FFmpeg源代码简单分析：libswscale的sws_scale()

=====================================================

FFmpeg的库函数源代码分析文章列表：

【架构图】

FFmpeg源代码结构图 - 解码

FFmpeg源代码结构图 - 编码

【通用】

FFmpeg 源代码简单分析：av_register_all()

FFmpeg 源代码简单分析：avcodec_register_all()

FFmpeg 源代码简单分析：内存的分配和释放（av_malloc()、av_free()等）

FFmpeg 源代码简单分析：常见结构体的初始化和销毁（AVFormatContext，AVFrame等）

FFmpeg 源代码简单分析：avio_open2()

FFmpeg 源代码简单分析：av_find_decoder()和av_find_encoder()

FFmpeg 源代码简单分析：avcodec_open2()

FFmpeg 源代码简单分析：avcodec_close()

【解码】

图解FFMPEG打开媒体的函数avformat_open_input

FFmpeg 源代码简单分析：avformat_open_input()

FFmpeg 源代码简单分析：avformat_find_stream_info()

FFmpeg 源代码简单分析：av_read_frame()

FFmpeg 源代码简单分析：avcodec_decode_video2()

FFmpeg 源代码简单分析：avformat_close_input()

【编码】

FFmpeg 源代码简单分析：avformat_alloc_output_context2()

FFmpeg 源代码简单分析：avformat_write_header()

FFmpeg 源代码简单分析：avcodec_encode_video()

FFmpeg 源代码简单分析：av_write_frame()

FFmpeg 源代码简单分析：av_write_trailer()

【其它】

FFmpeg源代码简单分析：日志输出系统（av_log()等）

FFmpeg源代码简单分析：结构体成员管理系统-AVClass

FFmpeg源代码简单分析：结构体成员管理系统-AVOption

FFmpeg源代码简单分析：libswscale的sws_getContext()

FFmpeg源代码简单分析：libswscale的sws_scale()

FFmpeg源代码简单分析：libavdevice的avdevice_register_all()

FFmpeg源代码简单分析：libavdevice的gdigrab

【脚本】

FFmpeg源代码简单分析：makefile

FFmpeg源代码简单分析：configure

【H.264】

FFmpeg的H.264解码器源代码简单分析：概述

=====================================================

本文继续上一篇文章《FFmpeg源代码分析：sws_getContext()》的内容，简单分析FFmpeg的图像处理（缩放，YUV/RGB格式转换）类库libswsscale中的sws_scale()函数。libswscale是一个主要用于处理图片像素数据的类库。可以完成图片像素格式的转换，图片的拉伸等工作。有关libswscale的使用可以参考文章：
《最简单的基于FFmpeg的libswscale的示例（YUV转RGB）》

该类库常用的函数数量很少，一般情况下就3个：

在分析sws_scale()的源代码之前，先简单回顾一下上篇文章中分析得到的两张图。

函数调用结构图

分析得到的libswscale的函数调用关系如下图所示。

Libswscale处理数据流程

Libswscale处理像素数据的流程可以概括为下图。



从图中可以看出，libswscale处理数据有两条最主要的方式：unscaled和scaled。unscaled用于处理不需要拉伸的像素数据（属于比较特殊的情况），scaled用于处理需要拉伸的像素数据。Unscaled只需要对图像像素格式进行转换；而Scaled则除了对像素格式进行转换之外，还需要对图像进行缩放。Scaled方式可以分成以下几个步骤：

sws_scale()

sws_scale()是用于转换像素的函数。它的声明位于libswscale\swscale.h，如下所示。

/**
* Scale the image slice in srcSlice and put the resulting scaled
* slice in the image in dst. A slice is a sequence of consecutive
* rows in an image.
*
* Slices have to be provided in sequential order, either in
* top-bottom or bottom-top order. If slices are provided in
* non-sequential order the behavior of the function is undefined.
*
* @param c         the scaling context previously created with
*                  sws_getContext()
* @param srcSlice  the array containing the pointers to the planes of
*                  the source slice
* @param srcStride the array containing the strides for each plane of
*                  the source image
* @param srcSliceY the position in the source image of the slice to
*                  process, that is the number (counted starting from
*                  zero) in the image of the first row of the slice
* @param srcSliceH the height of the source slice, that is the number
*                  of rows in the slice
* @param dst       the array containing the pointers to the planes of
*                  the destination image
* @param dstStride the array containing the strides for each plane of
*                  the destination image
* @return          the height of the output slice
*/
int sws_scale(struct SwsContext *c, const uint8_t *const srcSlice[],
const int srcStride[], int srcSliceY, int srcSliceH,
uint8_t *const dst[], const int dstStride[]);

sws_scale()的定义位于libswscale\swscale.c，如下所示。

/**
* swscale wrapper, so we don't need to export the SwsContext.
* Assumes planar YUV to be in YUV order instead of YVU.
*/
int sws_scale(struct SwsContext *c,
const uint8_t * const srcSlice[],
const int srcStride[], int srcSliceY,
int srcSliceH, uint8_t *const dst[],
const int dstStride[])
{
int i, ret;
const uint8_t *src2[4];
uint8_t *dst2[4];
uint8_t *rgb0_tmp = NULL;
//检查输入参数
if (!srcStride || !dstStride || !dst || !srcSlice) {
av_log(c, AV_LOG_ERROR, "One of the input parameters to sws_scale() is NULL, please check the calling code\n");
return 0;
}
if (c->cascaded_context[0] && srcSliceY == 0 && srcSliceH == c->cascaded_context[0]->srcH) {
ret = sws_scale(c->cascaded_context[0],
srcSlice, srcStride, srcSliceY, srcSliceH,
c->cascaded_tmp, c->cascaded_tmpStride);
if (ret < 0)
return ret;
ret = sws_scale(c->cascaded_context[1],
(const uint8_t * const * )c->cascaded_tmp, c->cascaded_tmpStride, 0, c->cascaded_context[0]->dstH,
dst, dstStride);
return ret;
}

memcpy(src2, srcSlice, sizeof(src2));
memcpy(dst2, dst, sizeof(dst2));

// do not mess up sliceDir if we have a "trailing" 0-size slice
if (srcSliceH == 0)
return 0;
//检查
if (!check_image_pointers(srcSlice, c->srcFormat, srcStride)) {
av_log(c, AV_LOG_ERROR, "bad src image pointers\n");
return 0;
}
if (!check_image_pointers((const uint8_t* const*)dst, c->dstFormat, dstStride)) {
av_log(c, AV_LOG_ERROR, "bad dst image pointers\n");
return 0;
}

if (c->sliceDir == 0 && srcSliceY != 0 && srcSliceY + srcSliceH != c->srcH) {
av_log(c, AV_LOG_ERROR, "Slices start in the middle!\n");
return 0;
}
if (c->sliceDir == 0) {
if (srcSliceY == 0) c->sliceDir = 1; else c->sliceDir = -1;
}
//使用调色板palette的特殊处理？应该不常见
if (usePal(c->srcFormat)) {
for (i = 0; i < 256; i++) {
int r, g, b, y, u, v, a = 0xff;
if (c->srcFormat == AV_PIX_FMT_PAL8) {
uint32_t p = ((const uint32_t *)(srcSlice[1]))[i];
a = (p >> 24) & 0xFF;
r = (p >> 16) & 0xFF;
g = (p >>  8) & 0xFF;
b =  p        & 0xFF;
} else if (c->srcFormat == AV_PIX_FMT_RGB8) {
r = ( i >> 5     ) * 36;
g = ((i >> 2) & 7) * 36;
b = ( i       & 3) * 85;
} else if (c->srcFormat == AV_PIX_FMT_BGR8) {
b = ( i >> 6     ) * 85;
g = ((i >> 3) & 7) * 36;
r = ( i       & 7) * 36;
} else if (c->srcFormat == AV_PIX_FMT_RGB4_BYTE) {
r = ( i >> 3     ) * 255;
g = ((i >> 1) & 3) * 85;
b = ( i       & 1) * 255;
} else if (c->srcFormat == AV_PIX_FMT_GRAY8 || c->srcFormat == AV_PIX_FMT_GRAY8A) {
r = g = b = i;
} else {
av_assert1(c->srcFormat == AV_PIX_FMT_BGR4_BYTE);
b = ( i >> 3     ) * 255;
g = ((i >> 1) & 3) * 85;
r = ( i       & 1) * 255;
}
#define RGB2YUV_SHIFT 15
#define BY ( (int) (0.114 * 219 / 255 * (1 << RGB2YUV_SHIFT) + 0.5))
#define BV (-(int) (0.081 * 224 / 255 * (1 << RGB2YUV_SHIFT) + 0.5))
#define BU ( (int) (0.500 * 224 / 255 * (1 << RGB2YUV_SHIFT) + 0.5))
#define GY ( (int) (0.587 * 219 / 255 * (1 << RGB2YUV_SHIFT) + 0.5))
#define GV (-(int) (0.419 * 224 / 255 * (1 << RGB2YUV_SHIFT) + 0.5))
#define GU (-(int) (0.331 * 224 / 255 * (1 << RGB2YUV_SHIFT) + 0.5))
#define RY ( (int) (0.299 * 219 / 255 * (1 << RGB2YUV_SHIFT) + 0.5))
#define RV ( (int) (0.500 * 224 / 255 * (1 << RGB2YUV_SHIFT) + 0.5))
#define RU (-(int) (0.169 * 224 / 255 * (1 << RGB2YUV_SHIFT) + 0.5))

