Add optional ability to load/save image rank property from/to xmp
sidecar "xmp.Rating" and color property from xmp "xmp.Label" ignoring
the ones provided in the processing params file.
This behavior is disabled by default and an option under settings ->
file browser has been added to enable it.
When enabled:
* On load:
* rank and color are not read from processing params.
* rank is mapped from xmp sidecar file rating entry.
* color is mapped from xmp sidecar file label entry.
* On save:
* rank and color are saved to the xmp sidecar
* rank and color are also saved to the processing param (pp3) files to
keep them in sync
Rating mapping:
Since rating can be also -1 but rank only goes from 0 to 5, the -1 value
is ignored like already done when importing from embedded xmp data.
Color mapping:
XMP has no color concept, usually programs like digikam uses the label
field to write a color name ("Red", "Orange"). The problem is that this
isn't standardized and label can be any string. Additionally Rawtherapee
has 5 specific colors while other programs can have different colors
with different name so they won't be shown if they don't map to the 5
color names supported by rawtherapee. On save only the 5 color supported
by rawtherapee wil be saved.
Trash is kept only in the profile files for multiple reasons:
* There's no trash concept in xmp, there's the rejected concept assigned
to a rating == -1.
* We could map rejected to trash but in rawtherapee rank and trash are two different values and
an image can have both rank >= 0 and trashed set to true.
Using an unique value like rating for rank and trash (when -1) will
require changing the current rawtherapee logic.
* Also digikam only handles ratings from 0 to 5 (no -1) and handles
trash in its own internal way without reflecting it in the xmp
sidecar.
If a png has a transparency chunk (tRNS), after converting the
transparency to alpha we should also strip alpha (color_type is not
updated with PNG_COLOR_MASK_ALPHA flag set) or the row parsing will
return also the alpha channel causing a memory overflow.
Read the correct tag for the as-shot multipliers (index 0x69). The port
from LibRaw incorrectly used the tag at index 0x69 + 0x64, which is the
daylight WB multipliers. This fix affects Canon cameras with color data
versions 34 (EOS R3) and 48 (EOS R6m2, EOS R7, and EOS R10).
* Avoidgamumethod set to Munsell-only
* Set appimage.yml and windows.yml to la_gamutmunsell
* Remove procparams conversion XYZ absolute tu Munsell only
* Neutralize appimage.yml and windows.yml
raw files, usually dng files, can provide a dcp profile by providing the
related tags.
This patch will make the "Use embedded" options in the color management
panel selectable if the source raw file embeds a dcp profile.
In this case the DCP panel will be enabled like when reading an external
dcp profile.
Additional changes:
* Rename "Use embedded, if possible" to just "Use embedded" since the
option is already disabled when no embedded profile is found.
* Update the "Use embedded" options tooltip as it's now not related to
only non raw images.
Also use the same values for the Sony ILCE-7RM5 which should have the
same sensor. Copy the PDAF pattern to the ILCE-7RM4, which has the same
size sensor.
* Local adjustments - Show additional settings - link with complexity (#6899)
* Change hishow -additional seeting - with complexity
* Modify windows.yml and appimage.yml
* Fixed bug in duplicate spot
* Remove pre-dev builds
* Update camconst.json white levels for 1DxII
* Fix warnings: conversion to double/float, unused variables, register keyword
* Fix crash when opening image in editor
Do not access uninitialized raw image data. The raw data is requested
when the demosaic mode is set to None and the cursor is moved over the
image in the editor. It can occur before the raw data is loaded.
* Fix sRGB working profile crash
The sRGB working profile cannot be found under some conditions because
the profile name is stored as a Glib::ustring and the same strings may
not be equal when using different locales. Use std::string whenever
comparing profile names.
---------
Co-authored-by: Desmis <jdesmis@gmail.com>
Co-authored-by: CarVac <airplaniac2002@gmail.com>
Co-authored-by: Alexander Gruzintsev <0v3rt1r3d@gmail.com>
* Fix defringe banding & desaturation race condition
Use a copy of the image instead of the image itself for the source data.
Affects Defringe, Color Appearance & Lighting Hot/bad pixel filter, and
Contrast by Detail Levels Reduce artifacts.
* Fix CodeQL warning
Fix multiplication overflow.
* Fix line in middle of 180 degree rotated image
For images with an odd number of rows, a 180 degree rotation resulted in
one row of un-rotated pixels because it was effectively rotated twice.
* Replace verbose swap with std::swap
* Improve Dynamic Range Compression - for some images with very high DR
* Clean code
* Improve TM fattal with saturation control in LA
* Saturation control fattal in LA
* Re-order paramsedit
* Change history_msg_tmo_satur with saturation
---------
Co-authored-by: U-PCSPECIALIST01\jdesm <jdesmis@gmail.com>
* Improve Dynamic Range Compression - for some images with very high DR
* Clean code
* Improve TM fattal with saturation control in LA
* Saturation control fattal in LA
* Re-order paramsedit
* Change history_msg_tmo_satur with saturation
* Improve Itcwb with non-raw files
* Change pre-dev builds wbrefinement
* Change template in pre-dev
* Improvment improccordinator.cc
* Forgotten observer convert
* Reenable wbauto autogrey as 5.8
* Remove wrong code
* Missing getrgbloc references
* Fixed bug due to bias in queu with temperaure correlation issue 6911
* Simpleprocess queue compatibility tif-jpg
* Preserve AWB edits from 5.9
In 5.9 for non-raw files,
1. RGB grey uses the unit multipliers with temperature bias applied.
2. Temperature correlation uses the equivalent of temperature 5000,
green 1, and red/blue equalizer 1.
* Refactor temperature correlation AWB code
* Fix inaccurate RGB grey WB preview after using ITC
The RGB grey automatic white balance algorithm caches the multipliers.
Temperature correlation automatic white balance also caches results to
the same location, but never uses it. This causes the RGB grey method to
produce incorrect results in the editor. Removing the temperature
correlation cache fixes the issue and does not have side-effects.
---------
Co-authored-by: U-PCSPECIALIST01\jdesm <jdesmis@gmail.com>
* Improve Itcwb with non-raw files
* Change pre-dev builds wbrefinement
* Change template in pre-dev
* Improvment improccordinator.cc
* Forgotten observer convert
* Reenable wbauto autogrey as 5.8
* Remove wrong code
* Missing getrgbloc references
* Fixed bug due to bias in queu with temperaure correlation issue 6911
* Simpleprocess queue compatibility tif-jpg
* Preserve AWB edits from 5.9
In 5.9 for non-raw files,
1. RGB grey uses the unit multipliers with temperature bias applied.
2. Temperature correlation uses the equivalent of temperature 5000,
green 1, and red/blue equalizer 1.
* Refactor temperature correlation AWB code
* Fix inaccurate RGB grey WB preview after using ITC
The RGB grey automatic white balance algorithm caches the multipliers.
Temperature correlation automatic white balance also caches results to
the same location, but never uses it. This causes the RGB grey method to
produce incorrect results in the editor. Removing the temperature
correlation cache fixes the issue and does not have side-effects.
---------
Co-authored-by: Lawrence Lee <45837045+Lawrence37@users.noreply.github.com>
