* Copyright (C) 2025 Huawei Device Co., Ltd.
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
#include "securec.h"
#include "avcodec_errors.h"
#include "avcodec_log.h"
#include "avcodec_trace.h"
#include "avc_encoder_convert.h"
#if defined(ARMV8)
#include <arm_neon.h>
#endif
namespace OHOS {
namespace MediaAVCodec {
namespace Codec {
#ifdef __cplusplus
extern "C" {
#endif
namespace {
constexpr OHOS::HiviewDFX::HiLogLabel LABEL = {LOG_CORE, LOG_DOMAIN_FRAMEWORK, "AvcEncoder"};
}
static const int16_t BT601_MATRIX[2][3][3] = {
{{76, 150, 29}, {-43, -85, 128}, {128, -107, -21}},
{{66, 129, 25}, {-38, -74, 112}, {112, -94, -18}},
};
static const int16_t BT709_MATRIX[2][3][3] = {
{{54, 183, 18}, {-29, -99, 128}, {128, -116, -12}},
{{47, 157, 16}, {-26, -86, 112}, {112, -102, -10}},
};
inline uint8_t Clip3(uint8_t min, uint8_t value, uint8_t max)
{
if (value < min) {
return min;
}
if (value > max) {
return max;
}
return value;
}
#if defined(ARMV8)
int32_t ConvertRgbToYuv420Neon(uint8_t *dstData, int32_t width, int32_t height,
int32_t bufferSize, RgbImageData &rgbData)
{
const uint8_t *srcData = rgbData.data;
int32_t srcStride = rgbData.stride;
int32_t bytesPerPixel = rgbData.bytesPerPixel;
COLOR_MATRIX colorMatrix = rgbData.matrix;
COLOR_RANGE colorRange = rgbData.range;
int32_t dstStride = width;
if (dstData == nullptr || srcData == nullptr) {
return AVCS_ERR_INVALID_VAL;
}
if ((dstStride * height * 3 / 2) > bufferSize) {
AVCODEC_LOGE("conversion buffer is too small for converting from RGB to YUV");
return AVCS_ERR_NO_MEMORY;
}
colorRange = (colorRange != RANGE_FULL && colorRange != RANGE_LIMITED) ? RANGE_LIMITED : colorRange;
const int16_t(*weights)[3] =
(colorMatrix == MATRIX_BT709) ? BT709_MATRIX[colorRange - 1] : BT601_MATRIX[colorRange - 1];
uint8_t zeroLvl = colorRange == RANGE_FULL ? 0 : 16;
const uint8x16_t u8_16 = vdupq_n_u8(zeroLvl);
const uint16x8_t mask = vdupq_n_s16(255);
const int8x8_t s8_rounding = vdup_n_s8(-128);
uint8x8_t scalar_Yr = vdup_n_u8((uint8_t)weights[0][0]);
uint8x8_t scalar_Yg = vdup_n_u8((uint8_t)weights[0][1]);
uint8x8_t scalar_Yb = vdup_n_u8((uint8_t)weights[0][2]);
int16x8_t u2_scalar = vdupq_n_s16(weights[1][1]);
int16x8_t v2_scalar = vdupq_n_s16(weights[2][1]);
int16x8_t u3_scalar = vdupq_n_s16(weights[1][2]);
int16x8_t v3_scalar = vdupq_n_s16(weights[2][2]);
int patch = width >> 4;
int widthLess = width - (patch << 4);
const int isPad = (widthLess > 0);
const uint8_t *rgbaBuf = srcData;
int rgbaOffset = 0;
int stridePadding = srcStride - width;
int yStride = 0;
int uStride = yStride >> 1;
int lumaOffset = 0;
int chromaUOffset = dstStride * height;
int chromaVOffset = chromaUOffset + (dstStride >> 1) * (height >> 1);
const int batchOffset = 16;
for (int j = 0; j < height; ++j) {
for (int i = 0; i < patch + isPad; ++i) {
uint8x16x3_t pixelRgbOri;
if (bytesPerPixel == RGBA_COLINC) {
uint8x16x4_t pixelRgbaOri = vld4q_u8(rgbaBuf + rgbaOffset);
pixelRgbOri.val[0] = pixelRgbaOri.val[0];
pixelRgbOri.val[1] = pixelRgbaOri.val[1];
pixelRgbOri.val[2] = pixelRgbaOri.val[2];
} else {
pixelRgbOri = vld3q_u8(rgbaBuf + rgbaOffset);
}
uint8x8x2_t rOri;
uint8x8x2_t gOri;
uint8x8x2_t bOri;
rOri.val[0] = vget_low_u8(pixelRgbOri.val[0]);
rOri.val[1] = vget_high_u8(pixelRgbOri.val[0]);