y = av_clip_uint8((RY * r + GY * g + BY * b + ( 33 << (RGB2YUV_SHIFT - 1))) >> RGB2YUV_SHIFT);
u = av_clip_uint8((RU * r + GU * g + BU * b + (257 << (RGB2YUV_SHIFT - 1))) >> RGB2YUV_SHIFT);
v = av_clip_uint8((RV * r + GV * g + BV * b + (257 << (RGB2YUV_SHIFT - 1))) >> RGB2YUV_SHIFT);
c->pal_yuv[i]= y + (u<<8) + (v<<16) + ((unsigned)a<<24);

switch (c->dstFormat) {
case AV_PIX_FMT_BGR32:
#if !HAVE_BIGENDIAN
case AV_PIX_FMT_RGB24:
#endif
c->pal_rgb[i]=  r + (g<<8) + (b<<16) + ((unsigned)a<<24);
break;
case AV_PIX_FMT_BGR32_1:
#if HAVE_BIGENDIAN
case AV_PIX_FMT_BGR24:
#endif
c->pal_rgb[i]= a + (r<<8) + (g<<16) + ((unsigned)b<<24);
break;
case AV_PIX_FMT_RGB32_1:
#if HAVE_BIGENDIAN
case AV_PIX_FMT_RGB24:
#endif
c->pal_rgb[i]= a + (b<<8) + (g<<16) + ((unsigned)r<<24);
break;
case AV_PIX_FMT_RGB32:
#if !HAVE_BIGENDIAN
case AV_PIX_FMT_BGR24:
#endif
default:
c->pal_rgb[i]=  b + (g<<8) + (r<<16) + ((unsigned)a<<24);
}
}
}
//Alpha的特殊处理?
if (c->src0Alpha && !c->dst0Alpha && isALPHA(c->dstFormat)) {
uint8_t *base;
int x,y;
rgb0_tmp = av_malloc(FFABS(srcStride[0]) * srcSliceH + 32);
if (!rgb0_tmp)
return AVERROR(ENOMEM);

base = srcStride[0] < 0 ? rgb0_tmp - srcStride[0] * (srcSliceH-1) : rgb0_tmp;
for (y=0; y<srcSliceH; y++){
memcpy(base + srcStride[0]*y, src2[0] + srcStride[0]*y, 4*c->srcW);
for (x=c->src0Alpha-1; x<4*c->srcW; x+=4) {
base[ srcStride[0]*y + x] = 0xFF;
}
}
src2[0] = base;
}
//XYZ的特殊处理?
if (c->srcXYZ && !(c->dstXYZ && c->srcW==c->dstW && c->srcH==c->dstH)) {
uint8_t *base;
rgb0_tmp = av_malloc(FFABS(srcStride[0]) * srcSliceH + 32);
if (!rgb0_tmp)
return AVERROR(ENOMEM);

base = srcStride[0] < 0 ? rgb0_tmp - srcStride[0] * (srcSliceH-1) : rgb0_tmp;

xyz12Torgb48(c, (uint16_t*)base, (const uint16_t*)src2[0], srcStride[0]/2, srcSliceH);
src2[0] = base;
}

if (!srcSliceY && (c->flags & SWS_BITEXACT) && c->dither == SWS_DITHER_ED && c->dither_error[0])
for (i = 0; i < 4; i++)
memset(c->dither_error[i], 0, sizeof(c->dither_error[0][0]) * (c->dstW+2));

// copy strides, so they can safely be modified
// sliceDir: 1 = top-to-bottom; -1 = bottom-to-top;
if (c->sliceDir == 1) {
// slices go from top to bottom
int srcStride2[4] = { srcStride[0], srcStride[1], srcStride[2],
srcStride[3] };
int dstStride2[4] = { dstStride[0], dstStride[1], dstStride[2],
dstStride[3] };

reset_ptr(src2, c->srcFormat);
reset_ptr((void*)dst2, c->dstFormat);

/* reset slice direction at end of frame */
if (srcSliceY + srcSliceH == c->srcH)
c->sliceDir = 0;
//关键：调用
ret = c->swscale(c, src2, srcStride2, srcSliceY, srcSliceH, dst2,
dstStride2);
} else {
// slices go from bottom to top => we flip the image internally
int srcStride2[4] = { -srcStride[0], -srcStride[1], -srcStride[2],
-srcStride[3] };
int dstStride2[4] = { -dstStride[0], -dstStride[1], -dstStride[2],
-dstStride[3] };

src2[0] += (srcSliceH - 1) * srcStride[0];
if (!usePal(c->srcFormat))
src2[1] += ((srcSliceH >> c->chrSrcVSubSample) - 1) * srcStride[1];
src2[2] += ((srcSliceH >> c->chrSrcVSubSample) - 1) * srcStride[2];
src2[3] += (srcSliceH - 1) * srcStride[3];
dst2[0] += ( c->dstH                         - 1) * dstStride[0];
dst2[1] += ((c->dstH >> c->chrDstVSubSample) - 1) * dstStride[1];
dst2[2] += ((c->dstH >> c->chrDstVSubSample) - 1) * dstStride[2];
dst2[3] += ( c->dstH                         - 1) * dstStride[3];

reset_ptr(src2, c->srcFormat);
reset_ptr((void*)dst2, c->dstFormat);

/* reset slice direction at end of frame */
if (!srcSliceY)
c->sliceDir = 0;
//关键：调用
ret = c->swscale(c, src2, srcStride2, c->srcH-srcSliceY-srcSliceH,
srcSliceH, dst2, dstStride2);
}

if (c->dstXYZ && !(c->srcXYZ && c->srcW==c->dstW && c->srcH==c->dstH)) {
/* replace on the same data */
rgb48Toxyz12(c, (uint16_t*)dst2[0], (const uint16_t*)dst2[0], dstStride[0]/2, ret);
}

av_free(rgb0_tmp);
return ret;
}

从sws_scale()的定义可以看出，它封装了SwsContext中的swscale()（注意这个函数中间没有“_”）。函数最重要的一句代码就是“c->swscale()”。除此之外，函数还做了一些增加“兼容性”的一些处理。函数的主要步骤如下所示。

