在浏览了一段时间的libtiff源代码后，我在tif_dirwrite.c中发现了以下内容：
/*
* Copyright (c) 1988-1997 Sam Leffler
* Copyright (c) 1991-1997 Silicon Graphics, Inc.
*
* Permission to use, copy, modify, distribute, and sell this software and
* its documentation for any purpose is hereby granted without fee, provided
* that (i) the above copyright notices and this permission notice appear in
* all copies of the software and related documentation, and (ii) the names of
* Sam Leffler and Silicon Graphics may not be used in any advertising or
* publicity relating to the software without the specific, prior written
* permission of Sam Leffler and Silicon Graphics.
*/
...
if (!n)
return(0);
/*
* Total hack: if this directory includes a SubIFD
* tag then force the next <n> directories to be
* written as ``sub directories'' of this one. This
* is used to write things like thumbnails and
* image masks that one wants to keep out of the
* normal directory linkage access mechanism.
*/
tif->tif_flags|=TIFF_INSUBIFD;
tif->tif_nsubifd=tif->tif_dir.td_nsubifd;
if (tif->tif_dir.td_nsubifd==1)
tif->tif_subifdoff=0;
else
tif->tif_subifdoff=m;
return(1);
...
我在这里包含版权信息，因为我不确定在此处发布库中的代码时是否需要。
所以，回答自己的问题（如何在主图像IFD的子IFD中编写缩略图）：
//...
//For the sake of this demo we will assume that I have opened a
//TIFF (TIFF* created_TIFF) in write mode and have included the correct header
//files
//set all of your TIFF fields for the main image
//...
//Define the number of sub-IFDs you are going to write
//(assuming here that we are only writing one thumbnail for the image):
int number_of_sub_IFDs = 1;
toff_t sub_IFDs_offsets[1] = { 0UL };
//set the TIFFTAG_SUBIFD field:
if(!TIFFSetField(created_TIFF, TIFFTAG_SUBIFD, number_of_sub_IFDs,
sub_IFDs_offsets))
{
//there was an error setting the field
}
//Write your main image raster data to the TIFF (using whatever means you need,
//such as TIFFWriteRawStrip, TIFFWriteEncodedStrip, TIFFWriteEncodedTile, etc.)
//...
//Write your main IFD like so:
TIFFWriteDirectory(created_TIFF);
//Now here is the trick: like the comment in the libtiff source states, the
//next n directories written will be sub-IFDs of the main IFD (where n is
//number_of_sub_IFDs specified when you set the TIFFTAG_SUBIFD field)
//Set up your sub-IFD
if(!TIFFSetField(created_TIFF, TIFFTAG_SUBFILETYPE, FILETYPE_REDUCEDIMAGE))
{
//there was an error setting the field
}
//set the rest of the required tags here, as well as any extras you would like
//(remember, these refer to the thumbnail, not the main image)
//...
//Write this sub-IFD:
TIFFWriteDirectory(created_TIFF);
//Assuming you are only writing one sub-IFD and are done with the file, you
//can close it now. If you specified more than one sub-IFD, you need repeat
//the above code (starting where we set TIFFTAG_SUBFILETYPE) for each of your
//sub-IFDs
TIFFClose(created_TIFF);
我希望这能帮助到某些人，让他们不必像我一样花费很多精力来弄清楚如何做这件事。真的很遗憾libtiff文档记录得如此之差，特别是考虑到它的广泛使用。
我认为@KSletmoe指出了一个关键点，即在编写任何子IFD之前，应该在IFD0中添加SubIFD标签（330），但仍存在问题。
SubIFD标签的定义是“指向子IFD的偏移量”。因此，如果您没有正确设置偏移量，则tiff解析器无法正确解析整个tiff。
有两种处理这种情况的方法。
首先，预先计算每个IFD/SubIFD的大小，然后在设置时填入SubIFD标签，而不是设置为0x0。
或者，您可以像下面这样正常编写每个IFD，然后返回到IFD0并使用由libtiff计算的最终偏移量添加SubIFD标签。
//Write down every thing you need in a tiff
// ... IFD0
TIFFWriteDirectory(tif);
// ... IFD1
TIFFWriteDirectory(tif);
// ... IFD2
TIFFWriteDirectory(tif);
//set current Dir as IFD0 in the end
TIFFSetDirectory(tif, 0);
//get next dir offset
sub_offset[1] = TIFFGetNextDirOff(tif, 2);
sub_offset[0] = TIFFGetNextDirOff(tif, 1);
//only for clean next dir offset in IFD0
no_use_offset = TIFFGetNextDirOff(tif, 0);
TIFFSetField(tif, TIFFTAG_SUBIFD, 2, sub_offset);
你可能需要在 tif_dir.c 中添加以下函数
uint64 TIFFGetNextDirOff(TIFF* tif, uint16 dirn)
{
uint64 nextdir;
uint16 n;
if (!(tif->tif_flags&TIFF_BIGTIFF))
nextdir = tif->tif_header.classic.tiff_diroff;
else
nextdir = tif->tif_header.big.tiff_diroff;
for (n = dirn; n > 0 && nextdir != 0; n--)
if (!TIFFAdvanceDirectory(tif, &nextdir, NULL))
return (0);
/*!!Watchout!! Here will reset the next dir offset to 0*/
tif->tif_nextdiroff = 0;
return nextdir;
}