1. 18 Nov, 2017 1 commit
  2. 05 Nov, 2017 1 commit
  3. 02 Nov, 2017 1 commit
  4. 26 Oct, 2017 1 commit
    • Tobias Ellinghaus's avatar
      Don't use export's conf to pass data to colorout · 1425e4f3
      Tobias Ellinghaus authored
      Abusing the configuration system by setting override profile/intent in
      the export namespace from all over the code to control what colorout
      does is bad style and prone to race conditions. It also messes with
      exporting images.
      
      Fixes #11787
      1425e4f3
  5. 22 Oct, 2017 1 commit
    • johannes hanika's avatar
      local laplacian: make roi aware in darkroom mode · 911133c8
      johannes hanika authored
      this only affects zooming in darkroom mode.
      the upper level of the laplacian pyramid are stolen
      from the preview pipeline. because they are at a different
      resolution, trilinear filtering is done to braid the data
      into the new output pyramids.
      
      this can result in some artifacts on very fine geometry
      for unlucky ratios between full and preview resolution.
      i think it's already a lot better than without though.
      
      the opencl code path is not fixed yet.
      
      i did not benchmark it, but since the full pipeline
      actually computes /less/ it should be faster modulo
      synchronisation with the preview pipe.
      911133c8
  6. 20 Oct, 2017 1 commit
  7. 07 Oct, 2017 1 commit
    • Tobias Ellinghaus's avatar
      Fix a crash from spot's legacy_params · 4c0754d0
      Tobias Ellinghaus authored
      I am not sure if the crash is really fixed by this, and if it maybe has
      side effects with legacy_params not working as intended, but the old
      code assumed that it was only run while in darkroom view, which is not
      necessarily true for legacy_params which is also run when generating
      thumbnails.
      Another part of the crash was libs/history.c getting the
      DT_SIGNAL_DEVELOP_HISTORY_CHANGE signal while in lighttable and doing
      bad things because of that. I didn't change that, but the crash seems to
      not happen any more. I assume it's just luck and that we have to revisit
      our signal system some day to either not emit those signals while not in
      darkroom, don't run callbacks from libs not visible at the moment, or at
      least check the current view before assuming that we are in darkroom and
      accessing darktable.develop.
      4c0754d0
  8. 01 Oct, 2017 1 commit
  9. 26 Sep, 2017 2 commits
    • Pascal Obry's avatar
      borders: sync opencl version. · f0cf3c41
      Pascal Obry authored
      The code for OpenCL had some discrepancies.
      f0cf3c41
    • Pascal Obry's avatar
      borders: fix frame size for the larger borders. · aea1d691
      Pascal Obry authored
      We do not want to extend the frame for the larger borders. This issue
      arise for a frame which do not have the same aspect than the image.
      
      To reproduce, open a portrait image, in the framing module:
      
         border size       : 10%
         aspect            : square
         frame line size   : 25%
         frame line offset : 100%
      
      The whole right part of the image is black.
      aea1d691
  10. 16 Sep, 2017 1 commit
  11. 12 Sep, 2017 1 commit
  12. 08 Sep, 2017 1 commit
    • Tobias Ellinghaus's avatar
      Make white balance sliders colored · 468ae24c
      Tobias Ellinghaus authored
      The temperature slider follows the inversed logic we have for Kelvin
      values, i.e., you set the color of the image and not the color cast you
      want to achieve. The tint slider however works like the rgb ones: it
      reflects the result on the image. This is open for discussion!
      468ae24c
  13. 07 Sep, 2017 1 commit
    • Tobias Ellinghaus's avatar
      Fix #11723: Fix a threading issue in rsvg/cairo · 9f10c211
      Tobias Ellinghaus authored
      This is mitigating the bug by running less of watermark's process() in
      parallel and putting more into a mutex. Not ideal. Eventually we might
      want to look at moving more of the setup code into commit_params().
      9f10c211
  14. 31 Aug, 2017 1 commit
  15. 21 Aug, 2017 1 commit
  16. 22 Jun, 2017 1 commit
  17. 21 Jun, 2017 2 commits
  18. 19 Jun, 2017 1 commit
  19. 17 Jun, 2017 1 commit
  20. 13 Jun, 2017 2 commits
  21. 12 Jun, 2017 1 commit
  22. 10 Jun, 2017 1 commit
  23. 09 Jun, 2017 2 commits
  24. 07 Jun, 2017 1 commit
  25. 06 Jun, 2017 2 commits
  26. 01 Jun, 2017 1 commit
  27. 30 May, 2017 1 commit
  28. 27 May, 2017 1 commit
  29. 17 May, 2017 2 commits
  30. 15 May, 2017 2 commits
  31. 10 May, 2017 1 commit
    • Roman Lebedev's avatar
      Clahe iop: process(): fix gcc7 build. · 7f8281c3
      Roman Lebedev authored
      [ 91%] Building C object src/iop/CMakeFiles/clahe.dir/introspection_clahe.c.o
      In file included from /home/lebedevri/darktable/build-gcc7/src/iop/introspection_clahe.c:2:0:
      /home/lebedevri/darktable/src/iop/clahe.c: In function 'process._omp_fn.1':
      /home/lebedevri/darktable/src/iop/clahe.c:200:7: error: assuming signed overflow does not occur when assuming that (X + c) < X is always false [-Werror=strict-overflow]
             for(int b = 0; b < hMin; b++)
             ^~~
      7f8281c3
  32. 09 May, 2017 1 commit
  33. 08 May, 2017 1 commit