1.检查输入的图像参数的合理性。

这一步骤首先检查输入输出的参数是否为空，然后通过调用check_image_pointers()检查输入输出图像的内存是否正确分配。check_image_pointers()的定义如下所示。

static int check_image_pointers(const uint8_t * const data[4], enum AVPixelFormat pix_fmt,
const int linesizes[4])
{
const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
int i;

for (i = 0; i < 4; i++) {
int plane = desc->comp[i].plane;
if (!data[plane] || !linesizes[plane])
return 0;
}

return 1;
}

从check_image_pointers()的定义可以看出，在特定像素格式前提下，如果该像素格式应该包含像素的分量为空，就返回0，否则返回1。
2.如果输入像素数据中使用了“调色板”（palette），则进行一些相应的处理。这一步通过函数usePal()来判定。usePal()的定义如下。

static av_always_inline int usePal(enum AVPixelFormat pix_fmt)
{
const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
av_assert0(desc);
return (desc->flags & AV_PIX_FMT_FLAG_PAL) || (desc->flags & AV_PIX_FMT_FLAG_PSEUDOPAL);
}

从定义可以看出该函数通过判定AVPixFmtDescriptor中的flag是否包含AV_PIX_FMT_FLAG_PAL来断定像素格式是否使用了“调色板”。
3.其它一些特殊格式的处理，比如说Alpha，XYZ等的处理（这方面没有研究过）。
4.如果输入的图像的扫描方式是从底部到顶部的（一般情况下是从顶部到底部），则将图像进行反转。
5.调用SwsContext中的swscale()。

SwsContext中的swscale()

swscale这个变量的类型是SwsFunc，实际上就是一个函数指针。它是整个类库的核心。当我们从外部调用swscale()函数的时候，实际上就是调用了SwsContext中的这个名称为swscale的变量（注意外部函数接口和这个内部函数指针的名字是一样的，但不是一回事）。
可以看一下SwsFunc这个类型的定义：

typedef int (*SwsFunc)(struct SwsContext *context, const uint8_t *src[],
int srcStride[], int srcSliceY, int srcSliceH,
uint8_t *dst[], int dstStride[]);

可以看出SwsFunc的定义的参数类型和libswscale类库外部接口函数swscale()的参数类型一模一样。
在libswscale中，该指针的指向可以分成2种情况：

在调用sws_getContext()初始化SwsContext的时候，会在其子函数sws_init_context()中对swscale指针进行赋值。如果图像没有进行拉伸，则会调用ff_get_unscaled_swscale()对其进行赋值；如果图像进行了拉伸，则会调用ff_getSwsFunc()对其进行赋值。下面分别看一下这2种情况。

没有拉伸--专有的像素转换函数

如果图像没有进行拉伸，则会调用ff_get_unscaled_swscale()对SwsContext的swscale进行赋值。上篇文章中记录了这个函数，在这里回顾一下。

ff_get_unscaled_swscale()

ff_get_unscaled_swscale()的定义如下。

void ff_get_unscaled_swscale(SwsContext *c)
{
const enum AVPixelFormat srcFormat = c->srcFormat;
const enum AVPixelFormat dstFormat = c->dstFormat;
const int flags = c->flags;
const int dstH = c->dstH;
int needsDither;

needsDither = isAnyRGB(dstFormat) &&
c->dstFormatBpp < 24 &&
(c->dstFormatBpp < c->srcFormatBpp || (!isAnyRGB(srcFormat)));

/* yv12_to_nv12 */
if ((srcFormat == AV_PIX_FMT_YUV420P || srcFormat == AV_PIX_FMT_YUVA420P) &&
(dstFormat == AV_PIX_FMT_NV12 || dstFormat == AV_PIX_FMT_NV21)) {
c->swscale = planarToNv12Wrapper;
}
/* nv12_to_yv12 */
if (dstFormat == AV_PIX_FMT_YUV420P &&
(srcFormat == AV_PIX_FMT_NV12 || srcFormat == AV_PIX_FMT_NV21)) {
c->swscale = nv12ToPlanarWrapper;
}
/* yuv2bgr */
if ((srcFormat == AV_PIX_FMT_YUV420P || srcFormat == AV_PIX_FMT_YUV422P ||
srcFormat == AV_PIX_FMT_YUVA420P) && isAnyRGB(dstFormat) &&
!(flags & SWS_ACCURATE_RND) && (c->dither == SWS_DITHER_BAYER || c->dither == SWS_DITHER_AUTO) && !(dstH & 1)) {
c->swscale = ff_yuv2rgb_get_func_ptr(c);
}

if (srcFormat == AV_PIX_FMT_YUV410P && !(dstH & 3) &&
(dstFormat == AV_PIX_FMT_YUV420P || dstFormat == AV_PIX_FMT_YUVA420P) &&
!(flags & SWS_BITEXACT)) {
c->swscale = yvu9ToYv12Wrapper;
}

/* bgr24toYV12 */
if (srcFormat == AV_PIX_FMT_BGR24 &&
(dstFormat == AV_PIX_FMT_YUV420P || dstFormat == AV_PIX_FMT_YUVA420P) &&
!(flags & SWS_ACCURATE_RND))
c->swscale = bgr24ToYv12Wrapper;

/* RGB/BGR -> RGB/BGR (no dither needed forms) */
if (isAnyRGB(srcFormat) && isAnyRGB(dstFormat) && findRgbConvFn(c)
&& (!needsDither || (c->flags&(SWS_FAST_BILINEAR|SWS_POINT))))
c->swscale = rgbToRgbWrapper;

if ((srcFormat == AV_PIX_FMT_GBRP && dstFormat == AV_PIX_FMT_GBRAP) ||
(srcFormat == AV_PIX_FMT_GBRAP && dstFormat == AV_PIX_FMT_GBRP))
c->swscale = planarRgbToplanarRgbWrapper;

#define isByteRGB(f) (             \
f == AV_PIX_FMT_RGB32   || \
f == AV_PIX_FMT_RGB32_1 || \
f == AV_PIX_FMT_RGB24   || \
f == AV_PIX_FMT_BGR32   || \
f == AV_PIX_FMT_BGR32_1 || \
f == AV_PIX_FMT_BGR24)

if (srcFormat == AV_PIX_FMT_GBRP && isPlanar(srcFormat) && isByteRGB(dstFormat))
c->swscale = planarRgbToRgbWrapper;

if ((srcFormat == AV_PIX_FMT_RGB48LE  || srcFormat == AV_PIX_FMT_RGB48BE  ||
srcFormat == AV_PIX_FMT_BGR48LE  || srcFormat == AV_PIX_FMT_BGR48BE  ||
srcFormat == AV_PIX_FMT_RGBA64LE || srcFormat == AV_PIX_FMT_RGBA64BE ||
srcFormat == AV_PIX_FMT_BGRA64LE || srcFormat == AV_PIX_FMT_BGRA64BE) &&
(dstFormat == AV_PIX_FMT_GBRP9LE  || dstFormat == AV_PIX_FMT_GBRP9BE  ||
dstFormat == AV_PIX_FMT_GBRP10LE || dstFormat == AV_PIX_FMT_GBRP10BE ||
dstFormat == AV_PIX_FMT_GBRP12LE || dstFormat == AV_PIX_FMT_GBRP12BE ||
dstFormat == AV_PIX_FMT_GBRP14LE || dstFormat == AV_PIX_FMT_GBRP14BE ||
dstFormat == AV_PIX_FMT_GBRP16LE || dstFormat == AV_PIX_FMT_GBRP16BE ||
dstFormat == AV_PIX_FMT_GBRAP16LE || dstFormat == AV_PIX_FMT_GBRAP16BE ))
c->swscale = Rgb16ToPlanarRgb16Wrapper;

if ((srcFormat == AV_PIX_FMT_GBRP9LE  || srcFormat == AV_PIX_FMT_GBRP9BE  ||
srcFormat == AV_PIX_FMT_GBRP16LE || srcFormat == AV_PIX_FMT_GBRP16BE ||
srcFormat == AV_PIX_FMT_GBRP10LE || srcFormat == AV_PIX_FMT_GBRP10BE ||
srcFormat == AV_PIX_FMT_GBRP12LE || srcFormat == AV_PIX_FMT_GBRP12BE ||
srcFormat == AV_PIX_FMT_GBRP14LE || srcFormat == AV_PIX_FMT_GBRP14BE ||
srcFormat == AV_PIX_FMT_GBRAP16LE || srcFormat == AV_PIX_FMT_GBRAP16BE) &&
(dstFormat == AV_PIX_FMT_RGB48LE  || dstFormat == AV_PIX_FMT_RGB48BE  ||
dstFormat == AV_PIX_FMT_BGR48LE  || dstFormat == AV_PIX_FMT_BGR48BE  ||
dstFormat == AV_PIX_FMT_RGBA64LE || dstFormat == AV_PIX_FMT_RGBA64BE ||
dstFormat == AV_PIX_FMT_BGRA64LE || dstFormat == AV_PIX_FMT_BGRA64BE))
c->swscale = planarRgb16ToRgb16Wrapper;

