Bugs
Layer panel: can toggle on/off checkbox using mouse outside of checkbox.
Save prompt on exit: cannot cancel; prompt pops up again. Remove cancel button.
Export overwrite prompt: yes and no buttons are inconsistently reversed.
Confirmed, I will have a look
Correct, in will swap them
I suspect this is a GTK “feature”, but I’ll check anyhow
Thanks!
During the past weeks I have worked on fixing several bugs, some of which were long-standing ones:
- a reproducible crash when closing image tabs. The fix involved quite some re-wiring of the image deletion code…
- an issue with the image updating when drag-n-dropping layers, where in some cases the layers above the dragged one would not be updated
- the file extension shown in the file chooser of the image export dialog was always “.jpg” regardless of the image format actually chosen. Now “.tif” is correctly appended when saving to TIFF format
Still on my TODO list…
This really seems to be something coming from GTK. I have no idea how to fix it…
In addition to bug fixes, I am working on an enhanced unsharp mask tool. For the moment the code is only available in a dedicated enhanced-usm branch, but I plan to merge the changes into the stable branch quite soon…
1 Like
Okay, it is a minor quibble really. But it can be a real time waster having to wait for the changes to process and cache.
I have fixed the behavior of the cancel button. Now it interrupts the application exit and brings you back to the edited image…
1 Like
Here is a summary of recent activities:
- the enhanced USM branch is now merged into stable, so it is available in the “standard” development packages
- talking about packages, I have re-named and re-organized the development packages so that finding the latest one for your system should be much easier now:
- I have fixed few bugs that were introduced by recent changes to the processing pipeline. In particular, the output of tools with interactive control points in the preview area (like the path mask and the gradient) was not correctly updated when adding/moving the control points
- I have fixed a discrepancy in colors in the RAW developer output between 1:1 and 1:2 or lower zoom levels
Next step will be to merge the simplified pipeline branch into stable, after which I will further focus on bug fixes and UI improvements. The goal would be to have a new release at the beginning of next year…
4 Likes
I am working on the UI layout of the tools dialogs, trying to achieve a more compact and polished look. Here are some examples of the result I have so far:
What do you think? Any suggestions?
1 Like
The label for your combo box needs padding/margin on the left.
I don’t mind the margin / padding.
I would like to see a bit more consistency between panels.
– The graphs look different: one is square, centred and has stuff above it.
– The graphs aren’t labelled.
– Text are sometimes title case, sentence case or lowercase.
Hi.
Not sure if I post this in this topic, but PhotoFlow_git_stable_linux_20191124_1815_bada1.AppImage is crashing when I try to export to jpeg:
PF::Histogram::update(): called
Histogram::update(): image=0x7fc6d45fa660
image->Bands=3
image->BandFmt=6
image size: 189x126
ImageToMapProc::render(): profile=0x25e27a0
ImageToMapProc::render(): colorspace=2
ImageToMapProc::render(): profile=0x25e27a0
ImageToMapProc::render(): colorspace=2
[Histogram::update_histogram] min=0 max=1
[Histogram::update_histogram] emitting signal_queue_draw
PF::Image::do_update(): pipeline #2 updated.
[ImageEditor::on_image_updated_async] emitting signal_image_updated
ImageEditor::on_image_updated() called.
ImageToMapProc::render(): profile=0x25e27a0
ImageToMapProc::render(): colorspace=2
ImageToMapProc::render(): profile=0x25e27a0
ImageToMapProc::render(): colorspace=2
ExportDialog::on_show() called.
last_saved:
(photoflow.bin:18722): Gdk-WARNING **: 22:11:48.590: gdk_window_set_icon_list: icons too large
File selected: /home/gustavo/Pictures/playraw/pools of light/DSCF1978.jpg
PF::Image::export_merged(): request.data=0x3337f50
PF::Image::export_merged(): waiting for export_done....
Attempt to unlock mutex that was not locked
Saving image to file /home/gustavo/Pictures/playraw/pools of light/DSCF1978.jpg...
/tmp/.mount_PhotoFaiSP0y/AppRun: line 50: 18722 Aborted (core dumped) "$APPDIR/usr/bin/photoflow.wrapper" "$@"
I repeated this attempt two more times with the same result.
The raw is from here. Follows pfi file:
DSCF1978.pfi (41.3 KB)
I am looking into this, not reproducible on MacOS. Is the jpeg file completely written before the crash?
1 Like
Nothing is written, actually.
I have fixed the issue, could you please test the latest package from today? For the records, the problem was introduced by this commit.
It works here, thanks!
1 Like
I absolutely agree! I am currently going through the various tools and trying to clean-up/harmonize the interface.
I have committed a number of improvements to the UI widgets and to the layout of the tool dialogs, trying to get a more consistent, compact and polished look.
Here is a screenshot that shows few examples:
As usual, binary packages are available for download from the continuous integration page.
4 Likes
I think there’s an image being worked on in there, somewhere… 
Been following your progress; I think it’s time for me to try Photoflow again… Nice selection of tools!
Edit: so, which of the CI AppImages is best for an overall try-out?
As @paperdigits posted, you should take the packages from the “stable” branch. The others were test branches that are now merged into stable. I have now removed the corresponding packages, just to avoid confusion.
3 Likes
Hello Andrea!
Just tested this new build on Windows 10.
One big advantage of PhotoFlow on Windows 10 is that it starts blazingly fast: only a few seconds to show the GUIs 
Some feature requests:
- Add some more basic shortcuts (e.g. ctr+o: to open an image);
- Add a way to show the name of the first image you open in the GUIs.
At present, you are forced to open at least 2 images to show the name of the first image, in its tab.
A minor problem is the duplication of 2 different X icons to close the windows.
At present, only one of the 2 works: the red one, on the bottom.
A crash which always occurs is duly reported here: