Add new comment

LibRaw 0.15.0-Alpha1

First public LibRaw 0.15 Alpha is available: LibRaw 0.15.0 Alpha1.

This version is provided to get developers' feedback (e.g. someone uses dropped document mode processing) and for testing purposes. Some features of 0.14 are missing and will be available in the next Alpha (see TODO section below).

Alpha1 is tested under Unix and Windows with usual large testing subset (about 350 RAW files). Changelog:

  1. API Changes:
    Removed, because as it seems no one uses it:
    • LibRaw::dcraw_document_mode_processing (and C-API call)
    • Data field imgdata.color.color_flags
    Removed, because internal logic changed:
    • LIBRAW_DECODER_4COMPONENTS
      Now unpack() unpacks RAW data to only two formats possible:
      • imgdata.rawdata.raw_image[] - flat buffer with bayer data (one value per pixel), including masked pixels.
      • imgdata.rawdata.color_image - 4-components array (similar to imgdata.image) with 3 or 4 components per pixel and without masked pixels.
    Changed:
    • imgdata.color.cblack[4] instead of [8] - for now cblack[4..7] (count of pixels used for black level calculation) is not used, so removed
    • LibRaw::fc() renamed to LibRaw::fcol() to maintain compatibility with dcraw internals
    Temporary not working properly or disabled (to be fixed in next Alpha-versions, see also TODO section below)
    • imgdata.params.cropbox[] - cropping does not work as of now, so it is disabled
    • LibRaw::subtract_black() - doesn't works properly, so not used in LibRaw internals, to be restored in next Alpha
    Added:
    • WF debanding code (banding suppression): int imgdata.params.wf_debanding - enables debanding float imgdata.params.wf_deband_threshold[4] - set threshold levels for debanding (in pixel values, so usual value is several units)
    • LibRaw::open_wfile(wchar_t *filename) (WIN32 only) - opens RAW filename passed using Windows unicode encoding.
    • imgdata.params.user_cblack[4] - allows to set per-channel black levels
    • User-provided data streams (derived from LibRaw_abstract_datastream) should implement jpeg_src() method to provide data source for JPEG decoder used in lossy DNG processing. See src/libraw_datastream.cpp for sample implementation.
  2. Imported dcraw 9.16/1.450, added support for following:
    • Adobe DNG: Fast Load DNG (LightRoom 4.x),lossy-compressed DNG (LR 4.x, you need to compile with -DUSE_JPEG and link with libjpeg 8)
    • Canon: G1 X, SX220 HS, EOS 5D Mark III, EOS 650D, EOS 1D-X
    • Casio: EX-Z8
    • Fujifilm: X-S1, HS30EXR, X1-Pro
    • Nikon: D4, D3200, D800, D800E
    • Olympus: E-M5
    • Panasonic: DMC-GF5
    • Samsung: NX20, NX210, ?>445@6:0 =>2>3> firmware NX100
    • Sigma: SD15,SD1, SD1 Merill, DP1, DP1S, DP1X, DP2, DP2S, DP2X (in LibRaw-demosaic-pack-GPL2)
    • Sony: NEX-F3, SLT-A37, SLT-A57
  3. TODO (additional to 'not working' section above)
    • Sony ARW2 Hack - turns off division by 4 of RAW values unpacked from sony ARW2 files.
      int imgdata.params.sony_arw2_hack;
    • Some optimizations introduced in previous LibRaw versions are temporarily removed:
      • OpenMP
      • One pass data copy + black subtraction + fuji data rotation (LibRaw::raw2image_ex() )
      • Callbacks for more efficient data handling (SSE2 or multithreaded) introduced in 0.15-PreAlpha
      • Fast LJPEG decoder
    • * Black level subtraction for Phase One files is performed on unpack() stage, not on postprocessing one.