if (av_pix_fmt_desc_get(srcFormat)->comp[0].depth_minus1 == 7 &&
isPackedRGB(srcFormat) && dstFormat == AV_PIX_FMT_GBRP)
c->swscale = rgbToPlanarRgbWrapper;

if (isBayer(srcFormat)) {
if (dstFormat == AV_PIX_FMT_RGB24)
c->swscale = bayer_to_rgb24_wrapper;
else if (dstFormat == AV_PIX_FMT_YUV420P)
c->swscale = bayer_to_yv12_wrapper;
else if (!isBayer(dstFormat)) {
av_log(c, AV_LOG_ERROR, "unsupported bayer conversion\n");
av_assert0(0);
}
}

/* bswap 16 bits per pixel/component packed formats */
if (IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_BAYER_BGGR16) ||
IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_BAYER_RGGB16) ||
IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_BAYER_GBRG16) ||
IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_BAYER_GRBG16) ||
IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_BGR444) ||
IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_BGR48)  ||
IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_BGRA64) ||
IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_BGR555) ||
IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_BGR565) ||
IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_BGRA64) ||
IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_GRAY16) ||
IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_YA16)   ||
IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_GBRP9)  ||
IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_GBRP10) ||
IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_GBRP12) ||
IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_GBRP14) ||
IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_GBRP16) ||
IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_GBRAP16) ||
IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_RGB444) ||
IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_RGB48)  ||
IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_RGBA64) ||
IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_RGB555) ||
IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_RGB565) ||
IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_RGBA64) ||
IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_XYZ12)  ||
IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_YUV420P9)  ||
IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_YUV420P10) ||
IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_YUV420P12) ||
IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_YUV420P14) ||
IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_YUV420P16) ||
IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_YUV422P9)  ||
IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_YUV422P10) ||
IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_YUV422P12) ||
IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_YUV422P14) ||
IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_YUV422P16) ||
IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_YUV444P9)  ||
IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_YUV444P10) ||
IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_YUV444P12) ||
IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_YUV444P14) ||
IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_YUV444P16))
c->swscale = packed_16bpc_bswap;

if (usePal(srcFormat) && isByteRGB(dstFormat))
c->swscale = palToRgbWrapper;

if (srcFormat == AV_PIX_FMT_YUV422P) {
if (dstFormat == AV_PIX_FMT_YUYV422)
c->swscale = yuv422pToYuy2Wrapper;
else if (dstFormat == AV_PIX_FMT_UYVY422)
c->swscale = yuv422pToUyvyWrapper;
}

/* LQ converters if -sws 0 or -sws 4*/
if (c->flags&(SWS_FAST_BILINEAR|SWS_POINT)) {
/* yv12_to_yuy2 */
if (srcFormat == AV_PIX_FMT_YUV420P || srcFormat == AV_PIX_FMT_YUVA420P) {
if (dstFormat == AV_PIX_FMT_YUYV422)
c->swscale = planarToYuy2Wrapper;
else if (dstFormat == AV_PIX_FMT_UYVY422)
c->swscale = planarToUyvyWrapper;
}
}
if (srcFormat == AV_PIX_FMT_YUYV422 &&
(dstFormat == AV_PIX_FMT_YUV420P || dstFormat == AV_PIX_FMT_YUVA420P))
c->swscale = yuyvToYuv420Wrapper;
if (srcFormat == AV_PIX_FMT_UYVY422 &&
(dstFormat == AV_PIX_FMT_YUV420P || dstFormat == AV_PIX_FMT_YUVA420P))
c->swscale = uyvyToYuv420Wrapper;
if (srcFormat == AV_PIX_FMT_YUYV422 && dstFormat == AV_PIX_FMT_YUV422P)
c->swscale = yuyvToYuv422Wrapper;
if (srcFormat == AV_PIX_FMT_UYVY422 && dstFormat == AV_PIX_FMT_YUV422P)
c->swscale = uyvyToYuv422Wrapper;

#define isPlanarGray(x) (isGray(x) && (x) != AV_PIX_FMT_YA8 && (x) != AV_PIX_FMT_YA16LE && (x) != AV_PIX_FMT_YA16BE)
/* simple copy */
if ( srcFormat == dstFormat ||
(srcFormat == AV_PIX_FMT_YUVA420P && dstFormat == AV_PIX_FMT_YUV420P) ||
(srcFormat == AV_PIX_FMT_YUV420P && dstFormat == AV_PIX_FMT_YUVA420P) ||
(isPlanarYUV(srcFormat) && isPlanarGray(dstFormat)) ||
(isPlanarYUV(dstFormat) && isPlanarGray(srcFormat)) ||
(isPlanarGray(dstFormat) && isPlanarGray(srcFormat)) ||
(isPlanarYUV(srcFormat) && isPlanarYUV(dstFormat) &&
c->chrDstHSubSample == c->chrSrcHSubSample &&
c->chrDstVSubSample == c->chrSrcVSubSample &&
dstFormat != AV_PIX_FMT_NV12 && dstFormat != AV_PIX_FMT_NV21 &&
srcFormat != AV_PIX_FMT_NV12 && srcFormat != AV_PIX_FMT_NV21))
{
if (isPacked(c->srcFormat))
c->swscale = packedCopyWrapper;
else /* Planar YUV or gray */
c->swscale = planarCopyWrapper;
}

if (ARCH_PPC)
ff_get_unscaled_swscale_ppc(c);
//     if (ARCH_ARM)
//         ff_get_unscaled_swscale_arm(c);
}

从代码中可以看出，它根据输入输出像素格式的不同，选择了不同的转换函数。例如YUV420P转换NV12的时候，就会将planarToNv12Wrapper()赋值给SwsContext的swscale指针。

有拉伸--swscale()

如果图像进行了拉伸，则会调用ff_getSwsFunc()对SwsContext的swscale进行赋值。上篇文章中记录了这个函数，在这里回顾一下。

SwsFunc ff_getSwsFunc(SwsContext *c)
{
sws_init_swscale(c);

if (ARCH_PPC)
ff_sws_init_swscale_ppc(c);
if (ARCH_X86)
ff_sws_init_swscale_x86(c);

return swscale;
}

注意，sws_init_context()对SwsContext的swscale进行赋值的语句是：

c->swscale = ff_getSwsFunc(c);

即把ff_getSwsFunc()的返回值赋值给SwsContext的swscale指针；而ff_getSwsFunc()的返回值是一个静态函数，名称就叫做“swscale”。
下面我们看一下这个swscale()静态函数的定义。

