any_NEON;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
ARGB1555ToYMatrixRow = ARGB1555ToYMatrixRow_NEON;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGB1555TOUVMATRIXROW_WASMSIMD)
|
|
if (TestCpuFlag(kCpuHasWASMSIMD)) {
|
|
ARGB1555ToUVMatrixRow = ARGB1555ToUVMatrixRow_Any_WASMSIMD;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
ARGB1555ToUVMatrixRow = ARGB1555ToUVMatrixRow_WASMSIMD;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGB1555TOYMATRIXROW_WASMSIMD)
|
|
if (TestCpuFlag(kCpuHasWASMSIMD)) {
|
|
ARGB1555ToYMatrixRow = ARGB1555ToYMatrixRow_Any_WASMSIMD;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
ARGB1555ToYMatrixRow = ARGB1555ToYMatrixRow_WASMSIMD;
|
|
}
|
|
}
|
|
#endif
|
|
|
|
if (!src_argb1555 || !dst_y || !dst_u || !dst_v || width <= 0 ||
|
|
height == 0 || height == INT_MIN) {
|
|
return -1;
|
|
}
|
|
// Negative height means invert the image.
|
|
if (height < 0) {
|
|
height = -height;
|
|
src_argb1555 = src_argb1555 + (ptrdiff_t)(height - 1) * src_stride_argb1555;
|
|
src_stride_argb1555 = -src_stride_argb1555;
|
|
}
|
|
|
|
for (y = 0; y < height - 1; y += 2) {
|
|
ARGB1555ToUVMatrixRow(src_argb1555, src_stride_argb1555, dst_u, dst_v,
|
|
width, &kArgbI601Constants);
|
|
ARGB1555ToYMatrixRow(src_argb1555, dst_y, width, &kArgbI601Constants);
|
|
ARGB1555ToYMatrixRow(src_argb1555 + src_stride_argb1555,
|
|
dst_y + dst_stride_y, width, &kArgbI601Constants);
|
|
src_argb1555 += src_stride_argb1555 * 2;
|
|
dst_y += dst_stride_y * 2;
|
|
dst_u += dst_stride_u;
|
|
dst_v += dst_stride_v;
|
|
}
|
|
if (height & 1) {
|
|
ARGB1555ToYMatrixRow(src_argb1555, dst_y, width, &kArgbI601Constants);
|
|
ARGB1555ToUVMatrixRow(src_argb1555, 0, dst_u, dst_v, width,
|
|
&kArgbI601Constants);
|
|
}
|
|
return 0;
|
|
}
|
|
// Convert ARGB4444 to I420.
|
|
LIBYUV_API
|
|
int ARGB4444ToI420(const uint8_t* src_argb4444,
|
|
int src_stride_argb4444,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
int y;
|
|
void (*ARGB4444ToUVMatrixRow)(
|
|
const uint8_t* src_argb4444, int src_stride_argb4444, uint8_t* dst_u,
|
|
uint8_t* dst_v, int width, const struct ArgbConstants* c) =
|
|
ARGB4444ToUVMatrixRow_C;
|
|
void (*ARGB4444ToYMatrixRow)(const uint8_t* src_argb4444, uint8_t* dst_y,
|
|
int width, const struct ArgbConstants* c) =
|
|
ARGB4444ToYMatrixRow_C;
|
|
|
|
#if defined(HAS_ARGB4444TOYMATRIXROW_AVX2)
|
|
if (TestCpuFlag(kCpuHasAVX2)) {
|
|
ARGB4444ToYMatrixRow = ARGB4444ToYMatrixRow_Any_AVX2;
|
|
if (IS_ALIGNED(width, 32)) {
|
|
ARGB4444ToYMatrixRow = ARGB4444ToYMatrixRow_AVX2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGB4444TOUVMATRIXROW_AVX2)
|
|
if (TestCpuFlag(kCpuHasAVX2)) {
|
|
ARGB4444ToUVMatrixRow = ARGB4444ToUVMatrixRow_Any_AVX2;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
