This adds the PSD metadata plug-in procedure call to the JPEG
plug-in, as part of implementing issue #7549.
Also implements the import half of issue #1842.
JPEGs only store image-level metadata like paths.
CMYK profile is now stored in GimpImage on load
(rather than being discarded) and it's used for export rather than
the default simulation profile stored in Preferences
We already had import support through littleCMS. We now use fully
babl/GEGL which makes our code more straightforward and identical,
whichever the input format.
The export support is totally new. It comes with a checkbox to propose
selecting CMYK export and a label displaying the CMYK profile which will
be used.
Now this whole implementation has a few drawbacks so far, but it will be
a good first sample for future CMYK-related improvements to come:
* The export profile I am using is what we call the "simulation
profile" from the GimpColorConfig. This corresponds to the default
"Soft-proofing" profile as set in Preferences. In particular, this is
not the actual soft-proofing profile for this image which might have
been changed through the View menu because this information is
currently and unfortunately unavailable to plug-ins. It is not the
"Preferred CMYK Profile" either, as set in Preferences.
TODOS:
- We really need to straighten the soft-proof profile core concept by
storing it in the image and making it visible to plug-in.
- Another interesting improvement could be to create a
GimpColorProfile procedure argument which would be mapped to a color
profile chooser widget, allowing people to choose profiles in
plug-ins. For an export plug-in in particular, it could allow to
select a profile different from the soft-proof one at export time.
* When we export, if no profile is choosen, babl will use a naive
profile. It would be nice to store this naive profile into the JPEG if
the "Save color profile" option is checked (same as we store a generic
sRGB profile when no RGB profile is set).
* When we import, we just import the image as sRGB. Since CMYK gamuts
are not necessarily within sRGB (some part of the spectrum is usually
well within, but other well outside), other than the basic conversion
accuracy issue, we may lose colors. It would be much nicer to be able
to select an output RGB profile. Optionally if we could create a RGB
color space which is made to contain the whole input CMYK profile
color space, without explicit choice step, it would be nice too.
* I am using babl's "cmyk" format, not the expected "CMYK" format.
"cmyk" is meant to be an inverted CMYK where 0.0 is full ink coverage
and 1.0 none. Nevertheless when loading the resulting JPEG in other
software (editors or viewers alike), the normal CMYK would always
display inverted colors and the inverted cmyk would look fine.
Finally I found a docs from libjpeg-turbo library, explaining that
Photoshop was wrongly inverting CMYK color data while it should not.
This text dates back from 1994, looking at the commit date which
introduced this paragraph. In the 28 years since then, could this
color inversion have become the de-facto standard for JPEG because one
of the main editor would just output all its JPEG files this way?
See: dfc63d42ee/libjpeg.txt (L1425-L1438)
While doing this cleanup, I found at least several other string leaks
in: file-compressor, file-gegl, file-pdf-save, file-raw-data, file-xwd,
jpeg-load, psd-save…
So it's quite worth it!
Note: in file-pdf-save, there is a global variable file_name which seems
to be happily leaked without caring (didn't look in details, but looks
so). I didn't fix this one which will require a bit more in-depth logics
care.
Commit aba721ae68 fixed its intended bug but brought a new: each time
the preview was updated, a new display was created. This fixes this new
bug. Hopefully it's good now!
The port had a slight error, because in gimp-2-10, the display_ID
actually had 3 states: 0 when gimp_export_image() kept the original
image to which we just add a preview layer, -1 when it created a new
image which we wanted to put in its own display, and the display ID
itself when created.
With the new API where display variable is an object, we can only have 2
cases. So I create an additional variable separate_display to make the
distinction.
In PNG, JPEG and TIFF export plug-ins which use the new API, use our new
function to set widget sensitivity.
Note that part of it is similar to commit 6a2910bd3b on `gimp-2-10`,
making "Save GeoTIFF data" checkbox insensitive when there are no such
data available (this feature was late on the main branch as we rushed
for 2.10.24 release).