static int swscale(SwsContext *c, const uint8_t *src[],
int srcStride[], int srcSliceY,
int srcSliceH, uint8_t *dst[], int dstStride[])
{
/* load a few things into local vars to make the code more readable?
* and faster */
//注意一下这些参数
//以亮度为准
const int srcW                   = c->srcW;
const int dstW                   = c->dstW;
const int dstH                   = c->dstH;
//以色度为准
const int chrDstW                = c->chrDstW;
const int chrSrcW                = c->chrSrcW;
const int lumXInc                = c->lumXInc;
const int chrXInc                = c->chrXInc;
const enum AVPixelFormat dstFormat = c->dstFormat;
const int flags                  = c->flags;
int32_t *vLumFilterPos           = c->vLumFilterPos;
int32_t *vChrFilterPos           = c->vChrFilterPos;
int32_t *hLumFilterPos           = c->hLumFilterPos;
int32_t *hChrFilterPos           = c->hChrFilterPos;
int16_t *hLumFilter              = c->hLumFilter;
int16_t *hChrFilter              = c->hChrFilter;
int32_t *lumMmxFilter            = c->lumMmxFilter;
int32_t *chrMmxFilter            = c->chrMmxFilter;
const int vLumFilterSize         = c->vLumFilterSize;
const int vChrFilterSize         = c->vChrFilterSize;
const int hLumFilterSize         = c->hLumFilterSize;
const int hChrFilterSize         = c->hChrFilterSize;
int16_t **lumPixBuf              = c->lumPixBuf;
int16_t **chrUPixBuf             = c->chrUPixBuf;
int16_t **chrVPixBuf             = c->chrVPixBuf;
int16_t **alpPixBuf              = c->alpPixBuf;
const int vLumBufSize            = c->vLumBufSize;
const int vChrBufSize            = c->vChrBufSize;
uint8_t *formatConvBuffer        = c->formatConvBuffer;
uint32_t *pal                    = c->pal_yuv;
yuv2planar1_fn yuv2plane1        = c->yuv2plane1;
yuv2planarX_fn yuv2planeX        = c->yuv2planeX;
yuv2interleavedX_fn yuv2nv12cX   = c->yuv2nv12cX;
yuv2packed1_fn yuv2packed1       = c->yuv2packed1;
yuv2packed2_fn yuv2packed2       = c->yuv2packed2;
yuv2packedX_fn yuv2packedX       = c->yuv2packedX;
yuv2anyX_fn yuv2anyX             = c->yuv2anyX;
const int chrSrcSliceY           =                srcSliceY >> c->chrSrcVSubSample;
const int chrSrcSliceH           = FF_CEIL_RSHIFT(srcSliceH,   c->chrSrcVSubSample);
int should_dither                = is9_OR_10BPS(c->srcFormat) ||
is16BPS(c->srcFormat);
int lastDstY;

/* vars which will change and which we need to store back in the context */
int dstY         = c->dstY;
int lumBufIndex  = c->lumBufIndex;
int chrBufIndex  = c->chrBufIndex;
int lastInLumBuf = c->lastInLumBuf;
int lastInChrBuf = c->lastInChrBuf;

if (!usePal(c->srcFormat)) {
pal = c->input_rgb2yuv_table;
}

if (isPacked(c->srcFormat)) {
src[0] =
src[1] =
src[2] =
src[3] = src[0];
srcStride[0] =
srcStride[1] =
srcStride[2] =
srcStride[3] = srcStride[0];
}
srcStride[1] <<= c->vChrDrop;
srcStride[2] <<= c->vChrDrop;

DEBUG_BUFFERS("swscale() %p[%d] %p[%d] %p[%d] %p[%d] -> %p[%d] %p[%d] %p[%d] %p[%d]\n",
src[0], srcStride[0], src[1], srcStride[1],
src[2], srcStride[2], src[3], srcStride[3],
dst[0], dstStride[0], dst[1], dstStride[1],
dst[2], dstStride[2], dst[3], dstStride[3]);
DEBUG_BUFFERS("srcSliceY: %d srcSliceH: %d dstY: %d dstH: %d\n",
srcSliceY, srcSliceH, dstY, dstH);
DEBUG_BUFFERS("vLumFilterSize: %d vLumBufSize: %d vChrFilterSize: %d vChrBufSize: %d\n",
vLumFilterSize, vLumBufSize, vChrFilterSize, vChrBufSize);

if (dstStride[0]&15 || dstStride[1]&15 ||
dstStride[2]&15 || dstStride[3]&15) {
static int warnedAlready = 0; // FIXME maybe move this into the context
if (flags & SWS_PRINT_INFO && !warnedAlready) {
av_log(c, AV_LOG_WARNING,
"Warning: dstStride is not aligned!\n"
"         ->cannot do aligned memory accesses anymore\n");
warnedAlready = 1;
}
}

if (   (uintptr_t)dst[0]&15 || (uintptr_t)dst[1]&15 || (uintptr_t)dst[2]&15
|| (uintptr_t)src[0]&15 || (uintptr_t)src[1]&15 || (uintptr_t)src[2]&15
|| dstStride[0]&15 || dstStride[1]&15 || dstStride[2]&15 || dstStride[3]&15
|| srcStride[0]&15 || srcStride[1]&15 || srcStride[2]&15 || srcStride[3]&15
) {
static int warnedAlready=0;
int cpu_flags = av_get_cpu_flags();
if (HAVE_MMXEXT && (cpu_flags & AV_CPU_FLAG_SSE2) && !warnedAlready){
av_log(c, AV_LOG_WARNING, "Warning: data is not aligned! This can lead to a speedloss\n");
warnedAlready=1;
}
}

/* Note the user might start scaling the picture in the middle so this
* will not get executed. This is not really intended but works
* currently, so people might do it. */
if (srcSliceY == 0) {
lumBufIndex  = -1;
chrBufIndex  = -1;
dstY         = 0;
lastInLumBuf = -1;
lastInChrBuf = -1;
}

if (!should_dither) {
c->chrDither8 = c->lumDither8 = sws_pb_64;
}
lastDstY = dstY;
//逐行循环，一次循环代表处理一行
//注意dstY和dstH两个变量
for (; dstY < dstH; dstY++) {
//色度的和亮度之间的关系
const int chrDstY = dstY >> c->chrDstVSubSample;
uint8_t *dest[4]  = {
dst[0] + dstStride[0] * dstY,
dst[1] + dstStride[1] * chrDstY,
dst[2] + dstStride[2] * chrDstY,
(CONFIG_SWSCALE_ALPHA && alpPixBuf) ? dst[3] + dstStride[3] * dstY : NULL,
};
int use_mmx_vfilter= c->use_mmx_vfilter;

// First line needed as input
const int firstLumSrcY  = FFMAX(1 - vLumFilterSize, vLumFilterPos[dstY]);
const int firstLumSrcY2 = FFMAX(1 - vLumFilterSize, vLumFilterPos[FFMIN(dstY | ((1 << c->chrDstVSubSample) - 1), dstH - 1)]);
// First line needed as input
const int firstChrSrcY  = FFMAX(1 - vChrFilterSize, vChrFilterPos[chrDstY]);

// Last line needed as input
int lastLumSrcY  = FFMIN(c->srcH,    firstLumSrcY  + vLumFilterSize) - 1;
int lastLumSrcY2 = FFMIN(c->srcH,    firstLumSrcY2 + vLumFilterSize) - 1;
int lastChrSrcY  = FFMIN(c->chrSrcH, firstChrSrcY  + vChrFilterSize) - 1;
int enough_lines;

// handle holes (FAST_BILINEAR & weird filters)
if (firstLumSrcY > lastInLumBuf)
lastInLumBuf = firstLumSrcY - 1;
if (firstChrSrcY > lastInChrBuf)
lastInChrBuf = firstChrSrcY - 1;
av_assert0(firstLumSrcY >= lastInLumBuf - vLumBufSize + 1);
av_assert0(firstChrSrcY >= lastInChrBuf - vChrBufSize + 1);

DEBUG_BUFFERS("dstY: %d\n", dstY);
DEBUG_BUFFERS("\tfirstLumSrcY: %d lastLumSrcY: %d lastInLumBuf: %d\n",
firstLumSrcY, lastLumSrcY, lastInLumBuf);
DEBUG_BUFFERS("\tfirstChrSrcY: %d lastChrSrcY: %d lastInChrBuf: %d\n",
firstChrSrcY, lastChrSrcY, lastInChrBuf);

// Do we have enough lines in this slice to output the dstY line
enough_lines = lastLumSrcY2 < srcSliceY + srcSliceH &&
lastChrSrcY < FF_CEIL_RSHIFT(srcSliceY + srcSliceH, c->chrSrcVSubSample);

if (!enough_lines) {
lastLumSrcY = srcSliceY + srcSliceH - 1;
lastChrSrcY = chrSrcSliceY + chrSrcSliceH - 1;
DEBUG_BUFFERS("buffering slice: lastLumSrcY %d lastChrSrcY %d\n",
lastLumSrcY, lastChrSrcY);
}