gOri.val[0] = vget_low_u8(pixelRgbOri.val[1]);
gOri.val[1] = vget_high_u8(pixelRgbOri.val[1]);
bOri.val[0] = vget_low_u8(pixelRgbOri.val[2]);
bOri.val[1] = vget_high_u8(pixelRgbOri.val[2]);
uint16x8x2_t yTemp;
uint8x16_t pixelY;
yTemp.val[0] = vmull_u8(rOri.val[0], scalar_Yr);
yTemp.val[1] = vmull_u8(rOri.val[1], scalar_Yr);
yTemp.val[0] = vmlal_u8(yTemp.val[0], gOri.val[0], scalar_Yg);
yTemp.val[1] = vmlal_u8(yTemp.val[1], gOri.val[1], scalar_Yg);
yTemp.val[0] = vmlal_u8(yTemp.val[0], bOri.val[0], scalar_Yb);
yTemp.val[1] = vmlal_u8(yTemp.val[1], bOri.val[1], scalar_Yb);
pixelY = vcombine_u8(vqshrn_n_u16(yTemp.val[0], 8), vqshrn_n_u16(yTemp.val[1], 8));
pixelY = vaddq_u8(pixelY, u8_16);
vst1q_u8(dstData + lumaOffset, pixelY);
if (j % 2 == 0) {
int16x8_t rUpdate = vreinterpretq_s16_u16(vandq_u16(vreinterpretq_u16_u8(pixelRgbOri.val[0]), mask));
int16x8_t gUpdate = vreinterpretq_s16_u16(vandq_u16(vreinterpretq_u16_u8(pixelRgbOri.val[1]), mask));
int16x8_t bUpdate = vreinterpretq_s16_u16(vandq_u16(vreinterpretq_u16_u8(pixelRgbOri.val[2]), mask));
int16x8_t signedU;
int16x8_t signedV;
uint8x8_t outputUV[2];
signedU = vmulq_n_s16(rUpdate, weights[1][0]);
signedV = vmulq_n_s16(rUpdate, weights[2][0]);
signedU = vmlaq_s16(signedU, gUpdate, u2_scalar);
signedV = vmlaq_s16(signedV, gUpdate, v2_scalar);
signedU = vmlaq_s16(signedU, bUpdate, u3_scalar);
signedV = vmlaq_s16(signedV, bUpdate, v3_scalar);
outputUV[0] = vreinterpret_u8_s8(vadd_s8(vqshrn_n_s16(signedU, 8), s8_rounding));
outputUV[1] = vreinterpret_u8_s8(vadd_s8(vqshrn_n_s16(signedV, 8), s8_rounding));
vst1_u8(dstData + chromaUOffset, outputUV[0]);
vst1_u8(dstData + chromaVOffset, outputUV[1]);
chromaUOffset += (i == 0 && isPad) ? (widthLess >> 1) : (batchOffset >> 1);
chromaVOffset += (i == 0 && isPad) ? (widthLess >> 1) : (batchOffset >> 1);
}
if (i == 0 && isPad) {
rgbaOffset += bytesPerPixel * widthLess;
lumaOffset += widthLess;
} else {
rgbaOffset += bytesPerPixel * batchOffset;
lumaOffset += batchOffset;
}
}
if ((j & 1) == 0) {
chromaUOffset += uStride;
chromaVOffset += uStride;
}
lumaOffset += yStride;
rgbaOffset += stridePadding * bytesPerPixel;
}
return AVCS_ERR_OK;
}
#endif
int32_t ConvertRgbToYuv420(uint8_t *dstData, int32_t width, int32_t height,
int32_t bufferSize, RgbImageData &rgbData)
{
const uint8_t *srcData = rgbData.data;
int32_t srcStride = rgbData.stride;
int32_t bytesPerPixel = rgbData.bytesPerPixel;
COLOR_MATRIX colorMatrix = rgbData.matrix;
COLOR_RANGE colorRange = rgbData.range;
int32_t dstStride = width;
if (dstData == nullptr || srcData == nullptr) {
return AVCS_ERR_INVALID_VAL;
}
if ((dstStride * height * 3 / 2) > bufferSize) {
AVCODEC_LOGE("conversion buffer is too small for converting from RGB to YUV");
return AVCS_ERR_NO_MEMORY;
}
uint8_t *dstU = dstData + dstStride * height;
uint8_t *dstV = dstU + (dstStride >> 1) * (height >> 1);
const uint8_t *pRed = srcData;
const uint8_t *pGreen = srcData + 1;
const uint8_t *pBlue = srcData + 2;
colorRange = (colorRange != RANGE_FULL && colorRange != RANGE_LIMITED) ? RANGE_LIMITED : colorRange;
const int16_t(*weights)[3] =
(colorMatrix == MATRIX_BT709) ? BT709_MATRIX[colorRange - 1] : BT601_MATRIX[colorRange - 1];
uint8_t zeroLvl = colorRange == RANGE_FULL ? 0 : 16;
uint8_t maxLvlLuma = colorRange == RANGE_FULL ? 255 : 235;