… GimpSaveProcedureDialog.
See issue #6092 and the discussion with Jacob. Basically we are trying
to improve the metadata situation with more edit abilities and
awareness, while in the same time having the export dialogs less of a
mess (the "Comment" input in particular will most likely move to the
metadata editor itself; I left it for now, until the move is done).
The "(edit)" link will basically just run "plug-in-metadata-editor".
Also as a side note: I realized that gimp_pdb_run_procedure() runs
procedures synchronously and wait for a result, which is fine for quick
non-interactive plug-ins, but freezes the calling process otherwise.
Actually even when we want synchronous result, we should allow for GUI
events to be processed (otherwise the OS just thinks the calling export
plug-in is a zombie and proposes to kill it). This API should probably
be improved (and an alternative async version added as well).
This format name is a public facing name for a file format, such as
"PNG", "JPEG", or "C-source". Since it is public facing, the function
recommends to localize it too.
This is an optional name, yet is made mandatory if you want to use
GimpSaveProcedureDialog because it will be used for the dialog title
(ensuring that all support format have a similar export dialog title).
Following this change, gimp_save_procedure_dialog_new() does not ask for
a title anymore (if anyone absolutely wants to set a custom title,
setting the "title" property on the dialog is always possible anyway,
but a generic and consistent title should be set as a default).
Also updating the 3 plug-ins which were already using the now-changed
API.
This is nearly 600 lines less for basically the same logics! Removed
code is in particular all the GUI code is favor of the new GUI
generation.
I also cleaned a lot of stuff, removing many global variables or ugly
pieces of code. I also removed a lot of redundant code of things which
are now generic, such as handling of "gimp-comment" parasite (this is
now handled by GimpSaveProcedure and GimpImageMetadata) as well as
saving previous run's values (this is also handled generically).
Note that Advanced Options used to be in an expander. For now I chose to
get them immediately visible (still in their own "Advanced Options"
section, but it's now a normal frame, not an expander hidden by default)
since lately we got some input that many "advanced options" in various
dialogs should not be hidden away. So let's try like this for now (even
though it packs quite a lot of options in the same dialog!).
I thoroughly tested, yet that were so many changes that bugs may have
sneaked in. Please anyone, test JPEG export!
Renaming the temporary function gimp_scale_entry_new2() into
gimp_scale_entry_new() now that the original code is entirely gone. This
is now a fully-fledged widget with a nice and proper introspectable API.
I got rid of gimp_scale_entry_set_sensitive(), as we can now use the
generic gtk_widget_set_sensitive(), and I ported the 2 plug-ins using
this function.
I also created gimp_scale_entry_set_value() to set the value (nicer than
setting object properties).
and in an attack of madness, changes almost all file plug-in
code to use GFile instead of filenames, which means passing
the GFile down to the bottom and get its filename at the very
end where it's actually needed.
jpeg-save.c:56: warning: "DEFAULT_QUALITY" redefined
56 | #define DEFAULT_QUALITY 90.0
|
In file included from C:/msys64/mingw64/x86_64-w64-mingw32/include/windows.h:71,
from C:/msys64/mingw64/x86_64-w64-mingw32/include/rpc.h:16,
from C:/msys64/mingw64/x86_64-w64-mingw32/include/objbase.h:7,
from C:/msys64/mingw64/x86_64-w64-mingw32/include/shlwapi.h:16,
from C:/msys64/mingw64/include/jmorecfg.h:19,
from C:/msys64/mingw64/include/jpeglib.h:31,
from jpeg-save.c:33:
C:/msys64/mingw64/x86_64-w64-mingw32/include/wingdi.h:1140: note: this is the location of the previous definition
1140 | #define DEFAULT_QUALITY 0
We now have both variants, one returning a GList, and another
returning an array. Turns out that while a list is often nicer,
sometimes a random-access array really keeps the code much simpler.
Adapt all plug-ins, and clean up a bit (like use g_list_reverse() once
instead of iterating the list reversed).
