Add new comment

General Notes on API


  1. LibRaw editions
  2. Error Code Conventions and Error Handling
  3. Nonstandard Situations That Are Not Errors
  4. Input Layer Abstraction
  5. Thread Safety
  6. The Use of C++
  7. Parameters of the LibRaw::imgdata.params Structure Affecting the Behavior of open_file/unpack/unpack_thumb
  8. Memory Usage
    1. Stack Usage
    2. Dynamic Memory Management
    3. Dynamic Memory Usage
      1. Memory Buffer for the RAW Image
      2. Memory for the Postprocessed Image
      3. Memory for the Decoded Thumbnail
      4. Memory for RAW Unpacking
      5. Memory for Postprocessing
      6. Memory for File Writing
      7. Unpacking into memory buffer
  9. Incompatibilities with dcraw
    1. Automatic maximum search/brightness adjustment
    2. Processing of Thumbnails from Kodak cameras

LibRaw Versions

Since version 0.9, there is only one LibRaw variants. Older versions have three separate editions (normal, -Lite and -Commercial versions).

Error Code Conventions and Error Handling

The following conventions concern the returned errors:

  1. All functions that can return an error code have integer type of return data.
  2. If there is no error, the return value is 0 (LIBRAW_SUCCESS).
  3. If an error has happened in a system call, the return value is errno (a positive number), which can be analyzed using strerror() or similar means.
  4. All LibRaw's own error codes are negative; each of these errors belongs to one of two types:
    Non-fatal errors
    Non-fatal errors do not forbid execution of other functions in the processing succession (e.g., unpack_thumb() can easily return the code corresponding to "preview is absent" but this does not prevent further call of unpack().
    Fatal errors
    In the case of fatal errors (memory shortage, input data error, data unpacking failure), the current stage of processing is terminated and all allocated resources are freed.
    If an attempt to continue processing is made, all subsequent API calls will return the LIBRAW_OUT_OF_ORDER_CALL error.
    At the same time, the LibRaw instance in which a fatal error has occurred can process the next RAW files in the usual way (by calling open_file() (or other input methods), then unpack(), etc.).
  5. The macro LIBRAW_FATAL_ERROR(error code) checks if an error is fatal or not.
  6. The error codes are listed and deciphered here.

Nonstandard Situations That Are Not Errors

If the program has encountered a nonstandard situation that does not prevent retrieval of some data from a file, it sends a signal by setting the corresponding bit in imgdata.process_warnings. The possible types of warnings are listed and deciphered here.

Input Layer Abstraction

LibRaw uses objects derived from LibRaw_abstract_datastream for data input. Semantics of these objects is similar to 'file with arbitrary seek' object: both read and seek operations are used.

Some RAW formats requires temporary switch to another data stream created on top on memory buffer for metadata read. Methods for doing so are implemented in base class LibRaw_abstract_datastream by internal data field substream. Look into source code of LibRaw_file_datastream class in libraw/libraw_datastream.h file for more details.
When implementing own datastream classes, you need to take substream into account and pass control to methods of this field if it is active (not NULL).

If datastream implementation knows name of input file, it should provide fname() call. This name will be used in error callbacks and in guessing name of JPEG file with metadata (for RAW files with external metadata).

For external metadata support input class should implement subfile_open()/subfile_close() methods. возврашают код ошибки.
Sample of these methods implementation may be found in LibRaw_file_datastream class (look into libraw/libraw_datastream.h file for details).

Thread safety

Thread safety is ensured if a LibRaw object is created and used within one thread. At the same time, the number of threads (each with its own LibRaw object) is not limited in any way (except by memory requirements).

If a LibRaw object is created in one execution thread and used in another, external synchronization is necessary.

There is two libraries under Unix environment (Linux/FreeBSD/MacOS): libraw_r.a (thread-safe) and libraw.a (single-threaded, slightly faster).

Thread-safe library version stores intermediate unpacker data into LibRaw class data. So, several copies of LibRaw, working in parallel, is possible.

Not thread-safe library uses global variable for intermediate data store which is faster but not reenterant. This non-thread-safe library still may be used in multi-threaded apps, but only if exactly one LibRaw class copy exists in program.

Windows version is similar to multi-threaded Unix one.

The Use of C++

Exception situations within LibRaw are handled using the C++ exception mechanism. All exceptions are caught inside the library functions and should not penetrate outside.

Memory is allocated/freed using functions malloc(calloc)/free rather than new/delete.

If C API is used, references to C++ calls new/delete still remain, and so linking with libstdc++(Unix)/....(Windows) is necessary.

Parameters of the LibRaw::imgdata.params Structure Affecting the Behavior of open_file/unpack/unpack_thumb

Most data fields of structure LibRaw::imgdata.params affect only data postprocessing, but there are some exceptions, which have been inherited by the current version of LibRaw from/ dcraw source texts (these dependences will be gradually removed).

imgdata.params.use_camera_matrix and imgdata.params.use_camera_wb
These fields affect loading of RAW data for cameras with a color matrix.
Attention! If parameter imgdata.params.use_camera_matrix is not set by the user, it is copied from imgdata.params.use_camera_wb at the stage of file opening.
If this parameter is greater than or equal to zero, assignment
          = imgdata.params.user_flip
is performed at the open_file() stage.
This parameter makes it possible to select the number of the extracted image for data formats in which storage of several RAW images in one data file is possible.
Affects RAW data loading for Phase One and Sinar backs. Also, it this parameter is set then image bitmap will be reduced by half in each dimension. In later case, all 4 components of bitmap will be filled during data extraction phase.
imgdata.params.threshold, imgdata.params.aber
If these parameters used, then half-sized bitmap will be used for data unpacking. See above for details.
Affects loading of white balance matrix for Leaf backs.

