arenasys
commented
2 years ago Yea, its not ready to be done automatically without side effects, so i left it as a manual only option.
The issue is changing the input folder without changing the staging folder can currently have unexpected results.
if you have input_a/1.png and input_b/1.png they will both share the staging metadata file 1.png.json, overriding each other. Also it may be unexpected to have the output data overridden.
Asking for each of the 3 folders from the user is annoying and confusing so theres two options i can think of.
- fix the staging folder layout so each input folder is isolated, and keep the output folder shared.
- have the program automatically create staging and output folders in whatever input folder you select, this could be annoying to some users that dont want their inputs to be messed with, but theres always the option to manually specify.
flesnuk
I didn't find a way to change the input folder without manually editing the start.sh arguments. Maybe a good time would be after packaging to the output folder. (Infering that you finished with the current dataset and want to go to next one). Possible by adding a button in the package pop-up instead of displaying the File explorer automatically to avoid confussion. (by thinking it is requesting the output folder)
But the best option is having a toolbar with the typical File, View, etc. improving the usability/friendliness of the program.