Use the drawable's space for the load/save formats to avoid unwanted
color conversions. Also cleaned up the loading code a bit, it was
doing weird duplicated stuff for the preview and !preview cases.
... in JPEG export.
Same as the WebP export, which is quite similar (8-bit max format), when
no profile was explicitly set, we want to convert any data from storage
format to non-linear (unlike when exporting high bit depth formats, such
as TIFF).
We only make an exception for 8-bit linear. This commit adds this
exception.
GIMP should not convert assigned profile to sRGB just because we stored
as linear on the XCF. In other words, we should not look at the image
precision to decide whether to export as linear (previously only 8-bit
linear images), but at the profile TRC. There are basically 3 cases:
(1) We don't save a profile, then convert to sRGB whatever the source
precision (because readers would assume sRGB for a no-profile jpeg).
(2) We save the default profiles: convert to sRGB because it's usually
a better choice for 8-bit formats and even working at 32-bit float
*linear* doesn't mean you want to export as 8-bit int *linear*. As the
image creator made no explicit export choice, we make an acceptable
default one.
(3) We save an explicitly assigned profile: keep the profile TRC, don't
convert!
Note that this apparently won't work perfectly right now, as GIMP
replaces the original TRC with the linear default TRC when converting to
linear. So the expected TRC is lost in such case when you have not
explicitly reset the correct profile. Yet this is on GIMP side and this
part of the issue should be fixed with the space invasion merge. For
now, this is how the plug-in should work.
This is based on my late discussion with Ell. Please everyone, and Ell
especially, review! :-)
(cherry picked from commit c5f7bac2ba)
... linear itself AND if we export the profile.
In most cases we want to save 8-bit image formats (here JPEG) as
non-linear, even though the work image may have been linear itself (yet
with higher bit depth). The reasons are shadow posterization on low bit
depth, and the fact that JPEG compression was designed for perceptually
uniform RGB and introduces shadow artifacts with linear RGB (see #1070,
message by Elle Stone). The only exception is when the creator was
working explicitly on 8-bit linear (not higher bit depth) AND if we
export the profile (otherwise most loaders around assume sRGB). In such
a case, let's consider the creator knows what one is doing and keep the
exported image linear.
Similar logics is already used in PNG exporter (though a bit of a
variant since PNG supports 16-bit so it is instead: 8-bit linear without
profile is promoted to 16-bit non-linear, and kept 8-bit linear with
profile).
... produces jpeg in linear color space.
The problem was that we were anyway always exporting to non-linear while
attaching a linear profile. A first approach would have been to export
in linear instead when the work image is linear, as proposed by mitch in
a first version of the patch. Yet as Elle Stone notes, it is not a great
idea to export 8-bit images as linear.
Instead let's continue to always export as non-linear, as we were
already doing. Yet when we also save a profile, and this one was
originally linear, let's convert it to sRGB TRC before exporting.
...export dialogs
Move most stuff out of the "Advanced" expander, only nerdish encoding
options are left there.
Issue #701: Add a "Save color profile" toggle which honors the default
value configured in preferences and always saves the profile when
enabled.
Various plug-ins exporting metadata should now follow preferences, which
would override any default. Of course these preferences can still be
overriden by saved settings (global parasite), previous run settings,
and finally through the GUI when interactive.
This is a privacy concern. Whereas importing metadata is usually a good
idea, exporting it should be a conscious action. A lot of private data
can be leaked through metadata and many people don't realize it (which
also usually means they don't need it). On the other hand, the people
who realize it are the ones who would explicitly edit the metadata and
check what they want to be exported or not.
This is only a first step. Some people may want to always export the
metadata and for these people, there should be abilities to change the
default.
Set use_orig_quality when both the quality and the subsampling are the
same as in the originally-imported jpeg.
Also improve subsampling initial selection: use the original subsampling
unless the default one is the best or the original one is the worst.
The current code was wrong and would often use the default subsampling
even when worse than the original one.
