Awesome Privacy

Recent Commits

• Pascal Obry (25 Feb 26)

  Merge pull request #20383 from jenshannoschwalm/opencl_maintenance_56_3 Third round of OpenCL maintenance for 5.6

• Pascal Obry (25 Feb 26)

  Merge pull request #20404 from victoryforce/translations-260225 Translations update: en@truecase and Ukrainian

• Victor Forsiuk (25 Feb 26)

  Ukrainian translation update

• Victor Forsiuk (25 Feb 26)

  en@truecase update

• Jeronimo Pellegrini (25 Feb 26)

  Updates to pt_BR

• Martin Straeten (25 Feb 26)

  Update de.po to recent pot File + some concistency improvements ( there'll be more unti the next release ;) )

• Hanno Schwalm (11 Feb 26)

  Remove protecting mutex in darktable.gui 1. There is no reason for the protecting mutex as all gui data changing happens in main thread. 2. We don't need to set darktable.gui to NULL as all pointers were set to NULL at the very start.

• Pascal Obry (24 Feb 26)

  Update French translation.

• Pascal Obry (24 Feb 26)

  Update darktable.pot for translators.

• Pascal Obry (24 Feb 26)

  Merge pull request #20397 from victoryforce/gui-string-fixes [GUI] A few more string fixes

• Victor Forsiuk (24 Feb 26)

  Add an explanation of what functionality will be available when gPhoto2 is enabled

• Hanno Schwalm (24 Feb 26)

  Updated OpenCL roundups for width&height The roundup magics for width&height are mainly relevant for kernels called without locals as good values generally improve performance. Profiling these magics is simply not worth the effort, we can do a very good guess based on maximum workgroup size for the device. Tests to do this per kernel via dt_opencl_get_kernel_work_group_size() shows that the overhead decreases performance for those simple kernels so we go the easy way. Please note: We still write calculated data to the per-device conf for now to avoid confusion for people reading that conf but very likely we will go for a cl version bump for less options offered.

• Victor Forsiuk (24 Feb 26)

  Correct sentence capitalization in context where it is expected

• Victor Forsiuk (24 Feb 26)

  Add mention of PNG in the "raster mask files root folder" preference tooltip

• Hanno Schwalm (24 Feb 26)

  Updated OpenCL events handling The maximum number of OpenCL events that can be handled within the driver/device is not exposed in the OpenCL API, in case this exceeds device-internal resources an error code would be returned which is handled and reported by darktable OpenCL interface. We still try to keep event resources within sensible limits to reduce stress. The per-device dt_opencl_device_t struct now only has a flag use_events, the max number of events is defined as DT_OPENCL_EVENTS and the log has been updated.

• Hanno Schwalm (23 Feb 26)

  Additional OpenCL clmem logs 1. For improved debugging of OpenCL performance we want information about clmem read/write/copy actions via the -d verbose switch 2. due to float->double conversion issues we sometimes had bad negative timings.

• Hanno Schwalm (23 Feb 26)

  Some OpenCL kernel maintenance & improvements 1. Use CLFARRAY for calc_Y0_mask (dual blend), denoiseprofile and hazeremoval for a simpler interface 2. Use clipf() macro in some more places 3. The weight function in atrous.cl gets some subtle performance boost by being an inline. Tested also for fast exp() variant, there seems to be no performance gain as the native function is equally performant. 4. Use OpenCL mix() function instead of _interpolatef() 5. Use two macros in rcd demosaicer 6. Make use of dt_fast_hypot()

• Hanno Schwalm (16 Feb 26)

  OpenCL kernel calling maintenance Consequently using kernel calling _args() variants.

• Hanno Schwalm (15 Feb 26)

  Introduce dt_opencl_enqueue_kernel_2d_local_args() This macro with it's backend dt_opencl_enqueue_kernel_2d_local_args_internal() is used to call the kernel including locals, sizes and all kernel parameters for simplification.

• Daniele Pighin (24 Feb 26)

  Use `keep_on_top` on mac, as in GTK3/Quartz `transient_for` forces the windows to be on the same screen. Also, removes the focus-follow-mouse behavior, as it is adding too much complexity and its really only needed to show the tooltip. Users are responsible to focus the 2nd window if so they want by clicking on it.

• Jeronimo Pellegrini (24 Feb 26)

  Updates to pt_BR

• Pascal Obry (24 Feb 26)

  Update integration test sub-module.

• Daniele Pighin (23 Feb 26)

  Prevent flickering when pinning an image in the 2nd window.

• Pascal Obry (23 Feb 26)

  Merge pull request #20386 from jenshannoschwalm/camera_handling_56 Camera handling for 5.6

• Pascal Obry (23 Feb 26)

  Update integration test sub-module.

• Hanno Schwalm (23 Feb 26)

  Fix blending log 1. If we use the -d perf option we want correct reports if there was and blending. We test for requested blending in pixelpipe code now. 2. The OpenCL guided filter internal tiling had some superfluous log info that was used when implementing the code.

• Hanno Schwalm (23 Feb 26)

  Fix export size restriction The calculation of max_possible_scale while exporting should only check for mem requirements for finalscale upscale, all limitations while processing the pipe should be handled internally via tiling or possibly avoiding a module's process() code.

• Pascal Obry (23 Feb 26)

  Merge pull request #19963 from masterpiga/pin-photo-in-2nd-window Adds the ability to pin the currently developed photo to the 2nd darkroom window

• Daniele Pighin (22 Feb 26)

  - Make 2nd window a transient of the main window, so that it always stays on top, and drop the dedicated "keep on top" button. - Implement focus following so that tooltips on the 2nd window show also if the window is not explicitly focused.

• Daniele Pighin (22 Feb 26)

  Make 2nd window buttons appear when the 2nd window is hovered.