I have started to re-work the visualization of the grid and handles in the crop tool. This is where I stand so far:
Next steps will be to make the grid configurable, and allow non-regular patterns like the golden rule, and to change the mouse pointer according to which function is triggered by the mouse movement (activation of one handle, dragging in the central area, etc…).
What do you think?
This is a followup to my distortion correction question. When I opened your PFI and exported a TIF, the distortion was different from your JPG output: Landscape near Altaussee - #30 by Carmelo_DrRaw. I don’t know the cause ATM. Could simply be a different Lensfun version. PFIs don’t currently hold export information (I think I asked about that before) but I don’t think export would change the distortion or crop; or G’MIC resize would change it.
I am using PhotoFlow_git_stable_w64_20200112_1509_22bdf by the way. Another problem that I have is rotate doesn’t auto crop when I export. 20191001T162108.pfi. Edit looks like it works after I toggle auto crop off then on again.
It could be a lensfun version mismatch between my own version on MacOS and the Windows version. I am now looking into this. Could you provide me an example of the output you obtain, with the corresponding PFI file, so that I can see if I can reproduce it exactly?
Sounds like a bug in the logic than handles the auto-crop button. I am trying to reproduce this as well.
No need to upload the PFI. I used yours with PhotoFlow_git_stable_w64_20200112_1509_22bdf. Exported using the default settings and resized using G’MIC.
My machine
Your Play Raw
I have introduced a number of improvements in the optical corrections, namely:
better auto-matching of camera/lens combo using on EXIF data. This is based on a looser association heuristics in the LensFun code, so it needs to be tested to see if this is not introducing too many false matches
the vignetting correction has been moved after the highlights reconstruction. Otherwise the highlights recovery was sometimes confused by the increase in luminance introduced by the vignetting correction
the focus distance now defaults to 1000 m when not available in the EXIF data. Before it was set to zero, thus de-activating some of the corrections.
The new code is available in both the stable and v0.3.1-rc1 branches.
Using 6abe5 on macOS I get a lens mismatch on a recent Play Raw (Sailboat at the Baltic Sea):
This lens is actually not present in the LensFun DB (and not precisely described in the EXIF data).
The previous, more strict code would give an “unknown lens”, while the new one shows the closest match… I have to see if there is some sort of “match score” criteria that we can use to reject clearly wrong matches.
That’s true, it is not present in Lensfun.
Up to now I didn’t bother as the lens has very less distortion…
Hi there,
I just wonder whether “insert image as layer” menu might allow resizing (or even stretching, tiling) the inserted image. Then, it can be a real time saver for those looking for such a tool.
That’s a good suggestion… I could introduce an option in the image layer to select how to match the input image size. Options could be:
Each could have an horizontal/vertical alignment option…
Is this what you had in mind?
One of my first feedback was about this. I prefer to import first, then adjust. Import is possible but moving layers on canvas, cropping, tiling, resizing and defining borders aren’t available or unstable.
Exactly!
It can be done elsewhere in a rather complex way but in current Photoflow, this may become second nature (as playing with tansparency and selecting various overlays are already there).
After quite some time of silent programming, here is an update of PhotoFlow’s features:
I have finally fixed a long-standing issue with EXIF data in exported TIFF files, which were for example seen as multi-page TIFFs by GIMP. Now EXIF data in TIFF files is saved correctly, apart from a small remaining issue with the copyright string being null that I am going to fix soon.
the stable branch now offers the possibility to save 16-bit or 32-bit EXR files. The ICC primaries are stored in the EXR file so that colors can be interpreted correctly, and EXIF data is also stored in a way that is recognized by at least GIMP and Darktable. The image data is automatically linearized upon saving if it is not already in a linear colorspace.
@Carmelo_DrRaw This is a gentle reminder to place the continuous releases to a more prominent position on GitHub. I use a bookmark but it shouldn’t be difficult for others to find.
This is what visitors see:
README.md also links to this ancient version.
PS Old typo (missing “)”).
Thanks for the reminder! I have finally managed to find out how to refresh a git tag, so that the continuous release gets moved to the top of the list. Now it’s done:
The typo will be fixed in the next packages…
It should be possible to open the file with the latest package from 2020-08-04, which includes an update of the raw decoding engine. Could you please check?
Thanks!
stable, rc1 or either are fine? Where there is choice there can be confusion.
@afre only stable for the moment. I will back-port the updated raw engine once I am sure there are no side-effects…
As usual, decompression takes a long time. 
Will find out later in the afternoon.
