Libtiff:从像素阵列生成 TIFF/EP Profile 2 原始图像

Fab*_*n K 5 image libtiff dcraw dng

我正在尝试从任意像素阵列生成 TIFF/EP Profile 2 原始 (.dng) 图像。该像素阵列代表拜耳模式(CFA)。

我研究了 TIFF/EP 文件规范,并通过使用 libtiff,包含了我认为生成完整 .dng 文件所需的所有标签。但是,我无法使用 dcraw 转换创建的文件(dcraw 显示无法解码文件)。

一个问题可能是由两个标签引起的,这两个标签在 TIFF/EP 规范中被声明为强制标签,但似乎并未在 libtiff 中实现:SensingMethod 和 TIFF/EPStandardID 标签。我真的必须包含它们吗(我已经看到示例代码忽略了这些标签,但仍然报告可以正常工作),如果是这样,我如何手动将它们添加到 libtiff 中?此外,设置 SubIFD 标记会产生错误消息“断言失败:*pa<=0xFFFFFFFFUL,文件 tif_dirwrite.c,第 1869 行”

总而言之,我不认为我的问题只是由于这三个标签造成的,而且我相信存在根本性的错误。也许你们中的某个人可以看看我的代码并提供一些线索?我不得不说 libtiff 的文档相当差,所以我的代码受到了极少数示例代码之一的启发:elphel_dng.c

非常感谢!法比安

附言。我将生成的文件上传到Dropbox

C++

#include "tiffio.h"
#include <iostream>
using namespace std;

int main(void)
{
TIFF *tif = TIFFOpen("8bitRaw.dng", "w");

const int sampleperpixel = 1;
const int width = 4;
const int height = 4;

static const short bayerPatternDimensions[] = { 2,2 };
static const double cam_xyz[] = { 2.0413690, -0.5649464, -0.3446944, -0.9692660, 1.8760108, 0.0415560,0.0134474, -0.1183897, 1.0154096 }; //AdobeRGB
static const double neutral[] = { 1.0, 1.0, 1.0 };
long max_white = 255;
long sub_offset = 0;
int row, i;
float gamma = 80;

//Arbitrary Bayer pixel values:
unsigned char image[height][width*sampleperpixel] = {};
image[0][0] = 5;    image[0][1] = 165;   image[0][2] = 0;   image[0][3] = 255;
image[1][0] = 0;    image[1][1] = 21;    image[1][2] = 0;   image[1][3] = 10;
image[2][0] = 0;    image[2][1] = 0;     image[2][2] = 30;  image[2][3] = 5;
image[3][0] = 21;   image[3][1] = 120;   image[3][2] = 1;   image[3][3] = 254;

//Black Thumbnail pixel values:
unsigned char Thumbnail_RGB_Array[] = { 0,0,0,0,0,0,0,0,0,0,0,0 };

//Linearization Table:
unsigned short curve[256];
for (i = 0; i < 256; i++)
    curve[i] = 255 * pow(i / 255.0, 100.0 / gamma) + 0.5;


TIFFSetField(tif, TIFFTAG_SUBFILETYPE, 1);
TIFFSetField(tif, TIFFTAG_IMAGEWIDTH, 4);
TIFFSetField(tif, TIFFTAG_IMAGELENGTH, 4);
TIFFSetField(tif, TIFFTAG_BITSPERSAMPLE, 8);
TIFFSetField(tif, TIFFTAG_SAMPLESPERPIXEL, 3);
TIFFSetField(tif, TIFFTAG_ORIENTATION, ORIENTATION_TOPLEFT);
TIFFSetField(tif, TIFFTAG_COMPRESSION, COMPRESSION_NONE);
TIFFSetField(tif, TIFFTAG_PLANARCONFIG, PLANARCONFIG_CONTIG);
TIFFSetField(tif, TIFFTAG_PHOTOMETRIC, PHOTOMETRIC_RGB);
TIFFSetField(tif, TIFFTAG_XRESOLUTION, 75.0);
TIFFSetField(tif, TIFFTAG_YRESOLUTION, 75.0);
TIFFSetField(tif, TIFFTAG_RESOLUTIONUNIT, RESUNIT_INCH);

