tannerhelland
commented
4 years ago Hi @mizka76. I apologize that it has taken me so long to reply to this issue. COVID-19 issues are impacting my time for PhotoDemon (a small concern compared to what many are dealing with, of course).
In the current nightly builds, I believe I have fixed problems with text entry in French locales. Please let me know if you encounter any other troubles and I will try to fix them right away!
As for the content-aware resize results, I believe the tool is functioning as intended. Seam carving is the name of the algorithm PhotoDemon uses for this tool, and that algorithm is primarily designed for minor size adjustments on real-world photographs. Regular or repeating patterns are likely to produce poor results.
Comparing PhotoDemon's output to some other open-source tools (e.g. http://nparashuram.com/seamcarving/) shows very similar results on repeating patterns (i.e. not good). I do not know how to improve the tool's behavior on patterns at present.
Do you know of any software that produces better content-aware results on repeating patterns? If so, please let me know as I would be very interested to study them to see what they do differently.
Thank you!
Version information
PhotoDemon version 8 alpha 1879)"
Windows 10 FRENCH version on a fresh install
"reglages >niveaux " ( adjustments/level) opens a message box (-2 is not a good value)
during 1s and , after this ,
when I enter a value > a 0.9 to 0.7 or
when I use the cursor just beside to increment the number the software crashes immediately.
typing a value in "custom filter" makes a crash too .
I think ( i dont know anything in programming) that it could be the same problem than you fixed a few days earlier ( with resize and the numbers in french versions) DebugReport_0.log
a second bug ( probably) is "image resize content aware " that gives som strange results like a non regular distorted image . the image I send show some result with an image of a regular pattern . Hope this helps you , and that you understand what i try to explain ( excuse my approximative english) Merci beaucoup pour ce que vous avez déjà fait.