// Do horizontal scaling
//水平拉伸
//亮度
while (lastInLumBuf < lastLumSrcY) {
const uint8_t *src1[4] = {
src[0] + (lastInLumBuf + 1 - srcSliceY) * srcStride[0],
src[1] + (lastInLumBuf + 1 - srcSliceY) * srcStride[1],
src[2] + (lastInLumBuf + 1 - srcSliceY) * srcStride[2],
src[3] + (lastInLumBuf + 1 - srcSliceY) * srcStride[3],
};
lumBufIndex++;
av_assert0(lumBufIndex < 2 * vLumBufSize);
av_assert0(lastInLumBuf + 1 - srcSliceY < srcSliceH);
av_assert0(lastInLumBuf + 1 - srcSliceY >= 0);
//关键：拉伸
hyscale(c, lumPixBuf[lumBufIndex], dstW, src1, srcW, lumXInc,
hLumFilter, hLumFilterPos, hLumFilterSize,
formatConvBuffer, pal, 0);
if (CONFIG_SWSCALE_ALPHA && alpPixBuf)
hyscale(c, alpPixBuf[lumBufIndex], dstW, src1, srcW,
lumXInc, hLumFilter, hLumFilterPos, hLumFilterSize,
formatConvBuffer, pal, 1);
lastInLumBuf++;
DEBUG_BUFFERS("\t\tlumBufIndex %d: lastInLumBuf: %d\n",
lumBufIndex, lastInLumBuf);
}
//水平拉伸
//色度
while (lastInChrBuf < lastChrSrcY) {
const uint8_t *src1[4] = {
src[0] + (lastInChrBuf + 1 - chrSrcSliceY) * srcStride[0],
src[1] + (lastInChrBuf + 1 - chrSrcSliceY) * srcStride[1],
src[2] + (lastInChrBuf + 1 - chrSrcSliceY) * srcStride[2],
src[3] + (lastInChrBuf + 1 - chrSrcSliceY) * srcStride[3],
};
chrBufIndex++;
av_assert0(chrBufIndex < 2 * vChrBufSize);
av_assert0(lastInChrBuf + 1 - chrSrcSliceY < (chrSrcSliceH));
av_assert0(lastInChrBuf + 1 - chrSrcSliceY >= 0);
// FIXME replace parameters through context struct (some at least)
//关键：拉伸
if (c->needs_hcscale)
hcscale(c, chrUPixBuf[chrBufIndex], chrVPixBuf[chrBufIndex],
chrDstW, src1, chrSrcW, chrXInc,
hChrFilter, hChrFilterPos, hChrFilterSize,
formatConvBuffer, pal);
lastInChrBuf++;
DEBUG_BUFFERS("\t\tchrBufIndex %d: lastInChrBuf: %d\n",
chrBufIndex, lastInChrBuf);
}
// wrap buf index around to stay inside the ring buffer
if (lumBufIndex >= vLumBufSize)
lumBufIndex -= vLumBufSize;
if (chrBufIndex >= vChrBufSize)
chrBufIndex -= vChrBufSize;
if (!enough_lines)
break;  // we can't output a dstY line so let's try with the next slice

#if HAVE_MMX_INLINE
updateMMXDitherTables(c, dstY, lumBufIndex, chrBufIndex,
lastInLumBuf, lastInChrBuf);
#endif
if (should_dither) {
c->chrDither8 = ff_dither_8x8_128[chrDstY & 7];
c->lumDither8 = ff_dither_8x8_128[dstY    & 7];
}
if (dstY >= dstH - 2) {
/* hmm looks like we can't use MMX here without overwriting
* this array's tail */
ff_sws_init_output_funcs(c, &yuv2plane1, &yuv2planeX, &yuv2nv12cX,
&yuv2packed1, &yuv2packed2, &yuv2packedX, &yuv2anyX);
use_mmx_vfilter= 0;
}

{
const int16_t **lumSrcPtr  = (const int16_t **)(void*) lumPixBuf  + lumBufIndex + firstLumSrcY - lastInLumBuf + vLumBufSize;
const int16_t **chrUSrcPtr = (const int16_t **)(void*) chrUPixBuf + chrBufIndex + firstChrSrcY - lastInChrBuf + vChrBufSize;
const int16_t **chrVSrcPtr = (const int16_t **)(void*) chrVPixBuf + chrBufIndex + firstChrSrcY - lastInChrBuf + vChrBufSize;
const int16_t **alpSrcPtr  = (CONFIG_SWSCALE_ALPHA && alpPixBuf) ?
(const int16_t **)(void*) alpPixBuf + lumBufIndex + firstLumSrcY - lastInLumBuf + vLumBufSize : NULL;
int16_t *vLumFilter = c->vLumFilter;
int16_t *vChrFilter = c->vChrFilter;

if (isPlanarYUV(dstFormat) ||
(isGray(dstFormat) && !isALPHA(dstFormat))) { // YV12 like
const int chrSkipMask = (1 << c->chrDstVSubSample) - 1;

vLumFilter +=    dstY * vLumFilterSize;
vChrFilter += chrDstY * vChrFilterSize;

//                 av_assert0(use_mmx_vfilter != (
//                                yuv2planeX == yuv2planeX_10BE_c
//                             || yuv2planeX == yuv2planeX_10LE_c
//                             || yuv2planeX == yuv2planeX_9BE_c
//                             || yuv2planeX == yuv2planeX_9LE_c
//                             || yuv2planeX == yuv2planeX_16BE_c
//                             || yuv2planeX == yuv2planeX_16LE_c
//                             || yuv2planeX == yuv2planeX_8_c) || !ARCH_X86);

if(use_mmx_vfilter){
vLumFilter= (int16_t *)c->lumMmxFilter;
vChrFilter= (int16_t *)c->chrMmxFilter;
}
//输出一行水平拉伸过的像素
//亮度
//是否垂直拉伸？
if (vLumFilterSize == 1) {
//亮度-不垂直拉伸-分量模式（planar）-输出一行水平拉伸的像素
yuv2plane1(lumSrcPtr[0], dest[0], dstW, c->lumDither8, 0);
} else {
//亮度-垂直拉伸-分量模式（planar）-输出一行水平拉伸的像素
yuv2planeX(vLumFilter, vLumFilterSize,
lumSrcPtr, dest[0],
dstW, c->lumDither8, 0);
}
//色度
//是否垂直拉伸？
if (!((dstY & chrSkipMask) || isGray(dstFormat))) {
if (yuv2nv12cX) {
yuv2nv12cX(c, vChrFilter,
vChrFilterSize, chrUSrcPtr, chrVSrcPtr,
dest[1], chrDstW);
} else if (vChrFilterSize == 1) {
//色度-不垂直拉伸-分量模式（planar）-输出一行水平拉伸的像素
//注意是2个分量
yuv2plane1(chrUSrcPtr[0], dest[1], chrDstW, c->chrDither8, 0);
yuv2plane1(chrVSrcPtr[0], dest[2], chrDstW, c->chrDither8, 3);
} else {
//色度-垂直拉伸-分量模式（planar）-输出一行水平拉伸的像素
//注意是2个分量
yuv2planeX(vChrFilter,
vChrFilterSize, chrUSrcPtr, dest[1],
chrDstW, c->chrDither8, 0);
yuv2planeX(vChrFilter,
vChrFilterSize, chrVSrcPtr, dest[2],
chrDstW, c->chrDither8, use_mmx_vfilter ? (c->uv_offx2 >> 1) : 3);
}
}