TIFFSetField(tif, TIFFTAG_MAKE, "DummyMake");
TIFFSetField(tif, TIFFTAG_MODEL, "DummyModel");
TIFFSetField(tif, TIFFTAG_SOFTWARE, "DummySoftware");
TIFFSetField(tif, TIFFTAG_ORIGINALRAWFILENAME, 1, "DummyName.dng");
TIFFSetField(tif, TIFFTAG_UNIQUECAMERAMODEL, "DummyUniqueModel");
TIFFSetField(tif, TIFFTAG_IMAGEDESCRIPTION, "DummyImageDescription");
TIFFSetField(tif, TIFFTAG_COPYRIGHT, "DummyCopyright");
TIFFSetField(tif, TIFFTAG_DATETIME, "2016:06:30 11:11:15");
TIFFSetField(tif, TIFFTAG_DNGVERSION, "\01\01\00\00");
TIFFSetField(tif, TIFFTAG_DNGBACKWARDVERSION, "\01\00\00\00");
TIFFSetField(tif, TIFFTAG_COLORMATRIX1, 9, cam_xyz);
TIFFSetField(tif, TIFFTAG_ASSHOTNEUTRAL, 3, neutral);
TIFFSetField(tif, TIFFTAG_CALIBRATIONILLUMINANT1, 21);
//SensingMethodTag and TIFF/EPStandardID tag: Libtiff doesn't seem to know these:
//TIFFSetField(tif, 37399, 2);
//TIFFSetField(tif, 37398, "\01\01\00\00");
//Yields an error:
//TIFFSetField(tif, TIFFTAG_SUBIFD, 1, &sub_offset);

//Write a black 4x4 pixel dummy thumbnail:
for (row = 0; row < 4 ; row++) 
    TIFFWriteScanline(tif, Thumbnail_RGB_Array, row, 0);
TIFFWriteDirectory(tif);


//Now write main raw image:
TIFFSetField(tif, TIFFTAG_SUBFILETYPE, 0);
TIFFSetField(tif, TIFFTAG_IMAGEWIDTH, width);
TIFFSetField(tif, TIFFTAG_IMAGELENGTH, height);
TIFFSetField(tif, TIFFTAG_BITSPERSAMPLE, 8);
TIFFSetField(tif, TIFFTAG_PHOTOMETRIC, PHOTOMETRIC_CFA);
TIFFSetField(tif, TIFFTAG_SAMPLESPERPIXEL, 1);
TIFFSetField(tif, TIFFTAG_PLANARCONFIG, PLANARCONFIG_CONTIG);
TIFFSetField(tif, TIFFTAG_CFAREPEATPATTERNDIM, bayerPatternDimensions);
TIFFSetField(tif, TIFFTAG_CFAPATTERN, "\01\00\02\01");
TIFFSetField(tif, TIFFTAG_CFAPLANECOLOR, 3, "\00\01\02");
TIFFSetField(tif, TIFFTAG_LINEARIZATIONTABLE, 256, curve);
TIFFSetField(tif, TIFFTAG_WHITELEVEL, 1, &max_white);

TIFFWriteScanline(tif, image[0], 0, 0);
TIFFWriteScanline(tif, image[1], 1, 0);
TIFFWriteScanline(tif, image[2], 2, 0);
TIFFWriteScanline(tif, image[3], 3, 0);
TIFFClose(tif);
return 0;
}
Run Code Online (Sandbox Code Playgroud)

Van*_*sta 3

我有几乎相同的要求,并从您离开的地方继续。下面是一些生成 128x128 原始拜耳 10 位 DNG 图像的代码,该图像与 dng_validate、dcraw、lightroom 和 resolve 兼容。