ARGB4444ToUVMatrixRow = ARGB4444ToUVMatrixRow_AVX2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGB4444TOUVMATRIXROW_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
ARGB4444ToUVMatrixRow = ARGB4444ToUVMatrixRow_Any_NEON;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
ARGB4444ToUVMatrixRow = ARGB4444ToUVMatrixRow_NEON;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGB4444TOYMATRIXROW_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
ARGB4444ToYMatrixRow = ARGB4444ToYMatrixRow_Any_NEON;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
ARGB4444ToYMatrixRow = ARGB4444ToYMatrixRow_NEON;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGB4444TOUVMATRIXROW_WASMSIMD)
|
|
if (TestCpuFlag(kCpuHasWASMSIMD)) {
|
|
ARGB4444ToUVMatrixRow = ARGB4444ToUVMatrixRow_Any_WASMSIMD;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
ARGB4444ToUVMatrixRow = ARGB4444ToUVMatrixRow_WASMSIMD;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_ARGB4444TOYMATRIXROW_WASMSIMD)
|
|
if (TestCpuFlag(kCpuHasWASMSIMD)) {
|
|
ARGB4444ToYMatrixRow = ARGB4444ToYMatrixRow_Any_WASMSIMD;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
ARGB4444ToYMatrixRow = ARGB4444ToYMatrixRow_WASMSIMD;
|
|
}
|
|
}
|
|
#endif
|
|
|
|
if (!src_argb4444 || !dst_y || !dst_u || !dst_v || width <= 0 ||
|
|
height == 0 || height == INT_MIN) {
|
|
return -1;
|
|
}
|
|
// Negative height means invert the image.
|
|
if (height < 0) {
|
|
height = -height;
|
|
src_argb4444 = src_argb4444 + (ptrdiff_t)(height - 1) * src_stride_argb4444;
|
|
src_stride_argb4444 = -src_stride_argb4444;
|
|
}
|
|
|
|
for (y = 0; y < height - 1; y += 2) {
|
|
ARGB4444ToUVMatrixRow(src_argb4444, src_stride_argb4444, dst_u, dst_v,
|
|
width, &kArgbI601Constants);
|
|
ARGB4444ToYMatrixRow(src_argb4444, dst_y, width, &kArgbI601Constants);
|
|
ARGB4444ToYMatrixRow(src_argb4444 + src_stride_argb4444,
|
|
dst_y + dst_stride_y, width, &kArgbI601Constants);
|
|
src_argb4444 += src_stride_argb4444 * 2;
|
|
dst_y += dst_stride_y * 2;
|
|
dst_u += dst_stride_u;
|
|
dst_v += dst_stride_v;
|
|
}
|
|
if (height & 1) {
|
|
ARGB4444ToYMatrixRow(src_argb4444, dst_y, width, &kArgbI601Constants);
|
|
ARGB4444ToUVMatrixRow(src_argb4444, 0, dst_u, dst_v, width,
|
|
&kArgbI601Constants);
|
|
}
|
|
return 0;
|
|
}
|
|
// Convert RGB24 to J400.
|
|
LIBYUV_API
|
|
int RGB24ToJ400(const uint8_t* src_rgb24,
|
|
int src_stride_rgb24,
|
|
uint8_t* dst_yj,
|
|
int dst_stride_yj,
|
|
int width,
|
|
int height) {
|
|
int y;
|
|
void (*RGBToYMatrixRow)(const uint8_t* src_rgb, uint8_t* dst_y, int width,
|
|
const struct ArgbConstants* c) = RGBToYMatrixRow_C;
|
|
if (!src_rgb24 || !dst_yj || width <= 0 || height == 0 || height == INT_MIN) {
|
|
return -1;
|
|
}
|
|
if (height < 0) {
|
|
height = -height;
|
|
src_rgb24 = src_rgb24 + (ptrdiff_t)(height - 1) * src_stride_rgb24;
|
|
src_stride_rgb24 = -src_stride_rgb24;
|
|
}
|
|
// Coalesce rows.
|
|
if (src_stride_rgb24 == width * 3 && dst_stride_yj == width &&
|
|
(ptrdiff_t)width * height <= INT_MAX) {
|
|
width *= height;
|
|
height = 1;
|
|
src_stride_rgb24 = dst_stride_yj = 0;
|
|
}
|
|
|
|
#if defined(HAS_RGBTOYMATRIXROW_SSSE3)
|
|
if (TestCpuFlag(kCpuHasSSSE3)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_SSSE3;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_SSSE3;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_AVX2)
|
|
if (TestCpuFlag(kCpuHasAVX2)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_AVX2;
|
|
if (IS_ALIGNED(width, 32)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_AVX2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_AVX512BW)
|
|
if (TestCpuFlag(kCpuHasAVX512BW)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_AVX512BW;
|
|
if (IS_ALIGNED(width, 64)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_AVX512BW;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_NEON;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_NEON;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_NEON_DOTPROD)
|
|
if (TestCpuFlag(kCpuHasNeonDotProd)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_NEON_DotProd;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_NEON_DotProd;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_LSX)
|
|
if (TestCpuFlag(kCpuHasLSX)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_LSX;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_LSX;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_LASX)
|
|
if (TestCpuFlag(kCpuHasLASX)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_LASX;
|
|
if (IS_ALIGNED(width, 32)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_LASX;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_RVV)
|
|
if (TestCpuFlag(kCpuHasRVV)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_RVV;
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_WASMSIMD)
|
|
if (TestCpuFlag(kCpuHasWASMSIMD)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_WASMSIMD;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_WASMSIMD;
|
|
}
|
|
}
|
|
#endif
|
|
|
|
for (y = 0; y < height; ++y) {
|
|
RGBToYMatrixRow(src_rgb24, dst_yj, width, &kArgbJPEGConstants);
|
|
src_rgb24 += src_stride_rgb24;
|
|
dst_yj += dst_stride_yj;
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
// Convert RAW to J400.
|
|
LIBYUV_API
|
|
int RAWToJ400(const uint8_t* src_raw,
|
|
int src_stride_raw,
|
|
uint8_t* dst_yj,
|
|
int dst_stride_yj,
|
|
int width,
|
|
int height) {
|
|
int y;
|
|
void (*RGBToYMatrixRow)(const uint8_t* src_rgb, uint8_t* dst_y, int width,
|
|
const struct ArgbConstants* c) = RGBToYMatrixRow_C;
|
|
if (!src_raw || !dst_yj || width <= 0 || height == 0 || height == INT_MIN) {
|
|
return -1;
|
|
}
|
|
if (height < 0) {
|
|
height = -height;
|
|
src_raw = src_raw + (ptrdiff_t)(height - 1) * src_stride_raw;
|
|
src_stride_raw = -src_stride_raw;
|
|
}
|
|
// Coalesce rows.
|
|
if (src_stride_raw == width * 3 && dst_stride_yj == width &&
|
|
(ptrdiff_t)width * height <= INT_MAX) {
|
|
width *= height;
|
|
height = 1;
|
|
src_stride_raw = dst_stride_yj = 0;
|
|
}
|
|
|
|
#if defined(HAS_RGBTOYMATRIXROW_SSSE3)
|
|
if (TestCpuFlag(kCpuHasSSSE3)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_SSSE3;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_SSSE3;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_AVX2)
|
|
if (TestCpuFlag(kCpuHasAVX2)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_AVX2;
|
|
if (IS_ALIGNED(width, 32)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_AVX2;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_AVX512BW)
|
|
if (TestCpuFlag(kCpuHasAVX512BW)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_AVX512BW;
|
|
if (IS_ALIGNED(width, 64)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_AVX512BW;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_NEON)
|
|
if (TestCpuFlag(kCpuHasNEON)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_NEON;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_NEON;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_NEON_DOTPROD)
|
|
if (TestCpuFlag(kCpuHasNeonDotProd)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_NEON_DotProd;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_NEON_DotProd;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_LSX)
|
|
if (TestCpuFlag(kCpuHasLSX)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_LSX;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_LSX;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_LASX)
|
|
if (TestCpuFlag(kCpuHasLASX)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_LASX;
|
|
if (IS_ALIGNED(width, 32)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_LASX;
|
|
}
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_RVV)
|
|
if (TestCpuFlag(kCpuHasRVV)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_RVV;
|
|
}
|
|
#endif
|
|
#if defined(HAS_RGBTOYMATRIXROW_WASMSIMD)
|
|
if (TestCpuFlag(kCpuHasWASMSIMD)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_Any_WASMSIMD;
|
|
if (IS_ALIGNED(width, 16)) {
|
|
RGBToYMatrixRow = RGBToYMatrixRow_WASMSIMD;
|
|
}
|
|
}
|
|
#endif
|
|
|
|
for (y = 0; y < height; ++y) {
|
|
RGBToYMatrixRow(src_raw, dst_yj, width, &kAbgrJPEGConstants);
|
|
src_raw += src_stride_raw;
|
|
dst_yj += dst_stride_yj;
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
// Convert Android420 to I420.
|
|
LIBYUV_API
|
|
int Android420ToI420(const uint8_t* src_y,
|
|
int src_stride_y,
|
|
const uint8_t* src_u,
|
|
int src_stride_u,
|
|
const uint8_t* src_v,
|
|
int src_stride_v,
|
|
int src_pixel_stride_uv,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
return Android420ToI420Rotate(src_y, src_stride_y, src_u, src_stride_u, src_v,
|
|
src_stride_v, src_pixel_stride_uv, dst_y,
|
|
dst_stride_y, dst_u, dst_stride_u, dst_v,
|
|
dst_stride_v, width, height, kRotate0);
|
|
}
|
|
|
|
// depth is source bits measured from lsb; For msb use 16
|
|
static int Biplanar16bitTo8bit(const uint16_t* src_y,
|
|
int src_stride_y,
|
|
const uint16_t* src_uv,
|
|
int src_stride_uv,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_uv,
|
|
int dst_stride_uv,
|
|
int width,
|
|
int height,
|
|
int subsample_x,
|
|
int subsample_y,
|
|
int depth) {
|
|
if ((!src_y && dst_y) || !src_uv || !dst_uv || width <= 0 || height == 0 ||
|
|
height == INT_MIN) {
|
|
return -1;
|
|
}
|
|
int uv_width = SUBSAMPLE(width, subsample_x, subsample_x);
|
|
int uv_height = SUBSAMPLE(height, subsample_y, subsample_y);
|
|
int scale = 1 << (24 - depth);
|
|
// Negative height means invert the image.
|
|
if (height < 0) {
|
|
height = -height;
|
|
uv_height = -uv_height;
|
|
src_y = src_y + (ptrdiff_t)(height - 1) * src_stride_y;
|
|
src_uv = src_uv + (ptrdiff_t)(uv_height - 1) * src_stride_uv;
|
|
src_stride_y = -src_stride_y;
|
|
src_stride_uv = -src_stride_uv;
|
|
}
|
|
|
|
// Convert Y plane.
|
|
if (dst_y) {
|
|
Convert16To8Plane(src_y, src_stride_y, dst_y, dst_stride_y, scale, width,
|
|
height);
|
|
}
|
|
// Convert UV planes.
|
|
Convert16To8Plane(src_uv, src_stride_uv, dst_uv, dst_stride_uv, scale,
|
|
uv_width * 2, uv_height);
|
|
return 0;
|
|
}
|
|
|
|
// Convert 10 bit P010 to 8 bit NV12.