if (CONFIG_SWSCALE_ALPHA && alpPixBuf) {
if(use_mmx_vfilter){
vLumFilter= (int16_t *)c->alpMmxFilter;
}
if (vLumFilterSize == 1) {
yuv2plane1(alpSrcPtr[0], dest[3], dstW,
c->lumDither8, 0);
} else {
yuv2planeX(vLumFilter,
vLumFilterSize, alpSrcPtr, dest[3],
dstW, c->lumDither8, 0);
}
}
} else if (yuv2packedX) {
av_assert1(lumSrcPtr  + vLumFilterSize - 1 < (const int16_t **)lumPixBuf  + vLumBufSize * 2);
av_assert1(chrUSrcPtr + vChrFilterSize - 1 < (const int16_t **)chrUPixBuf + vChrBufSize * 2);
if (c->yuv2packed1 && vLumFilterSize == 1 &&
vChrFilterSize <= 2) { // unscaled RGB
int chrAlpha = vChrFilterSize == 1 ? 0 : vChrFilter[2 * dstY + 1];
//不垂直拉伸-打包模式（packed）-输出一行水平拉伸的像素
yuv2packed1(c, *lumSrcPtr, chrUSrcPtr, chrVSrcPtr,
alpPixBuf ? *alpSrcPtr : NULL,
dest[0], dstW, chrAlpha, dstY);
} else if (c->yuv2packed2 && vLumFilterSize == 2 &&
vChrFilterSize == 2) { // bilinear upscale RGB
int lumAlpha = vLumFilter[2 * dstY + 1];
int chrAlpha = vChrFilter[2 * dstY + 1];
lumMmxFilter[2] =
lumMmxFilter[3] = vLumFilter[2 * dstY]    * 0x10001;
chrMmxFilter[2] =
chrMmxFilter[3] = vChrFilter[2 * chrDstY] * 0x10001;
yuv2packed2(c, lumSrcPtr, chrUSrcPtr, chrVSrcPtr,
alpPixBuf ? alpSrcPtr : NULL,
dest[0], dstW, lumAlpha, chrAlpha, dstY);
} else { // general RGB
//垂直拉伸-打包模式（packed）-输出一行水平拉伸的像素
yuv2packedX(c, vLumFilter + dstY * vLumFilterSize,
lumSrcPtr, vLumFilterSize,
vChrFilter + dstY * vChrFilterSize,
chrUSrcPtr, chrVSrcPtr, vChrFilterSize,
alpSrcPtr, dest[0], dstW, dstY);
}
} else {
av_assert1(!yuv2packed1 && !yuv2packed2);
yuv2anyX(c, vLumFilter + dstY * vLumFilterSize,
lumSrcPtr, vLumFilterSize,
vChrFilter + dstY * vChrFilterSize,
chrUSrcPtr, chrVSrcPtr, vChrFilterSize,
alpSrcPtr, dest, dstW, dstY);
}
}
}
if (isPlanar(dstFormat) && isALPHA(dstFormat) && !alpPixBuf) {
int length = dstW;
int height = dstY - lastDstY;

if (is16BPS(dstFormat) || isNBPS(dstFormat)) {
const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(dstFormat);
fillPlane16(dst[3], dstStride[3], length, height, lastDstY,
1, desc->comp[3].depth_minus1,
isBE(dstFormat));
} else
fillPlane(dst[3], dstStride[3], length, height, lastDstY, 255);
}

#if HAVE_MMXEXT_INLINE
if (av_get_cpu_flags() & AV_CPU_FLAG_MMXEXT)
__asm__ volatile ("sfence" ::: "memory");
#endif
emms_c();

/* store changed local vars back in the context */
c->dstY         = dstY;
c->lumBufIndex  = lumBufIndex;
c->chrBufIndex  = chrBufIndex;
c->lastInLumBuf = lastInLumBuf;
c->lastInChrBuf = lastInChrBuf;

return dstY - lastDstY;
}

可以看出swscale()是一行一行的进行图像缩放工作的。其中每行数据的处理按照“先水平拉伸，然后垂直拉伸”的方式进行处理。具体的实现函数如下所示：

1. 水平拉伸
2. 垂直拉伸

下面具体看看这几个函数的定义。

hyscale()

水平亮度拉伸函数hyscale()的定义位于libswscale\swscale.c，如下所示。

// *** horizontal scale Y line to temp buffer
static av_always_inline void hyscale(SwsContext *c, int16_t *dst, int dstWidth,
const uint8_t *src_in[4],
int srcW, int xInc,
const int16_t *hLumFilter,
const int32_t *hLumFilterPos,
int hLumFilterSize,
uint8_t *formatConvBuffer,
uint32_t *pal, int isAlpha)
{
void (*toYV12)(uint8_t *, const uint8_t *, const uint8_t *, const uint8_t *, int, uint32_t *) =
isAlpha ? c->alpToYV12 : c->lumToYV12;
void (*convertRange)(int16_t *, int) = isAlpha ? NULL : c->lumConvertRange;
const uint8_t *src = src_in[isAlpha ? 3 : 0];

if (toYV12) {
toYV12(formatConvBuffer, src, src_in[1], src_in[2], srcW, pal);
src = formatConvBuffer;
} else if (c->readLumPlanar && !isAlpha) {
//读取
c->readLumPlanar(formatConvBuffer, src_in, srcW, c->input_rgb2yuv_table);
//赋值
src = formatConvBuffer;
} else if (c->readAlpPlanar && isAlpha) {
c->readAlpPlanar(formatConvBuffer, src_in, srcW, NULL);
src = formatConvBuffer;
}

if (!c->hyscale_fast) {
//亮度-水平拉伸
c->hyScale(c, dst, dstWidth, src, hLumFilter,
hLumFilterPos, hLumFilterSize);
} else { // fast bilinear upscale / crap downscale
c->hyscale_fast(c, dst, dstWidth, src, srcW, xInc);
}
//如果需要取值范围的转换（0-255和16-235之间）
if (convertRange)
convertRange(dst, dstWidth);
}

从hyscale()的源代码可以看出，它的流程如下所示。
1.转换成Y（亮度）
如果SwsContext的toYV12()函数存在，调用用该函数将数据转换为Y。如果该函数不存在，则调用SwsContext的readLumPlanar()读取Y。
2.拉伸
拉伸通过SwsContext的hyScale ()函数完成。如果存在hyscale_fast()方法的话，系统会优先调用hyscale_fast()。
3.转换范围（如果需要的话）
如果需要转换亮度的取值范围（例如需要进行16-235的MPEG标准与0-255的JPEG标准之间的转换），则会调用SwsContext的lumConvertRange ()函数。
上述几个步骤的涉及到的函数在上一篇文章中几经介绍过了，在这里重复一下。

toYV12() [SwsContext ->lumToYV12()]

toYV12()的实现函数是在ff_sws_init_input_funcs()中初始化的。在这里举几种具体的输入像素格式。

输入格式为YUYV422/ YVYU422
ff_sws_init_input_funcs()中，输入像素格式为YUYV422/ YVYU422的时候，toYV12()指向yuy2ToY_c()函数。源代码如下所示。

case AV_PIX_FMT_YUYV422:
case AV_PIX_FMT_YVYU422:
case AV_PIX_FMT_YA8:
c->lumToYV12 = yuy2ToY_c;
break;

yuy2ToY_c()的定义如下所示。

static void yuy2ToY_c(uint8_t *dst, const uint8_t *src, const uint8_t *unused1, const uint8_t *unused2,  int width,
uint32_t *unused)
{
int i;
for (i = 0; i < width; i++)
dst[i] = src[2 * i];
}

从yuy2ToY_c()的定义可以看出，该函数取出了所有的Y值（Y值在src[]数组中的下标为偶数）。

输入格式为RGB24
ff_sws_init_input_funcs()中，输入像素格式为RGB24的时候，toYV12()指向yuy2ToY_c()函数。源代码如下所示。

case AV_PIX_FMT_RGB24:
c->lumToYV12 = rgb24ToY_c;
break;

rgb24ToY_c()的定义如下所示。

static void rgb24ToY_c(uint8_t *_dst, const uint8_t *src, const uint8_t *unused1, const uint8_t *unused2, int width,
uint32_t *rgb2yuv)
{
int16_t *dst = (int16_t *)_dst;
int32_t ry = rgb2yuv[RY_IDX], gy = rgb2yuv[GY_IDX], by = rgb2yuv[BY_IDX];
int i;
for (i = 0; i < width; i++) {
int r = src[i * 3 + 0];
int g = src[i * 3 + 1];
int b = src[i * 3 + 2];

dst[i] = ((ry*r + gy*g + by*b + (32<<(RGB2YUV_SHIFT-1)) + (1<<(RGB2YUV_SHIFT-7)))>>(RGB2YUV_SHIFT-6));
}
}