Memory Usage

Stack Usage

An instance of the LibRaw class has its own size about 800 Kb; if constructions like LibRaw imageProcessor; are used, this memory is stack-allocated.

Methods of class LibRaw (and C API calls) may allocate up to 130-140 Kb of data on the stack (to place auto variables) during their work.

Thus, the work of one LibRaw instance may require about 1MB of stack memory. This is not a problem for most contemporary architectures. However, when working in a multithreaded environment, one should not forget to allocate a sufficient amount of memory for the thread stack.

In the case of dynamic allocation (LibRaw *iProcessor = new LibRaw;), the requirements to stack memory will decrease by ~800 Kb, which is the size of a class instance). If C API is used, the LibRaw instance is allocated dynamically.

Dynamic Memory Management

LibRaw keeps record of all allocated dynamic memory blocks; in the case of an exceptional situation (fatal error), they are all freed. The code for keeping this record is fairly primitive and not designed to consider allocation of many blocks (in the normal situation, allocation takes place from 2 to 6 times during file processing); this fact should be taken into account by developers trying to add new methods to LibRaw.

Dynamic Memory Usage

LibRaw uses dynamic memory

  • for the decoded image;
  • for the decoded thumbnail;
  • for the postprocessed image;
  • for the ICC profile retrieved from the RAW file (if available);
  • for temporary data at the stage of RAW file unpacking;
  • for temporary data at the stage of postprocessing and result output;
  • for reading of the RAW source file (only under Win32).

Memory buffer for the RAW image

Decoded RAW data are stored:

  • one 16-bit value per pixel for "bayer" images. The masked pixels (black or dark or masked frame) are stored with image data.
  • Free or four 16-bit values for full-color images (Foveon, Linear DNG, Canon sRAW etc.).
  • one,three, or four 32-bit floating point values per pixel for floating-point data.

The buffer for RAW data is allocated by unpack() call and freed upon calling recycle().

Memory for the Postprocessed Image

On postprocessing stage each pixel contains four 16-bit values, one for each possible color channel (some sensors are actually 4-color).

The buffer for the decoded image is allocated upon calling raw2image() or dcraw_process()

The buffer freed upon calling recycle() or free_image() calls.

Memory for the Decoded Thumbnail

Memory for the thumbmail is allocated upon calling unpack_thumb() and freed upon calling recycle(). The size of the allocated buffer is precisely adjusted to the thumbnail size, i.e., up to several Mb.

Memory for RAW Unpacking

Memory for temporary buffer needed during RAW data unpacking may be allocated during the work of unpack() and freed before completion of this function. The sizes of the allocated buffers are small, up to tens of Kb.

Memory for Postprocessing

During image postprocessing (inherited from dcraw), memory for the histogram (128 Kb) is allocated. This memory is allocated upon calling dcraw_process() and freed upon calling recycle().

In addition, during the work of dcraw_process() and during the usage of some available possibilities, like

  • rotation of images from FUJI cameras;
  • correction of chromatic aberrations;
  • image size changes (including correction of non-square pixels);
  • highlight recovery;

a temporary buffer with the size equal to the size of the resultant image (6-8 bytes per pixel for various processing stages) will be allocated. As soon as the intermediate substage of processing is completed, the buffer with the previous copy of the image will be freed.
If postprocessing is not used, then temporary buffers are not allocated.

Memory for File Writing

Upon calling dcraw_ppm_tiff_writer(), memory for a single row of the output image is allocated. The allocated memory is freed before the end of this call.

Unpacking into memory buffer

Functions dcraw_make_mem_image() dcraw_make_mem_thumb() (and complementary calls in C-API) allocates memory for entire output datasets (full RGB bitmap and thumbnail, respectively).To free allocated memory use dcraw_clear_mem() function.

Incompatibilities with dcraw

Automatic maximum search/brightness adjustment

Many camera formats really use less data range, than possible by format nature (bit count). If data maximum estimated incorrectly (too low) this may resuls in colored highlights ('pink clouds') because of data cut at wrong level.

To prevent this, LibRaw uses real data maximum from current file if this maximum is larger than format maximum multiplied by imdata.params.adjust_maximum_thr value (default is 0.75).

To turn off this feature (and repeat dcraw.c pink clouds) set imdata.params.adjust_maximum_thr to 0.0

Processing of Thumbnails from Kodak cameras

In some Kodak cameras, the preview (thumbnail) is stored in the form of uncorrected image. During its extraction using dcraw -e, the white balance, color conversion, and other settings are the same as those used for extraction of the main RAW data (including defect removal and dark frame subtraction, which is erroneous, since the image size is different).
In LibRaw::unpack_thumb() calls, the white balance taken from the camera ("as shot") is used and no settings from imgdata.params are considered.

For all other cameras, thumbnails are extracted "as is," without any color conversions, both in dcraw and in LibRaw.