uint8_t maxLvlChroma = colorRange == RANGE_FULL ? 255 : 240;
for (int32_t y = 0; y < height; ++y) {
for (int32_t x = 0; x < width; ++x) {
uint8_t r = *pRed;
uint8_t g = *pGreen;
uint8_t b = *pBlue;
uint8_t luma = ((r * weights[0][0] + g * weights[0][1] + b * weights[0][2]) >> 8) + zeroLvl;
dstData[x] = Clip3(zeroLvl, luma, maxLvlLuma);
if ((x & 1) == 0 && (y & 1) == 0) {
uint8_t u = ((r * weights[1][0] + g * weights[1][1] + b * weights[1][2]) >> 8) + 128;
uint8_t v = ((r * weights[2][0] + g * weights[2][1] + b * weights[2][2]) >> 8) + 128;
dstU[x >> 1] = Clip3(zeroLvl, u, maxLvlChroma);
dstV[x >> 1] = Clip3(zeroLvl, v, maxLvlChroma);
}
pRed += bytesPerPixel;
pGreen += bytesPerPixel;
pBlue += bytesPerPixel;
}
if ((y & 1) == 0) {
dstU += dstStride >> 1;
dstV += dstStride >> 1;
}
pRed -= bytesPerPixel * width;
pGreen -= bytesPerPixel * width;
pBlue -= bytesPerPixel * width;
pRed += bytesPerPixel * srcStride;
pGreen += bytesPerPixel * srcStride;
pBlue += bytesPerPixel * srcStride;
dstData += dstStride;
}
return AVCS_ERR_OK;
}
int32_t ConvertNv12ToYuv420(uint8_t *dstData, int32_t width, int32_t height,
int32_t bufferSize, YuvImageData &yuvData)
{
if (bufferSize < (width * height * 3 / 2)) {
AVCODEC_LOGE("conversion buffer is too small for converting from NV12 to YUV");
return AVCS_ERR_NO_MEMORY;
}
int32_t dstStride = width;
uint8_t *dstY = dstData;
uint8_t *dstU = dstData + dstStride * height;
uint8_t *dstV = dstU + (dstStride >> 1) * (height >> 1);
int32_t srcStride = yuvData.stride;
const uint8_t *srcY = yuvData.data;
const uint8_t *srcU = srcY + yuvData.uvOffset;
const uint8_t *srcV = srcY + yuvData.uvOffset + 1;
int32_t x;
int32_t y;
for (y = 0; y < height; y++) {
errno_t ret = memcpy_s(dstY, width, srcY, width);
CHECK_AND_RETURN_RET_LOG(ret == EOK, AVCS_ERR_UNKNOWN, "memcpy_s failed");
dstY += dstStride;
srcY += srcStride;
}
height = height >> 1;
width = width >> 1;
dstStride = dstStride >> 1;
for (y = 0; y < height; y++) {
for (x = 0; x < width; x++) {
dstU[x] = srcU[x << 1];
dstV[x] = srcV[x << 1];
}
dstU += dstStride;
dstV += dstStride;
srcU += srcStride;
srcV += srcStride;
}
return AVCS_ERR_OK;
}
int32_t ConvertNv21ToYuv420(uint8_t *dstData, int32_t width, int32_t height,
int32_t bufferSize, YuvImageData &yuvData)
{
if (bufferSize < (width * height * 3 / 2)) {
AVCODEC_LOGE("conversion buffer is too small for converting from NV21 to YUV");
return AVCS_ERR_NO_MEMORY;
}
int32_t dstStride = width;
uint8_t *dstY = dstData;
uint8_t *dstU = dstData + dstStride * height;
uint8_t *dstV = dstU + (dstStride >> 1) * (height >> 1);
int32_t srcStride = yuvData.stride;
const uint8_t *srcY = yuvData.data;
const uint8_t *srcU = srcY + yuvData.uvOffset + 1;
const uint8_t *srcV = srcY + yuvData.uvOffset;
int32_t x;
int32_t y;
for (y = 0; y < height; y++) {
errno_t ret = memcpy_s(dstY, width, srcY, width);
CHECK_AND_RETURN_RET_LOG(ret == EOK, AVCS_ERR_UNKNOWN, "memcpy_s failed");
dstY += dstStride;
srcY += srcStride;
}
height = height >> 1;
width = width >> 1;
dstStride = dstStride >> 1;
for (y = 0; y < height; y++) {
for (x = 0; x < width; x++) {
dstU[x] = srcU[x << 1];
dstV[x] = srcV[x << 1];
}
dstU += dstStride;
dstV += dstStride;
srcU += srcStride;
srcV += srcStride;
}
return AVCS_ERR_OK;
}
#ifdef __cplusplus
}
#endif
}
}
}