从rgb24ToY_c()的定义可以看出，该函数通过R、G、B三个元素计算Y的值。其中R、G、B的系数取自于数组rgb2yuv[]（这个地方还没有研究）；RGB2YUV_SHIFT似乎代表了转换后YUV的位数，取值为15（这个地方也还没有深入看）。

SwsContext -> hyScale ()

SwsContext -> hyScale ()的实现函数是在sws_init_swscale ()中初始化的。可以回顾一下sws_init_swscale ()的定义，如下所示。

static av_cold void sws_init_swscale(SwsContext *c)
{
enum AVPixelFormat srcFormat = c->srcFormat;

ff_sws_init_output_funcs(c, &c->yuv2plane1, &c->yuv2planeX,
&c->yuv2nv12cX, &c->yuv2packed1,
&c->yuv2packed2, &c->yuv2packedX, &c->yuv2anyX);

ff_sws_init_input_funcs(c);

if (c->srcBpc == 8) {
if (c->dstBpc <= 14) {
c->hyScale = c->hcScale = hScale8To15_c;
if (c->flags & SWS_FAST_BILINEAR) {
c->hyscale_fast = ff_hyscale_fast_c;
c->hcscale_fast = ff_hcscale_fast_c;
}
} else {
c->hyScale = c->hcScale = hScale8To19_c;
}
} else {
c->hyScale = c->hcScale = c->dstBpc > 14 ? hScale16To19_c
: hScale16To15_c;
}

ff_sws_init_range_convert(c);

if (!(isGray(srcFormat) || isGray(c->dstFormat) ||
srcFormat == AV_PIX_FMT_MONOBLACK || srcFormat == AV_PIX_FMT_MONOWHITE))
c->needs_hcscale = 1;
}

从sws_init_swscale ()的定义可以看出，ff_sws_init_input_funcs()和ff_sws_init_range_convert()之间的代码完成了hyScale()的初始化。根据srcBpc和dstBpc取值的不同，有几种不同的拉伸函数。根据我的理解，srcBpc代表了输入的每个像素单个分量的位数，dstBpc代表了输出的每个像素单个分量的位数。最常见的像素单个分量的位数是8位。从代码中可以看出，在输入像素单个分量的位数为8位，而且输出像素单个分量的位数也为8位的时候，SwsContext 的 hyScale ()会指向hScale8To15_c()函数。

hScale8To15_c()

hScale8To15_c()的定义如下所示。有关这个方面的代码还没有详细研究，日后再作补充。

// bilinear / bicubic scaling
static void hScale8To15_c(SwsContext *c, int16_t *dst, int dstW,
const uint8_t *src, const int16_t *filter,
const int32_t *filterPos, int filterSize)
{
int i;
for (i = 0; i < dstW; i++) {
int j;
int srcPos = filterPos[i];
int val    = 0;
for (j = 0; j < filterSize; j++) {
val += ((int)src[srcPos + j]) * filter[filterSize * i + j];
}
dst[i] = FFMIN(val >> 7, (1 << 15) - 1); // the cubic equation does overflow ...
}
}

lumConvertRange () [SwsContext -> lumConvertRange()]

SwsContext -> hyScale ()的实现函数是在ff_sws_init_range_convert()中初始化的。可以回顾一下ff_sws_init_range_convert ()的定义，如下所示。

av_cold void ff_sws_init_range_convert(SwsContext *c)
{
c->lumConvertRange = NULL;
c->chrConvertRange = NULL;
if (c->srcRange != c->dstRange && !isAnyRGB(c->dstFormat)) {
if (c->dstBpc <= 14) {
if (c->srcRange) {
c->lumConvertRange = lumRangeFromJpeg_c;
c->chrConvertRange = chrRangeFromJpeg_c;
} else {
c->lumConvertRange = lumRangeToJpeg_c;
c->chrConvertRange = chrRangeToJpeg_c;
}
} else {
if (c->srcRange) {
c->lumConvertRange = lumRangeFromJpeg16_c;
c->chrConvertRange = chrRangeFromJpeg16_c;
} else {
c->lumConvertRange = lumRangeToJpeg16_c;
c->chrConvertRange = chrRangeToJpeg16_c;
}
}
}
}

SwsContext 的lumConvertRange()函数主要用于JPEG标准像素取值范围（0-255）和MPEG标准像素取值范围（16-235）之间的转换。有关这方面的分析在上一篇文章中一斤详细叙述过，在这里不再重复。简单看一下其中的一个函数。

lumRangeFromJpeg_c()

把亮度从JPEG标准转换为MPEG标准（0-255转换为16-235）的函数lumRangeFromJpeg_c()的定义如下所示。

static void lumRangeFromJpeg_c(int16_t *dst, int width)
{
int i;
for (i = 0; i < width; i++)
dst[i] = (dst[i] * 14071 + 33561947) >> 14;
}

其实这个函数就是做了一个（0-255）到（16-235）的映射。它将亮度值“0”映射成“16”，“255”映射成“235”，因此我们可以代入一个“255”看看转换后的数值是否为“235”。在这里需要注意，dst中存储的像素数值是15bit的亮度值。因此我们需要将8bit的数值“255”左移7位后带入。经过计算，255左移7位后取值为32640，计算后得到的数值为30080，右移7位后得到的8bit亮度值即为235。

hcscale()

水平色度拉伸函数hcscale()的定义位于libswscale\swscale.c，如下所示。

static av_always_inline void hcscale(SwsContext *c, int16_t *dst1,
int16_t *dst2, int dstWidth,
const uint8_t *src_in[4],
int srcW, int xInc,
const int16_t *hChrFilter,
const int32_t *hChrFilterPos,
int hChrFilterSize,
uint8_t *formatConvBuffer, uint32_t *pal)
{
const uint8_t *src1 = src_in[1], *src2 = src_in[2];
if (c->chrToYV12) {
uint8_t *buf2 = formatConvBuffer +
FFALIGN(srcW*2+78, 16);
//转换
c->chrToYV12(formatConvBuffer, buf2, src_in[0], src1, src2, srcW, pal);
src1= formatConvBuffer;
src2= buf2;
} else if (c->readChrPlanar) {
uint8_t *buf2 = formatConvBuffer +
FFALIGN(srcW*2+78, 16);
//读取
c->readChrPlanar(formatConvBuffer, buf2, src_in, srcW, c->input_rgb2yuv_table);
//赋值
src1 = formatConvBuffer;
src2 = buf2;
}

if (!c->hcscale_fast) {
//色度-水平拉伸
c->hcScale(c, dst1, dstWidth, src1, hChrFilter, hChrFilterPos, hChrFilterSize);
c->hcScale(c, dst2, dstWidth, src2, hChrFilter, hChrFilterPos, hChrFilterSize);
} else { // fast bilinear upscale / crap downscale
c->hcscale_fast(c, dst1, dst2, dstWidth, src1, src2, srcW, xInc);
}
//如果需要取值范围的转换（0-255和16-235之间）
if (c->chrConvertRange)
c->chrConvertRange(dst1, dst2, dstWidth);
}

从hcscale()的源代码可以看出，它的流程如下所示。
1.转换成UV
该功能通过SwsContext的chrToYV12 ()函数完成。如果该函数不存在，则调用SwsContext的readChrPlanar ()读取UV。

2.拉伸
拉伸通过SwsContext的hcScale ()函数完成。如果存在hcscale_fast()方法的话，系统会优先调用hcscale_fast ()。

3.转换范围（如果需要的话）
如果需要转换色度的取值范围（例如色度取值范围从0-255转换为16-240），则会调用SwsContext的chrConvertRange ()函数。

hcscale()的原理和hyScale ()的原理基本上是一样的，在这里既不再详细研究了。

还有几个函数没有分析，但是时间有限，以后有机会再进行补充。

雷霄骅
leixiaohua1020@126.com http://blog.csdn.net/leixiaohua1020