Menu ▾ ▴

#23 lcms 2.0 writes floating point CLUTs incorrectly

v1.0 (example)
open-fixed
nobody
Interface (example) (2)
5
2010-07-20
2010-07-09
Robert Minsk
No

Type_LUTA2B_Write and Type_LUTB2A_Write call WriteCLUT with a precision of 2 because it only checks the Lut->SaveAs8Bits flag. WriteCLUT should be called with a precession of 4 which would cause an error "Unknow precision of '%d'". Right now it tries to write floats as 16-bits.

Also the error message needs to be fixed to "Unknown" instead of "Unknow".

Should it try to convert the CLUT to 16 or 8 bits based on the precision?

Discussion

  • Marti Maria Saguer

    Marti Maria Saguer - 2010-07-10
    • status: open --> open-accepted
     

  • Marti Maria Saguer

    Marti Maria Saguer - 2010-07-10

    LutA2B and LubBToA only supports 8 and 16 bits, according the ICC spec. Floating point data can only be saved by using multiprofile elements.
    But the CLUT code certainly misses a check on floating point data.

     

  • Marti Maria Saguer

    Marti Maria Saguer - 2010-07-20

    Fixed in git. Will be released on 2.1.

     

  • Marti Maria Saguer

    Marti Maria Saguer - 2010-07-20
    • status: open-accepted --> open-fixed