Some PhF Layout suggestions

I can’t see this button. Here is sample: loaded raw file, selected mask layer with “gradient”.

In order to associate an opacity mask to a tool you have to proceed in the following way:

  • add an adjustment layer, for example a “curves” adjustment
  • double-click on the white rectangle to the right of the layer name - this will get you into the layer mask editing mode
  • add some initial layer to your mask, for example a gradient, and toggle the “show mask” checkbox to see the mask painted on top of the image
  • add some additional layer to further refine the mask, for example a “threshold” or a “path mask” tool
  • change the blend mode of the topmost mask layer to whatever you need to achieve the desired combination of masks - you can multiply the masks, intersect them or combine them
  • click on the “back” button to exit the layer mask editing mode

Hope this is clear…

OK, thanks. It’s a great technique :smile:.
Another question: GIMP says to me that exported floating point tif has 3 layers, why? And what it means? Layer 0 is imported OK, but 1 and 2:

image

afre, about sliders range. For example - in a basic adjustments slider “saturation” has range ± 99, and normal using – not more than 5, I think

This is an old issue, that I have not managed yet to fix… it is related to how the EXIF data is embedded into TIFF files by PhF.

Why RawTherapee writes normal floating point tiff?

You meant PhotoFlow?

No, he is right: RT and DT write correct TIFF files, while the PhF TIFF output for some reason contains three layers… it is related to the EXIF data, but I have not yet figured out where the problem exactly is…

Ah, ok, got it wrong. My bad.

PhF from git has (in Arch linux) option for compiler “threat warning as errors” from some optimistic developer. So we are without last version.

@Chawoosh are you referring to the AUR package? I just checked and couldn’t see any custom compile option in the PKGBUILD…

May be in the “cofigure” file. I don’t have time to search, pardon me.
Effect is somewhere in a middle:

Blockquote Scanning dependencies of target rawspeed
[ 4%] Building CXX object src/external/rawspeed/CMakeFiles/rawspeed.dir/src/librawspeed/common/ChecksumFile.cpp.o
[ 5%] Building CXX object src/external/rawspeed/CMakeFiles/rawspeed.dir/src/librawspeed/common/Common.cpp.o
[ 5%] Building CXX object src/external/rawspeed/CMakeFiles/rawspeed.dir/src/librawspeed/common/Cpuid.cpp.o
[ 5%] Building CXX object src/external/rawspeed/CMakeFiles/rawspeed.dir/src/librawspeed/common/DngOpcodes.cpp.o
In file included from /home/chawoosh/.cache/aurman/photoflow/src/photoflow/src/external/rawspeed/src/librawspeed/common/DngOpcodes.cpp:25:
/home/chawoosh/.cache/aurman/photoflow/src/photoflow/src/external/rawspeed/src/librawspeed/common/Mutex.h:80:38: error: ‘const’ attribute on function returning ‘void’ [-Werror=attributes]
void attribute((const)) Lock() const ACQUIRE() {
^~~~~
/home/chawoosh/.cache/aurman/photoflow/src/photoflow/src/external/rawspeed/src/librawspeed/common/Mutex.h:85:40: error: ‘const’ attribute on function returning ‘void’ [-Werror=attributes]
void attribute((const)) Unlock() const RELEASE() {
^~~~~
cc1plus: all warnings being treated as errors
make[2]: *** [src/external/rawspeed/CMakeFiles/rawspeed.dir/build.make:102: src/external/rawspeed/CMakeFiles/rawspeed.dir/src/librawspeed/common/DngOpcodes.cpp.o] Error 1
make[1]: *** [CMakeFiles/Makefile2:572: src/external/rawspeed/CMakeFiles/rawspeed.dir/all] Error 2

Blockquote

I think it has to do with thumbnails from the raw; if you just copy ALL the metadata over it includes the thumbnails as layers.

Filmulator was at one point writing 4-layer TIFFs, but then I just decided to not write metadata because I didn’t want to spend the time to properly deal with metadata.

(As a result of the copy-everything implementation, it also just writes the camera jpeg thumbnail to the output jpegs, which means that the thumbs aren’t reflective of what Filmulator outputs)

1 Like

The fact is that I am already removing the thumbnails from the EXIF data which I load from the raw files, or at least this is what I expect this piece of code to do. But something is clearly still not done properly. I know that DT has quite some code to cleanup the EXIF data prior to saving the TIFF files. I have to look into that…

1 Like

PhF still can’t be assembled - somewhere in target rawspeed cc find flag to threat warning as errors. Previous target (lensfun) compiled with warnings and linked. In my system conf files there isn’t any CFlags.
O! Just now I see this:

In file included from /home/chawoosh/.cache/aurman/photoflow/src/photoflow/src/external/rawspeed/src/librawspeed/common/DngOpcodes.cpp:25:
/home/chawoosh/.cache/aurman/photoflow/src/photoflow/src/external/rawspeed/src/librawspeed/common/Mutex.h:80:38: error: ‘const’ attribute on function returning ‘void’ [-Werror=attributes]
void attribute((const)) Lock() const ACQUIRE() {
^~~~~
/home/chawoosh/.cache/aurman/photoflow/src/photoflow/src/external/rawspeed/src/librawspeed/common/Mutex.h:85:40: error: ‘const’ attribute on function returning ‘void’ [-Werror=attributes]
void attribute((const)) Unlock() const RELEASE() {
^~~~~
cc1plus: all warnings being treated as errors

I find option -Werror in many make files and in “compile_commands.json” - great number, in any directive.

I have hopefully fixed all warnings, the code should compile with -Werror now…

Thanx, all’s compiled and works.

One think about usability: in geometry and rotate are needed horizontal and vertical (removable and movable) levels.

This is actually already possible: when you add a scale/rotate layer, you can click twice on the preview to create a guide line. The image will then be rotated in such a way that the line will end up being horizontal or vertical, depending which one is closer to the initial angle.

I am experimenting with the layout of the tools dialogs, trying to save some vertical space.
I have introduced a tab widget, and moved the blend mode, opacity slider and input source selector into a separate tab. Here is how it would look like:

37

01

What do you think?