Fuji X-Trans is so special and requires different processing in most cases, so COLOR() really used only in bayer processing code, but not in X-Trans.
BTW, this need to be fixed and COLOR() should return correct color even for X-Trans files too. I'll reply to this thread again after pushing fix to github repo.
Thanks for you reply. And I have another question.when I read the part of source code in libraw_cxx.cpp, the part of source code is :
char name[100], value[100];
utf2char(P[i].name,name);
utf2char(P[i].value,value);
if (!strcmp (name, "ISO"))
imgdata.other.iso_speed = atoi(value);
I am confused with this part of source code. when you developed this libraw library,how can you read "ISO speed" from Raw file to iso_speed variable of libraw?
Have not used XCode GUI mode.
To re-build LibRaw from command line, start with OS X tuned version (http://www.libraw.org/data/LibRaw-0.16.2-MacOSX.zip). To rebuild it just use
make -f Makefile.dist
It should works fine
Could you possibly explain the steps required to compile those samples? Preferably using xcode?
Also building a command line tool from the modified sample file, how could I do this on a linux system?
I have no experience with working with c++.
I essentially need to extract out the thumbnail from a raw but be able to specify the output directory of the file.
Since I am watching the dir, when any new files are added I process them, but do not want to add the thumbnail to the same directory for several reasons. dcraw does not seem to have a flag to specify an output directory when extracting the thumbnail.
It also is creating strange images when we apply the color profile.
I would love to be able to dig into libraw and start creating dedicated command line tools for our processing, if you could direct me in the right direction to things I would need to learn and tools I would use.
Also, since this is going to run on raspberry pi, would you be able to recommend a build process that would work for a device like that?
I know this may be a lot to ask. Sorry for the newbie questions. Any help is greatly appreciated.
LibRaw's dcraw_emu intended to be used instead of dcraw in most (but not all) cases. If you need some specific command-line tool, start with simple_dcraw sample and modify it at your taste.
I was able to associate the libraw.dll and libraw.h with matlab using loadlibrary function. It builds a function table and I am able to open a raw file and so on. However, I haven't gotten to the point where I can open the bayer pixel data. It has something to do with how pointers are used. In any case, I wanted to let you know it is possible to interface with Matlab.
You may use LibRaw to read RAW image and convert it to RGB bitmap (in memory).
To create JPEG image from this bitmap you'll need another software (LibJPEG or so)
Got the file, thanks.
This is very special case:
color.maximum is calculated from metadata for these files, this is just 0xfffc - black_level
And after LibRaw::unpack() real data maximum is within this range.
But on postprocessing stage, PhaseOne data are linearized using other metadata contained in file (complex process in phase_one_flat_field() call).
After that stage, real (linearized) data values may have values above color.maximum
The best way it to share it somewhere (Dropbox, google) and send me link.
Or just send via E-mail
my personal e-mail is lexa@lexa.ru
Sure, how to send you the file (46,9 MB)?
This is possible and normal for some formats. For example, for DNG files color.maximum is read from corresponding tag, that may contain anything.
BTW, could you provide sample file for analysis?
LibRaw do not write RAW data in any format. It can store processing results in .ppm or .tiff, that's all.
Consider Adobe DNG SDK, it can read and write DNG files.
Could you please specify more details
- environment (OS, etc)
- compiler used
- interpolation mode used
Thank you!
You may use fcol() instead of FC() in COLOR() definition:
int COLOR(int row, int col) { return libraw_internal_data.internal_output_params.fuji_width? FCF(row,col):fcol(row,col);}This is not so efficient because fcol() is not inlined, but if you do not use COLOR() for all pixels of image, it will work.
fcol do not take into account Fuji SuperCCD specifics.
What is LibRaw::fcol, can it be used instead?
Fuji X-Trans is so special and requires different processing in most cases, so COLOR() really used only in bayer processing code, but not in X-Trans.
BTW, this need to be fixed and COLOR() should return correct color even for X-Trans files too. I'll reply to this thread again after pushing fix to github repo.
For most (TIFF/EP) raw files, EXIF data is pointed by EXIF (34665) tag
Make and model for these files are parsed in other place.
Can you tell me the method to find the exif metadata location in Raw files?
How about other parameters like "camera model" ,"camera make", like this Exif metadata, how can you read it to the libraw datastructure?
This piece of code is related only to foveon images with own (non-EXIF) metadata format
Thanks for you reply. And I have another question.when I read the part of source code in libraw_cxx.cpp, the part of source code is :
char name[100], value[100];
utf2char(P[i].name,name);
utf2char(P[i].value,value);
if (!strcmp (name, "ISO"))
imgdata.other.iso_speed = atoi(value);
I am confused with this part of source code. when you developed this libraw library,how can you read "ISO speed" from Raw file to iso_speed variable of libraw?
Have not used XCode GUI mode.
To re-build LibRaw from command line, start with OS X tuned version (http://www.libraw.org/data/LibRaw-0.16.2-MacOSX.zip). To rebuild it just use
make -f Makefile.dist
It should works fine
Thank you Alex.
Could you possibly explain the steps required to compile those samples? Preferably using xcode?
Also building a command line tool from the modified sample file, how could I do this on a linux system?
I have no experience with working with c++.
I essentially need to extract out the thumbnail from a raw but be able to specify the output directory of the file.
Since I am watching the dir, when any new files are added I process them, but do not want to add the thumbnail to the same directory for several reasons. dcraw does not seem to have a flag to specify an output directory when extracting the thumbnail.
It also is creating strange images when we apply the color profile.
I would love to be able to dig into libraw and start creating dedicated command line tools for our processing, if you could direct me in the right direction to things I would need to learn and tools I would use.
Also, since this is going to run on raspberry pi, would you be able to recommend a build process that would work for a device like that?
I know this may be a lot to ask. Sorry for the newbie questions. Any help is greatly appreciated.
Thank you!
LibRaw's dcraw_emu intended to be used instead of dcraw in most (but not all) cases. If you need some specific command-line tool, start with simple_dcraw sample and modify it at your taste.
Thank you for this resource, as always. - Kirk
Unfortunately, I know nothing about matlab extensions.
I was able to associate the libraw.dll and libraw.h with matlab using loadlibrary function. It builds a function table and I am able to open a raw file and so on. However, I haven't gotten to the point where I can open the bayer pixel data. It has something to do with how pointers are used. In any case, I wanted to let you know it is possible to interface with Matlab.
Sorry for delayed answer (all forum posts are pre-moderated due to forum spam).
You may use LibRaw to read RAW data and convert it to RGB bitmap.
To pack to JPEG you'll need another library (libjpeg or so)
You may use LibRaw to read RAW image and convert it to RGB bitmap (in memory).
To create JPEG image from this bitmap you'll need another software (LibJPEG or so)
Pages