我已经删除了几乎所有内容并只保留了最少的内容以使 dng_validate 满意。

#include <tiffio.h>
#include <math.h>

int main (void)
{
   const int width = 128;
   const int height = 128;

   static const short bayerPatternDimensions[] = { 2, 2 };
   static const float ColorMatrix1[] =
   {
      1.0, 0.0, 0.0,
      0.0, 1.0, 0.0,
      0.0, 0.0, 1.0,
   };

   static const float AsShotNeutral[] =
   {
      1.0, 1.0, 1.0,
   };

   int row, i;
   float gamma = 80;

   #define GR 0x80, 0x20, 0x08, 0x02, 0x00,
   #define WH 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
   #define BL 0x00, 0x00, 0x00, 0x00, 0x00,
   #define BLK(N) N N N N N N N N

   // Arbitrary Bayer pixel values:
   unsigned char image[] =
   {
      BLK(WH)
      BLK(GR)
      BLK(BL)
      BLK(WH)
      BLK(GR)
      BLK(BL)
      BLK(WH)
      BLK(GR)
      BLK(BL)
   };

   TIFF* tif = TIFFOpen ("out.dng", "w");
   TIFFSetField (tif, TIFFTAG_DNGVERSION, "\01\01\00\00");
   TIFFSetField (tif, TIFFTAG_SUBFILETYPE, 0);
   TIFFSetField (tif, TIFFTAG_COMPRESSION, COMPRESSION_NONE);
   TIFFSetField (tif, TIFFTAG_IMAGEWIDTH, width);
   TIFFSetField (tif, TIFFTAG_IMAGELENGTH, height);
   TIFFSetField (tif, TIFFTAG_BITSPERSAMPLE, 10);
   TIFFSetField (tif, TIFFTAG_ROWSPERSTRIP, 1);
   TIFFSetField (tif, TIFFTAG_ORIENTATION, ORIENTATION_TOPLEFT);
   TIFFSetField (tif, TIFFTAG_PHOTOMETRIC, PHOTOMETRIC_CFA);
   TIFFSetField (tif, TIFFTAG_SAMPLESPERPIXEL, 1);
   TIFFSetField (tif, TIFFTAG_PLANARCONFIG, PLANARCONFIG_CONTIG);
   TIFFSetField (tif, TIFFTAG_SAMPLEFORMAT, SAMPLEFORMAT_UINT);
   TIFFSetField (tif, TIFFTAG_CFAREPEATPATTERNDIM, bayerPatternDimensions);
   TIFFSetField (tif, TIFFTAG_CFAPATTERN, "\00\01\01\02");
   TIFFSetField (tif, TIFFTAG_MAKE, "DNG");
   TIFFSetField (tif, TIFFTAG_UNIQUECAMERAMODEL, "DNG");
   TIFFSetField (tif, TIFFTAG_COLORMATRIX1, 9, ColorMatrix1);
   TIFFSetField (tif, TIFFTAG_ASSHOTNEUTRAL, 3, AsShotNeutral);
   TIFFSetField (tif, TIFFTAG_CFALAYOUT, 1);
   TIFFSetField (tif, TIFFTAG_CFAPLANECOLOR, 3, "\00\01\02");

   unsigned char* cur = image;

   for (row = 0; row < height;)
   {
      for (i = 0; i < 32; ++i, ++row)
         TIFFWriteScanline (tif, cur, row, 0);
      cur += 40;
   }

   TIFFClose (tif);
   return 0;
}
Run Code Online (Sandbox Code Playgroud)

dcraw 非常顽固地无法工作,我必须在 gdb 中获取其源代码和跟踪才能找到错误情况:

if (!load_raw || height < 22 || width < 22 ||
  tiff_bps > 16 || tiff_samples > 6 || colors > 4)
  is_raw = 0;
Run Code Online (Sandbox Code Playgroud)

因此,您第一次尝试创建 4x4 图像将因此自动失败。

  • 对于现代 libtiff 版本,您必须提供 CFAPattern 的大小。例如:`TIFFSetField(tif, TIFFTAG_CFAPATTERN, 4, "\00\01\01\02");` (2认同)