|
|
// Depth set to 16 because P010 uses 10 msb and this function keeps the upper 8
|
|
// bits of the specified number of bits.
|
|
LIBYUV_API
|
|
int P010ToNV12(const uint16_t* src_y,
|
|
int src_stride_y,
|
|
const uint16_t* src_uv,
|
|
int src_stride_uv,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_uv,
|
|
int dst_stride_uv,
|
|
int width,
|
|
int height) {
|
|
return Biplanar16bitTo8bit(src_y, src_stride_y, src_uv, src_stride_uv, dst_y,
|
|
dst_stride_y, dst_uv, dst_stride_uv, width, height,
|
|
1, 1, 16);
|
|
}
|
|
|
|
static int Planar8bitTo8bit(const uint8_t* src_y,
|
|
int src_stride_y,
|
|
const uint8_t* src_u,
|
|
int src_stride_u,
|
|
const uint8_t* src_v,
|
|
int src_stride_v,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height,
|
|
int subsample_x,
|
|
int subsample_y,
|
|
int scale_y,
|
|
int bias_y,
|
|
int scale_uv,
|
|
int bias_uv) {
|
|
if ((!src_y && dst_y) || !src_u || !src_v || !dst_u || !dst_v || width <= 0 ||
|
|
height == 0 || height == INT_MIN) {
|
|
return -1;
|
|
}
|
|
int uv_width = SUBSAMPLE(width, subsample_x, subsample_x);
|
|
int uv_height = SUBSAMPLE(height, subsample_y, subsample_y);
|
|
// Negative height means invert the image.
|
|
if (height < 0) {
|
|
height = -height;
|
|
uv_height = -uv_height;
|
|
src_y = src_y + (ptrdiff_t)(height - 1) * src_stride_y;
|
|
src_u = src_u + (ptrdiff_t)(uv_height - 1) * src_stride_u;
|
|
src_v = src_v + (ptrdiff_t)(uv_height - 1) * src_stride_v;
|
|
src_stride_y = -src_stride_y;
|
|
src_stride_u = -src_stride_u;
|
|
src_stride_v = -src_stride_v;
|
|
}
|
|
|
|
// Convert Y plane.
|
|
if (dst_y) {
|
|
Convert8To8Plane(src_y, src_stride_y, dst_y, dst_stride_y, scale_y, bias_y,
|
|
width, height);
|
|
}
|
|
// Convert UV planes.
|
|
Convert8To8Plane(src_u, src_stride_u, dst_u, dst_stride_u, scale_uv, bias_uv,
|
|
uv_width, uv_height);
|
|
Convert8To8Plane(src_v, src_stride_v, dst_v, dst_stride_v, scale_uv, bias_uv,
|
|
uv_width, uv_height);
|
|
return 0;
|
|
}
|
|
|
|
LIBYUV_API
|
|
int J420ToI420(const uint8_t* src_y,
|
|
int src_stride_y,
|
|
const uint8_t* src_u,
|
|
int src_stride_u,
|
|
const uint8_t* src_v,
|
|
int src_stride_v,
|
|
uint8_t* dst_y,
|
|
int dst_stride_y,
|
|
uint8_t* dst_u,
|
|
int dst_stride_u,
|
|
uint8_t* dst_v,
|
|
int dst_stride_v,
|
|
int width,
|
|
int height) {
|
|
return Planar8bitTo8bit(src_y, src_stride_y, src_u, src_stride_u, src_v,
|
|
src_stride_v, dst_y, dst_stride_y, dst_u,
|
|
dst_stride_u, dst_v, dst_stride_v, width, height, 1,
|
|
1, 220, 16, 225, 16);
|
|
}
|
|
|
|
#ifdef __cplusplus
|
|
} // extern "C"
|
|
} // namespace libyuv
|
|
#endif
|